From b642730be93149baa7556e5791393168ab396175 Mon Sep 17 00:00:00 2001
From: José Fonseca <jrfonseca@tungstengraphics.com>
Date: Fri, 15 Feb 2008 17:35:24 +0900
Subject: Code reorganization: move files into their places.

This is in a separate commit to ensure renames are properly preserved.
---
 src/gallium/drivers/cell/spu/spu_vertex_shader.h | 63 ++++++++++++++++++++++++
 1 file changed, 63 insertions(+)
 create mode 100644 src/gallium/drivers/cell/spu/spu_vertex_shader.h

(limited to 'src/gallium/drivers/cell/spu/spu_vertex_shader.h')

diff --git a/src/gallium/drivers/cell/spu/spu_vertex_shader.h b/src/gallium/drivers/cell/spu/spu_vertex_shader.h
new file mode 100644
index 0000000000..b5bf31e67d
--- /dev/null
+++ b/src/gallium/drivers/cell/spu/spu_vertex_shader.h
@@ -0,0 +1,63 @@
+#ifndef SPU_VERTEX_SHADER_H
+#define SPU_VERTEX_SHADER_H
+
+#include "pipe/p_format.h"
+#include "spu_exec.h"
+
+struct spu_vs_context;
+
+typedef void (*spu_fetch_func)(qword *out, const qword *in, qword defaults,
+			       const qword *shuffle_data);
+typedef void (*spu_full_fetch_func)( struct spu_vs_context *draw,
+				     struct spu_exec_machine *machine,
+				     const unsigned *elts,
+				     unsigned count );
+
+struct spu_vs_context {
+   struct pipe_viewport_state viewport;
+
+   struct {
+      uint64_t src_ptr[PIPE_ATTRIB_MAX];
+      unsigned pitch[PIPE_ATTRIB_MAX];
+      unsigned size[PIPE_ATTRIB_MAX];
+      enum pipe_format format[PIPE_ATTRIB_MAX];
+      unsigned nr_attrs;
+      boolean dirty;
+
+      spu_fetch_func fetch[PIPE_ATTRIB_MAX];
+      spu_full_fetch_func fetch_func;
+   } vertex_fetch;
+   
+   /* Clip derived state:
+    */
+   float plane[12][4];
+   unsigned nr_planes;
+
+   struct spu_exec_machine machine;
+   const float (*constants)[4];
+
+   unsigned num_vs_outputs;
+};
+
+extern void spu_update_vertex_fetch(struct spu_vs_context *draw);
+
+static INLINE void spu_vertex_fetch(struct spu_vs_context *draw,
+				    struct spu_exec_machine *machine,
+				    const unsigned *elts,
+				    unsigned count)
+{
+   if (draw->vertex_fetch.dirty) {
+      spu_update_vertex_fetch(draw);
+      draw->vertex_fetch.dirty = 0;
+   }
+   
+   (*draw->vertex_fetch.fetch_func)(draw, machine, elts, count);
+}
+
+struct cell_command_vs;
+
+extern void
+spu_execute_vertex_shader(struct spu_vs_context *draw,
+			  const struct cell_command_vs *vs);
+
+#endif /* SPU_VERTEX_SHADER_H */
-- 
cgit v1.2.3


From 3320b1874e810583f95b93a89697b2955987b84f Mon Sep 17 00:00:00 2001
From: Ian Romanick <idr@us.ibm.com>
Date: Fri, 15 Feb 2008 11:03:54 -0800
Subject: Cell: Enable code gen for SPE attribute fetch

Doubles are still unsupported.
---
 src/gallium/drivers/cell/common.h                 |  15 +-
 src/gallium/drivers/cell/ppu/Makefile             |   1 +
 src/gallium/drivers/cell/ppu/cell_context.h       |   4 +
 src/gallium/drivers/cell/ppu/cell_vertex_fetch.c  |  15 +-
 src/gallium/drivers/cell/ppu/cell_vertex_shader.c |  21 +-
 src/gallium/drivers/cell/spu/spu_main.c           |  22 +-
 src/gallium/drivers/cell/spu/spu_vertex_fetch.c   | 477 +---------------------
 src/gallium/drivers/cell/spu/spu_vertex_shader.h  |   6 +-
 8 files changed, 71 insertions(+), 490 deletions(-)

(limited to 'src/gallium/drivers/cell/spu/spu_vertex_shader.h')

diff --git a/src/gallium/drivers/cell/common.h b/src/gallium/drivers/cell/common.h
index 4de514c358..74b131fbef 100644
--- a/src/gallium/drivers/cell/common.h
+++ b/src/gallium/drivers/cell/common.h
@@ -90,6 +90,7 @@
 #define CELL_CMD_STATE_VS_ARRAY_INFO 16
 #define CELL_CMD_STATE_BLEND         17
 #define CELL_CMD_VS_EXECUTE          18
+#define CELL_CMD_STATE_ATTRIB_FETCH  19
 
 
 #define CELL_NUM_BUFFERS 4
@@ -128,13 +129,19 @@ struct cell_command_clear_surface
  */
 struct cell_array_info
 {
-    uint64_t base;      /**< Base address of the 0th element. */
-    uint attr;          /**< Attribute that this state is for. */
-    uint pitch;         /**< Byte pitch from one entry to the next. */
-    uint format;        /**< Pipe format of each entry. */
+   uint64_t base;      /**< Base address of the 0th element. */
+   uint attr;          /**< Attribute that this state is for. */
+   uint pitch;         /**< Byte pitch from one entry to the next. */
+   uint size;
+   uint function_offset;
 } ALIGN16_ATTRIB;
 
 
+struct cell_attribute_fetch_code {
+   uint64_t base;
+   uint size;
+};
+
 struct cell_shader_info
 {
    unsigned num_outputs;
diff --git a/src/gallium/drivers/cell/ppu/Makefile b/src/gallium/drivers/cell/ppu/Makefile
index a4c3f29e8a..196ab777f5 100644
--- a/src/gallium/drivers/cell/ppu/Makefile
+++ b/src/gallium/drivers/cell/ppu/Makefile
@@ -34,6 +34,7 @@ SOURCES = \
 	cell_surface.c \
 	cell_texture.c \
 	cell_vbuf.c \
+	cell_vertex_fetch.c \
 	cell_vertex_shader.c \
 	cell_winsys.c
 
diff --git a/src/gallium/drivers/cell/ppu/cell_context.h b/src/gallium/drivers/cell/ppu/cell_context.h
index 6196c0c72f..91f8e542a2 100644
--- a/src/gallium/drivers/cell/ppu/cell_context.h
+++ b/src/gallium/drivers/cell/ppu/cell_context.h
@@ -36,6 +36,7 @@
 #include "draw/draw_vbuf.h"
 #include "cell_winsys.h"
 #include "cell/common.h"
+#include "ppc/rtasm/spe_asm.h"
 
 
 struct cell_vbuf_render;
@@ -111,6 +112,9 @@ struct cell_context
    /** [4] to ensure 16-byte alignment for each status word */
    uint buffer_status[CELL_MAX_SPUS][CELL_NUM_BUFFERS][4] ALIGN16_ATTRIB;
 
+
+   struct spe_function attrib_fetch;
+   unsigned attrib_fetch_offsets[PIPE_ATTRIB_MAX];
 };
 
 
diff --git a/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c b/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c
index f2432f4ff5..f10689a959 100644
--- a/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c
+++ b/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c
@@ -27,10 +27,10 @@
 #include "pipe/p_context.h"
 #include "pipe/p_format.h"
 
-#include "pipe/draw/draw_context.h"
-#include "pipe/draw/draw_private.h"
+#include "../auxiliary/draw/draw_context.h"
+#include "../auxiliary/draw/draw_private.h"
 
-#include "pipe/cell/ppu/cell_context.h"
+#include "cell_context.h"
 #include "ppc/rtasm/spe_asm.h"
 
 typedef uint64_t register_mask;
@@ -380,13 +380,4 @@ void cell_update_vertex_fetch(struct draw_context *draw)
 	     cell->attrib_fetch_offsets[function_index[i]];
       }
    }
-   
-   static first_time = 1;
-   if (first_time) {
-      first_time = 0;
-      const unsigned instructions = p->csr - p->store;
-      for (i = 0; i < instructions; i++) {
-	 printf("\t.long\t0x%08x\n", p->store[i]);
-      }
-   }
 }
diff --git a/src/gallium/drivers/cell/ppu/cell_vertex_shader.c b/src/gallium/drivers/cell/ppu/cell_vertex_shader.c
index 0ba4506505..6a1d3bc20a 100644
--- a/src/gallium/drivers/cell/ppu/cell_vertex_shader.c
+++ b/src/gallium/drivers/cell/ppu/cell_vertex_shader.c
@@ -55,14 +55,32 @@ cell_vertex_shader_queue_flush(struct draw_context *draw)
    uint64_t *batch;
    struct cell_array_info *array_info;
    unsigned i, j;
+   struct cell_attribute_fetch_code *cf;
 
    assert(draw->vs.queue_nr != 0);
 
    /* XXX: do this on statechange: 
     */
    draw_update_vertex_fetch(draw);
+   cell_update_vertex_fetch(draw);
+
+
+   batch = cell_batch_alloc(cell, sizeof(batch[0]) + sizeof(*cf));
+   batch[0] = CELL_CMD_STATE_ATTRIB_FETCH;
+   cf = (struct cell_attribute_fetch_code *) (&batch[1]);
+   cf->base = cell->attrib_fetch.store;
+   cf->size = ROUNDUP16((unsigned)((void *) cell->attrib_fetch.csr 
+				   - (void *) cell->attrib_fetch.store));
+
 
    for (i = 0; i < draw->vertex_fetch.nr_attrs; i++) {
+      const enum pipe_format format = draw->vertex_element[i].src_format;
+      const unsigned count = ((pf_size_x(format) != 0)
+			      + (pf_size_y(format) != 0)
+			      + (pf_size_z(format) != 0)
+			      + (pf_size_w(format) != 0));
+      const unsigned size = pf_size_x(format) * count;
+
       batch = cell_batch_alloc(cell, sizeof(batch[0]) + sizeof(*array_info));
 
       batch[0] = CELL_CMD_STATE_VS_ARRAY_INFO;
@@ -72,7 +90,8 @@ cell_vertex_shader_queue_flush(struct draw_context *draw)
       array_info->base = (uintptr_t) draw->vertex_fetch.src_ptr[i];
       array_info->attr = i;
       array_info->pitch = draw->vertex_fetch.pitch[i];
-      array_info->format = draw->vertex_element[i].src_format;
+      array_info->size = size;
+      array_info->function_offset = cell->attrib_fetch_offsets[i];
    }
 
    batch = cell_batch_alloc(cell, sizeof(batch[0])
diff --git a/src/gallium/drivers/cell/spu/spu_main.c b/src/gallium/drivers/cell/spu/spu_main.c
index 1e7243b863..fcbf0f841e 100644
--- a/src/gallium/drivers/cell/spu/spu_main.c
+++ b/src/gallium/drivers/cell/spu/spu_main.c
@@ -54,6 +54,9 @@ struct spu_global spu;
 
 struct spu_vs_context draw;
 
