From fd31f92cea0ce8613a22d8f4b3c75b340bcc5689 Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keithw@vmware.com>
Date: Wed, 22 Jul 2009 00:39:00 +0100
Subject: gallium: simplify tgsi_full_immediate struct

Remove the need to have a pointer in this struct by just including
the immediate data inline.  Having a pointer in the struct introduces
complications like needing to alloc/free the data pointed to, uncertainty
about who owns the data, etc.  There doesn't seem to be a need for it,
and it is unlikely to make much difference plus or minus to performance.

Added some asserts as we now will trip up on immediates with more
than four elements.  There were actually already quite a few such asserts,
but the >4 case could be used in the future to specify indexable immediate
ranges, such as lookup tables.
---
 src/mesa/state_tracker/st_mesa_to_tgsi.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

(limited to 'src/mesa/state_tracker')

diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c
index e150dff9bb..6380cd6b2a 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.c
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c
@@ -225,11 +225,15 @@ static struct tgsi_full_immediate
 make_immediate(const float *value, uint size)
 {
    struct tgsi_full_immediate imm;
+   unsigned i;
 
    imm = tgsi_default_full_immediate();
    imm.Immediate.NrTokens += size;
    imm.Immediate.DataType = TGSI_IMM_FLOAT32;
-   imm.u.Pointer = value;
+
+   for (i = 0; i < size; i++)
+      imm.u[i].Float = value[i];
+
    return imm;
 }
 
-- 
cgit v1.2.3


From a7427b0f7b2325b8dcc560d57cb894df25ebef93 Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Mon, 27 Jul 2009 15:10:28 -0600
Subject: st/mesa: silence warning

---
 src/mesa/state_tracker/st_cb_texture.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'src/mesa/state_tracker')

diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c
index 15f84b6638..ee71c012c6 100644
--- a/src/mesa/state_tracker/st_cb_texture.c
+++ b/src/mesa/state_tracker/st_cb_texture.c
@@ -1047,7 +1047,8 @@ st_TexSubimage(GLcontext *ctx, GLint dims, GLenum target, GLint level,
       _mesa_image_image_stride(packing, width, height, format, type);
    GLint i;
    const GLubyte *src;
-   enum pipe_transfer_usage transfer_usage;
+   /* init to silence warning only: */
+   enum pipe_transfer_usage transfer_usage = PIPE_TRANSFER_WRITE;
 
    DBG("%s target %s level %d offset %d,%d %dx%d\n", __FUNCTION__,
        _mesa_lookup_enum_by_nr(target),
-- 
cgit v1.2.3


From 2420b283b783751d4def3a3a2a0ed8bf7bb7b6a8 Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keithw@vmware.com>
Date: Tue, 28 Jul 2009 18:45:22 +0100
Subject: mesa/st: recognize no-op scissor state when checking clear_with_quads

Some apps enable scissor but set the rectangle to the dimensions of
the window.  Don't let this force us onto a slower clear path.
---
 src/mesa/state_tracker/st_cb_clear.c | 24 ++++++++++++++++++++----
 1 file changed, 20 insertions(+), 4 deletions(-)

(limited to 'src/mesa/state_tracker')

diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c
index 668acbccb8..8a8c99f7e1 100644
--- a/src/mesa/state_tracker/st_cb_clear.c
+++ b/src/mesa/state_tracker/st_cb_clear.c
@@ -280,7 +280,11 @@ clear_with_quad(GLcontext *ctx,
 static INLINE GLboolean
 check_clear_color_with_quad(GLcontext *ctx, struct gl_renderbuffer *rb)
 {
-   if (ctx->Scissor.Enabled)
+   if (ctx->Scissor.Enabled &&
+       (ctx->Scissor.X != 0 ||
+        ctx->Scissor.Y != 0 ||
+        ctx->Scissor.Width < rb->Width ||
+        ctx->Scissor.Height < rb->Height))
       return TRUE;
 
    if (!ctx->Color.ColorMask[0] ||
@@ -300,7 +304,11 @@ check_clear_depth_stencil_with_quad(GLcontext *ctx, struct gl_renderbuffer *rb)
    GLboolean maskStencil
       = (ctx->Stencil.WriteMask[0] & stencilMax) != stencilMax;
 
-   if (ctx->Scissor.Enabled)
+   if (ctx->Scissor.Enabled &&
+       (ctx->Scissor.X != 0 ||
+        ctx->Scissor.Y != 0 ||
+        ctx->Scissor.Width < rb->Width ||
+        ctx->Scissor.Height < rb->Height))
       return TRUE;
 
    if (maskStencil)
@@ -319,7 +327,11 @@ check_clear_depth_with_quad(GLcontext *ctx, struct gl_renderbuffer *rb)
    const struct st_renderbuffer *strb = st_renderbuffer(rb);
    const GLboolean isDS = pf_is_depth_and_stencil(strb->surface->format);
 
-   if (ctx->Scissor.Enabled)
+   if (ctx->Scissor.Enabled &&
+       (ctx->Scissor.X != 0 ||
+        ctx->Scissor.Y != 0 ||
+        ctx->Scissor.Width < rb->Width ||
+        ctx->Scissor.Height < rb->Height))
       return TRUE;
 
    if (isDS && 
@@ -345,7 +357,11 @@ check_clear_stencil_with_quad(GLcontext *ctx, struct gl_renderbuffer *rb)
    if (maskStencil) 
       return TRUE;
 
-   if (ctx->Scissor.Enabled)
+   if (ctx->Scissor.Enabled &&
+       (ctx->Scissor.X != 0 ||
+        ctx->Scissor.Y != 0 ||
+        ctx->Scissor.Width < rb->Width ||
+        ctx->Scissor.Height < rb->Height))
       return TRUE;
 
    /* This is correct, but it is necessary to look at the depth clear
-- 
cgit v1.2.3


From 684282953937a37541f26c6e51ceec4134c62dfb Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keithw@vmware.com>
Date: Wed, 29 Jul 2009 12:47:23 +0100
Subject: mesa/st: short-circuit glFinish calls on WIN32 only

Windows opengl32.dll calls glFinish prior to every swapbuffers, which
makes it pretty hard to get decent performance...

Work around by mapping finish to flush on PIPE_OS_WINDOWS.  This is
conformant, though it might confuse poorly-written benchmarks which
attempt to measure a single event rather than figuring out the rate of
continuous processing.
---
 src/mesa/state_tracker/st_cb_flush.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

(limited to 'src/mesa/state_tracker')

diff --git a/src/mesa/state_tracker/st_cb_flush.c b/src/mesa/state_tracker/st_cb_flush.c
index fbaffd154f..f7de4e7a4d 100644
--- a/src/mesa/state_tracker/st_cb_flush.c
+++ b/src/mesa/state_tracker/st_cb_flush.c
@@ -152,4 +152,16 @@ void st_init_flush_functions(struct dd_function_table *functions)
 {
    functions->Flush = st_glFlush;
    functions->Finish = st_glFinish;
+
+   /* Windows opengl32.dll calls glFinish prior to every swapbuffers.
+    * This is unnecessary and degrades performance.  Luckily we have some
+    * scope to work around this, as the externally-visible behaviour of
+    * Finish() is identical to Flush() in all cases - no differences in
+    * rendering or ReadPixels are visible if we opt not to wait here.
+    *
+    * Only set this up on windows to avoid suprise elsewhere.
+    */
+#ifdef PIPE_OS_WINDOWS
+   functions->Finish = st_glFlush;
+#endif
 }
-- 
cgit v1.2.3


From b9889517f5e7eceb14bd103bbdff9ff47ce6ed25 Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Wed, 29 Jul 2009 16:27:03 -0600
Subject: gallium: fix SSE shadow texture instructions

When sampling a 2D shadow map we need 3 texcoord components, not 2.
The third component (distance from light source) is compared against
the texture sample to return the result (visible vs. occluded).

Also, enable proper handling of TGSI_TEXTURE_SHADOW targets in Mesa->TGSI
translation.  There's a possibility for breakage in gallium drivers if
they fail to handle the TGSI_TEXTURE_SHADOW1D / TGSI_TEXTURE_SHADOW2D /
TGSI_TEXTURE_SHADOWRECT texture targets for TGSI_OPCODE_TEX/TXP instructions,
but that should be easy to fix.

With these changes, progs/demos/shadowtex.c renders properly again with
softpipe.
---
 src/gallium/auxiliary/tgsi/tgsi_sse2.c   | 6 +++---
 src/mesa/state_tracker/st_mesa_to_tgsi.c | 6 ------
 2 files changed, 3 insertions(+), 9 deletions(-)

(limited to 'src/mesa/state_tracker')

diff --git a/src/gallium/auxiliary/tgsi/tgsi_sse2.c b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
index 781ea1e75c..16848f7cc5 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_sse2.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
@@ -1467,15 +1467,15 @@ emit_tex( struct x86_function *func,
 
    switch (inst->InstructionExtTexture.Texture) {
    case TGSI_TEXTURE_1D:
-   case TGSI_TEXTURE_SHADOW1D:
       count = 1;
       break;
    case TGSI_TEXTURE_2D:
    case TGSI_TEXTURE_RECT:
-   case TGSI_TEXTURE_SHADOW2D:
-   case TGSI_TEXTURE_SHADOWRECT:
       count = 2;
       break;
+   case TGSI_TEXTURE_SHADOW1D:
+   case TGSI_TEXTURE_SHADOW2D:
+   case TGSI_TEXTURE_SHADOWRECT:
    case TGSI_TEXTURE_3D:
    case TGSI_TEXTURE_CUBE:
       count = 3;
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c
index 6380cd6b2a..c8fb39d558 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.c
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c
@@ -158,12 +158,6 @@ map_texture_target(
     GLuint textarget,
     GLboolean shadow )
 {
-#if 1
-   /* XXX remove this line after we've checked that the rest of gallium
-    * can handle the TGSI_TEXTURE_SHADOWx tokens.
-    */
-   shadow = GL_FALSE;
-#endif
    switch( textarget ) {
    case TEXTURE_1D_INDEX:
       if (shadow)
-- 
cgit v1.2.3


From 9d0b8d72d8d704ff4d8e10448b60cbb42f07eecb Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Wed, 29 Jul 2009 20:07:41 -0600
Subject: mesa: add new FRAG_ATTRIB_FACE and FRAG_ATTRIB_PNTC fragment program
 inputs

Previously, the FOGC attribute contained the fragment fog coord, front/back-
face flag and the gl_PointCoord.xy values.  Now each of those things are
separate fragment program attributes.  This simplifies quite a few things in
Mesa and gallium.

Need to test i965 driver and fix up point coord handling in the gallium/draw
module...
---
 src/mesa/drivers/dri/i965/brw_wm_fp.c    | 61 ++++++++++++++++++--------------
 src/mesa/main/mtypes.h                   |  9 ++---
 src/mesa/shader/arbprogparse.c           |  7 ----
 src/mesa/shader/programopt.c             |  1 -
 src/mesa/shader/slang/slang_codegen.c    |  4 +--
 src/mesa/shader/slang/slang_link.c       | 14 --------
 src/mesa/state_tracker/st_atom_shader.c  | 17 ---------
 src/mesa/state_tracker/st_mesa_to_tgsi.c | 21 -----------
 src/mesa/state_tracker/st_program.c      | 42 ++++++++--------------
 src/mesa/swrast/s_fragprog.c             |  5 ++-
 src/mesa/swrast/s_points.c               | 26 +++++++-------
 11 files changed, 69 insertions(+), 138 deletions(-)

(limited to 'src/mesa/state_tracker')

diff --git a/src/mesa/drivers/dri/i965/brw_wm_fp.c b/src/mesa/drivers/dri/i965/brw_wm_fp.c
index b9e8dd2e96..606baab967 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_fp.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_fp.c
@@ -376,14 +376,6 @@ static void emit_interp( struct brw_wm_compile *c,
       }
       break;
    case FRAG_ATTRIB_FOGC:
-      /* The FOGC input is really special.  When a program uses glFogFragCoord,
-       * the results returned are supposed to be (f,0,0,1).  But for Mesa GLSL,
-       * the glFrontFacing and glPointCoord values are also stashed in FOGC.
-       * So, write the interpolated fog value to X, then either 0, 1, or the
-       * stashed values to Y, Z, W.  Note that this means that
-       * glFogFragCoord.yzw can be wrong in those cases!
-       */
-
       /* Interpolate the fog coordinate */
       emit_op(c,
 	      WM_PINTERP,
@@ -393,26 +385,40 @@ static void emit_interp( struct brw_wm_compile *c,
 	      deltas,
 	      get_pixel_w(c));
 
-      /* Move the front facing value into FOGC.y if it's needed. */
-      if (c->fp->program.UsesFrontFacing) {
-	 emit_op(c,
-		 WM_FRONTFACING,
-		 dst_mask(dst, WRITEMASK_Y),
-		 0,
-		 src_undef(),
-		 src_undef(),
-		 src_undef());
-      } else {
-	 emit_op(c,
-		 OPCODE_MOV,
-		 dst_mask(dst, WRITEMASK_Y),
-		 0,
-		 src_swizzle1(interp, SWIZZLE_ZERO),
-		 src_undef(),
-		 src_undef());
-      }
+      emit_op(c,
+	      OPCODE_MOV,
+	      dst_mask(dst, WRITEMASK_YZW),
+	      0,
+	      src_swizzle(interp,
+			  SWIZZLE_ZERO,
+			  SWIZZLE_ZERO,
+			  SWIZZLE_ZERO,
+			  SWIZZLE_ONE),
+	      src_undef(),
+	      src_undef());
+      break;
+
+   case FRAG_ATTRIB_FACE:
+      /* XXX review/test this case */
+      emit_op(c,
+              WM_FRONTFACING,
+              dst_mask(dst, WRITEMASK_X),
+              0,
+              src_undef(),
+              src_undef(),
+              src_undef());
+      break;
+
+   case FRAG_ATTRIB_PNTC:
+      /* XXX review/test this case */
+      emit_op(c,
+	      WM_PINTERP,
+	      dst_mask(dst, WRITEMASK_XY),
+	      0,
+	      interp,
+	      deltas,
+	      get_pixel_w(c));
 