+static unsigned char attribute_fetch_code_buffer[136 * PIPE_ATTRIB_MAX]
+    ALIGN16_ATTRIB;
+
 /**
  * Tell the PPU that this SPU has finished copying a buffer to
  * local store and that it may be reused by the PPU.
@@ -306,7 +309,8 @@ cmd_state_vs_array_info(const struct cell_array_info *vs_info)
    ASSERT(attr < PIPE_ATTRIB_MAX);
    draw.vertex_fetch.src_ptr[attr] = vs_info->base;
    draw.vertex_fetch.pitch[attr] = vs_info->pitch;
-   draw.vertex_fetch.format[attr] = vs_info->format;
+   draw.vertex_fetch.size[attr] = vs_info->size;
+   draw.vertex_fetch.code_offset[attr] = vs_info->function_offset;
    draw.vertex_fetch.dirty = 1;
 }
 
@@ -433,6 +437,22 @@ cmd_batch(uint opcode)
          cmd_state_vs_array_info((struct cell_array_info *) &buffer[pos+1]);
          pos += (1 + ROUNDUP8(sizeof(struct cell_array_info)) / 8);
          break;
+      case CELL_CMD_STATE_ATTRIB_FETCH: {
+         struct cell_attribute_fetch_code *code =
+             (struct cell_attribute_fetch_code *) &buffer[pos+1];
+
+              mfc_get(attribute_fetch_code_buffer,
+                      (unsigned int) code->base,  /* src */
+                      code->size,
+                      TAG_BATCH_BUFFER,
+                      0, /* tid */
+                      0  /* rid */);
+         wait_on_mask(1 << TAG_BATCH_BUFFER);
+
+         draw.vertex_fetch.code = attribute_fetch_code_buffer;
+         pos += (1 + ROUNDUP8(sizeof(struct cell_attribute_fetch_code)) / 8);
+         break;
+      }
       default:
          printf("SPU %u: bad opcode: 0x%llx\n", spu.init.id, buffer[pos]);
          ASSERT(0);
diff --git a/src/gallium/drivers/cell/spu/spu_vertex_fetch.c b/src/gallium/drivers/cell/spu/spu_vertex_fetch.c
index 45e3c26c00..55c6c28717 100644
--- a/src/gallium/drivers/cell/spu/spu_vertex_fetch.c
+++ b/src/gallium/drivers/cell/spu/spu_vertex_fetch.c
@@ -1,6 +1,7 @@
 /**************************************************************************
  * 
  * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * (C) Copyright IBM Corporation 2008
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -28,10 +29,10 @@
  /*
   * Authors:
   *   Keith Whitwell <keith@tungstengraphics.com>
+  *   Ian Romanick <idr@us.ibm.com>
   */
 
 #include <spu_mfcio.h>
-#include <transpose_matrix4x4.h>
 
 #include "pipe/p_util.h"
 #include "pipe/p_state.h"
@@ -59,6 +60,10 @@
 
 #define DRAW_DBG 0
 
+typedef void (*spu_fetch_func)(qword *out, const qword *in,
+			       const qword *shuffle_data);
+
+
 static const qword fetch_shuffle_data[] = {
    /* Shuffle used by CVT_64_FLOAT
     */
@@ -97,22 +102,6 @@ static const qword fetch_shuffle_data[] = {
 };
 
 
-static INLINE void
-trans4x4(qword row0, qword row1, qword row2, qword row3, qword *out,
-         const qword *shuffle)
-{
-   qword t1 = si_shufb(row0, row2, shuffle[3]);
-   qword t2 = si_shufb(row0, row2, shuffle[4]);
-   qword t3 = si_shufb(row1, row3, shuffle[3]);
-   qword t4 = si_shufb(row1, row3, shuffle[4]);
-
-   out[0] = si_shufb(t1, t3, shuffle[3]);
-   out[1] = si_shufb(t1, t3, shuffle[4]);
-   out[2] = si_shufb(t2, t4, shuffle[3]);
-   out[3] = si_shufb(t2, t4, shuffle[4]);
-}
-
-
 /**
  * Fetch between 1 and 32 bytes from an unaligned address
  */
@@ -151,446 +140,6 @@ fetch_unaligned(qword *dst, unsigned ea, unsigned size)
 }
 
 
-#define CVT_32_FLOAT(q, s)    (*(q))
-
-static INLINE qword
-CVT_64_FLOAT(const qword *qw, const qword *shuffle)
-{
-   qword a = si_frds(qw[0]);
-   qword b = si_frds(si_rotqbyi(qw[0], 8));
-   qword c = si_frds(qw[1]);
-   qword d = si_frds(si_rotqbyi(qw[1], 8));
-
-   qword ab = si_shufb(a, b, shuffle[0]);
-   qword cd = si_shufb(c, d, si_rotqbyi(shuffle[0], 8));
-   
-   return si_or(ab, cd);
-}
-
-
-static INLINE qword
-CVT_8_USCALED(const qword *qw, const qword *shuffle)
-{
-   return si_cuflt(si_shufb(*qw, *qw, shuffle[1]), 0);
-}
-
-
-static INLINE qword
-CVT_16_USCALED(const qword *qw, const qword *shuffle)
-{
-   return si_cuflt(si_shufb(*qw, *qw, shuffle[2]), 0);
-}
-
-
-static INLINE qword
-CVT_32_USCALED(const qword *qw, const qword *shuffle)
-{
-   (void) shuffle;
-   return si_cuflt(*qw, 0);
-}
-
-static INLINE qword
-CVT_8_SSCALED(const qword *qw, const qword *shuffle)
-{
-   return si_csflt(si_shufb(*qw, *qw, shuffle[1]), 0);
-}
-
-
-static INLINE qword
-CVT_16_SSCALED(const qword *qw, const qword *shuffle)
-{
-   return si_csflt(si_shufb(*qw, *qw, shuffle[2]), 0);
-}
-
-
-static INLINE qword
-CVT_32_SSCALED(const qword *qw, const qword *shuffle)
-{
-   (void) shuffle;
-   return si_csflt(*qw, 0);
-}
-
-
-static INLINE qword
-CVT_8_UNORM(const qword *qw, const qword *shuffle)
-{
-   const qword scale = (qword) spu_splats(1.0f / 255.0f);
-   return si_fm(CVT_8_USCALED(qw, shuffle), scale);
-}
-
-
-static INLINE qword
-CVT_16_UNORM(const qword *qw, const qword *shuffle)
-{
-   const qword scale = (qword) spu_splats(1.0f / 65535.0f);
-   return si_fm(CVT_16_USCALED(qw, shuffle), scale);
-}
-
-
-static INLINE qword
-CVT_32_UNORM(const qword *qw, const qword *shuffle)
-{
-   const qword scale = (qword) spu_splats(1.0f / 4294967295.0f);
-   return si_fm(CVT_32_USCALED(qw, shuffle), scale);
-}
-
-
-static INLINE qword
-CVT_8_SNORM(const qword *qw, const qword *shuffle)
-{
-   const qword scale = (qword) spu_splats(1.0f / 127.0f);
-   return si_fm(CVT_8_SSCALED(qw, shuffle), scale);
-}
-
-
-static INLINE qword
-CVT_16_SNORM(const qword *qw, const qword *shuffle)
-{
-   const qword scale = (qword) spu_splats(1.0f / 32767.0f);
-   return si_fm(CVT_16_SSCALED(qw, shuffle), scale);
-}
-
-
-static INLINE qword
-CVT_32_SNORM(const qword *qw, const qword *shuffle)
-{
-   const qword scale = (qword) spu_splats(1.0f / 2147483647.0f);
-   return si_fm(CVT_32_SSCALED(qw, shuffle), scale);
-}
-
-#define SZ_4 si_il(0U)
-#define SZ_3 si_fsmbi(0x000f)
-#define SZ_2 si_fsmbi(0x00ff)
-#define SZ_1 si_fsmbi(0x0fff)
-
-/**
- * Fetch a float[4] vertex attribute from memory, doing format/type
- * conversion as needed.
- *
- * This is probably needed/dupliocated elsewhere, eg format
- * conversion, texture sampling etc.
- */
-#define FETCH_ATTRIB( NAME, SZ, CVT, N )			\
-static void							\
-fetch_##NAME(qword *out, const qword *in, qword defaults, \
-                const qword *shuffle)	\
-{								\
-   qword tmp[4];						\
-								\
-   tmp[0] = si_selb(CVT(in + (0 * N), shuffle), defaults, SZ);		\
-   tmp[1] = si_selb(CVT(in + (1 * N), shuffle), defaults, SZ);		\
-   tmp[2] = si_selb(CVT(in + (2 * N), shuffle), defaults, SZ);		\
-   tmp[3] = si_selb(CVT(in + (3 * N), shuffle), defaults, SZ);		\
-   trans4x4(tmp[0], tmp[1], tmp[2], tmp[3], out, shuffle);		\
-}
-
-
-FETCH_ATTRIB( R64G64B64A64_FLOAT,   SZ_4, CVT_64_FLOAT, 2 )
-FETCH_ATTRIB( R64G64B64_FLOAT,      SZ_3, CVT_64_FLOAT, 2 )
-FETCH_ATTRIB( R64G64_FLOAT,         SZ_2, CVT_64_FLOAT, 2 )
-FETCH_ATTRIB( R64_FLOAT,            SZ_1, CVT_64_FLOAT, 2 )
-
-FETCH_ATTRIB( R32G32B32A32_FLOAT,   SZ_4, CVT_32_FLOAT, 1 )
-FETCH_ATTRIB( R32G32B32_FLOAT,      SZ_3, CVT_32_FLOAT, 1 )
-FETCH_ATTRIB( R32G32_FLOAT,         SZ_2, CVT_32_FLOAT, 1 )
-FETCH_ATTRIB( R32_FLOAT,            SZ_1, CVT_32_FLOAT, 1 )
-
-FETCH_ATTRIB( R32G32B32A32_USCALED, SZ_4, CVT_32_USCALED, 1 )
-FETCH_ATTRIB( R32G32B32_USCALED,    SZ_3, CVT_32_USCALED, 1 )
-FETCH_ATTRIB( R32G32_USCALED,       SZ_2, CVT_32_USCALED, 1 )
-FETCH_ATTRIB( R32_USCALED,          SZ_1, CVT_32_USCALED, 1 )
-
-FETCH_ATTRIB( R32G32B32A32_SSCALED, SZ_4, CVT_32_SSCALED, 1 )
-FETCH_ATTRIB( R32G32B32_SSCALED,    SZ_3, CVT_32_SSCALED, 1 )
-FETCH_ATTRIB( R32G32_SSCALED,       SZ_2, CVT_32_SSCALED, 1 )
-FETCH_ATTRIB( R32_SSCALED,          SZ_1, CVT_32_SSCALED, 1 )
-
-FETCH_ATTRIB( R32G32B32A32_UNORM, SZ_4, CVT_32_UNORM, 1 )
-FETCH_ATTRIB( R32G32B32_UNORM,    SZ_3, CVT_32_UNORM, 1 )
-FETCH_ATTRIB( R32G32_UNORM,       SZ_2, CVT_32_UNORM, 1 )
-FETCH_ATTRIB( R32_UNORM,          SZ_1, CVT_32_UNORM, 1 )
-
-FETCH_ATTRIB( R32G32B32A32_SNORM, SZ_4, CVT_32_SNORM, 1 )
-FETCH_ATTRIB( R32G32B32_SNORM,    SZ_3, CVT_32_SNORM, 1 )
-FETCH_ATTRIB( R32G32_SNORM,       SZ_2, CVT_32_SNORM, 1 )
-FETCH_ATTRIB( R32_SNORM,          SZ_1, CVT_32_SNORM, 1 )
-
-FETCH_ATTRIB( R16G16B16A16_USCALED, SZ_4, CVT_16_USCALED, 1 )
-FETCH_ATTRIB( R16G16B16_USCALED,    SZ_3, CVT_16_USCALED, 1 )
-FETCH_ATTRIB( R16G16_USCALED,       SZ_2, CVT_16_USCALED, 1 )
-FETCH_ATTRIB( R16_USCALED,          SZ_1, CVT_16_USCALED, 1 )
-
-FETCH_ATTRIB( R16G16B16A16_SSCALED, SZ_4, CVT_16_SSCALED, 1 )
-FETCH_ATTRIB( R16G16B16_SSCALED,    SZ_3, CVT_16_SSCALED, 1 )
-FETCH_ATTRIB( R16G16_SSCALED,       SZ_2, CVT_16_SSCALED, 1 )
-FETCH_ATTRIB( R16_SSCALED,          SZ_1, CVT_16_SSCALED, 1 )
-
-FETCH_ATTRIB( R16G16B16A16_UNORM, SZ_4, CVT_16_UNORM, 1 )
-FETCH_ATTRIB( R16G16B16_UNORM,    SZ_3, CVT_16_UNORM, 1 )
-FETCH_ATTRIB( R16G16_UNORM,       SZ_2, CVT_16_UNORM, 1 )
-FETCH_ATTRIB( R16_UNORM,          SZ_1, CVT_16_UNORM, 1 )
-
-FETCH_ATTRIB( R16G16B16A16_SNORM, SZ_4, CVT_16_SNORM, 1 )
-FETCH_ATTRIB( R16G16B16_SNORM,    SZ_3, CVT_16_SNORM, 1 )
-FETCH_ATTRIB( R16G16_SNORM,       SZ_2, CVT_16_SNORM, 1 )
-FETCH_ATTRIB( R16_SNORM,          SZ_1, CVT_16_SNORM, 1 )
-
-FETCH_ATTRIB( R8G8B8A8_USCALED,   SZ_4, CVT_8_USCALED, 1 )
-FETCH_ATTRIB( R8G8B8_USCALED,     SZ_3, CVT_8_USCALED, 1 )
-FETCH_ATTRIB( R8G8_USCALED,       SZ_2, CVT_8_USCALED, 1 )
-FETCH_ATTRIB( R8_USCALED,         SZ_1, CVT_8_USCALED, 1 )
-
-FETCH_ATTRIB( R8G8B8A8_SSCALED,  SZ_4, CVT_8_SSCALED, 1 )
-FETCH_ATTRIB( R8G8B8_SSCALED,    SZ_3, CVT_8_SSCALED, 1 )
-FETCH_ATTRIB( R8G8_SSCALED,      SZ_2, CVT_8_SSCALED, 1 )
-FETCH_ATTRIB( R8_SSCALED,        SZ_1, CVT_8_SSCALED, 1 )
-
-FETCH_ATTRIB( R8G8B8A8_UNORM,  SZ_4, CVT_8_UNORM, 1 )
-FETCH_ATTRIB( R8G8B8_UNORM,    SZ_3, CVT_8_UNORM, 1 )
-FETCH_ATTRIB( R8G8_UNORM,      SZ_2, CVT_8_UNORM, 1 )
-FETCH_ATTRIB( R8_UNORM,        SZ_1, CVT_8_UNORM, 1 )
-
-FETCH_ATTRIB( R8G8B8A8_SNORM,  SZ_4, CVT_8_SNORM, 1 )
-FETCH_ATTRIB( R8G8B8_SNORM,    SZ_3, CVT_8_SNORM, 1 )
-FETCH_ATTRIB( R8G8_SNORM,      SZ_2, CVT_8_SNORM, 1 )
-FETCH_ATTRIB( R8_SNORM,        SZ_1, CVT_8_SNORM, 1 )
-
-FETCH_ATTRIB( A8R8G8B8_UNORM,       SZ_4, CVT_8_UNORM, 1 )
-
-
-
-static spu_fetch_func get_fetch_func( enum pipe_format format )
-{
-   switch (format) {
-   case PIPE_FORMAT_R64_FLOAT:
-      return fetch_R64_FLOAT;
-   case PIPE_FORMAT_R64G64_FLOAT:
-      return fetch_R64G64_FLOAT;
-   case PIPE_FORMAT_R64G64B64_FLOAT:
-      return fetch_R64G64B64_FLOAT;
-   case PIPE_FORMAT_R64G64B64A64_FLOAT:
-      return fetch_R64G64B64A64_FLOAT;
-
-   case PIPE_FORMAT_R32_FLOAT:
-      return fetch_R32_FLOAT;
-   case PIPE_FORMAT_R32G32_FLOAT:
-      return fetch_R32G32_FLOAT;
-   case PIPE_FORMAT_R32G32B32_FLOAT:
-      return fetch_R32G32B32_FLOAT;
-   case PIPE_FORMAT_R32G32B32A32_FLOAT:
-      return fetch_R32G32B32A32_FLOAT;
-
-   case PIPE_FORMAT_R32_UNORM:
-      return fetch_R32_UNORM;
-   case PIPE_FORMAT_R32G32_UNORM:
-      return fetch_R32G32_UNORM;
-   case PIPE_FORMAT_R32G32B32_UNORM:
-      return fetch_R32G32B32_UNORM;
-   case PIPE_FORMAT_R32G32B32A32_UNORM:
-      return fetch_R32G32B32A32_UNORM;
-
-   case PIPE_FORMAT_R32_USCALED:
-      return fetch_R32_USCALED;
-   case PIPE_FORMAT_R32G32_USCALED:
-      return fetch_R32G32_USCALED;
-   case PIPE_FORMAT_R32G32B32_USCALED:
-      return fetch_R32G32B32_USCALED;
-   case PIPE_FORMAT_R32G32B32A32_USCALED:
-      return fetch_R32G32B32A32_USCALED;
-
-   case PIPE_FORMAT_R32_SNORM:
-      return fetch_R32_SNORM;
-   case PIPE_FORMAT_R32G32_SNORM:
-      return fetch_R32G32_SNORM;
-   case PIPE_FORMAT_R32G32B32_SNORM:
-      return fetch_R32G32B32_SNORM;
-   case PIPE_FORMAT_R32G32B32A32_SNORM:
-      return fetch_R32G32B32A32_SNORM;
-
-   case PIPE_FORMAT_R32_SSCALED:
-      return fetch_R32_SSCALED;
-   case PIPE_FORMAT_R32G32_SSCALED:
-      return fetch_R32G32_SSCALED;
-   case PIPE_FORMAT_R32G32B32_SSCALED:
-      return fetch_R32G32B32_SSCALED;
-   case PIPE_FORMAT_R32G32B32A32_SSCALED:
-      return fetch_R32G32B32A32_SSCALED;
-
-   case PIPE_FORMAT_R16_UNORM:
-      return fetch_R16_UNORM;
-   case PIPE_FORMAT_R16G16_UNORM:
-      return fetch_R16G16_UNORM;
-   case PIPE_FORMAT_R16G16B16_UNORM:
-      return fetch_R16G16B16_UNORM;
-   case PIPE_FORMAT_R16G16B16A16_UNORM:
-      return fetch_R16G16B16A16_UNORM;
-
-   case PIPE_FORMAT_R16_USCALED:
-      return fetch_R16_USCALED;
-   case PIPE_FORMAT_R16G16_USCALED:
-      return fetch_R16G16_USCALED;
-   case PIPE_FORMAT_R16G16B16_USCALED:
-      return fetch_R16G16B16_USCALED;
-   case PIPE_FORMAT_R16G16B16A16_USCALED:
-      return fetch_R16G16B16A16_USCALED;
-
-   case PIPE_FORMAT_R16_SNORM:
-      return fetch_R16_SNORM;
-   case PIPE_FORMAT_R16G16_SNORM:
-      return fetch_R16G16_SNORM;
-   case PIPE_FORMAT_R16G16B16_SNORM:
-      return fetch_R16G16B16_SNORM;
-   case PIPE_FORMAT_R16G16B16A16_SNORM:
-      return fetch_R16G16B16A16_SNORM;
-
-   case PIPE_FORMAT_R16_SSCALED:
-      return fetch_R16_SSCALED;
-   case PIPE_FORMAT_R16G16_SSCALED:
-      return fetch_R16G16_SSCALED;
-   case PIPE_FORMAT_R16G16B16_SSCALED:
-      return fetch_R16G16B16_SSCALED;
-   case PIPE_FORMAT_R16G16B16A16_SSCALED:
-      return fetch_R16G16B16A16_SSCALED;
-
-   case PIPE_FORMAT_R8_UNORM:
-      return fetch_R8_UNORM;
-   case PIPE_FORMAT_R8G8_UNORM:
-      return fetch_R8G8_UNORM;
-   case PIPE_FORMAT_R8G8B8_UNORM:
-      return fetch_R8G8B8_UNORM;
-   case PIPE_FORMAT_R8G8B8A8_UNORM:
-      return fetch_R8G8B8A8_UNORM;
-
-   case PIPE_FORMAT_R8_USCALED:
-      return fetch_R8_USCALED;
-   case PIPE_FORMAT_R8G8_USCALED:
-      return fetch_R8G8_USCALED;
-   case PIPE_FORMAT_R8G8B8_USCALED:
-      return fetch_R8G8B8_USCALED;
-   case PIPE_FORMAT_R8G8B8A8_USCALED:
-      return fetch_R8G8B8A8_USCALED;
-
-   case PIPE_FORMAT_R8_SNORM:
-      return fetch_R8_SNORM;
-   case PIPE_FORMAT_R8G8_SNORM:
-      return fetch_R8G8_SNORM;
-   case PIPE_FORMAT_R8G8B8_SNORM:
-      return fetch_R8G8B8_SNORM;
-   case PIPE_FORMAT_R8G8B8A8_SNORM:
-      return fetch_R8G8B8A8_SNORM;
-
-   case PIPE_FORMAT_R8_SSCALED:
-      return fetch_R8_SSCALED;
-   case PIPE_FORMAT_R8G8_SSCALED:
-      return fetch_R8G8_SSCALED;
-   case PIPE_FORMAT_R8G8B8_SSCALED:
-      return fetch_R8G8B8_SSCALED;
-   case PIPE_FORMAT_R8G8B8A8_SSCALED:
-      return fetch_R8G8B8A8_SSCALED;
-
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
-      return fetch_A8R8G8B8_UNORM;
-
-   case 0:
-      return NULL;		/* not sure why this is needed */
-
-   default:
-      assert(0);
-      return NULL;
-   }
-}
-
-
-static unsigned get_vertex_size( enum pipe_format format )
-{
-   switch (format) {
-   case PIPE_FORMAT_R64_FLOAT:
-      return 8;
-   case PIPE_FORMAT_R64G64_FLOAT:
-      return 2 * 8;
-   case PIPE_FORMAT_R64G64B64_FLOAT:
-      return 3 * 8;
-   case PIPE_FORMAT_R64G64B64A64_FLOAT:
-      return 4 * 8;
-
-   case PIPE_FORMAT_R32_SSCALED:
-   case PIPE_FORMAT_R32_SNORM:
-   case PIPE_FORMAT_R32_USCALED:
-   case PIPE_FORMAT_R32_UNORM:
-   case PIPE_FORMAT_R32_FLOAT:
-      return 4;
-   case PIPE_FORMAT_R32G32_SSCALED:
-   case PIPE_FORMAT_R32G32_SNORM:
-   case PIPE_FORMAT_R32G32_USCALED:
-   case PIPE_FORMAT_R32G32_UNORM:
-   case PIPE_FORMAT_R32G32_FLOAT:
-      return 2 * 4;
-   case PIPE_FORMAT_R32G32B32_SSCALED:
-   case PIPE_FORMAT_R32G32B32_SNORM:
-   case PIPE_FORMAT_R32G32B32_USCALED:
-   case PIPE_FORMAT_R32G32B32_UNORM:
-   case PIPE_FORMAT_R32G32B32_FLOAT:
-      return 3 * 4;
-   case PIPE_FORMAT_R32G32B32A32_SSCALED:
-   case PIPE_FORMAT_R32G32B32A32_SNORM:
-   case PIPE_FORMAT_R32G32B32A32_USCALED:
-   case PIPE_FORMAT_R32G32B32A32_UNORM:
-   case PIPE_FORMAT_R32G32B32A32_FLOAT:
-      return 4 * 4;
-
-   case PIPE_FORMAT_R16_SSCALED:
-   case PIPE_FORMAT_R16_SNORM:
-   case PIPE_FORMAT_R16_UNORM:
-   case PIPE_FORMAT_R16_USCALED:
-      return 2;
-   case PIPE_FORMAT_R16G16_SSCALED:
-   case PIPE_FORMAT_R16G16_SNORM:
-   case PIPE_FORMAT_R16G16_USCALED:
-   case PIPE_FORMAT_R16G16_UNORM:
-      return 2 * 2;
-   case PIPE_FORMAT_R16G16B16_SSCALED:
-   case PIPE_FORMAT_R16G16B16_SNORM:
-   case PIPE_FORMAT_R16G16B16_USCALED:
-   case PIPE_FORMAT_R16G16B16_UNORM:
-      return 3 * 2;
-   case PIPE_FORMAT_R16G16B16A16_SSCALED:
-   case PIPE_FORMAT_R16G16B16A16_SNORM:
-   case PIPE_FORMAT_R16G16B16A16_USCALED:
-   case PIPE_FORMAT_R16G16B16A16_UNORM:
-      return 4 * 2;
-
-   case PIPE_FORMAT_R8_SSCALED:
-   case PIPE_FORMAT_R8_SNORM:
-   case PIPE_FORMAT_R8_USCALED:
-   case PIPE_FORMAT_R8_UNORM:
-      return 1;
-   case PIPE_FORMAT_R8G8_SSCALED:
-   case PIPE_FORMAT_R8G8_SNORM:
-   case PIPE_FORMAT_R8G8_USCALED:
-   case PIPE_FORMAT_R8G8_UNORM:
-      return 2 * 1;
-   case PIPE_FORMAT_R8G8B8_SSCALED:
-   case PIPE_FORMAT_R8G8B8_SNORM:
-   case PIPE_FORMAT_R8G8B8_USCALED:
-   case PIPE_FORMAT_R8G8B8_UNORM:
-      return 3 * 1;
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
-   case PIPE_FORMAT_R8G8B8A8_SSCALED:
-   case PIPE_FORMAT_R8G8B8A8_SNORM:
-   case PIPE_FORMAT_R8G8B8A8_USCALED:
-   case PIPE_FORMAT_R8G8B8A8_UNORM:
-      return 4 * 1;
-
-   case 0:
-      return 0;		/* not sure why this is needed */
-
-   default:
-      assert(0);
-      return 0;
-   }
-}
-
-
 /**
  * Fetch vertex attributes for 'count' vertices.
  */
@@ -612,10 +161,10 @@ static void generic_vertex_fetch(struct spu_vs_context *draw,
    /* loop over vertex attributes (vertex shader inputs)
     */
    for (attr = 0; attr < nr_attrs; attr++) {
-      const qword default_values = (qword)(vec_float4){ 0.0, 0.0, 0.0, 1.0 };
       const unsigned pitch = draw->vertex_fetch.pitch[attr];
       const uint64_t src = draw->vertex_fetch.src_ptr[attr];
-      const spu_fetch_func fetch = draw->vertex_fetch.fetch[attr];
+      const spu_fetch_func fetch = (spu_fetch_func)
+	  (draw->vertex_fetch.code + draw->vertex_fetch.code_offset[attr]);
       unsigned i;
       unsigned idx;
       const unsigned bytes_per_entry = draw->vertex_fetch.size[attr];
@@ -644,8 +193,7 @@ static void generic_vertex_fetch(struct spu_vs_context *draw,
 