-      /* Should do the PointCoord thing here. */
       emit_op(c,
 	      OPCODE_MOV,
 	      dst_mask(dst, WRITEMASK_ZW),
@@ -425,6 +431,7 @@ static void emit_interp( struct brw_wm_compile *c,
 	      src_undef(),
 	      src_undef());
       break;
+
    default:
       emit_op(c,
 	      WM_PINTERP,
diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
index d0309f5e90..da01c58c2d 100644
--- a/src/mesa/main/mtypes.h
+++ b/src/mesa/main/mtypes.h
@@ -227,7 +227,9 @@ typedef enum
    FRAG_ATTRIB_TEX5 = 9,
    FRAG_ATTRIB_TEX6 = 10,
    FRAG_ATTRIB_TEX7 = 11,
-   FRAG_ATTRIB_VAR0 = 12,  /**< shader varying */
+   FRAG_ATTRIB_FACE = 12,  /**< front/back face */
+   FRAG_ATTRIB_PNTC = 13,  /**< sprite/point coord */
+   FRAG_ATTRIB_VAR0 = 14,  /**< shader varying */
    FRAG_ATTRIB_MAX = (FRAG_ATTRIB_VAR0 + MAX_VARYING)
 } gl_frag_attrib;
 
@@ -239,6 +241,8 @@ typedef enum
 #define FRAG_BIT_COL0  (1 << FRAG_ATTRIB_COL0)
 #define FRAG_BIT_COL1  (1 << FRAG_ATTRIB_COL1)
 #define FRAG_BIT_FOGC  (1 << FRAG_ATTRIB_FOGC)
+#define FRAG_BIT_FACE  (1 << FRAG_ATTRIB_FACE)
+#define FRAG_BIT_PNTC  (1 << FRAG_ATTRIB_PNTC)
 #define FRAG_BIT_TEX0  (1 << FRAG_ATTRIB_TEX0)
 #define FRAG_BIT_TEX1  (1 << FRAG_ATTRIB_TEX1)
 #define FRAG_BIT_TEX2  (1 << FRAG_ATTRIB_TEX2)
@@ -1834,9 +1838,6 @@ struct gl_fragment_program
    struct gl_program Base;   /**< base class */
    GLenum FogOption;
    GLboolean UsesKill;          /**< shader uses KIL instruction */
-   GLboolean UsesPointCoord;    /**< shader uses gl_PointCoord */
-   GLboolean UsesFrontFacing;   /**< shader used gl_FrontFacing */
-   GLboolean UsesFogFragCoord;  /**< shader used gl_FogFragCoord */
 };
 
 
diff --git a/src/mesa/shader/arbprogparse.c b/src/mesa/shader/arbprogparse.c
index bc65aba39a..f428ee541b 100644
--- a/src/mesa/shader/arbprogparse.c
+++ b/src/mesa/shader/arbprogparse.c
@@ -3974,13 +3974,6 @@ _mesa_parse_arb_fragment_program(GLcontext* ctx, GLenum target,
    if (program->FogOption)
       program->Base.InputsRead |= FRAG_BIT_FOGC;
 
-   /* XXX: assume that ARB fragment programs don't have access to the
-    * FrontFacing and PointCoord values stuffed into the fog
-    * coordinate in GLSL shaders.
-    */
-   if (program->Base.InputsRead & FRAG_BIT_FOGC)
-      program->UsesFogFragCoord = GL_TRUE;
-      
    if (program->Base.Instructions)
       _mesa_free(program->Base.Instructions);
    program->Base.Instructions = ap.Base.Instructions;
diff --git a/src/mesa/shader/programopt.c b/src/mesa/shader/programopt.c
index ac5fe0f691..f70c75cec8 100644
--- a/src/mesa/shader/programopt.c
+++ b/src/mesa/shader/programopt.c
@@ -396,7 +396,6 @@ _mesa_append_fog_code(GLcontext *ctx, struct gl_fragment_program *fprog)
    fprog->Base.Instructions = newInst;
    fprog->Base.NumInstructions = inst - newInst;
    fprog->Base.InputsRead |= FRAG_BIT_FOGC;
-   fprog->UsesFogFragCoord = GL_TRUE;
    /* XXX do this?  fprog->FogOption = GL_NONE; */
 }
 
diff --git a/src/mesa/shader/slang/slang_codegen.c b/src/mesa/shader/slang/slang_codegen.c
index 2b7e781f98..4fe68eafa7 100644
--- a/src/mesa/shader/slang/slang_codegen.c
+++ b/src/mesa/shader/slang/slang_codegen.c
@@ -381,8 +381,8 @@ _slang_input_index(const char *name, GLenum target, GLuint *swizzleOut)
       { "gl_TexCoord", FRAG_ATTRIB_TEX0, SWIZZLE_NOOP },
       /* note: we're packing several quantities into the fogcoord vector */
       { "gl_FogFragCoord", FRAG_ATTRIB_FOGC, SWIZZLE_XXXX },
-      { "gl_FrontFacing", FRAG_ATTRIB_FOGC, SWIZZLE_YYYY }, /*XXX*/
-      { "gl_PointCoord", FRAG_ATTRIB_FOGC, SWIZZLE_ZWWW },
+      { "gl_PointCoord", FRAG_ATTRIB_PNTC, SWIZZLE_XYZW },
+      { "gl_FrontFacing", FRAG_ATTRIB_FACE, SWIZZLE_XXXX },
       { NULL, 0, SWIZZLE_NOOP }
    };
    GLuint i;
diff --git a/src/mesa/shader/slang/slang_link.c b/src/mesa/shader/slang/slang_link.c
index f6032d1e9a..6c19fc895b 100644
--- a/src/mesa/shader/slang/slang_link.c
+++ b/src/mesa/shader/slang/slang_link.c
@@ -484,20 +484,6 @@ _slang_update_inputs_outputs(struct gl_program *prog)
       for (j = 0; j < numSrc; j++) {
          if (inst->SrcReg[j].File == PROGRAM_INPUT) {
             prog->InputsRead |= 1 << inst->SrcReg[j].Index;
-            if (prog->Target == GL_FRAGMENT_PROGRAM_ARB &&
-                inst->SrcReg[j].Index == FRAG_ATTRIB_FOGC) {
-               /* The fragment shader FOGC input is used for fog,
-                * front-facing and sprite/point coord.
-                */
-               struct gl_fragment_program *fp = fragment_program(prog);
-               const GLint swz = GET_SWZ(inst->SrcReg[j].Swizzle, 0);
-               if (swz == SWIZZLE_X)
-                  fp->UsesFogFragCoord = GL_TRUE;
-               else if (swz == SWIZZLE_Y)
-                  fp->UsesFrontFacing = GL_TRUE;
-               else if (swz == SWIZZLE_Z || swz == SWIZZLE_W)
-                  fp->UsesPointCoord = GL_TRUE;
-            }
          }
          else if (inst->SrcReg[j].File == PROGRAM_ADDRESS) {
             maxAddrReg = MAX2(maxAddrReg, (GLuint) (inst->SrcReg[j].Index + 1));
diff --git a/src/mesa/state_tracker/st_atom_shader.c b/src/mesa/state_tracker/st_atom_shader.c
index 8b3bb5cc03..ee649be885 100644
--- a/src/mesa/state_tracker/st_atom_shader.c
+++ b/src/mesa/state_tracker/st_atom_shader.c
@@ -139,23 +139,6 @@ find_translated_vp(struct st_context *st,
          if (fragInputsRead & (1 << inAttr)) {
             stfp->input_to_slot[inAttr] = numIn;
             numIn++;
-            if (((1 << inAttr) & FRAG_BIT_FOGC)) {
-               /* leave placeholders for the
-                * extra registers we extract from fog */
-               if (stfp->Base.UsesFrontFacing) {
-                  if (!stfp->Base.UsesFogFragCoord)
-                     --stfp->input_to_slot[inAttr];
-                  else
-                     ++numIn;
-               }
-               if (stfp->Base.UsesPointCoord) {
-                  if (!stfp->Base.UsesFrontFacing &&
-                      !stfp->Base.UsesFogFragCoord)
-                     stfp->input_to_slot[inAttr] -= 2;
-                  else
-                     ++numIn;
-               }
-            }
          }
          else {
             stfp->input_to_slot[inAttr] = UNUSED;
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c
index 6380cd6b2a..210ca82c42 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.c
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c
@@ -112,27 +112,6 @@ map_register_file_index(
 {
    switch( file ) {
    case TGSI_FILE_INPUT:
-      if (procType == TGSI_PROCESSOR_FRAGMENT &&
-          index == FRAG_ATTRIB_FOGC) {
-         if (GET_SWZ(*swizzle, 0) == SWIZZLE_X) {
-            /* do nothing we're, ok */
-         } else if (GET_SWZ(*swizzle, 0) == SWIZZLE_Y) {
-            /* replace the swizzle with xxxx */
-            *swizzle = MAKE_SWIZZLE4(SWIZZLE_X,
-                                     SWIZZLE_X,
-                                     SWIZZLE_X,
-                                     SWIZZLE_X);
-            /* register after fog */
-            return inputMapping[index] + 1;
-         } else {
-            *swizzle = MAKE_SWIZZLE4(SWIZZLE_Z,
-                                     SWIZZLE_W,
-                                     SWIZZLE_Z,
-                                     SWIZZLE_W);
-            /* register after frontface */
-            return inputMapping[index] + 2;
-         }
-      }
       /* inputs are mapped according to the user-defined map */
       return inputMapping[index];
 
diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c
index 806e0ca8f6..d2da20ae42 100644
--- a/src/mesa/state_tracker/st_program.c
+++ b/src/mesa/state_tracker/st_program.c
@@ -458,34 +458,20 @@ st_translate_fragment_program(struct st_context *st,
             stfp->input_semantic_index[slot] = 1;
             interpMode[slot] = TGSI_INTERPOLATE_LINEAR;
             break;
-         case FRAG_ATTRIB_FOGC: {
-            int extra_decls = 0;
-            if (stfp->Base.UsesFogFragCoord) {
-               stfp->input_semantic_name[slot] = TGSI_SEMANTIC_FOG;
-               stfp->input_semantic_index[slot] = 0;
-               interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE;
-               input_flags[slot] = stfp->Base.Base.InputFlags[attr];
-               ++extra_decls;
-	    }
-            if (stfp->Base.UsesFrontFacing) {
-               GLint idx = slot + extra_decls;
-               stfp->input_semantic_name[idx] = TGSI_SEMANTIC_FACE;
-               stfp->input_semantic_index[idx] = 0;
-               interpMode[idx] = TGSI_INTERPOLATE_CONSTANT;
-               input_flags[idx] = stfp->Base.Base.InputFlags[attr];
-               ++extra_decls;
-            }
-            if (stfp->Base.UsesPointCoord) {
-               GLint idx = slot + extra_decls;
-               stfp->input_semantic_name[idx] = TGSI_SEMANTIC_GENERIC;
-               stfp->input_semantic_index[idx] = num_generic++;
-               interpMode[idx] = TGSI_INTERPOLATE_PERSPECTIVE;
-               input_flags[idx] = stfp->Base.Base.InputFlags[attr];
-               ++extra_decls;
-            }
-            fs_num_inputs += extra_decls - 1;
-            continue;
-         }
+         case FRAG_ATTRIB_FOGC:
+            stfp->input_semantic_name[slot] = TGSI_SEMANTIC_FOG;
+            stfp->input_semantic_index[slot] = 0;
+            interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE;
+            break;
+         case FRAG_ATTRIB_FACE:
+            stfp->input_semantic_name[slot] = TGSI_SEMANTIC_FACE;
+            stfp->input_semantic_index[slot] = num_generic++;
+            interpMode[slot] = TGSI_INTERPOLATE_CONSTANT;
+            break;
+         case FRAG_ATTRIB_PNTC:
+            stfp->input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
+            stfp->input_semantic_index[slot] = num_generic++;
+            interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE;
             break;
          case FRAG_ATTRIB_TEX0:
          case FRAG_ATTRIB_TEX1:
diff --git a/src/mesa/swrast/s_fragprog.c b/src/mesa/swrast/s_fragprog.c
index b71fb9eae9..613a91b0ec 100644
--- a/src/mesa/swrast/s_fragprog.c
+++ b/src/mesa/swrast/s_fragprog.c
@@ -157,9 +157,8 @@ init_machine(GLcontext *ctx, struct gl_program_machine *machine,
 
    /* if running a GLSL program (not ARB_fragment_program) */
    if (ctx->Shader.CurrentProgram) {
-      /* Store front/back facing value in register FOGC.Y */
-      machine->Attribs[FRAG_ATTRIB_FOGC][col][1] = 1.0 - span->facing;
-      /* Note FOGC.ZW is gl_PointCoord if drawing a sprite */
+      /* Store front/back facing value */
+      machine->Attribs[FRAG_ATTRIB_FACE][col][0] = 1.0 - span->facing;
    }
 
    machine->CurElement = col;
diff --git a/src/mesa/swrast/s_points.c b/src/mesa/swrast/s_points.c
index 0a3ad97a71..64c9cda516 100644
--- a/src/mesa/swrast/s_points.c
+++ b/src/mesa/swrast/s_points.c
@@ -139,7 +139,8 @@ sprite_point(GLcontext *ctx, const SWvertex *vert)
       }
 