       /* Convert all 4 vertices to vectors of float.
        */
-      (*fetch)(&machine->Inputs[attr].xyzw[0].q, in, default_values,
-               fetch_shuffle_data);
+      (*fetch)(&machine->Inputs[attr].xyzw[0].q, in, fetch_shuffle_data);
    }
 }
 
@@ -662,12 +210,5 @@ void spu_update_vertex_fetch( struct spu_vs_context *draw )
    }
 
 
-   for (i = 0; i < draw->vertex_fetch.nr_attrs; i++) {
-      draw->vertex_fetch.fetch[i] =
-          get_fetch_func(draw->vertex_fetch.format[i]);
-      draw->vertex_fetch.size[i] =
-          get_vertex_size(draw->vertex_fetch.format[i]);
-   }
-
    draw->vertex_fetch.fetch_func = generic_vertex_fetch;
 }
diff --git a/src/gallium/drivers/cell/spu/spu_vertex_shader.h b/src/gallium/drivers/cell/spu/spu_vertex_shader.h
index b5bf31e67d..0fb0bc28d0 100644
--- a/src/gallium/drivers/cell/spu/spu_vertex_shader.h
+++ b/src/gallium/drivers/cell/spu/spu_vertex_shader.h
@@ -6,8 +6,6 @@
 
 struct spu_vs_context;
 
-typedef void (*spu_fetch_func)(qword *out, const qword *in, qword defaults,
-			       const qword *shuffle_data);
 typedef void (*spu_full_fetch_func)( struct spu_vs_context *draw,
 				     struct spu_exec_machine *machine,
 				     const unsigned *elts,
@@ -20,12 +18,12 @@ struct spu_vs_context {
       uint64_t src_ptr[PIPE_ATTRIB_MAX];
       unsigned pitch[PIPE_ATTRIB_MAX];
       unsigned size[PIPE_ATTRIB_MAX];
-      enum pipe_format format[PIPE_ATTRIB_MAX];
+      unsigned code_offset[PIPE_ATTRIB_MAX];
       unsigned nr_attrs;
       boolean dirty;
 
-      spu_fetch_func fetch[PIPE_ATTRIB_MAX];
       spu_full_fetch_func fetch_func;
+      void *code;
    } vertex_fetch;
    
    /* Clip derived state:
-- 
cgit v1.2.3


From 66be2810c3be07dd1ee45a60cfc632725837f2cd Mon Sep 17 00:00:00 2001
From: Ian Romanick <idr@us.ibm.com>
Date: Mon, 18 Feb 2008 18:55:39 -0800
Subject: Cell: emit vertex shaders and uniforms more intelligently

---
 src/gallium/drivers/cell/common.h                 | 22 ++++----
 src/gallium/drivers/cell/ppu/cell_state_emit.c    | 18 ++++++
 src/gallium/drivers/cell/ppu/cell_vertex_shader.c | 17 +++---
 src/gallium/drivers/cell/spu/spu_main.c           |  9 +++
 src/gallium/drivers/cell/spu/spu_vertex_fetch.c   |  6 +-
 src/gallium/drivers/cell/spu/spu_vertex_shader.c  | 68 ++++++++++-------------
 src/gallium/drivers/cell/spu/spu_vertex_shader.h  |  5 ++
 7 files changed, 85 insertions(+), 60 deletions(-)

(limited to 'src/gallium/drivers/cell/spu/spu_vertex_shader.h')

diff --git a/src/gallium/drivers/cell/common.h b/src/gallium/drivers/cell/common.h
index 74b131fbef..cf892206c6 100644
--- a/src/gallium/drivers/cell/common.h
+++ b/src/gallium/drivers/cell/common.h
@@ -87,10 +87,12 @@
 #define CELL_CMD_STATE_TEXTURE       13
 #define CELL_CMD_STATE_VERTEX_INFO   14
 #define CELL_CMD_STATE_VIEWPORT      15
-#define CELL_CMD_STATE_VS_ARRAY_INFO 16
-#define CELL_CMD_STATE_BLEND         17
-#define CELL_CMD_VS_EXECUTE          18
-#define CELL_CMD_STATE_ATTRIB_FETCH  19
+#define CELL_CMD_STATE_UNIFORMS      16
+#define CELL_CMD_STATE_VS_ARRAY_INFO 17
+#define CELL_CMD_STATE_BIND_VS       18
+#define CELL_CMD_STATE_BLEND         19
+#define CELL_CMD_STATE_ATTRIB_FETCH  20
+#define CELL_CMD_VS_EXECUTE          21
 
 
 #define CELL_NUM_BUFFERS 4
@@ -144,14 +146,13 @@ struct cell_attribute_fetch_code {
 
 struct cell_shader_info
 {
-   unsigned num_outputs;
-
    uint64_t declarations;
-   unsigned num_declarations;
    uint64_t instructions;
-   unsigned num_instructions;
-   uint64_t uniforms;
    uint64_t  immediates;
+
+   unsigned num_outputs;
+   unsigned num_declarations;
+   unsigned num_instructions;
    unsigned num_immediates;
 } ALIGN16_ATTRIB;
 
@@ -160,10 +161,9 @@ struct cell_shader_info
 struct cell_command_vs
 {
    uint64_t opcode;       /**< CELL_CMD_VS_EXECUTE */
-   struct cell_shader_info   shader;
+   uint64_t vOut[SPU_VERTS_PER_BATCH];
    unsigned num_elts;
    unsigned elts[SPU_VERTS_PER_BATCH];
-   uint64_t vOut[SPU_VERTS_PER_BATCH];
    float plane[12][4];
    unsigned nr_planes;
    unsigned nr_attrs;
diff --git a/src/gallium/drivers/cell/ppu/cell_state_emit.c b/src/gallium/drivers/cell/ppu/cell_state_emit.c
index 5d2a786449..49c0d130c5 100644
--- a/src/gallium/drivers/cell/ppu/cell_state_emit.c
+++ b/src/gallium/drivers/cell/ppu/cell_state_emit.c
@@ -31,6 +31,8 @@
 #include "cell_state_emit.h"
 #include "cell_batch.h"
 #include "cell_texture.h"
+#include "draw/draw_context.h"
+#include "draw/draw_private.h"
 
 
 static void
@@ -100,4 +102,20 @@ cell_emit_state(struct cell_context *cell)
       emit_state_cmd(cell, CELL_CMD_STATE_VERTEX_INFO,
                      &cell->vertex_info, sizeof(struct vertex_info));
    }
+   
+   if (cell->dirty & CELL_NEW_VS) {
+      const struct draw_context *const draw = cell->draw;
+      struct cell_shader_info info;
+
+      info.num_outputs = draw->num_vs_outputs;
+      info.declarations = (uintptr_t) draw->machine.Declarations;
+      info.num_declarations = draw->machine.NumDeclarations;
+      info.instructions = (uintptr_t) draw->machine.Instructions;
+      info.num_instructions = draw->machine.NumInstructions;
+      info.immediates = (uintptr_t) draw->machine.Imms;
+      info.num_immediates = draw->machine.ImmLimit / 4;
+
+      emit_state_cmd(cell, CELL_CMD_STATE_BIND_VS,
+		     & info, sizeof(info));
+   }
 }
diff --git a/src/gallium/drivers/cell/ppu/cell_vertex_shader.c b/src/gallium/drivers/cell/ppu/cell_vertex_shader.c
index 6a1d3bc20a..64c7821c19 100644
--- a/src/gallium/drivers/cell/ppu/cell_vertex_shader.c
+++ b/src/gallium/drivers/cell/ppu/cell_vertex_shader.c
@@ -54,6 +54,7 @@ cell_vertex_shader_queue_flush(struct draw_context *draw)
    struct cell_command_vs *const vs = &cell_global.command[0].vs;
    uint64_t *batch;
    struct cell_array_info *array_info;
+   struct cell_shader_info *shader_info;
    unsigned i, j;
    struct cell_attribute_fetch_code *cf;
 
@@ -100,17 +101,17 @@ cell_vertex_shader_queue_flush(struct draw_context *draw)
    (void) memcpy(&batch[1], &draw->viewport,
                  sizeof(struct pipe_viewport_state));
 
+   {
+      uint64_t uniforms = (uintptr_t) draw->user.constants;
+
+      batch = cell_batch_alloc(cell, 2 *sizeof(batch[0]));
+      batch[0] = CELL_CMD_STATE_UNIFORMS;
+      batch[1] = uniforms;
+   }
+
    cell_batch_flush(cell);
 
    vs->opcode = CELL_CMD_VS_EXECUTE;
-   vs->shader.num_outputs = draw->num_vs_outputs;
-   vs->shader.declarations = (uintptr_t) draw->machine.Declarations;
-   vs->shader.num_declarations = draw->machine.NumDeclarations;
-   vs->shader.instructions = (uintptr_t) draw->machine.Instructions;
-   vs->shader.num_instructions = draw->machine.NumInstructions;
-   vs->shader.uniforms = (uintptr_t) draw->user.constants;
-   vs->shader.immediates = (uintptr_t) draw->machine.Imms;
-   vs->shader.num_immediates = draw->machine.ImmLimit / 4;
    vs->nr_attrs = draw->vertex_fetch.nr_attrs;
 
    (void) memcpy(vs->plane, draw->plane, sizeof(draw->plane));
diff --git a/src/gallium/drivers/cell/spu/spu_main.c b/src/gallium/drivers/cell/spu/spu_main.c
index fcbf0f841e..dbc3705c24 100644
--- a/src/gallium/drivers/cell/spu/spu_main.c
+++ b/src/gallium/drivers/cell/spu/spu_main.c
@@ -433,10 +433,19 @@ cmd_batch(uint opcode)
                        sizeof(struct pipe_viewport_state));
          pos += (1 + ROUNDUP8(sizeof(struct pipe_viewport_state)) / 8);
          break;
+      case CELL_CMD_STATE_UNIFORMS:
+         draw.constants = (float (*)[4]) (uintptr_t) buffer[pos + 1];
+         pos += 2;
+         break;
       case CELL_CMD_STATE_VS_ARRAY_INFO:
          cmd_state_vs_array_info((struct cell_array_info *) &buffer[pos+1]);
          pos += (1 + ROUNDUP8(sizeof(struct cell_array_info)) / 8);
          break;
+      case CELL_CMD_STATE_BIND_VS:
+         spu_bind_vertex_shader(&draw,
+                                (struct cell_shader_info *) &buffer[pos+1]);
+         pos += (1 + ROUNDUP8(sizeof(struct cell_shader_info)) / 8);
+         break;
       case CELL_CMD_STATE_ATTRIB_FETCH: {
          struct cell_attribute_fetch_code *code =
              (struct cell_attribute_fetch_code *) &buffer[pos+1];
diff --git a/src/gallium/drivers/cell/spu/spu_vertex_fetch.c b/src/gallium/drivers/cell/spu/spu_vertex_fetch.c
index 55c6c28717..e5d9910ff3 100644
--- a/src/gallium/drivers/cell/spu/spu_vertex_fetch.c
+++ b/src/gallium/drivers/cell/spu/spu_vertex_fetch.c
@@ -64,7 +64,7 @@ typedef void (*spu_fetch_func)(qword *out, const qword *in,
 			       const qword *shuffle_data);
 
 
-static const qword fetch_shuffle_data[] = {
+static const qword fetch_shuffle_data[5] ALIGN16_ATTRIB = {
    /* Shuffle used by CVT_64_FLOAT
     */
    {
@@ -108,7 +108,7 @@ static const qword fetch_shuffle_data[] = {
 static INLINE void
 fetch_unaligned(qword *dst, unsigned ea, unsigned size)
 {
-   qword tmp[4];
+   qword tmp[4] ALIGN16_ATTRIB;
    const int shift = ea & 0x0f;
    const unsigned aligned_start_ea = ea & ~0x0f;
    const unsigned aligned_end_ea = (ea + size) & ~0x0f;
@@ -169,7 +169,7 @@ static void generic_vertex_fetch(struct spu_vs_context *draw,
       unsigned idx;
       const unsigned bytes_per_entry = draw->vertex_fetch.size[attr];
       const unsigned quads_per_entry = (bytes_per_entry + 15) / 16;
-      qword in[2 * 4];
+      qword in[2 * 4] ALIGN16_ATTRIB;
 
 
       /* Fetch four attributes for four vertices.  
diff --git a/src/gallium/drivers/cell/spu/spu_vertex_shader.c b/src/gallium/drivers/cell/spu/spu_vertex_shader.c
index 3f5bf41aa2..8363efeeb6 100644
--- a/src/gallium/drivers/cell/spu/spu_vertex_shader.c
+++ b/src/gallium/drivers/cell/spu/spu_vertex_shader.c
@@ -165,63 +165,55 @@ run_vertex_program(struct spu_vs_context *draw,
 }
 
 
-static void
-spu_bind_vertex_shader(struct spu_vs_context *draw,
-		       void *uniforms,
-		       void *planes,
-		       unsigned nr_planes,
-		       unsigned num_outputs
-		       )
-{
-   draw->constants = (float (*)[4]) uniforms;
-
-   (void) memcpy(draw->plane, planes, sizeof(float) * 4 * nr_planes);
-   draw->nr_planes = nr_planes;
-   draw->num_vs_outputs = num_outputs;
-
-   /* specify the shader to interpret/execute */
-   spu_exec_machine_init(&draw->machine,
-			 PIPE_MAX_SAMPLERS,
-			 NULL /*samplers*/,
-			 PIPE_SHADER_VERTEX);
-}
-
-
 unsigned char immediates[(sizeof(float) * 4 * TGSI_EXEC_NUM_IMMEDIATES) + 32]
     ALIGN16_ATTRIB;
 
+
 void
-spu_execute_vertex_shader(struct spu_vs_context *draw,
-                          const struct cell_command_vs *vs)
+spu_bind_vertex_shader(struct spu_vs_context *draw,
+		       struct cell_shader_info *vs)
 {
-   unsigned i;
-
-   const uint64_t immediate_addr = vs->shader.immediates;
+   const unsigned immediate_addr = vs->immediates;
    const unsigned immediate_size = 
-       ROUNDUP16((sizeof(float) * 4 * vs->shader.num_immediates)
-                 + (immediate_addr & 0x0f));
+       ROUNDUP16((sizeof(float) * 4 * vs->num_immediates)
+		 + (immediate_addr & 0x0f));
+ 
 
    mfc_get(immediates, immediate_addr & ~0x0f, immediate_size,
            TAG_VERTEX_BUFFER, 0, 0);
 
    draw->machine.Instructions = (struct tgsi_full_instruction *)
-       vs->shader.instructions;
-   draw->machine.NumInstructions = vs->shader.num_instructions;
+       vs->instructions;
+   draw->machine.NumInstructions = vs->num_instructions;
 
    draw->machine.Declarations = (struct tgsi_full_declaration *)
-       vs->shader.declarations;
-   draw->machine.NumDeclarations = vs->shader.num_declarations;
+       vs->declarations;
+   draw->machine.NumDeclarations = vs->num_declarations;
 
-   draw->vertex_fetch.nr_attrs = vs->nr_attrs;
+   draw->num_vs_outputs = vs->num_outputs;
+
+   /* specify the shader to interpret/execute */
+   spu_exec_machine_init(&draw->machine,
+			 PIPE_MAX_SAMPLERS,
+			 NULL /*samplers*/,
+			 PIPE_SHADER_VERTEX);
 
    wait_on_mask(1 << TAG_VERTEX_BUFFER);
 
    (void) memcpy(& draw->machine.Imms, &immediates[immediate_addr & 0x0f],
-                 sizeof(float) * 4 * vs->shader.num_immediates);
+                 sizeof(float) * 4 * vs->num_immediates);
+}
 
-   spu_bind_vertex_shader(draw, vs->shader.uniforms,
-                          vs->plane, vs->nr_planes,
-                          vs->shader.num_outputs);
+
+void
+spu_execute_vertex_shader(struct spu_vs_context *draw,
+                          const struct cell_command_vs *vs)
+{
+   unsigned i;
+
+   (void) memcpy(draw->plane, vs->plane, sizeof(float) * 4 * vs->nr_planes);
+   draw->nr_planes = vs->nr_planes;
+   draw->vertex_fetch.nr_attrs = vs->nr_attrs;
 
    for (i = 0; i < vs->num_elts; i += 4) {
       const unsigned batch_size = MIN2(vs->num_elts - i, 4);
diff --git a/src/gallium/drivers/cell/spu/spu_vertex_shader.h b/src/gallium/drivers/cell/spu/spu_vertex_shader.h
index 0fb0bc28d0..54a4b8d9b9 100644
--- a/src/gallium/drivers/cell/spu/spu_vertex_shader.h
+++ b/src/gallium/drivers/cell/spu/spu_vertex_shader.h
@@ -1,6 +1,7 @@
 #ifndef SPU_VERTEX_SHADER_H
 #define SPU_VERTEX_SHADER_H
 
+#include "cell/common.h"
 #include "pipe/p_format.h"
 #include "spu_exec.h"
 
@@ -54,6 +55,10 @@ static INLINE void spu_vertex_fetch(struct spu_vs_context *draw,
 
 struct cell_command_vs;
 
+extern void
+spu_bind_vertex_shader(struct spu_vs_context *draw,
+		       struct cell_shader_info *vs);
+
 extern void
 spu_execute_vertex_shader(struct spu_vs_context *draw,
 			  const struct cell_command_vs *vs);
-- 
cgit v1.2.3


From 39038c11699bbc9baab744542e96d54e91cb452a Mon Sep 17 00:00:00 2001
From: Brian <brian.paul@tungstengraphics.com>
Date: Thu, 27 Mar 2008 17:41:55 -0600
Subject: gallium: replace PIPE_ATTRIB_MAX with PIPE_MAX_ATTRIBS