       ATTRIB_LOOP_BEGIN
-         if (attr >= FRAG_ATTRIB_TEX0 && attr < FRAG_ATTRIB_VAR0) {
+         if ((attr >= FRAG_ATTRIB_TEX0 && attr <= FRAG_ATTRIB_TEX7) ||
+            attr >= FRAG_ATTRIB_VAR0) {
             const GLuint u = attr - FRAG_ATTRIB_TEX0;
             /* a texcoord */
             if (ctx->Point.CoordReplace[u]) {
@@ -170,15 +171,15 @@ sprite_point(GLcontext *ctx, const SWvertex *vert)
                continue;
             }
          }
-         else if (attr == FRAG_ATTRIB_FOGC) {
-            /* GLSL gl_PointCoord is stored in fog.zw */
-            span.attrStart[FRAG_ATTRIB_FOGC][2] = 0.0;
-            span.attrStart[FRAG_ATTRIB_FOGC][3] = 0.0; /* t0 set below */
-            span.attrStepX[FRAG_ATTRIB_FOGC][2] = dsdx;
-            span.attrStepX[FRAG_ATTRIB_FOGC][3] = 0.0;
-            span.attrStepY[FRAG_ATTRIB_FOGC][2] = 0.0;
-            span.attrStepY[FRAG_ATTRIB_FOGC][3] = dtdy;
-            tCoords[numTcoords++] = FRAG_ATTRIB_FOGC;
+         else if (attr == FRAG_ATTRIB_PNTC) {
+            /* GLSL gl_PointCoord.xy (.zw undefined) */
+            span.attrStart[FRAG_ATTRIB_PNTC][0] = 0.0;
+            span.attrStart[FRAG_ATTRIB_PNTC][1] = 0.0; /* t0 set below */
+            span.attrStepX[FRAG_ATTRIB_PNTC][0] = dsdx;
+            span.attrStepX[FRAG_ATTRIB_PNTC][1] = 0.0;
+            span.attrStepY[FRAG_ATTRIB_PNTC][0] = 0.0;
+            span.attrStepY[FRAG_ATTRIB_PNTC][1] = dtdy;
+            tCoords[numTcoords++] = FRAG_ATTRIB_PNTC;
             continue;
          }
          /* use vertex's texcoord/attrib */
@@ -221,10 +222,7 @@ sprite_point(GLcontext *ctx, const SWvertex *vert)
          GLuint i;
          /* setup texcoord T for this row */
          for (i = 0; i < numTcoords; i++) {
-            if (tCoords[i] == FRAG_ATTRIB_FOGC)
-               span.attrStart[FRAG_ATTRIB_FOGC][3] = tcoord;
-            else
-               span.attrStart[tCoords[i]][1] = tcoord;
+            span.attrStart[tCoords[i]][1] = tcoord;
          }
 
          /* these might get changed by span clipping */
-- 
cgit v1.2.3


From f583745519b2b99ca637cdfa6201fd653c848fd6 Mon Sep 17 00:00:00 2001
From: Thomas Hellstrom <thellstrom@vmware.com>
Date: Thu, 30 Jul 2009 12:34:02 +0200
Subject: mesa st: Report unsupported render-to-texture formats.

If a texture image is bound to a framebuffer for render-to-texture, but
the hardware doesn't support rendering to its internal format,
report the framebuffer as incomplete with FRAMEBUFFER_UNSUPPORTED.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
---
 src/mesa/state_tracker/st_cb_fbo.c | 53 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 53 insertions(+)

(limited to 'src/mesa/state_tracker')

diff --git a/src/mesa/state_tracker/st_cb_fbo.c b/src/mesa/state_tracker/st_cb_fbo.c
index ecdb988033..a96602878e 100644
--- a/src/mesa/state_tracker/st_cb_fbo.c
+++ b/src/mesa/state_tracker/st_cb_fbo.c
@@ -445,6 +445,35 @@ st_finish_render_texture(GLcontext *ctx,
 }
 
 
+/**
+ * Validate a renderbuffer attachment for a particular usage.
+ */
+
+static GLboolean
+st_validate_attachment(struct pipe_screen *screen,
+		       const struct gl_renderbuffer_attachment *att,
+		       GLuint usage)
+{
+   const struct st_texture_object *stObj =
+      st_texture_object(att->Texture);
+
+   /**
+    * Only validate texture attachments for now, since
+    * st_renderbuffer_alloc_storage makes sure that
+    * the format is supported.
+    */
+
+   if (att->Type != GL_TEXTURE)
+      return GL_TRUE;
+
+   if (!stObj)
+      return GL_FALSE;
+
+   return screen->is_format_supported(screen, stObj->pt->format,
+				      PIPE_TEXTURE_2D,
+				      usage, 0);
+}
+
 /**
  * Check that the framebuffer configuration is valid in terms of what
  * the driver can support.
@@ -454,13 +483,37 @@ st_finish_render_texture(GLcontext *ctx,
 static void
 st_validate_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
 {
+   struct pipe_screen *screen = ctx->st->pipe->screen;
    const struct gl_renderbuffer *depthRb =
       fb->Attachment[BUFFER_DEPTH].Renderbuffer;
    const struct gl_renderbuffer *stencilRb =
       fb->Attachment[BUFFER_STENCIL].Renderbuffer;
+   GLuint i;
 
    if (stencilRb && depthRb && stencilRb != depthRb) {
       fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
+      return;
+   }
+
+   if (!st_validate_attachment(screen,
+			       &fb->Attachment[BUFFER_DEPTH],
+			       PIPE_TEXTURE_USAGE_DEPTH_STENCIL)) {
+      fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
+      return;
+   }
+   if (!st_validate_attachment(screen,
+			       &fb->Attachment[BUFFER_STENCIL],
+			       PIPE_TEXTURE_USAGE_DEPTH_STENCIL)) {
+      fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
+      return;
+   }
+   for (i = 0; i < ctx->Const.MaxColorAttachments; i++) {
+      if (!st_validate_attachment(screen,
+				  &fb->Attachment[BUFFER_COLOR0 + i],
+				  PIPE_TEXTURE_USAGE_RENDER_TARGET)) {
+	 fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
+	 return;
+      }
    }
 }
 
-- 
cgit v1.2.3


From 98ba6cfb4cc7440f0fb4f101ff714143720e175e Mon Sep 17 00:00:00 2001
From: Thomas Hellstrom <thellstrom@vmware.com>
Date: Thu, 30 Jul 2009 12:39:26 +0200
Subject: mesa st: Change the preferred internal format for some RGB formats.

For GL_RGB5, GL_RGB4 and GL_R3_G3_B2, prefer
PIPE_FORMAT_R5G6B5_UNORM over PIPE_FORMAT_A1R5G5B5_UNORM, since hardware
is more likely to support the previous format for rendering.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
---
 src/mesa/state_tracker/st_format.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

(limited to 'src/mesa/state_tracker')

diff --git a/src/mesa/state_tracker/st_format.c b/src/mesa/state_tracker/st_format.c
index b243c249e3..dcb90a3107 100644
--- a/src/mesa/state_tracker/st_format.c
+++ b/src/mesa/state_tracker/st_format.c
@@ -435,10 +435,10 @@ st_choose_format(struct pipe_context *pipe, GLenum internalFormat,
    case GL_RGB5:
    case GL_RGB4:
    case GL_R3_G3_B2:
-      if (screen->is_format_supported( screen, PIPE_FORMAT_A1R5G5B5_UNORM, target, tex_usage, geom_flags ))
-         return PIPE_FORMAT_A1R5G5B5_UNORM;
       if (screen->is_format_supported( screen, PIPE_FORMAT_R5G6B5_UNORM, target, tex_usage, geom_flags ))
          return PIPE_FORMAT_R5G6B5_UNORM;
+      if (screen->is_format_supported( screen, PIPE_FORMAT_A1R5G5B5_UNORM, target, tex_usage, geom_flags ))
+         return PIPE_FORMAT_A1R5G5B5_UNORM;
       return default_rgba_format( screen, target, tex_usage, geom_flags );
 
    case GL_ALPHA:
-- 
cgit v1.2.3


From cb90c43676c258419e4b617c908570891d3674cb Mon Sep 17 00:00:00 2001
From: Michal Krol <michal@vmware.com>
Date: Fri, 31 Jul 2009 18:12:53 +0200
Subject: Rename TGSI LOOP instruction to better match theri usage.

The LOOP/ENDLOOP pair is renamed to BGNFOR/ENDFOR as its behaviour
is similar to a C language for-loop.

The BGNLOOP2/ENDLOOP2 pair is renamed to BGNLOOP/ENDLOOP as now
there is no name collision.
---
 src/gallium/auxiliary/gallivm/tgsitollvm.cpp        | 16 ++++++++--------
 src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt | 12 ++++++------
 src/gallium/auxiliary/tgsi/tgsi_dump.c              |  4 ++--
 src/gallium/auxiliary/tgsi/tgsi_exec.c              |  8 ++++----
 src/gallium/auxiliary/tgsi/tgsi_info.c              |  8 ++++----
 src/gallium/auxiliary/tgsi/tgsi_sanity.c            |  6 +++---
 src/gallium/auxiliary/tgsi/tgsi_sse2.c              |  4 ++--
 src/gallium/drivers/cell/ppu/cell_gen_fp.c          |  4 ++--
 src/gallium/drivers/cell/spu/spu_exec.c             |  8 ++++----
 src/gallium/drivers/i965simple/brw_wm_glsl.c        |  6 +++---
 src/gallium/include/pipe/p_shader_tokens.h          |  8 ++++----
 src/gallium/state_trackers/vega/asm_filters.h       |  4 ++--
 src/mesa/drivers/dri/i965/brw_wm_glsl.c             |  2 +-
 src/mesa/state_tracker/st_mesa_to_tgsi.c            |  4 ++--
 14 files changed, 47 insertions(+), 47 deletions(-)

(limited to 'src/mesa/state_tracker')

diff --git a/src/gallium/auxiliary/gallivm/tgsitollvm.cpp b/src/gallium/auxiliary/gallivm/tgsitollvm.cpp
index 8d885e48be..bf84401e11 100644
--- a/src/gallium/auxiliary/gallivm/tgsitollvm.cpp
+++ b/src/gallium/auxiliary/gallivm/tgsitollvm.cpp
@@ -516,7 +516,7 @@ translate_instruction(llvm::Module *module,
       return;  //just update the state
    }
       break;
-   case TGSI_OPCODE_LOOP:
+   case TGSI_OPCODE_BGNFOR:
       break;
    case TGSI_OPCODE_REP:
       break;
@@ -532,7 +532,7 @@ translate_instruction(llvm::Module *module,
       return; //just update the state
    }
       break;
-   case TGSI_OPCODE_ENDLOOP:
+   case TGSI_OPCODE_ENDFOR:
       break;
    case TGSI_OPCODE_ENDREP:
       break;
@@ -574,7 +574,7 @@ translate_instruction(llvm::Module *module,
       break;
    case TGSI_OPCODE_ENDPRIM:
       break;
-   case TGSI_OPCODE_BGNLOOP2: {
+   case TGSI_OPCODE_BGNLOOP: {
       instr->beginLoop();
       storage->setCurrentBlock(instr->currentBlock());
       return;
@@ -587,7 +587,7 @@ translate_instruction(llvm::Module *module,
       return;
    }
       break;
-   case TGSI_OPCODE_ENDLOOP2: {
+   case TGSI_OPCODE_ENDLOOP: {
       instr->endLoop();
       storage->setCurrentBlock(instr->currentBlock());
       return;
@@ -890,7 +890,7 @@ translate_instructionir(llvm::Module *module,
    case TGSI_OPCODE_IF: {
    }
       break;
-   case TGSI_OPCODE_LOOP:
+   case TGSI_OPCODE_BGNFOR:
       break;
    case TGSI_OPCODE_REP:
       break;
@@ -900,7 +900,7 @@ translate_instructionir(llvm::Module *module,
    case TGSI_OPCODE_ENDIF: {
    }
       break;
-   case TGSI_OPCODE_ENDLOOP:
+   case TGSI_OPCODE_ENDFOR:
       break;
    case TGSI_OPCODE_ENDREP:
       break;
@@ -941,13 +941,13 @@ translate_instructionir(llvm::Module *module,
       break;
    case TGSI_OPCODE_ENDPRIM:
       break;
-   case TGSI_OPCODE_BGNLOOP2: {
+   case TGSI_OPCODE_BGNLOOP: {
    }
       break;
    case TGSI_OPCODE_BGNSUB: {
    }
       break;
-   case TGSI_OPCODE_ENDLOOP2: {
+   case TGSI_OPCODE_ENDLOOP: {
    }
       break;
    case TGSI_OPCODE_ENDSUB: {
diff --git a/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt b/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt
index 5f88cc2aca..802ec37118 100644
--- a/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt
+++ b/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt
@@ -665,14 +665,14 @@ TGSI Instruction Specification
   TBD
 
 
-1.9.8  LOOP - Loop
+1.9.8  BGNFOR - Begin a For-Loop
 
   dst.x = floor(src.x)
   dst.y = floor(src.y)
   dst.z = floor(src.z)
 
   if (dst.y <= 0)
-    pc = [matching ENDLOOP] + 1
+    pc = [matching ENDFOR] + 1
   endif
 
   Note: The destination must be a loop register.
@@ -694,13 +694,13 @@ TGSI Instruction Specification
   TBD
 
 
-1.9.12  ENDLOOP - End Loop
+1.9.12  ENDFOR - End a For-Loop
 
   dst.x = dst.x + dst.z
   dst.y = dst.y - 1.0
 
   if (dst.y > 0)
-    pc = [matching LOOP instruction] + 1
+    pc = [matching BGNFOR instruction] + 1
   endif
 
   Note: The destination must be a loop register.
@@ -856,7 +856,7 @@ TGSI Instruction Specification
 ----------
 
 
-1.13.1  BGNLOOP2 - Begin Loop
+1.13.1  BGNLOOP - Begin a Loop
 
   TBD
 
@@ -866,7 +866,7 @@ TGSI Instruction Specification
   TBD
 
 
-1.13.3  ENDLOOP2 - End Loop
+1.13.3  ENDLOOP - End a Loop
 
   TBD
 
diff --git a/src/gallium/auxiliary/tgsi/tgsi_dump.c b/src/gallium/auxiliary/tgsi/tgsi_dump.c
index 4ca3a16b8a..f36b1114a9 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_dump.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_dump.c
@@ -473,8 +473,8 @@ iter_instruction(
    switch (inst->Instruction.Opcode) {
    case TGSI_OPCODE_IF:
    case TGSI_OPCODE_ELSE:
-   case TGSI_OPCODE_BGNLOOP2:
-   case TGSI_OPCODE_ENDLOOP2:
+   case TGSI_OPCODE_BGNLOOP:
+   case TGSI_OPCODE_ENDLOOP:
    case TGSI_OPCODE_CAL:
       TXT( " :" );
       UID( inst->InstructionExtLabel.Label );
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c
index 0179bba5a2..5af0a94794 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c
@@ -3087,9 +3087,9 @@ exec_instruction(
       mach->Primitives[mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0]] = 0;
       break;
 
-   case TGSI_OPCODE_LOOP:
+   case TGSI_OPCODE_BGNFOR:
       /* fall-through (for now) */
-   case TGSI_OPCODE_BGNLOOP2:
+   case TGSI_OPCODE_BGNLOOP:
       /* push LoopMask and ContMasks */
       assert(mach->LoopStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
       mach->LoopStack[mach->LoopStackTop++] = mach->LoopMask;
@@ -3097,9 +3097,9 @@ exec_instruction(
       mach->ContStack[mach->ContStackTop++] = mach->ContMask;
       break;
 
-   case TGSI_OPCODE_ENDLOOP:
+   case TGSI_OPCODE_ENDFOR:
       /* fall-through (for now at least) */
-   case TGSI_OPCODE_ENDLOOP2:
+   case TGSI_OPCODE_ENDLOOP:
       /* Restore ContMask, but don't pop */
       assert(mach->ContStackTop > 0);
       mach->ContMask = mach->ContStack[mach->ContStackTop - 1];
diff --git a/src/gallium/auxiliary/tgsi/tgsi_info.c b/src/gallium/auxiliary/tgsi/tgsi_info.c
index 3a47e9b84d..cd8871e32d 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_info.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_info.c
@@ -106,11 +106,11 @@ static const struct tgsi_opcode_info opcode_info[TGSI_OPCODE_LAST] =
    { 1, 2, 1, 0, "TXL", TGSI_OPCODE_TXL },
    { 0, 0, 0, 0, "BRK", TGSI_OPCODE_BRK },
    { 0, 1, 0, 1, "IF", TGSI_OPCODE_IF },
-   { 1, 1, 0, 0, "LOOP", TGSI_OPCODE_LOOP },
+   { 1, 1, 0, 0, "BGNFOR", TGSI_OPCODE_BGNFOR },
    { 0, 1, 0, 0, "REP", TGSI_OPCODE_REP },
    { 0, 0, 0, 1, "ELSE", TGSI_OPCODE_ELSE },
    { 0, 0, 0, 0, "ENDIF", TGSI_OPCODE_ENDIF },
-   { 1, 0, 0, 0, "ENDLOOP", TGSI_OPCODE_ENDLOOP },
+   { 1, 0, 0, 0, "ENDFOR", TGSI_OPCODE_ENDFOR },
    { 0, 0, 0, 0, "ENDREP", TGSI_OPCODE_ENDREP },
    { 0, 1, 0, 0, "PUSHA", TGSI_OPCODE_PUSHA },
    { 1, 0, 0, 0, "POPA", TGSI_OPCODE_POPA },
@@ -130,9 +130,9 @@ static const struct tgsi_opcode_info opcode_info[TGSI_OPCODE_LAST] =
    { 0, 0, 0, 0, "CONT", TGSI_OPCODE_CONT },
    { 0, 0, 0, 0, "EMIT", TGSI_OPCODE_EMIT },
    { 0, 0, 0, 0, "ENDPRIM", TGSI_OPCODE_ENDPRIM },
-   { 0, 0, 0, 1, "BGNLOOP2", TGSI_OPCODE_BGNLOOP2 },
+   { 0, 0, 0, 1, "BGNLOOP", TGSI_OPCODE_BGNLOOP },
    { 0, 0, 0, 0, "BGNSUB", TGSI_OPCODE_BGNSUB },
-   { 0, 0, 0, 1, "ENDLOOP2", TGSI_OPCODE_ENDLOOP2 },
+   { 0, 0, 0, 1, "ENDLOOP", TGSI_OPCODE_ENDLOOP },
    { 0, 0, 0, 0, "ENDSUB", TGSI_OPCODE_ENDSUB },
    { 1, 1, 0, 0, "NOISE1", TGSI_OPCODE_NOISE1 },
    { 1, 1, 0, 0, "NOISE2", TGSI_OPCODE_NOISE2 },
diff --git a/src/gallium/auxiliary/tgsi/tgsi_sanity.c b/src/gallium/auxiliary/tgsi/tgsi_sanity.c
index cb62a95fc0..4fe8553c42 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_sanity.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_sanity.c
@@ -242,8 +242,8 @@ iter_instruction(
    }
 
    switch (inst->Instruction.Opcode) {
-   case TGSI_OPCODE_LOOP:
-   case TGSI_OPCODE_ENDLOOP:
+   case TGSI_OPCODE_BGNFOR:
+   case TGSI_OPCODE_ENDFOR:
       if (inst->FullDstRegisters[0].DstRegister.File != TGSI_FILE_LOOP ||
           inst->FullDstRegisters[0].DstRegister.Index != 0) {
          report_error(ctx, "Destination register must be LOOP[0]");
@@ -252,7 +252,7 @@ iter_instruction(
    }
 
    switch (inst->Instruction.Opcode) {
-   case TGSI_OPCODE_LOOP:
+   case TGSI_OPCODE_BGNFOR:
       if (inst->FullSrcRegisters[0].SrcRegister.File != TGSI_FILE_CONSTANT &&
           inst->FullSrcRegisters[0].SrcRegister.File != TGSI_FILE_IMMEDIATE) {
          report_error(ctx, "Source register file must be either CONST or IMM");
diff --git a/src/gallium/auxiliary/tgsi/tgsi_sse2.c b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
index 16848f7cc5..52186770e6 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_sse2.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
@@ -2531,7 +2531,7 @@ emit_instruction(
       return 0;
       break;
 
-   case TGSI_OPCODE_LOOP:
+   case TGSI_OPCODE_BGNFOR:
       return 0;
       break;
 
@@ -2547,7 +2547,7 @@ emit_instruction(
       return 0;
       break;
 
-   case TGSI_OPCODE_ENDLOOP:
+   case TGSI_OPCODE_ENDFOR:
       return 0;
       break;
 
diff --git a/src/gallium/drivers/cell/ppu/cell_gen_fp.c b/src/gallium/drivers/cell/ppu/cell_gen_fp.c
index 7cd5656a7e..58a8b5d0b0 100644
--- a/src/gallium/drivers/cell/ppu/cell_gen_fp.c
+++ b/src/gallium/drivers/cell/ppu/cell_gen_fp.c
@@ -1834,9 +1834,9 @@ emit_instruction(struct codegen *gen,
    case TGSI_OPCODE_ENDIF:
       return emit_ENDIF(gen, inst);
 
-   case TGSI_OPCODE_BGNLOOP2:
+   case TGSI_OPCODE_BGNLOOP:
       return emit_BGNLOOP(gen, inst);
-   case TGSI_OPCODE_ENDLOOP2:
+   case TGSI_OPCODE_ENDLOOP:
       return emit_ENDLOOP(gen, inst);
    case TGSI_OPCODE_BRK:
       return emit_BRK(gen, inst);
diff --git a/src/gallium/drivers/cell/spu/spu_exec.c b/src/gallium/drivers/cell/spu/spu_exec.c
index 570553e1d6..6db9501128 100644
--- a/src/gallium/drivers/cell/spu/spu_exec.c
+++ b/src/gallium/drivers/cell/spu/spu_exec.c
@@ -1758,9 +1758,9 @@ exec_instruction(
       mach->Primitives[mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0]] = 0;
       break;
 
-   case TGSI_OPCODE_LOOP:
+   case TGSI_OPCODE_BGNFOR:
       /* fall-through (for now) */
-   case TGSI_OPCODE_BGNLOOP2:
+   case TGSI_OPCODE_BGNLOOP:
       /* push LoopMask and ContMasks */
       ASSERT(mach->LoopStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
       mach->LoopStack[mach->LoopStackTop++] = mach->LoopMask;
@@ -1768,9 +1768,9 @@ exec_instruction(
       mach->ContStack[mach->ContStackTop++] = mach->ContMask;
       break;
 
-   case TGSI_OPCODE_ENDLOOP:
+   case TGSI_OPCODE_ENDFOR:
       /* fall-through (for now at least) */
-   case TGSI_OPCODE_ENDLOOP2:
+   case TGSI_OPCODE_ENDLOOP:
       /* Restore ContMask, but don't pop */
       ASSERT(mach->ContStackTop > 0);
       mach->ContMask = mach->ContStack[mach->ContStackTop - 1];
diff --git a/src/gallium/drivers/i965simple/brw_wm_glsl.c b/src/gallium/drivers/i965simple/brw_wm_glsl.c
index ab6410aa60..db75963932 100644
--- a/src/gallium/drivers/i965simple/brw_wm_glsl.c
+++ b/src/gallium/drivers/i965simple/brw_wm_glsl.c
@@ -947,7 +947,7 @@ static void brw_wm_emit_instruction( struct brw_wm_compile *c,
 #endif
 
       break;
-   case TGSI_OPCODE_LOOP:
+   case TGSI_OPCODE_BGNFOR:
       c->loop_inst[c->loop_insn++] = brw_DO(p, BRW_EXECUTE_8);
       break;
    case TGSI_OPCODE_BRK:
@@ -958,11 +958,11 @@ static void brw_wm_emit_instruction( struct brw_wm_compile *c,
       brw_CONT(p);
       brw_set_predicate_control(p, BRW_PREDICATE_NONE);
       break;
-   case TGSI_OPCODE_ENDLOOP:
+   case TGSI_OPCODE_ENDFOR:
       c->loop_insn--;
       c->inst0 = c->inst1 = brw_WHILE(p, c->loop_inst[c->loop_insn]);
       /* patch all the BREAK instructions from
-	 last BEGINLOOP */
+         last BGNFOR */
       while (c->inst0 > c->loop_inst[c->loop_insn]) {
 	 c->inst0--;
 	 if (c->inst0->header.opcode == BRW_OPCODE_BREAK) {
diff --git a/src/gallium/include/pipe/p_shader_tokens.h b/src/gallium/include/pipe/p_shader_tokens.h
index c4be604e5a..f0ba4fb308 100644
--- a/src/gallium/include/pipe/p_shader_tokens.h
+++ b/src/gallium/include/pipe/p_shader_tokens.h
@@ -238,11 +238,11 @@ union tgsi_immediate_data
 #define TGSI_OPCODE_TXL                 72
 #define TGSI_OPCODE_BRK                 73
 #define TGSI_OPCODE_IF                  74
-#define TGSI_OPCODE_LOOP                75
+#define TGSI_OPCODE_BGNFOR              75
 #define TGSI_OPCODE_REP                 76
 #define TGSI_OPCODE_ELSE                77
 #define TGSI_OPCODE_ENDIF               78
-#define TGSI_OPCODE_ENDLOOP             79
+#define TGSI_OPCODE_ENDFOR              79
 #define TGSI_OPCODE_ENDREP              80
 #define TGSI_OPCODE_PUSHA               81
 #define TGSI_OPCODE_POPA                82
@@ -262,9 +262,9 @@ union tgsi_immediate_data
 #define TGSI_OPCODE_CONT                96
 #define TGSI_OPCODE_EMIT                97
 #define TGSI_OPCODE_ENDPRIM             98
-#define TGSI_OPCODE_BGNLOOP2            99
+#define TGSI_OPCODE_BGNLOOP             99
 #define TGSI_OPCODE_BGNSUB              100
-#define TGSI_OPCODE_ENDLOOP2            101
+#define TGSI_OPCODE_ENDLOOP             101
 #define TGSI_OPCODE_ENDSUB              102
 #define TGSI_OPCODE_NOISE1              103
 #define TGSI_OPCODE_NOISE2              104
diff --git a/src/gallium/state_trackers/vega/asm_filters.h b/src/gallium/state_trackers/vega/asm_filters.h
index 49807b9ab4..9a49f2e12d 100644
--- a/src/gallium/state_trackers/vega/asm_filters.h
+++ b/src/gallium/state_trackers/vega/asm_filters.h
@@ -60,7 +60,7 @@ static const char convolution_asm[] =
    "DCL SAMP[0], CONSTANT\n"
    "0: MOV TEMP[0], CONST[0].xxxx\n"
    "1: MOV TEMP[1], CONST[0].xxxx\n"
-   "2: BGNLOOP2 :14\n"
+   "2: BGNLOOP :14\n"
    "3: SGE TEMP[0].z, TEMP[0].yyyy, CONST[1].xxxx\n"
    "4: IF TEMP[0].zzzz :7\n"
    "5: BRK\n"
@@ -72,7 +72,7 @@ static const char convolution_asm[] =
    "11: MOV TEMP[3], CONST[ADDR[0]+%d]\n"
    "12: MAD TEMP[1], TEMP[2], TEMP[3], TEMP[1]\n"
    "13: ADD TEMP[0].y, TEMP[0].yyyy, CONST[0].yyyy\n"
-   "14: ENDLOOP2 :2\n"
+   "14: ENDLOOP :2\n"
    "15: MAD OUT[0], TEMP[1], CONST[1].yyyy, CONST[1].zzzz\n"
    "16: END\n";
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm_glsl.c b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
index 19f777fe32..85a4237d5a 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_glsl.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
@@ -3007,7 +3007,7 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c)
  
                   loop_depth--;
                   inst0 = inst1 = brw_WHILE(p, loop_inst[loop_depth]);
-                  /* patch all the BREAK/CONT instructions from last BEGINLOOP */
+                  /* patch all the BREAK/CONT instructions from last BGNLOOP */
                   while (inst0 > loop_inst[loop_depth]) {
                      inst0--;
                      if (inst0->header.opcode == BRW_OPCODE_BREAK) {
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c
index c8fb39d558..dd7e40be08 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.c
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c
@@ -368,7 +368,7 @@ compile_instruction(
       fullinst->Instruction.Opcode = TGSI_OPCODE_ADD;
       break;
    case OPCODE_BGNLOOP:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_BGNLOOP2;
+      fullinst->Instruction.Opcode = TGSI_OPCODE_BGNLOOP;
       fullinst->InstructionExtLabel.Label = inst->BranchTarget + preamble_size;
       break;
    case OPCODE_BGNSUB:
@@ -426,7 +426,7 @@ compile_instruction(
       fullinst->Instruction.Opcode = TGSI_OPCODE_ENDIF;
       break;
    case OPCODE_ENDLOOP:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_ENDLOOP2;
+      fullinst->Instruction.Opcode = TGSI_OPCODE_ENDLOOP;
       fullinst->InstructionExtLabel.Label = inst->BranchTarget + preamble_size;
       break;
    case OPCODE_ENDSUB:
-- 
cgit v1.2.3