The later follows the naming scheme of other limits.
Keep the old definition until all possible usage is updated.
---
 src/gallium/auxiliary/draw/draw_context.c        |  4 ++--
 src/gallium/auxiliary/draw/draw_private.h        | 12 ++++++------
 src/gallium/auxiliary/draw/draw_pt_fetch_emit.c  |  2 +-
 src/gallium/auxiliary/draw/draw_vf.c             |  4 ++--
 src/gallium/auxiliary/draw/draw_vf.h             |  2 +-
 src/gallium/auxiliary/draw/draw_vs_exec.c        |  4 ++--
 src/gallium/auxiliary/draw/draw_vs_llvm.c        |  4 ++--
 src/gallium/auxiliary/draw/draw_vs_sse.c         |  4 ++--
 src/gallium/drivers/cell/ppu/cell_context.h      |  6 +++---
 src/gallium/drivers/cell/ppu/cell_draw_arrays.c  |  4 ++--
 src/gallium/drivers/cell/ppu/cell_state_vertex.c |  4 ++--
 src/gallium/drivers/cell/ppu/cell_vertex_fetch.c |  2 +-
 src/gallium/drivers/cell/spu/spu_main.c          |  4 ++--
 src/gallium/drivers/cell/spu/spu_vertex_shader.c |  4 ++--
 src/gallium/drivers/cell/spu/spu_vertex_shader.h |  8 ++++----
 src/gallium/drivers/failover/fo_context.h        |  4 ++--
 src/gallium/drivers/failover/fo_state_emit.c     |  4 ++--
 src/gallium/drivers/i915simple/i915_context.c    |  4 ++--
 src/gallium/drivers/i915simple/i915_context.h    |  2 +-
 src/gallium/drivers/i965simple/brw_clip.h        |  2 +-
 src/gallium/drivers/i965simple/brw_context.h     | 12 ++++++------
 src/gallium/drivers/i965simple/brw_state.c       |  2 +-
 src/gallium/drivers/i965simple/brw_wm.h          |  4 ++--
 src/gallium/drivers/softpipe/sp_context.h        |  6 +++---
 src/gallium/drivers/softpipe/sp_draw_arrays.c    |  4 ++--
 src/gallium/drivers/softpipe/sp_quad_fs.c        |  4 ++--
 src/gallium/drivers/softpipe/sp_state_vertex.c   |  4 ++--
 src/gallium/include/pipe/p_state.h               |  3 ++-
 src/mesa/state_tracker/st_draw.c                 |  2 +-
 29 files changed, 63 insertions(+), 62 deletions(-)

(limited to 'src/gallium/drivers/cell/spu/spu_vertex_shader.h')

diff --git a/src/gallium/auxiliary/draw/draw_context.c b/src/gallium/auxiliary/draw/draw_context.c
index 903cc26766..81858e01ca 100644
--- a/src/gallium/auxiliary/draw/draw_context.c
+++ b/src/gallium/auxiliary/draw/draw_context.c
@@ -238,7 +238,7 @@ draw_set_vertex_buffer(struct draw_context *draw,
                        const struct pipe_vertex_buffer *buffer)
 {
    draw_do_flush( draw, DRAW_FLUSH_VERTEX_CACHE/*STATE_CHANGE*/ );
-   assert(attr < PIPE_ATTRIB_MAX);
+   assert(attr < PIPE_MAX_ATTRIBS);
    draw->vertex_buffer[attr] = *buffer;
 }
 
@@ -249,7 +249,7 @@ draw_set_vertex_element(struct draw_context *draw,
                         const struct pipe_vertex_element *element)
 {
    draw_do_flush( draw, DRAW_FLUSH_VERTEX_CACHE/*STATE_CHANGE*/ );
-   assert(attr < PIPE_ATTRIB_MAX);
+   assert(attr < PIPE_MAX_ATTRIBS);
    draw->vertex_element[attr] = *element;
 }
 
diff --git a/src/gallium/auxiliary/draw/draw_private.h b/src/gallium/auxiliary/draw/draw_private.h
index 7007ee22c4..8eb2f515cb 100644
--- a/src/gallium/auxiliary/draw/draw_private.h
+++ b/src/gallium/auxiliary/draw/draw_private.h
@@ -228,8 +228,8 @@ struct draw_context
    /* pipe state that we need: */
    const struct pipe_rasterizer_state *rasterizer;
    struct pipe_viewport_state viewport;
-   struct pipe_vertex_buffer vertex_buffer[PIPE_ATTRIB_MAX];
-   struct pipe_vertex_element vertex_element[PIPE_ATTRIB_MAX];
+   struct pipe_vertex_buffer vertex_buffer[PIPE_MAX_ATTRIBS];
+   struct pipe_vertex_element vertex_element[PIPE_MAX_ATTRIBS];
    struct draw_vertex_shader *vertex_shader;
 
    uint num_vs_outputs;  /**< convenience, from vertex_shader */
@@ -242,7 +242,7 @@ struct draw_context
       unsigned eltSize;
 
       /** vertex arrays */
-      const void *vbuffer[PIPE_ATTRIB_MAX];
+      const void *vbuffer[PIPE_MAX_ATTRIBS];
 
       /** constant buffer (for vertex shader) */
       const void *constants;
@@ -275,9 +275,9 @@ struct draw_context
    /* Vertex fetch internal state
     */
    struct {
-      const ubyte *src_ptr[PIPE_ATTRIB_MAX];
-      unsigned pitch[PIPE_ATTRIB_MAX];
-      fetch_func fetch[PIPE_ATTRIB_MAX];
+      const ubyte *src_ptr[PIPE_MAX_ATTRIBS];
+      unsigned pitch[PIPE_MAX_ATTRIBS];
+      fetch_func fetch[PIPE_MAX_ATTRIBS];
       unsigned nr_attrs;
       full_fetch_func fetch_func;
       pt_fetch_func pt_fetch;
diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c b/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c
index 64ef83d800..9b098bc173 100644
--- a/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c
+++ b/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c
@@ -78,7 +78,7 @@ struct fetch_emit_middle_end {
       unsigned pitch;
       void (*fetch)( const void *from, float *attrib);
       void (*emit)( const float *attrib, float **out );
-   } fetch[PIPE_ATTRIB_MAX];
+   } fetch[PIPE_MAX_ATTRIBS];
    
    unsigned nr_fetch;
    unsigned hw_vertex_size;
diff --git a/src/gallium/auxiliary/draw/draw_vf.c b/src/gallium/auxiliary/draw/draw_vf.c
index f4e29a6293..7bb34ace7a 100644
--- a/src/gallium/auxiliary/draw/draw_vf.c
+++ b/src/gallium/auxiliary/draw/draw_vf.c
@@ -158,7 +158,7 @@ draw_vf_set_vertex_attributes( struct draw_vertex_fetch *vf,
    unsigned offset = 0;
    unsigned i, j;
 
-   assert(nr < PIPE_ATTRIB_MAX);
+   assert(nr < PIPE_MAX_ATTRIBS);
 
    for (j = 0, i = 0; i < nr; i++) {
       const unsigned format = map[i].format;
@@ -390,7 +390,7 @@ struct draw_vertex_fetch *draw_vf_create( void )
    struct draw_vertex_fetch *vf = CALLOC_STRUCT(draw_vertex_fetch);
    unsigned i;
 
-   for (i = 0; i < PIPE_ATTRIB_MAX; i++)
+   for (i = 0; i < PIPE_MAX_ATTRIBS; i++)
       vf->attr[i].vf = vf;
 
    vf->identity[0] = 0.0;
diff --git a/src/gallium/auxiliary/draw/draw_vf.h b/src/gallium/auxiliary/draw/draw_vf.h
index 011c8f0ff1..7555d1bd58 100644
--- a/src/gallium/auxiliary/draw/draw_vf.h
+++ b/src/gallium/auxiliary/draw/draw_vf.h
@@ -169,7 +169,7 @@ struct draw_vf_attr
 
 struct draw_vertex_fetch
 {
-   struct draw_vf_attr attr[PIPE_ATTRIB_MAX];
+   struct draw_vf_attr attr[PIPE_MAX_ATTRIBS];
    unsigned attr_count;
    unsigned vertex_stride;
 
diff --git a/src/gallium/auxiliary/draw/draw_vs_exec.c b/src/gallium/auxiliary/draw/draw_vs_exec.c
index 4e2fa72707..487d0ea7f4 100644
--- a/src/gallium/auxiliary/draw/draw_vs_exec.c
+++ b/src/gallium/auxiliary/draw/draw_vs_exec.c
@@ -99,8 +99,8 @@ vs_exec_run( struct draw_vertex_shader *shader,
    struct tgsi_exec_machine *machine = &draw->machine;
    unsigned int j;
 
-   ALIGN16_DECL(struct tgsi_exec_vector, inputs, PIPE_ATTRIB_MAX);
-   ALIGN16_DECL(struct tgsi_exec_vector, outputs, PIPE_ATTRIB_MAX);
+   ALIGN16_DECL(struct tgsi_exec_vector, inputs, PIPE_MAX_ATTRIBS);
+   ALIGN16_DECL(struct tgsi_exec_vector, outputs, PIPE_MAX_ATTRIBS);
    const float *scale = draw->viewport.scale;
    const float *trans = draw->viewport.translate;
 
diff --git a/src/gallium/auxiliary/draw/draw_vs_llvm.c b/src/gallium/auxiliary/draw/draw_vs_llvm.c
index bd983f2ddf..d29cb18efe 100644
--- a/src/gallium/auxiliary/draw/draw_vs_llvm.c
+++ b/src/gallium/auxiliary/draw/draw_vs_llvm.c
@@ -107,8 +107,8 @@ vs_llvm_run( struct draw_vertex_shader *base,
    struct tgsi_exec_machine *machine = &draw->machine;
    unsigned int j;
 
-   ALIGN16_DECL(struct tgsi_exec_vector, inputs, PIPE_ATTRIB_MAX);
-   ALIGN16_DECL(struct tgsi_exec_vector, outputs, PIPE_ATTRIB_MAX);
+   ALIGN16_DECL(struct tgsi_exec_vector, inputs, PIPE_MAX_ATTRIBS);
+   ALIGN16_DECL(struct tgsi_exec_vector, outputs, PIPE_MAX_ATTRIBS);
    const float *scale = draw->viewport.scale;
    const float *trans = draw->viewport.translate;
 
diff --git a/src/gallium/auxiliary/draw/draw_vs_sse.c b/src/gallium/auxiliary/draw/draw_vs_sse.c
index a4503c143e..bc910dc2d0 100644
--- a/src/gallium/auxiliary/draw/draw_vs_sse.c
+++ b/src/gallium/auxiliary/draw/draw_vs_sse.c
@@ -114,8 +114,8 @@ vs_sse_run( struct draw_vertex_shader *base,
    struct tgsi_exec_machine *machine = &draw->machine;
    unsigned int j;
 