From c1785c19ca0716a7e85777242949a0c33e28988f Mon Sep 17 00:00:00 2001
From: Patrice Mandin <patmandin@gmail.com>
Date: Sat, 1 Aug 2009 11:03:30 +0200
Subject: mesa st: Move logbase2 function to util/u_math.h

---
 src/gallium/auxiliary/util/u_math.h    | 23 ++++++++++++++++++-----
 src/mesa/state_tracker/st_cb_texture.c | 17 ++++-------------
 2 files changed, 22 insertions(+), 18 deletions(-)

(limited to 'src/mesa/state_tracker')

diff --git a/src/gallium/auxiliary/util/u_math.h b/src/gallium/auxiliary/util/u_math.h
index e5003af01d..30e6e2f6b3 100644
--- a/src/gallium/auxiliary/util/u_math.h
+++ b/src/gallium/auxiliary/util/u_math.h
@@ -216,23 +216,23 @@ util_fast_exp2(float x)
    int32_t ipart;
    float fpart, mpart;
    union fi epart;
-   
+
    if(x > 129.00000f)
       return 3.402823466e+38f;
-   
+
    if(x < -126.99999f)
       return 0.0f;
 
    ipart = (int32_t) x;
    fpart = x - (float) ipart;
-   
+
    /* same as
     *   epart.f = (float) (1 << ipart)
     * but faster and without integer overflow for ipart > 31 */
    epart.i = (ipart + 127 ) << 23;
-   
+
    mpart = pow2_table[POW2_TABLE_OFFSET + (int)(fpart * POW2_TABLE_SCALE)];
-   
+
    return epart.f * mpart;
 }
 
@@ -409,6 +409,19 @@ float_to_ubyte(float f)
 }
 
 
+/**
+ * Calc log base 2
+ */
+static INLINE unsigned
+util_logbase2(unsigned n)
+{
+   unsigned log2 = 0;
+   while (n >>= 1)
+      ++log2;
+   return log2;
+}
+
+
 
 #define CLAMP( X, MIN, MAX )  ( (X)<(MIN) ? (MIN) : ((X)>(MAX) ? (MAX) : (X)) )
 
diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c
index ee71c012c6..e8d7f70ad6 100644
--- a/src/mesa/state_tracker/st_cb_texture.c
+++ b/src/mesa/state_tracker/st_cb_texture.c
@@ -59,6 +59,7 @@
 #include "util/u_tile.h"
 #include "util/u_blit.h"
 #include "util/u_surface.h"
+#include "util/u_math.h"
 
 
 #define DBG if (0) printf
@@ -237,16 +238,6 @@ do_memcpy(void *dest, const void *src, size_t n)
 }
 
 
-static INLINE unsigned
-logbase2(unsigned n)
-{
-   unsigned log2 = 0;
-   while (n >>= 1)
-      ++log2;
-   return log2;
-}
-
-
 /**
  * Return default texture usage bitmask for the given texture format.
  */
@@ -340,9 +331,9 @@ guess_and_alloc_texture(struct st_context *st,
       lastLevel = firstLevel;
    }
    else {
-      GLuint l2width = logbase2(width);
-      GLuint l2height = logbase2(height);
-      GLuint l2depth = logbase2(depth);
+      GLuint l2width = util_logbase2(width);
+      GLuint l2height = util_logbase2(height);
+      GLuint l2depth = util_logbase2(depth);
       lastLevel = firstLevel + MAX2(MAX2(l2width, l2height), l2depth);
    }
 
-- 
cgit v1.2.3


From fc3d564daeacdbd76b97de2ffc10e15931a18c7a Mon Sep 17 00:00:00 2001
From: Patrice Mandin <patmandin@gmail.com>
Date: Sat, 1 Aug 2009 11:15:18 +0200
Subject: mesa st: Use POT texture for draw pixels operations if NPOT texture
 is not supported

---
 src/mesa/state_tracker/st_cb_drawpixels.c | 68 ++++++++++++++++++++++++++-----
 1 file changed, 58 insertions(+), 10 deletions(-)

(limited to 'src/mesa/state_tracker')

diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c
index 8b5094a04f..b39403129d 100644
--- a/src/mesa/state_tracker/st_cb_drawpixels.c
+++ b/src/mesa/state_tracker/st_cb_drawpixels.c
@@ -60,6 +60,7 @@
 #include "pipe/p_inlines.h"
 #include "util/u_tile.h"
 #include "util/u_draw_quad.h"
+#include "util/u_math.h"
 #include "shader/prog_instruction.h"
 #include "cso_cache/cso_context.h"
 
@@ -341,6 +342,7 @@ make_texture(struct st_context *st,
    enum pipe_format pipeFormat;
    GLuint cpp;
    GLenum baseFormat;
+   int ptw, pth;
 
    baseFormat = _mesa_base_format(format);
 
@@ -355,7 +357,28 @@ make_texture(struct st_context *st,
    if (!pixels)
       return NULL;
 
-   pt = st_texture_create(st, PIPE_TEXTURE_2D, pipeFormat, 0, width, height, 1,
+   /* Need to use POT texture? */
+   ptw = width;
+   pth = height;
+   if (!screen->get_param(screen, PIPE_CAP_NPOT_TEXTURES)) {
+      int l2pt, maxSize;
+
+      l2pt = util_logbase2(width);
+      if (1<<l2pt != width) {
+         ptw = 1<<(l2pt+1);
+      }
+      l2pt = util_logbase2(height);
+      if (1<<l2pt != height) {
+         pth = 1<<(l2pt+1);
+      }
+
+      /* Check against maximum texture size */
+      maxSize = 1 << (pipe->screen->get_param(pipe->screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS) - 1);
+      assert(ptw <= maxSize);
+      assert(pth <= maxSize);
+   }
+
+   pt = st_texture_create(st, PIPE_TEXTURE_2D, pipeFormat, 0, ptw, pth, 1,
                           PIPE_TEXTURE_USAGE_SAMPLER);
    if (!pt) {
       _mesa_unmap_drawpix_pbo(ctx, unpack);
@@ -420,7 +443,7 @@ make_texture(struct st_context *st,
 static void
 draw_quad(GLcontext *ctx, GLfloat x0, GLfloat y0, GLfloat z,
           GLfloat x1, GLfloat y1, const GLfloat *color,
-          GLboolean invertTex)
+          GLboolean invertTex, GLfloat maxXcoord, GLfloat maxYcoord)
 {
    struct st_context *st = ctx->st;
    struct pipe_context *pipe = ctx->st->pipe;
@@ -435,8 +458,9 @@ draw_quad(GLcontext *ctx, GLfloat x0, GLfloat y0, GLfloat z,
       const GLfloat clip_y0 = y0 / fb_height * 2.0f - 1.0f;
       const GLfloat clip_x1 = x1 / fb_width * 2.0f - 1.0f;
       const GLfloat clip_y1 = y1 / fb_height * 2.0f - 1.0f;
-      const GLfloat sLeft = 0.0f, sRight = 1.0f;
-      const GLfloat tTop = invertTex, tBot = 1.0f - tTop;
+      const GLfloat sLeft = 0.0f, sRight = maxXcoord;
+      const GLfloat tTop = invertTex ? maxYcoord : 0.0f;
+      const GLfloat tBot = invertTex ? 0.0f : maxYcoord;
       GLuint tex, i;
 
       /* upper-left */
@@ -608,7 +632,9 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z,
    y0 = (GLfloat) y;
    y1 = y + height * ctx->Pixel.ZoomY;
 
-   draw_quad(ctx, x0, y0, z, x1, y1, color, invertTex);
+   draw_quad(ctx, x0, y0, z, x1, y1, color, invertTex,
+	     (GLfloat) width / pt->width[0],
+	     (GLfloat) height / pt->height[0]);
 
    /* restore state */
    cso_restore_rasterizer(cso);
@@ -648,7 +674,7 @@ draw_stencil_pixels(GLcontext *ctx, GLint x, GLint y,
       usage = PIPE_TRANSFER_READ_WRITE;
    else
       usage = PIPE_TRANSFER_WRITE;
-   
+
    pt = st_cond_flush_get_tex_transfer(st_context(ctx), strb->texture, 0, 0, 0,
 				       usage, x, y,
 				       width, height);
@@ -841,7 +867,7 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy,
       usage = PIPE_TRANSFER_READ_WRITE;
    else
       usage = PIPE_TRANSFER_WRITE;
-   
+
    ptDraw = st_cond_flush_get_tex_transfer(st_context(ctx),
 					   rbDraw->texture, 0, 0, 0,
 					   usage, dstx, dsty,
@@ -849,7 +875,7 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy,
 
    assert(ptDraw->block.width == 1);
    assert(ptDraw->block.height == 1);
-   
+
    /* map the stencil buffer */
    drawMap = screen->transfer_map(screen, ptDraw);
 
@@ -923,6 +949,7 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy,
    struct pipe_texture *pt;
    GLfloat *color;
    enum pipe_format srcFormat, texFormat;
+   int ptw, pth;
 
    pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, NULL);
 
@@ -1004,13 +1031,34 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy,
          height -= -srcy;
          srcy = 0;
       }
-      
+
       if (height < 0)
          return;
    }
 
+   /* Need to use POT texture? */
+   ptw = width;
+   pth = height;
+   if (!screen->get_param(screen, PIPE_CAP_NPOT_TEXTURES)) {
+      int l2pt, maxSize;
+
+      l2pt = util_logbase2(width);
+      if (1<<l2pt != width) {
+         ptw = 1<<(l2pt+1);
+      }
+      l2pt = util_logbase2(height);
+      if (1<<l2pt != height) {
+         pth = 1<<(l2pt+1);
+      }
+
+      /* Check against maximum texture size */
+      maxSize = 1 << (pipe->screen->get_param(pipe->screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS) - 1);
+      assert(ptw <= maxSize);
+      assert(pth <= maxSize);
+   }
+
    pt = st_texture_create(ctx->st, PIPE_TEXTURE_2D, texFormat, 0,
-                          width, height, 1,
+                          ptw, pth, 1,
                           PIPE_TEXTURE_USAGE_SAMPLER);
    if (!pt)
       return;
-- 
cgit v1.2.3


From ed6125fe9b2a806103cc58a60027aa2aba64dd2f Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Mon, 3 Aug 2009 11:16:23 -0600
Subject: st/mesa: we don't support GL_NV_point_sprite (see comment)

---
 src/mesa/state_tracker/st_extensions.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

(limited to 'src/mesa/state_tracker')

diff --git a/src/mesa/state_tracker/st_extensions.c b/src/mesa/state_tracker/st_extensions.c
index e6728ae895..9f6564dbc5 100644
--- a/src/mesa/state_tracker/st_extensions.c
+++ b/src/mesa/state_tracker/st_extensions.c
@@ -222,7 +222,9 @@ void st_init_extensions(struct st_context *st)
 
    if (screen->get_param(screen, PIPE_CAP_POINT_SPRITE)) {
       ctx->Extensions.ARB_point_sprite = GL_TRUE;
-      ctx->Extensions.NV_point_sprite = GL_TRUE;
+      /* GL_NV_point_sprite is not supported by gallium because we don't
+       * support the GL_POINT_SPRITE_R_MODE_NV option.
+       */
    }
 
    if (screen->get_param(screen, PIPE_CAP_OCCLUSION_QUERY)) {
-- 
cgit v1.2.3


From f792137593b16b850a8a95dbb4859d49effb9f7c Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Wed, 5 Aug 2009 13:26:19 -0600
Subject: st/mesa: fix Y inversion and optimize st_BlitFramebuffer()

Need to check for Y inversion separately for src/dest buffers.

If both the src and dest regions are upside down, make them right-side
up for a better chance at a fast path.

progs/tests/copypixrate -blit is much faster now.
---
 src/mesa/state_tracker/st_cb_blit.c | 28 +++++++++++++++++++++++++---
 1 file changed, 25 insertions(+), 3 deletions(-)

(limited to 'src/mesa/state_tracker')

diff --git a/src/mesa/state_tracker/st_cb_blit.c b/src/mesa/state_tracker/st_cb_blit.c
index 2852623472..98610473b3 100644
--- a/src/mesa/state_tracker/st_cb_blit.c
+++ b/src/mesa/state_tracker/st_cb_blit.c
@@ -75,6 +75,11 @@ st_BlitFramebuffer(GLcontext *ctx,
                          ? PIPE_TEX_MIPFILTER_NEAREST
                          : PIPE_TEX_MIPFILTER_LINEAR);
 
+   if (!_mesa_clip_blit(ctx, &srcX0, &srcY0, &srcX1, &srcY1,
+                        &dstX0, &dstY0, &dstX1, &dstY1)) {
+      return; /* nothing to draw/blit */
+   }
+
    if (mask & GL_COLOR_BUFFER_BIT) {
       struct st_renderbuffer *srcRb = 
          st_renderbuffer(ctx->ReadBuffer->_ColorReadBuffer);
@@ -84,12 +89,29 @@ st_BlitFramebuffer(GLcontext *ctx,
       struct pipe_surface *dstSurf = dstRb->surface;
 
       if (st_fb_orientation(ctx->DrawBuffer) == Y_0_TOP) {
-         /* invert Y */
+         /* invert Y for dest */
+         dstY0 = dstRb->Base.Height - dstY0;
+         dstY1 = dstRb->Base.Height - dstY1;
+      }
+
+      if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) {
+         /* invert Y for src */
          srcY0 = srcRb->Base.Height - srcY0;
          srcY1 = srcRb->Base.Height - srcY1;
+      }
 
-         dstY0 = dstRb->Base.Height - dstY0;
-         dstY1 = dstRb->Base.Height - dstY1;
+      if (srcY0 > srcY1 && dstY0 > dstY1) {
+         /* Both src and dst are upside down.  Swap Y to make it
+          * right-side up to increase odds of using a fast path.
+          * Recall that all Gallium raster coords have Y=0=top.
+          */
+         GLint tmp;
+         tmp = srcY0;
+         srcY0 = srcY1;
+         srcY1 = tmp;
+         tmp = dstY0;
+         dstY0 = dstY1;
+         dstY1 = tmp;
       }
 
       util_blit_pixels(st->blit,
-- 
cgit v1.2.3


From a59579c983df651373435791f8f4fd249293b344 Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Wed, 5 Aug 2009 13:46:47 -0600
Subject: st/mesa: implement BlitFramebuffer() for depth/stencil (incomplete)

We now handle the case of blitting Z+stencil to/from combined Z/stencil
surfaces.  But Z-only or stencil-only and separate depth/stencil surfaces
are not yet implemented.
---
 src/mesa/state_tracker/st_cb_blit.c | 109 ++++++++++++++++++++++++++----------
 1 file changed, 80 insertions(+), 29 deletions(-)

(limited to 'src/mesa/state_tracker')

diff --git a/src/mesa/state_tracker/st_cb_blit.c b/src/mesa/state_tracker/st_cb_blit.c
index 98610473b3..c741940bcf 100644
--- a/src/mesa/state_tracker/st_cb_blit.c
+++ b/src/mesa/state_tracker/st_cb_blit.c
@@ -69,56 +69,107 @@ st_BlitFramebuffer(GLcontext *ctx,
                    GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
                    GLbitfield mask, GLenum filter)
 {
+   const GLbitfield depthStencil = (GL_DEPTH_BUFFER_BIT |
+                                    GL_STENCIL_BUFFER_BIT);
    struct st_context *st = ctx->st;
-
    const uint pFilter = ((filter == GL_NEAREST)
                          ? PIPE_TEX_MIPFILTER_NEAREST
                          : PIPE_TEX_MIPFILTER_LINEAR);
+   struct gl_framebuffer *readFB = ctx->ReadBuffer;
+   struct gl_framebuffer *drawFB = ctx->DrawBuffer;
 
    if (!_mesa_clip_blit(ctx, &srcX0, &srcY0, &srcX1, &srcY1,
                         &dstX0, &dstY0, &dstX1, &dstY1)) {
       return; /* nothing to draw/blit */
    }
 
+   if (st_fb_orientation(drawFB) == Y_0_TOP) {
+      /* invert Y for dest */
+      dstY0 = drawFB->Height - dstY0;
+      dstY1 = drawFB->Height - dstY1;
+   }
+
+   if (st_fb_orientation(readFB) == Y_0_TOP) {
+      /* invert Y for src */
+      srcY0 = readFB->Height - srcY0;
+      srcY1 = readFB->Height - srcY1;
+   }
+
+   if (srcY0 > srcY1 && dstY0 > dstY1) {
+      /* Both src and dst are upside down.  Swap Y to make it
+       * right-side up to increase odds of using a fast path.
+       * Recall that all Gallium raster coords have Y=0=top.
+       */
+      GLint tmp;
+      tmp = srcY0;
+      srcY0 = srcY1;
+      srcY1 = tmp;
+      tmp = dstY0;
+      dstY0 = dstY1;
+      dstY1 = tmp;
+   }
+
    if (mask & GL_COLOR_BUFFER_BIT) {
       struct st_renderbuffer *srcRb = 
-         st_renderbuffer(ctx->ReadBuffer->_ColorReadBuffer);
+         st_renderbuffer(readFB->_ColorReadBuffer);
       struct st_renderbuffer *dstRb = 
-         st_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]);
+         st_renderbuffer(drawFB->_ColorDrawBuffers[0]);
       struct pipe_surface *srcSurf = srcRb->surface;
       struct pipe_surface *dstSurf = dstRb->surface;
 
-      if (st_fb_orientation(ctx->DrawBuffer) == Y_0_TOP) {
-         /* invert Y for dest */
-         dstY0 = dstRb->Base.Height - dstY0;
-         dstY1 = dstRb->Base.Height - dstY1;
-      }
-
-      if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) {
-         /* invert Y for src */
-         srcY0 = srcRb->Base.Height - srcY0;
-         srcY1 = srcRb->Base.Height - srcY1;
-      }
-
-      if (srcY0 > srcY1 && dstY0 > dstY1) {
-         /* Both src and dst are upside down.  Swap Y to make it
-          * right-side up to increase odds of using a fast path.
-          * Recall that all Gallium raster coords have Y=0=top.
-          */
-         GLint tmp;
-         tmp = srcY0;
-         srcY0 = srcY1;
-         srcY1 = tmp;
-         tmp = dstY0;
-         dstY0 = dstY1;
-         dstY1 = tmp;
-      }
-
       util_blit_pixels(st->blit,
                        srcSurf, srcX0, srcY0, srcX1, srcY1,
                        dstSurf, dstX0, dstY0, dstX1, dstY1,
                        0.0, pFilter);
+   }
 
+   if (mask & depthStencil) {
+      /* depth and/or stencil blit */
+
+      /* get src/dst depth surfaces */
+      struct st_renderbuffer *srcDepthRb = 
+         st_renderbuffer(readFB->Attachment[BUFFER_DEPTH].Renderbuffer);
+      struct st_renderbuffer *dstDepthRb = 
+         st_renderbuffer(drawFB->Attachment[BUFFER_DEPTH].Renderbuffer);
+      struct pipe_surface *srcDepthSurf =
+         srcDepthRb ? srcDepthRb->surface : NULL;
+      struct pipe_surface *dstDepthSurf =
+         dstDepthRb ? dstDepthRb->surface : NULL;
+
+      /* get src/dst stencil surfaces */
+      struct st_renderbuffer *srcStencilRb = 
+         st_renderbuffer(readFB->Attachment[BUFFER_STENCIL].Renderbuffer);
+      struct st_renderbuffer *dstStencilRb = 
+         st_renderbuffer(drawFB->Attachment[BUFFER_STENCIL].Renderbuffer);
+      struct pipe_surface *srcStencilSurf =
+         srcStencilRb ? srcStencilRb->surface : NULL;
+      struct pipe_surface *dstStencilSurf =
+         dstStencilRb ? dstStencilRb->surface : NULL;
+
+      if ((mask & depthStencil) == depthStencil &&
+          srcDepthSurf == srcStencilSurf &&
+          dstDepthSurf == dstStencilSurf) {
+         /* Blitting depth and stencil values between combined
+          * depth/stencil buffers.  This is the ideal case for such buffers.
+          */
+         util_blit_pixels(st->blit,
+                          srcDepthSurf, srcX0, srcY0, srcX1, srcY1,
+                          dstDepthSurf, dstX0, dstY0, dstX1, dstY1,
+                          0.0, pFilter);
+      }
+      else {
+         /* blitting depth and stencil separately */
+
+         if (mask & GL_DEPTH_BUFFER_BIT) {
+            /* blit Z only */
+            _mesa_problem(ctx, "st_BlitFramebuffer(DEPTH) not completed");
+         }
+
+         if (mask & GL_STENCIL_BUFFER_BIT) {
+            /* blit stencil only */
+            _mesa_problem(ctx, "st_BlitFramebuffer(STENCIL) not completed");
+         }
+      }
    }
 }
 
-- 
cgit v1.2.3


From 54a7115fc27c640e2b3f1a362e8e07aac220556d Mon Sep 17 00:00:00 2001
From: Chia-I Wu <olvaffe@gmail.com>
Date: Wed, 5 Aug 2009 16:07:19 -0600
Subject: mesa/st: Add support for binding pipe surface to texture.

This commit adds functions to bind a pipe surface to a texture.  This
allows texturing directly from the surface.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
---
 src/mesa/state_tracker/st_cb_texture.c |   6 ++
 src/mesa/state_tracker/st_public.h     |   4 +-
 src/mesa/state_tracker/st_texture.c    | 108 ++++++++++++++++++++++++++++-----
 src/mesa/state_tracker/st_texture.h    |   5 ++
 4 files changed, 107 insertions(+), 16 deletions(-)

(limited to 'src/mesa/state_tracker')

diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c
index e8d7f70ad6..90a059ca69 100644
--- a/src/mesa/state_tracker/st_cb_texture.c
+++ b/src/mesa/state_tracker/st_cb_texture.c
@@ -523,6 +523,12 @@ st_TexImage(GLcontext * ctx,
    DBG("%s target %s level %d %dx%dx%d border %d\n", __FUNCTION__,
        _mesa_lookup_enum_by_nr(target), level, width, height, depth, border);
 
+   /* switch to "normal" */
+   if (stObj->surface_based) {
+      _mesa_clear_texture_object(ctx, texObj);
+      stObj->surface_based = GL_FALSE;
+   }
+
    /* gallium does not support texture borders, strip it off */
    if (border) {
       strip_texture_border(border, &width, &height, &depth, unpack, &unpackNB);
diff --git a/src/mesa/state_tracker/st_public.h b/src/mesa/state_tracker/st_public.h
index 04d3a3d7c2..a5fdac32d1 100644
--- a/src/mesa/state_tracker/st_public.h
+++ b/src/mesa/state_tracker/st_public.h
@@ -107,7 +107,9 @@ void st_swapbuffers(struct st_framebuffer *stfb,
                     struct pipe_surface **front_left,
                     struct pipe_surface **front_right);
 
-int st_set_teximage(struct pipe_texture *pt, int target);
+int st_bind_texture_surface(struct pipe_surface *ps, int target, int level,
+                            enum pipe_format format);
+int st_unbind_texture_surface(struct pipe_surface *ps, int target, int level);
 
 /** Redirect rendering into stfb's surface to a texture image */
 int st_bind_teximage(struct st_framebuffer *stfb, uint surfIndex,
diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c
index 10faa633ea..d58803991a 100644
--- a/src/mesa/state_tracker/st_texture.c
+++ b/src/mesa/state_tracker/st_texture.c
@@ -32,6 +32,7 @@
 #include "st_cb_fbo.h"
 #include "st_inlines.h"
 #include "main/enums.h"
+#include "main/texobj.h"
 #include "main/teximage.h"
 #include "main/texstore.h"
 
@@ -353,25 +354,95 @@ st_texture_image_copy(struct pipe_context *pipe,
    }
 }
 
-/** Bind a pipe surface for use as a texture image */
+
+/**
+ * Bind a pipe surface to a texture object.  After the call,
+ * the texture object is marked dirty and will be (re-)validated.
+ *
+ * If this is the first surface bound, the texture object is said to
+ * switch from normal to surface based.  It will be cleared first in
+ * this case.
+ *
+ * \param ps      pipe surface to be unbound
+ * \param target  texture target
+ * \param level   image level
+ * \param format  internal format of the texture
+ */
 int
-st_set_teximage(struct pipe_texture *pt, int target)
+st_bind_texture_surface(struct pipe_surface *ps, int target, int level,
+                        enum pipe_format format)
 {
    GET_CURRENT_CONTEXT(ctx);
    const GLuint unit = ctx->Texture.CurrentUnit;
    struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
    struct gl_texture_object *texObj;
    struct gl_texture_image *texImage;
+   struct st_texture_object *stObj;
    struct st_texture_image *stImage;
-   int internalFormat;
+   GLenum internalFormat;
 
-   switch (pt->format) {
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
-      internalFormat = GL_RGBA8;
+   switch (target) {
+   case ST_TEXTURE_2D:
+      target = GL_TEXTURE_2D;
+      break;
+   case ST_TEXTURE_RECT:
+      target = GL_TEXTURE_RECTANGLE_ARB;
       break;
    default:
       return 0;
-   };
+   }
+
+   /* map pipe format to base format for now */
+   if (pf_get_component_bits(format, PIPE_FORMAT_COMP_A) > 0)
+      internalFormat = GL_RGBA;
+   else
+      internalFormat = GL_RGB;
+
+   texObj = _mesa_select_tex_object(ctx, texUnit, target);
+   _mesa_lock_texture(ctx, texObj);
+
+   stObj = st_texture_object(texObj);
+   /* switch to surface based */
+   if (!stObj->surface_based) {
+      _mesa_clear_texture_object(ctx, texObj);
+      stObj->surface_based = GL_TRUE;
+   }
+
+   texImage = _mesa_get_tex_image(ctx, texObj, target, level);
+   stImage = st_texture_image(texImage);
+
+   _mesa_init_teximage_fields(ctx, target, texImage,
+                              ps->width, ps->height, 1, 0, internalFormat);
+   texImage->TexFormat = st_ChooseTextureFormat(ctx, internalFormat,
+                                                GL_RGBA, GL_UNSIGNED_BYTE);
+   _mesa_set_fetch_functions(texImage, 2);
+   pipe_texture_reference(&stImage->pt, ps->texture);
+
+   _mesa_dirty_texobj(ctx, texObj, GL_TRUE);
+   _mesa_unlock_texture(ctx, texObj);
+   
+   return 1;
+}
+
+
+/**
+ * Unbind a pipe surface from a texture object.  After the call,
+ * the texture object is marked dirty and will be (re-)validated.
+ *
+ * \param ps      pipe surface to be unbound
+ * \param target  texture target
+ * \param level   image level
+ */
+int
+st_unbind_texture_surface(struct pipe_surface *ps, int target, int level)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   const GLuint unit = ctx->Texture.CurrentUnit;
+   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
+   struct gl_texture_object *texObj;
+   struct gl_texture_image *texImage;
+   struct st_texture_object *stObj;
+   struct st_texture_image *stImage;
 
    switch (target) {
    case ST_TEXTURE_2D:
@@ -385,21 +456,28 @@ st_set_teximage(struct pipe_texture *pt, int target)
    }
 
    texObj = _mesa_select_tex_object(ctx, texUnit, target);
-   texImage = _mesa_get_tex_image(ctx, texObj, target, 0);
+
+   _mesa_lock_texture(ctx, texObj);
+
+   texImage = _mesa_get_tex_image(ctx, texObj, target, level);
+   stObj = st_texture_object(texObj);
    stImage = st_texture_image(texImage);
-   
-   _mesa_init_teximage_fields(ctx, GL_TEXTURE_2D, texImage, pt->width[0],
-                              pt->height[0], 1, 0, internalFormat);
 