-   ALIGN16_DECL(struct tgsi_exec_vector, inputs, PIPE_ATTRIB_MAX);
-   ALIGN16_DECL(struct tgsi_exec_vector, outputs, PIPE_ATTRIB_MAX);
+   ALIGN16_DECL(struct tgsi_exec_vector, inputs, PIPE_MAX_ATTRIBS);
+   ALIGN16_DECL(struct tgsi_exec_vector, outputs, PIPE_MAX_ATTRIBS);
    const float *scale = draw->viewport.scale;
    const float *trans = draw->viewport.translate;
 
diff --git a/src/gallium/drivers/cell/ppu/cell_context.h b/src/gallium/drivers/cell/ppu/cell_context.h
index 0442abddc1..7f656a9744 100644
--- a/src/gallium/drivers/cell/ppu/cell_context.h
+++ b/src/gallium/drivers/cell/ppu/cell_context.h
@@ -103,8 +103,8 @@ struct cell_context
    struct cell_texture *texture[PIPE_MAX_SAMPLERS];
    uint num_textures;
    struct pipe_viewport_state viewport;
-   struct pipe_vertex_buffer vertex_buffer[PIPE_ATTRIB_MAX];
-   struct pipe_vertex_element vertex_element[PIPE_ATTRIB_MAX];
+   struct pipe_vertex_buffer vertex_buffer[PIPE_MAX_ATTRIBS];
+   struct pipe_vertex_element vertex_element[PIPE_MAX_ATTRIBS];
 
    ubyte *cbuf_map[PIPE_MAX_COLOR_BUFS];
    ubyte *zsbuf_map;
@@ -141,7 +141,7 @@ struct cell_context
 
 
    struct spe_function attrib_fetch;
-   unsigned attrib_fetch_offsets[PIPE_ATTRIB_MAX];
+   unsigned attrib_fetch_offsets[PIPE_MAX_ATTRIBS];
 };
 
 
diff --git a/src/gallium/drivers/cell/ppu/cell_draw_arrays.c b/src/gallium/drivers/cell/ppu/cell_draw_arrays.c
index c839fb4d12..b896252f81 100644
--- a/src/gallium/drivers/cell/ppu/cell_draw_arrays.c
+++ b/src/gallium/drivers/cell/ppu/cell_draw_arrays.c
@@ -123,7 +123,7 @@ cell_draw_elements(struct pipe_context *pipe,
    /*
     * Map vertex buffers
     */
-   for (i = 0; i < PIPE_ATTRIB_MAX; i++) {
+   for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
       if (sp->vertex_buffer[i].buffer) {
          void *buf = pipe->winsys->buffer_map(pipe->winsys,
                                               sp->vertex_buffer[i].buffer,
@@ -151,7 +151,7 @@ cell_draw_elements(struct pipe_context *pipe,
    /*
     * unmap vertex/index buffers - will cause draw module to flush
     */
-   for (i = 0; i < PIPE_ATTRIB_MAX; i++) {
+   for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
       if (sp->vertex_buffer[i].buffer) {
          draw_set_mapped_vertex_buffer(draw, i, NULL);
          pipe->winsys->buffer_unmap(pipe->winsys, sp->vertex_buffer[i].buffer);
diff --git a/src/gallium/drivers/cell/ppu/cell_state_vertex.c b/src/gallium/drivers/cell/ppu/cell_state_vertex.c
index 563831b62d..37d25fb357 100644
--- a/src/gallium/drivers/cell/ppu/cell_state_vertex.c
+++ b/src/gallium/drivers/cell/ppu/cell_state_vertex.c
@@ -41,7 +41,7 @@ cell_set_vertex_element(struct pipe_context *pipe,
                             const struct pipe_vertex_element *attrib)
 {
    struct cell_context *cell = cell_context(pipe);
-   assert(index < PIPE_ATTRIB_MAX);
+   assert(index < PIPE_MAX_ATTRIBS);
    cell->vertex_element[index] = *attrib; /* struct copy */
    cell->dirty |= CELL_NEW_VERTEX;
 
@@ -55,7 +55,7 @@ cell_set_vertex_buffer(struct pipe_context *pipe,
                            const struct pipe_vertex_buffer *buffer)
 {
    struct cell_context *cell = cell_context(pipe);
-   assert(index < PIPE_ATTRIB_MAX);
+   assert(index < PIPE_MAX_ATTRIBS);
    cell->vertex_buffer[index] = *buffer; /* struct copy */
    cell->dirty |= CELL_NEW_VERTEX;
 
diff --git a/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c b/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c
index 4828a8023b..49d5443cde 100644
--- a/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c
+++ b/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c
@@ -263,7 +263,7 @@ void cell_update_vertex_fetch(struct draw_context *draw)
    struct cell_context *const cell =
        (struct cell_context *) draw->driver_private;
    struct spe_function *p = &cell->attrib_fetch;
-   unsigned function_index[PIPE_ATTRIB_MAX];
+   unsigned function_index[PIPE_MAX_ATTRIBS];
    unsigned unique_attr_formats;
    int out_ptr;
    int in_ptr;
diff --git a/src/gallium/drivers/cell/spu/spu_main.c b/src/gallium/drivers/cell/spu/spu_main.c
index fccff01e10..d7f46f8024 100644
--- a/src/gallium/drivers/cell/spu/spu_main.c
+++ b/src/gallium/drivers/cell/spu/spu_main.c
@@ -55,7 +55,7 @@ struct spu_global spu;
 
 struct spu_vs_context draw;
 
-static unsigned char attribute_fetch_code_buffer[136 * PIPE_ATTRIB_MAX]
+static unsigned char attribute_fetch_code_buffer[136 * PIPE_MAX_ATTRIBS]
     ALIGN16_ATTRIB;
 
 static unsigned char depth_stencil_code_buffer[4 * 64]
@@ -361,7 +361,7 @@ cmd_state_vs_array_info(const struct cell_array_info *vs_info)
 {
    const unsigned attr = vs_info->attr;
 
-   ASSERT(attr < PIPE_ATTRIB_MAX);
+   ASSERT(attr < PIPE_MAX_ATTRIBS);
    draw.vertex_fetch.src_ptr[attr] = vs_info->base;
    draw.vertex_fetch.pitch[attr] = vs_info->pitch;
    draw.vertex_fetch.size[attr] = vs_info->size;
diff --git a/src/gallium/drivers/cell/spu/spu_vertex_shader.c b/src/gallium/drivers/cell/spu/spu_vertex_shader.c
index 8363efeeb6..3119a78c06 100644
--- a/src/gallium/drivers/cell/spu/spu_vertex_shader.c
+++ b/src/gallium/drivers/cell/spu/spu_vertex_shader.c
@@ -86,8 +86,8 @@ run_vertex_program(struct spu_vs_context *draw,
    struct spu_exec_machine *machine = &draw->machine;
    unsigned int j;
 
-   ALIGN16_DECL(struct spu_exec_vector, inputs, PIPE_ATTRIB_MAX);
-   ALIGN16_DECL(struct spu_exec_vector, outputs, PIPE_ATTRIB_MAX);
+   ALIGN16_DECL(struct spu_exec_vector, inputs, PIPE_MAX_ATTRIBS);
+   ALIGN16_DECL(struct spu_exec_vector, outputs, PIPE_MAX_ATTRIBS);
    const float *scale = draw->viewport.scale;
    const float *trans = draw->viewport.translate;
 
diff --git a/src/gallium/drivers/cell/spu/spu_vertex_shader.h b/src/gallium/drivers/cell/spu/spu_vertex_shader.h
index 54a4b8d9b9..4c74f5e74d 100644
--- a/src/gallium/drivers/cell/spu/spu_vertex_shader.h
+++ b/src/gallium/drivers/cell/spu/spu_vertex_shader.h
@@ -16,10 +16,10 @@ struct spu_vs_context {
    struct pipe_viewport_state viewport;
 
    struct {
-      uint64_t src_ptr[PIPE_ATTRIB_MAX];
-      unsigned pitch[PIPE_ATTRIB_MAX];
-      unsigned size[PIPE_ATTRIB_MAX];
-      unsigned code_offset[PIPE_ATTRIB_MAX];
+      uint64_t src_ptr[PIPE_MAX_ATTRIBS];
+      unsigned pitch[PIPE_MAX_ATTRIBS];
+      unsigned size[PIPE_MAX_ATTRIBS];
+      unsigned code_offset[PIPE_MAX_ATTRIBS];
       unsigned nr_attrs;
       boolean dirty;
 
diff --git a/src/gallium/drivers/failover/fo_context.h b/src/gallium/drivers/failover/fo_context.h
index 8f3ad3ee79..4afe10c4b8 100644
--- a/src/gallium/drivers/failover/fo_context.h
+++ b/src/gallium/drivers/failover/fo_context.h
@@ -84,8 +84,8 @@ struct failover_context {
    struct pipe_scissor_state scissor;
    struct pipe_texture *texture[PIPE_MAX_SAMPLERS];
    struct pipe_viewport_state viewport;
-   struct pipe_vertex_buffer vertex_buffer[PIPE_ATTRIB_MAX];
-   struct pipe_vertex_element vertex_element[PIPE_ATTRIB_MAX];
+   struct pipe_vertex_buffer vertex_buffer[PIPE_MAX_ATTRIBS];
+   struct pipe_vertex_element vertex_element[PIPE_MAX_ATTRIBS];
 
    void *sw_sampler_state[PIPE_MAX_SAMPLERS];
    void *hw_sampler_state[PIPE_MAX_SAMPLERS];
diff --git a/src/gallium/drivers/failover/fo_state_emit.c b/src/gallium/drivers/failover/fo_state_emit.c
index 3de931e04e..bb89f925e9 100644
--- a/src/gallium/drivers/failover/fo_state_emit.c
+++ b/src/gallium/drivers/failover/fo_state_emit.c
@@ -104,7 +104,7 @@ failover_state_emit( struct failover_context *failover )
    }
 
    if (failover->dirty & FO_NEW_VERTEX_BUFFER) {
-      for (i = 0; i < PIPE_ATTRIB_MAX; i++) {
+      for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
 	 if (failover->dirty_vertex_buffer & (1<<i)) {
 	    failover->sw->set_vertex_buffer( failover->sw, i, 
 					     &failover->vertex_buffer[i] );
@@ -113,7 +113,7 @@ failover_state_emit( struct failover_context *failover )
    }
 
    if (failover->dirty & FO_NEW_VERTEX_ELEMENT) {
-      for (i = 0; i < PIPE_ATTRIB_MAX; i++) {
+      for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
 	 if (failover->dirty_vertex_element & (1<<i)) {
 	    failover->sw->set_vertex_element( failover->sw, i, 
 					      &failover->vertex_element[i] );
diff --git a/src/gallium/drivers/i915simple/i915_context.c b/src/gallium/drivers/i915simple/i915_context.c
index 15ff2360b7..fee33d82de 100644
--- a/src/gallium/drivers/i915simple/i915_context.c
+++ b/src/gallium/drivers/i915simple/i915_context.c
@@ -65,7 +65,7 @@ i915_draw_elements( struct pipe_context *pipe,
    /*
     * Map vertex buffers
     */
-   for (i = 0; i < PIPE_ATTRIB_MAX; i++) {
+   for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
       if (i915->vertex_buffer[i].buffer) {
          void *buf
             = pipe->winsys->buffer_map(pipe->winsys,
@@ -96,7 +96,7 @@ i915_draw_elements( struct pipe_context *pipe,
    /*
     * unmap vertex/index buffers
     */
-   for (i = 0; i < PIPE_ATTRIB_MAX; i++) {
+   for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
       if (i915->vertex_buffer[i].buffer) {
          pipe->winsys->buffer_unmap(pipe->winsys, i915->vertex_buffer[i].buffer);
          draw_set_mapped_vertex_buffer(draw, i, NULL);
diff --git a/src/gallium/drivers/i915simple/i915_context.h b/src/gallium/drivers/i915simple/i915_context.h
index 746f18ba38..8e707ea574 100644
--- a/src/gallium/drivers/i915simple/i915_context.h
+++ b/src/gallium/drivers/i915simple/i915_context.h
@@ -232,7 +232,7 @@ struct i915_context
    struct pipe_scissor_state scissor;
    struct i915_texture *texture[PIPE_MAX_SAMPLERS];
    struct pipe_viewport_state viewport;
-   struct pipe_vertex_buffer vertex_buffer[PIPE_ATTRIB_MAX];
+   struct pipe_vertex_buffer vertex_buffer[PIPE_MAX_ATTRIBS];
 
    unsigned dirty;
 
diff --git a/src/gallium/drivers/i965simple/brw_clip.h b/src/gallium/drivers/i965simple/brw_clip.h
index a89d08b791..d70fc094ff 100644
--- a/src/gallium/drivers/i965simple/brw_clip.h
+++ b/src/gallium/drivers/i965simple/brw_clip.h
@@ -116,7 +116,7 @@ struct brw_clip_compile {
    unsigned last_mrf;
 
    unsigned header_position_offset;
-   unsigned offset[PIPE_ATTRIB_MAX];
+   unsigned offset[PIPE_MAX_ATTRIBS];
 };
 
 #define ATTR_SIZE  (4*4)
diff --git a/src/gallium/drivers/i965simple/brw_context.h b/src/gallium/drivers/i965simple/brw_context.h
index b83a13c3b6..0c96ba1732 100644
--- a/src/gallium/drivers/i965simple/brw_context.h
+++ b/src/gallium/drivers/i965simple/brw_context.h
@@ -433,17 +433,17 @@ struct brw_cached_batch_item {
 
 
 
-/* Protect against a future where PIPE_ATTRIB_MAX > 32.  Wouldn't life
+/* Protect against a future where PIPE_MAX_ATTRIBS > 32.  Wouldn't life
  * be easier if C allowed arrays of packed elements?
  */
-#define ATTRIB_BIT_DWORDS  ((PIPE_ATTRIB_MAX+31)/32)
+#define ATTRIB_BIT_DWORDS  ((PIPE_MAX_ATTRIBS+31)/32)
 
 
 
 
 struct brw_vertex_info {
-   unsigned varying;  /* varying:1[PIPE_ATTRIB_MAX] */
-   unsigned sizes[ATTRIB_BIT_DWORDS * 2]; /* sizes:2[PIPE_ATTRIB_MAX] */
+   unsigned varying;  /* varying:1[PIPE_MAX_ATTRIBS] */
+   unsigned sizes[ATTRIB_BIT_DWORDS * 2]; /* sizes:2[PIPE_MAX_ATTRIBS] */
 };
 
 
@@ -496,9 +496,9 @@ struct brw_context
       /* Arrays with buffer objects to copy non-bufferobj arrays into
        * for upload:
        */
-      const struct pipe_vertex_buffer *vbo_array[PIPE_ATTRIB_MAX];
+      const struct pipe_vertex_buffer *vbo_array[PIPE_MAX_ATTRIBS];
 
-      struct brw_vertex_element_state inputs[PIPE_ATTRIB_MAX];
+      struct brw_vertex_element_state inputs[PIPE_MAX_ATTRIBS];
 
 #define BRW_NR_UPLOAD_BUFS 17
 #define BRW_UPLOAD_INIT_SIZE (128*1024)
diff --git a/src/gallium/drivers/i965simple/brw_state.c b/src/gallium/drivers/i965simple/brw_state.c
index f5efe9fc06..0d04a8a594 100644
--- a/src/gallium/drivers/i965simple/brw_state.c
+++ b/src/gallium/drivers/i965simple/brw_state.c
@@ -292,7 +292,7 @@ static void brw_set_vertex_element(struct pipe_context *pipe,
    /* flush ? */
    struct brw_context *brw = brw_context(pipe);
 
-   assert(index < PIPE_ATTRIB_MAX);
+   assert(index < PIPE_MAX_ATTRIBS);
    struct brw_vertex_element_state el;
    memset(&el, 0, sizeof(el));
 
diff --git a/src/gallium/drivers/i965simple/brw_wm.h b/src/gallium/drivers/i965simple/brw_wm.h
index a1ac0f504a..b29c4393f0 100644
--- a/src/gallium/drivers/i965simple/brw_wm.h
+++ b/src/gallium/drivers/i965simple/brw_wm.h
@@ -76,7 +76,7 @@ struct brw_wm_prog_key {
 
 #define PROGRAM_INTERNAL_PARAM
 #define MAX_NV_FRAGMENT_PROGRAM_INSTRUCTIONS 1024 /* 72 for GL_ARB_f_p */
-#define BRW_WM_MAX_INSN  (MAX_NV_FRAGMENT_PROGRAM_INSTRUCTIONS*3 + PIPE_ATTRIB_MAX + 3)
+#define BRW_WM_MAX_INSN  (MAX_NV_FRAGMENT_PROGRAM_INSTRUCTIONS*3 + PIPE_MAX_ATTRIBS + 3)
 #define BRW_WM_MAX_GRF   128		/* hardware limit */
 #define BRW_WM_MAX_VREG  (BRW_WM_MAX_INSN * 4)
 #define BRW_WM_MAX_REF   (BRW_WM_MAX_INSN * 12)
@@ -84,7 +84,7 @@ struct brw_wm_prog_key {
 #define BRW_WM_MAX_CONST 256
 #define BRW_WM_MAX_KILLS MAX_NV_FRAGMENT_PROGRAM_INSTRUCTIONS
 
-#define PAYLOAD_DEPTH     (PIPE_ATTRIB_MAX)
+#define PAYLOAD_DEPTH     (PIPE_MAX_ATTRIBS)
 
 #define MAX_IFSN 32
 #define MAX_LOOP_DEPTH 32
diff --git a/src/gallium/drivers/softpipe/sp_context.h b/src/gallium/drivers/softpipe/sp_context.h
index 19e6cfaf02..dc9d0e6d5d 100644
--- a/src/gallium/drivers/softpipe/sp_context.h
+++ b/src/gallium/drivers/softpipe/sp_context.h
@@ -77,8 +77,8 @@ struct softpipe_context {
    struct pipe_scissor_state scissor;
    struct pipe_texture *texture[PIPE_MAX_SAMPLERS];
    struct pipe_viewport_state viewport;
-   struct pipe_vertex_buffer vertex_buffer[PIPE_ATTRIB_MAX];
-   struct pipe_vertex_element vertex_element[PIPE_ATTRIB_MAX];
+   struct pipe_vertex_buffer vertex_buffer[PIPE_MAX_ATTRIBS];
+   struct pipe_vertex_element vertex_element[PIPE_MAX_ATTRIBS];
    unsigned dirty;
 
    unsigned num_samplers;
@@ -92,7 +92,7 @@ struct softpipe_context {
    /*
     * Mapped vertex buffers
     */
-   ubyte *mapped_vbuffer[PIPE_ATTRIB_MAX];
+   ubyte *mapped_vbuffer[PIPE_MAX_ATTRIBS];
    
    /** Mapped constant buffers */
    void *mapped_constants[PIPE_SHADER_TYPES];
diff --git a/src/gallium/drivers/softpipe/sp_draw_arrays.c b/src/gallium/drivers/softpipe/sp_draw_arrays.c
index 5b5a0fe573..ab54050d3f 100644
--- a/src/gallium/drivers/softpipe/sp_draw_arrays.c
+++ b/src/gallium/drivers/softpipe/sp_draw_arrays.c
@@ -125,7 +125,7 @@ softpipe_draw_elements(struct pipe_context *pipe,
    /*
     * Map vertex buffers
     */
-   for (i = 0; i < PIPE_ATTRIB_MAX; i++) {
+   for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
       if (sp->vertex_buffer[i].buffer) {
          void *buf
             = pipe->winsys->buffer_map(pipe->winsys,
@@ -153,7 +153,7 @@ softpipe_draw_elements(struct pipe_context *pipe,
    /*
     * unmap vertex/index buffers - will cause draw module to flush
     */
-   for (i = 0; i < PIPE_ATTRIB_MAX; i++) {
+   for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
       if (sp->vertex_buffer[i].buffer) {
          draw_set_mapped_vertex_buffer(draw, i, NULL);
          pipe->winsys->buffer_unmap(pipe->winsys, sp->vertex_buffer[i].buffer);
diff --git a/src/gallium/drivers/softpipe/sp_quad_fs.c b/src/gallium/drivers/softpipe/sp_quad_fs.c
index 861285101f..c10ad80e01 100644
--- a/src/gallium/drivers/softpipe/sp_quad_fs.c
+++ b/src/gallium/drivers/softpipe/sp_quad_fs.c
@@ -185,8 +185,8 @@ struct quad_stage *sp_quad_shade_stage( struct softpipe_context *softpipe )
    uint i;
 
    /* allocate storage for program inputs/outputs, aligned to 16 bytes */
-   qss->inputs = MALLOC(PIPE_ATTRIB_MAX * sizeof(*qss->inputs) + 16);
-   qss->outputs = MALLOC(PIPE_ATTRIB_MAX * sizeof(*qss->outputs) + 16);
+   qss->inputs = MALLOC(PIPE_MAX_ATTRIBS * sizeof(*qss->inputs) + 16);
+   qss->outputs = MALLOC(PIPE_MAX_ATTRIBS * sizeof(*qss->outputs) + 16);
    qss->machine.Inputs = align16(qss->inputs);
    qss->machine.Outputs = align16(qss->outputs);
 
diff --git a/src/gallium/drivers/softpipe/sp_state_vertex.c b/src/gallium/drivers/softpipe/sp_state_vertex.c
index f01a10de3b..c054e76d9b 100644
--- a/src/gallium/drivers/softpipe/sp_state_vertex.c
+++ b/src/gallium/drivers/softpipe/sp_state_vertex.c
@@ -42,7 +42,7 @@ softpipe_set_vertex_element(struct pipe_context *pipe,
                             const struct pipe_vertex_element *attrib)
 {
    struct softpipe_context *softpipe = softpipe_context(pipe);
-   assert(index < PIPE_ATTRIB_MAX);
+   assert(index < PIPE_MAX_ATTRIBS);
    softpipe->vertex_element[index] = *attrib; /* struct copy */
    softpipe->dirty |= SP_NEW_VERTEX;
 
@@ -56,7 +56,7 @@ softpipe_set_vertex_buffer(struct pipe_context *pipe,
                            const struct pipe_vertex_buffer *buffer)
 {
    struct softpipe_context *softpipe = softpipe_context(pipe);
-   assert(index < PIPE_ATTRIB_MAX);
+   assert(index < PIPE_MAX_ATTRIBS);
    softpipe->vertex_buffer[index] = *buffer; /* struct copy */
    softpipe->dirty |= SP_NEW_VERTEX;
 
diff --git a/src/gallium/include/pipe/p_state.h b/src/gallium/include/pipe/p_state.h
index a2bd8c6aaa..2490412126 100644
--- a/src/gallium/include/pipe/p_state.h
+++ b/src/gallium/include/pipe/p_state.h
@@ -54,7 +54,8 @@ extern "C" {
 #define PIPE_MAX_SAMPLERS     8
 #define PIPE_MAX_CLIP_PLANES  6
 #define PIPE_MAX_CONSTANT    32
-#define PIPE_ATTRIB_MAX      32
+#define PIPE_MAX_ATTRIBS     32
+#define PIPE_ATTRIB_MAX      32 /* XXX obsolete - remove */
 #define PIPE_MAX_COLOR_BUFS   8
 #define PIPE_MAX_TEXTURE_LEVELS  16
 #define PIPE_MAX_FEEDBACK_ATTRIBS 16
diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c
index 20af90df7d..4aca3311b7 100644
--- a/src/mesa/state_tracker/st_draw.c
+++ b/src/mesa/state_tracker/st_draw.c
@@ -606,7 +606,7 @@ st_feedback_draw_vbo(GLcontext *ctx,
    /*
     * unmap vertex/index buffers
     */
-   for (i = 0; i < PIPE_ATTRIB_MAX; i++) {
+   for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
       if (draw->vertex_buffer[i].buffer) {
          pipe->winsys->buffer_unmap(pipe->winsys,
                                     draw->vertex_buffer[i].buffer);
-- 
cgit v1.2.3