-   texImage->TexFormat = st_ChooseTextureFormat(ctx, internalFormat, GL_RGBA,
-                                                GL_UNSIGNED_BYTE);
-   _mesa_set_fetch_functions(texImage, 2);
+   /* Make sure the pipe surface is still bound.  The texture object is still
+    * considered surface based even if this is the last bound surface. */
+   if (stImage->pt == ps->texture) {
+      pipe_texture_reference(&stImage->pt, NULL);
+      _mesa_clear_texture_image(ctx, texImage);
 
-   pipe_texture_reference(&stImage->pt, pt);
+      _mesa_dirty_texobj(ctx, texObj, GL_TRUE);
+   }
 
+   _mesa_unlock_texture(ctx, texObj);
+   
    return 1;
 }
 
+
 /** Redirect rendering into stfb's surface to a texture image */
 int
 st_bind_teximage(struct st_framebuffer *stfb, uint surfIndex,
diff --git a/src/mesa/state_tracker/st_texture.h b/src/mesa/state_tracker/st_texture.h
index b9d447cb56..60868ce067 100644
--- a/src/mesa/state_tracker/st_texture.h
+++ b/src/mesa/state_tracker/st_texture.h
@@ -69,6 +69,11 @@ struct st_texture_object
    struct pipe_texture *pt;
 
    GLboolean teximage_realloc;
+
+   /* True if there is/was a surface bound to this texture object.  It helps
+    * track whether the texture object is surface based or not.
+    */
+   GLboolean surface_based;
 };
 
 
-- 
cgit v1.2.3


From 1359d69f11f2d81ad3a1d44b7f729ee243d7f53d Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Fri, 7 Aug 2009 10:49:38 -0600
Subject: st/mesa: remove redundant calls to _mesa_set_vp_override()

Called from core Mesa now.
---
 src/mesa/state_tracker/st_cb_drawpixels.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

(limited to 'src/mesa/state_tracker')

diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c
index b39403129d..d19a88fa7c 100644
--- a/src/mesa/state_tracker/st_cb_drawpixels.c
+++ b/src/mesa/state_tracker/st_cb_drawpixels.c
@@ -804,8 +804,8 @@ st_DrawPixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height,
       return;
    }
 
-   _mesa_set_vp_override( ctx, TRUE );
-   _mesa_update_state( ctx );
+   /* Mesa state should be up to date by now */
+   assert(ctx->NewState == 0x0);
 
    st_validate_state(st);
 
@@ -833,8 +833,6 @@ st_DrawPixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height,
          pipe_texture_reference(&pt, NULL);
       }
    }
-
-   _mesa_set_vp_override( ctx, FALSE );
 }
 
 
-- 
cgit v1.2.3


From 2708ddfb06a36d8568e2aa130bf1f7d551fcd309 Mon Sep 17 00:00:00 2001
From: Eric Anholt <eric@anholt.net>
Date: Tue, 11 Aug 2009 12:31:01 -0700
Subject: vbo: Avoid extra validation of DrawElements.

This saves mapping the index buffer to get a bounds on the indices that
drivers just drop on the floor in the VBO case (cache win), saves a bonus
walk of the indices in the CheckArrayBounds case, and other miscellaneous
validation.  On intel it's a particularly a large win (50-100% in my app)
because even though we let the indices stay in both CPU and GPU caches, we
still end up waiting for the GPU to be done with the buffer before reading
from it.

Drivers that want the min/max_index fields must now check index_bounds_valid
and use vbo_get_minmax_index before using them.
---
 src/mesa/drivers/dri/i965/brw_draw.c      |  53 +++-------
 src/mesa/drivers/dri/i965/brw_draw.h      |   1 +
 src/mesa/drivers/dri/r300/r300_draw.c     |   7 ++
 src/mesa/state_tracker/st_cb_rasterpos.c  |   2 +-
 src/mesa/state_tracker/st_draw.c          |   5 +
 src/mesa/state_tracker/st_draw.h          |   2 +
 src/mesa/state_tracker/st_draw_feedback.c |   4 +
 src/mesa/tnl/t_context.c                  |   2 +-
 src/mesa/tnl/t_draw.c                     |  18 +++-
 src/mesa/tnl/tnl.h                        |  10 ++
 src/mesa/vbo/vbo.h                        |   6 +-
 src/mesa/vbo/vbo_exec_array.c             | 158 ++++++++++++++++--------------
 src/mesa/vbo/vbo_exec_draw.c              |   1 +
 src/mesa/vbo/vbo_rebase.c                 |   1 +
 src/mesa/vbo/vbo_save_draw.c              |   1 +
 src/mesa/vbo/vbo_split_copy.c             |   1 +
 src/mesa/vbo/vbo_split_inplace.c          |   1 +
 17 files changed, 156 insertions(+), 117 deletions(-)

(limited to 'src/mesa/state_tracker')

diff --git a/src/mesa/drivers/dri/i965/brw_draw.c b/src/mesa/drivers/dri/i965/brw_draw.c
index 5152c3f3a5..8c94c904c1 100644
--- a/src/mesa/drivers/dri/i965/brw_draw.c
+++ b/src/mesa/drivers/dri/i965/brw_draw.c
@@ -422,54 +422,31 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
    return retval;
 }
 
-static GLboolean brw_need_rebase( GLcontext *ctx,
-				  const struct gl_client_array *arrays[],
-				  const struct _mesa_index_buffer *ib,
-				  GLuint min_index )
-{
-   if (min_index == 0) 
-      return GL_FALSE;
-
-   if (ib) {
-      if (!vbo_all_varyings_in_vbos(arrays))
-	 return GL_TRUE;
-      else
-	 return GL_FALSE;
-   }
-   else {
-      /* Hmm.  This isn't quite what I wanted.  BRW can actually
-       * handle the mixed case well enough that we shouldn't need to
-       * rebase.  However, it's probably not very common, nor hugely
-       * expensive to do it this way:
-       */
-      if (!vbo_all_varyings_in_vbos(arrays))
-	 return GL_TRUE;
-      else
-	 return GL_FALSE;
-   }
-}
-				  
-
 void brw_draw_prims( GLcontext *ctx,
 		     const struct gl_client_array *arrays[],
 		     const struct _mesa_prim *prim,
 		     GLuint nr_prims,
 		     const struct _mesa_index_buffer *ib,
+		     GLboolean index_bounds_valid,
 		     GLuint min_index,
 		     GLuint max_index )
 {
    GLboolean retval;
 
-   /* Decide if we want to rebase.  If so we end up recursing once
-    * only into this function.
-    */
-   if (brw_need_rebase( ctx, arrays, ib, min_index )) {
-      vbo_rebase_prims( ctx, arrays, 
-			prim, nr_prims, 
-			ib, min_index, max_index, 
-			brw_draw_prims );
-      
-      return;
+   if (!vbo_all_varyings_in_vbos(arrays)) {
+      if (!index_bounds_valid)
+	 vbo_get_minmax_index(ctx, prim, ib, &min_index, &max_index);
+
+      /* Decide if we want to rebase.  If so we end up recursing once
+       * only into this function.
+       */
+      if (min_index != 0) {
+	 vbo_rebase_prims(ctx, arrays,
+			  prim, nr_prims,
+			  ib, min_index, max_index,
+			  brw_draw_prims );
+	 return;
+      }
    }
 
    /* Make a first attempt at drawing:
diff --git a/src/mesa/drivers/dri/i965/brw_draw.h b/src/mesa/drivers/dri/i965/brw_draw.h
index 9aebbdb1b8..2a14db217f 100644
--- a/src/mesa/drivers/dri/i965/brw_draw.h
+++ b/src/mesa/drivers/dri/i965/brw_draw.h
@@ -39,6 +39,7 @@ void brw_draw_prims( GLcontext *ctx,
 		     const struct _mesa_prim *prims,
 		     GLuint nr_prims,
 		     const struct _mesa_index_buffer *ib,
+		     GLboolean index_bounds_valid,
 		     GLuint min_index,
 		     GLuint max_index );
 
diff --git a/src/mesa/drivers/dri/r300/r300_draw.c b/src/mesa/drivers/dri/r300/r300_draw.c
index fcfd309933..aedc6cfb2a 100644
--- a/src/mesa/drivers/dri/r300/r300_draw.c
+++ b/src/mesa/drivers/dri/r300/r300_draw.c
@@ -462,6 +462,7 @@ static void r300DrawPrims(GLcontext *ctx,
 			 const struct _mesa_prim *prim,
 			 GLuint nr_prims,
 			 const struct _mesa_index_buffer *ib,
+			 GLboolean index_bounds_valid,
 			 GLuint min_index,
 			 GLuint max_index)
 {
@@ -476,6 +477,12 @@ static void r300DrawPrims(GLcontext *ctx,
 	limits.max_indices = 65535;
 	limits.max_vb_size = 1024*1024;
 
+	/* This check should get folded into just the places that
+	 * min/max index are really needed.
+	 */
+	if (!index_bounds_valid)
+	   vbo_get_minmax_index(ctx, prim, ib, &min_index, &max_index);
+
 	if (min_index) {
 		vbo_rebase_prims( ctx, arrays, prim, nr_prims, ib, min_index, max_index, r300DrawPrims );
 		return;
diff --git a/src/mesa/state_tracker/st_cb_rasterpos.c b/src/mesa/state_tracker/st_cb_rasterpos.c
index 3bcccd0df4..d82b2a2035 100644
--- a/src/mesa/state_tracker/st_cb_rasterpos.c
+++ b/src/mesa/state_tracker/st_cb_rasterpos.c
@@ -251,7 +251,7 @@ st_RasterPos(GLcontext *ctx, const GLfloat v[4])
    rs->array[0].Ptr = (GLubyte *) v;
 
    /* draw the point */
-   st_feedback_draw_vbo(ctx, rs->arrays, &rs->prim, 1, NULL, 0, 1);
+   st_feedback_draw_vbo(ctx, rs->arrays, &rs->prim, 1, NULL, GL_TRUE, 0, 1);
 }
 
 
diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c
index 914a507bef..503a5f34a3 100644
--- a/src/mesa/state_tracker/st_draw.c
+++ b/src/mesa/state_tracker/st_draw.c
@@ -533,6 +533,7 @@ st_draw_vbo(GLcontext *ctx,
             const struct _mesa_prim *prims,
             GLuint nr_prims,
             const struct _mesa_index_buffer *ib,
+	    GLboolean index_bounds_valid,
             GLuint min_index,
             GLuint max_index)
 {
@@ -545,6 +546,10 @@ st_draw_vbo(GLcontext *ctx,
    unsigned num_vbuffers, num_velements;
    GLboolean userSpace;
 
+   /* Gallium probably doesn't want this in some cases. */
+   if (!index_bounds_valid)
+      vbo_get_minmax_index(ctx, prims, ib, &min_index, &max_index);
+
    /* sanity check for pointer arithmetic below */
    assert(sizeof(arrays[0]->Ptr[0]) == 1);
 
diff --git a/src/mesa/state_tracker/st_draw.h b/src/mesa/state_tracker/st_draw.h
index dcfe7e1536..3e0face656 100644
--- a/src/mesa/state_tracker/st_draw.h
+++ b/src/mesa/state_tracker/st_draw.h
@@ -47,6 +47,7 @@ st_draw_vbo(GLcontext *ctx,
             const struct _mesa_prim *prims,
             GLuint nr_prims,
             const struct _mesa_index_buffer *ib,
+	    GLboolean index_bounds_valid,
             GLuint min_index,
             GLuint max_index);
 
@@ -56,6 +57,7 @@ st_feedback_draw_vbo(GLcontext *ctx,
                      const struct _mesa_prim *prims,
                      GLuint nr_prims,
                      const struct _mesa_index_buffer *ib,
+		     GLboolean index_bounds_valid,
                      GLuint min_index,
                      GLuint max_index);
 
diff --git a/src/mesa/state_tracker/st_draw_feedback.c b/src/mesa/state_tracker/st_draw_feedback.c
index 2712c131c0..b2d682ef64 100644
--- a/src/mesa/state_tracker/st_draw_feedback.c
+++ b/src/mesa/state_tracker/st_draw_feedback.c
@@ -96,6 +96,7 @@ st_feedback_draw_vbo(GLcontext *ctx,
                      const struct _mesa_prim *prims,
                      GLuint nr_prims,
                      const struct _mesa_index_buffer *ib,
+		     GLboolean index_bounds_valid,
                      GLuint min_index,
                      GLuint max_index)
 {
@@ -114,6 +115,9 @@ st_feedback_draw_vbo(GLcontext *ctx,
 
    st_validate_state(ctx->st);
 
+   if (!index_bounds_valid)
+      vbo_get_minmax_index(ctx, prims, ib, &min_index, &max_index);
+
    /* must get these after state validation! */
    vp = ctx->st->vp;
    vs = &st->vp->state;
diff --git a/src/mesa/tnl/t_context.c b/src/mesa/tnl/t_context.c
index f69b122046..f2771cde09 100644
--- a/src/mesa/tnl/t_context.c
+++ b/src/mesa/tnl/t_context.c
@@ -81,7 +81,7 @@ _tnl_CreateContext( GLcontext *ctx )
    tnl->nr_blocks = 0;
 
    /* plug in the VBO drawing function */
-   vbo_set_draw_func(ctx, _tnl_draw_prims);
+   vbo_set_draw_func(ctx, _tnl_vbo_draw_prims);
 
    _math_init_transformation();
    _math_init_translate();
diff --git a/src/mesa/tnl/t_draw.c b/src/mesa/tnl/t_draw.c
index 2ec65d5323..c64c2c2077 100644
--- a/src/mesa/tnl/t_draw.c
+++ b/src/mesa/tnl/t_draw.c
@@ -360,6 +360,20 @@ static void unmap_vbos( GLcontext *ctx,
 }
 
 
+void _tnl_vbo_draw_prims(GLcontext *ctx,
+			 const struct gl_client_array *arrays[],
+			 const struct _mesa_prim *prim,
+			 GLuint nr_prims,
+			 const struct _mesa_index_buffer *ib,
+			 GLboolean index_bounds_valid,
+			 GLuint min_index,
+			 GLuint max_index)
+{
+   if (!index_bounds_valid)
+      vbo_get_minmax_index(ctx, prim, ib, &min_index, &max_index);
+
+   _tnl_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index);
+}
 
 /* This is the main entrypoint into the slimmed-down software tnl
  * module.  In a regular swtnl driver, this can be plugged straight
@@ -393,7 +407,7 @@ void _tnl_draw_prims( GLcontext *ctx,
        */
       vbo_rebase_prims( ctx, arrays, prim, nr_prims, ib, 
 			min_index, max_index,
-			_tnl_draw_prims );
+			_tnl_vbo_draw_prims );
       return;
    }
    else if (max_index > max) {
@@ -411,7 +425,7 @@ void _tnl_draw_prims( GLcontext *ctx,
        */
       vbo_split_prims( ctx, arrays, prim, nr_prims, ib, 
 		       0, max_index,
-		       _tnl_draw_prims,
+		       _tnl_vbo_draw_prims,
 		       &limits );
    }
    else {
diff --git a/src/mesa/tnl/tnl.h b/src/mesa/tnl/tnl.h
index 4d628aa9a6..9c66d3b019 100644
--- a/src/mesa/tnl/tnl.h
+++ b/src/mesa/tnl/tnl.h
@@ -81,6 +81,16 @@ _tnl_draw_prims( GLcontext *ctx,
 		 GLuint min_index,
 		 GLuint max_index);
 
+void
+_tnl_vbo_draw_prims( GLcontext *ctx,
+		     const struct gl_client_array *arrays[],
+		     const struct _mesa_prim *prim,
+		     GLuint nr_prims,
+		     const struct _mesa_index_buffer *ib,
+		     GLboolean index_bounds_valid,
+		     GLuint min_index,
+		     GLuint max_index);
+
 extern void
 _mesa_load_tracked_matrices(GLcontext *ctx);
 
diff --git a/src/mesa/vbo/vbo.h b/src/mesa/vbo/vbo.h
index 5362226c2f..5986e93576 100644
--- a/src/mesa/vbo/vbo.h
+++ b/src/mesa/vbo/vbo.h
@@ -69,6 +69,7 @@ typedef void (*vbo_draw_func)( GLcontext *ctx,
 			       const struct _mesa_prim *prims,
 			       GLuint nr_prims,
 			       const struct _mesa_index_buffer *ib,
+			       GLboolean index_bounds_valid,
 			       GLuint min_index,
 			       GLuint max_index );
 
@@ -112,7 +113,10 @@ void vbo_rebase_prims( GLcontext *ctx,
 		       GLuint min_index,
 		       GLuint max_index,
 		       vbo_draw_func draw );
-
+void
+vbo_get_minmax_index(GLcontext *ctx, const struct _mesa_prim *prim,
+		     const struct _mesa_index_buffer *ib,
+		     GLuint *min_index, GLuint *max_index);
 
 void vbo_use_buffer_objects(GLcontext *ctx);
 
diff --git a/src/mesa/vbo/vbo_exec_array.c b/src/mesa/vbo/vbo_exec_array.c
index f4b9b2f744..4fb4845f6b 100644
--- a/src/mesa/vbo/vbo_exec_array.c
+++ b/src/mesa/vbo/vbo_exec_array.c
@@ -41,15 +41,29 @@
 /**
  * Compute min and max elements for glDraw[Range]Elements() calls.
  */
-static void
-get_minmax_index(GLuint count, GLuint type, const GLvoid *indices,
-                 GLuint *min_index, GLuint *max_index)
+void
+vbo_get_minmax_index(GLcontext *ctx,
+		     const struct _mesa_prim *prim,
+		     const struct _mesa_index_buffer *ib,
+		     GLuint *min_index, GLuint *max_index)
 {
    GLuint i;
+   GLsizei count = prim->count;
+   const void *indices;
+
+   if (ib->obj->Name) {
+      const GLvoid *map = ctx->Driver.MapBuffer(ctx,
+                                                GL_ELEMENT_ARRAY_BUFFER_ARB,
+                                                GL_READ_ONLY,
+                                                ib->obj);
+      indices = ADD_POINTERS(map, ib->ptr);
+   } else {
+      indices = ib->ptr;
+   }
 
-   switch(type) {
+   switch (ib->type) {
    case GL_UNSIGNED_INT: {
-      const GLuint *ui_indices = (const GLuint *)indices;
+      const GLuint *ui_indices = (const GLuint *)ib->ptr;
       GLuint max_ui = ui_indices[count-1];
       GLuint min_ui = ui_indices[0];
       for (i = 0; i < count; i++) {
@@ -88,6 +102,12 @@ get_minmax_index(GLuint count, GLuint type, const GLvoid *indices,
       assert(0);
       break;
    }
+
+   if (ib->obj->Name != 0) {
+      ctx->Driver.UnmapBuffer(ctx,
+			      GL_ELEMENT_ARRAY_BUFFER_ARB,
+			      ib->obj);
+   }
 }
 
 
@@ -500,7 +520,7 @@ vbo_exec_DrawArrays(GLenum mode, GLint start, GLsizei count)
    prim[0].indexed = 0;
 
    vbo->draw_prims( ctx, exec->array.inputs, prim, 1, NULL,
-                    start, start + count - 1 );
+                    GL_TRUE, start, start + count - 1 );
 
 #if 0
    print_draw_arrays(ctx, exec, mode, start, count);
@@ -566,53 +586,19 @@ dump_element_buffer(GLcontext *ctx, GLenum type)
                            ctx->Array.ElementArrayBufferObj);
 }
 
-
-static void GLAPIENTRY
-vbo_exec_DrawRangeElements(GLenum mode,
-			   GLuint start, GLuint end,
-			   GLsizei count, GLenum type, const GLvoid *indices)
+/* Inner support for both _mesa_DrawElements and _mesa_DrawRangeElements */
+static void
+vbo_validated_drawrangeelements(GLcontext *ctx, GLenum mode,
+				GLboolean index_bounds_valid,
+				GLuint start, GLuint end,
+				GLsizei count, GLenum type,
+				const GLvoid *indices)
 {
-   GET_CURRENT_CONTEXT(ctx);
    struct vbo_context *vbo = vbo_context(ctx);
    struct vbo_exec_context *exec = &vbo->exec;
    struct _mesa_index_buffer ib;
    struct _mesa_prim prim[1];
 
-   if (!_mesa_validate_DrawRangeElements( ctx, mode, start, end, count,
-                                          type, indices ))
-      return;
-
-   if (end >= ctx->Array.ArrayObj->_MaxElement) {
-      /* the max element is out of bounds of one or more enabled arrays */
-      _mesa_warning(ctx, "glDraw[Range]Elements(start %u, end %u, count %d, "
-                    "type 0x%x, indices=%p)\n"
-                    "\tindex=%u is out of bounds (max=%u)  "
-                    "Element Buffer %u (size %d)",
-                    start, end, count, type, indices, end,
-                    ctx->Array.ArrayObj->_MaxElement - 1,
-                    ctx->Array.ElementArrayBufferObj->Name,
-                    ctx->Array.ElementArrayBufferObj->Size);
-
-      if (0)
-         dump_element_buffer(ctx, type);
-
-      if (0)
-         _mesa_print_arrays(ctx);
-      return;
-   }
-   else if (0) {
-      _mesa_printf("glDraw[Range]Elements"
-                   "(start %u, end %u, type 0x%x, count %d) ElemBuf %u\n",
-                   start, end, type, count,
-                   ctx->Array.ElementArrayBufferObj->Name);
-   }
-
-#if 0
-   check_draw_elements_data(ctx, count, type, indices);
-#else
-   (void) check_draw_elements_data;
-#endif
-
    FLUSH_CURRENT( ctx, 0 );
 
    if (ctx->NewState)
@@ -623,13 +609,13 @@ vbo_exec_DrawRangeElements(GLenum mode,
       return;
    }
 
-   bind_arrays( ctx );
-
    if (ctx->NewState)
       _mesa_update_state( ctx );
 
+   bind_arrays( ctx );
+
    ib.count = count;
-   ib.type = type; 
+   ib.type = type;
    ib.obj = ctx->Array.ElementArrayBufferObj;
    ib.ptr = indices;
 
@@ -673,44 +659,68 @@ vbo_exec_DrawRangeElements(GLenum mode,
     * for the latter case elsewhere.
     */
 
-   vbo->draw_prims( ctx, exec->array.inputs, prim, 1, &ib, start, end );
+   vbo->draw_prims( ctx, exec->array.inputs, prim, 1, &ib,
+		    index_bounds_valid, start, end );
 }
 
-
 static void GLAPIENTRY
-vbo_exec_DrawElements(GLenum mode, GLsizei count, GLenum type,
-                      const GLvoid *indices)
+vbo_exec_DrawRangeElements(GLenum mode,
+			   GLuint start, GLuint end,
+			   GLsizei count, GLenum type, const GLvoid *indices)
 {
    GET_CURRENT_CONTEXT(ctx);
-   GLuint min_index = 0;
-   GLuint max_index = 0;
 
-   if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices ))
+   if (!_mesa_validate_DrawRangeElements( ctx, mode, start, end, count,
+                                          type, indices ))
       return;
 
-   if (!vbo_validate_shaders(ctx)) {
-      _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawElements(bad shader)");
+   if (end >= ctx->Array.ArrayObj->_MaxElement) {
+      /* the max element is out of bounds of one or more enabled arrays */
+      _mesa_warning(ctx, "glDraw[Range]Elements(start %u, end %u, count %d, "
+                    "type 0x%x, indices=%p)\n"
+                    "\tindex=%u is out of bounds (max=%u)  "
+                    "Element Buffer %u (size %d)",
+                    start, end, count, type, indices, end,
+                    ctx->Array.ArrayObj->_MaxElement - 1,
+                    ctx->Array.ElementArrayBufferObj->Name,
+                    ctx->Array.ElementArrayBufferObj->Size);
+
+      if (0)
+         dump_element_buffer(ctx, type);
+
+      if (0)
+         _mesa_print_arrays(ctx);
       return;
    }
+   else if (0) {
+      _mesa_printf("glDraw[Range]Elements"
+                   "(start %u, end %u, type 0x%x, count %d) ElemBuf %u\n",
+                   start, end, type, count,
+                   ctx->Array.ElementArrayBufferObj->Name);
+   }
 
-   if (ctx->Array.ElementArrayBufferObj->Name) {
-      const GLvoid *map = ctx->Driver.MapBuffer(ctx,
-                                                GL_ELEMENT_ARRAY_BUFFER_ARB,
-                                                GL_READ_ONLY,
-                                                ctx->Array.ElementArrayBufferObj);
+#if 0
+   check_draw_elements_data(ctx, count, type, indices);
+#else
+   (void) check_draw_elements_data;
+#endif
+
+   vbo_validated_drawrangeelements(ctx, mode, GL_TRUE, start, end,
+				   count, type, indices);
+}
 
-      get_minmax_index(count, type, ADD_POINTERS(map, indices),
-                       &min_index, &max_index);
 
-      ctx->Driver.UnmapBuffer(ctx,
-			      GL_ELEMENT_ARRAY_BUFFER_ARB,
-			      ctx->Array.ElementArrayBufferObj);
-   }
-   else {
-      get_minmax_index(count, type, indices, &min_index, &max_index);
-   }
+static void GLAPIENTRY
+vbo_exec_DrawElements(GLenum mode, GLsizei count, GLenum type,
+                      const GLvoid *indices)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices ))
+      return;
 
-   vbo_exec_DrawRangeElements(mode, min_index, max_index, count, type, indices);
+   vbo_validated_drawrangeelements(ctx, mode, GL_FALSE, ~0, ~0,
+				   count, type, indices);
 }
 
 
diff --git a/src/mesa/vbo/vbo_exec_draw.c b/src/mesa/vbo/vbo_exec_draw.c
index 18419928b2..a988424d21 100644
--- a/src/mesa/vbo/vbo_exec_draw.c
+++ b/src/mesa/vbo/vbo_exec_draw.c
@@ -378,6 +378,7 @@ vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap )
 				       exec->vtx.prim, 
 				       exec->vtx.prim_count,
 				       NULL,
+				       GL_TRUE,
 				       0,
 				       exec->vtx.vert_count - 1);
 
diff --git a/src/mesa/vbo/vbo_rebase.c b/src/mesa/vbo/vbo_rebase.c
index ea87dede64..3bf7ef580f 100644
--- a/src/mesa/vbo/vbo_rebase.c
+++ b/src/mesa/vbo/vbo_rebase.c
@@ -208,6 +208,7 @@ void vbo_rebase_prims( GLcontext *ctx,
 	 prim, 
 	 nr_prims, 
 	 ib, 
+	 GL_TRUE,
 	 0, 
 	 max_index - min_index );
    
diff --git a/src/mesa/vbo/vbo_save_draw.c b/src/mesa/vbo/vbo_save_draw.c
index 5110648c28..d834fa1f2e 100644
--- a/src/mesa/vbo/vbo_save_draw.c
+++ b/src/mesa/vbo/vbo_save_draw.c
@@ -279,6 +279,7 @@ void vbo_save_playback_vertex_list( GLcontext *ctx, void *data )
 				    node->prim, 
 				    node->prim_count,
 				    NULL,
+				    GL_TRUE,
 				    0,	/* Node is a VBO, so this is ok */
 				    node->count - 1);
    }
diff --git a/src/mesa/vbo/vbo_split_copy.c b/src/mesa/vbo/vbo_split_copy.c
index d7ffebf607..3f8a222805 100644
--- a/src/mesa/vbo/vbo_split_copy.c
+++ b/src/mesa/vbo/vbo_split_copy.c
@@ -194,6 +194,7 @@ flush( struct copy_context *copy )
 	       copy->dstprim,
 	       copy->dstprim_nr,
 	       &copy->dstib,
+	       GL_TRUE,
 	       0,
 	       copy->dstbuf_nr );
 
diff --git a/src/mesa/vbo/vbo_split_inplace.c b/src/mesa/vbo/vbo_split_inplace.c
index 9628227e7c..da84eaa6ea 100644
--- a/src/mesa/vbo/vbo_split_inplace.c
+++ b/src/mesa/vbo/vbo_split_inplace.c
@@ -85,6 +85,7 @@ static void flush_vertex( struct split_context *split )
 		split->dstprim,
 		split->dstprim_nr,
 		NULL,
+		GL_TRUE,
 		min_index,
 		max_index);
 
-- 
cgit v1.2.3