From: Brian Paul <brian.paul@tungstengraphics.com>
Date: Thu, 11 Sep 2008 00:22:00 +0000 (-0600)
Subject: cell: checkpoint commit of new per-fragment processing
X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=284ab5a6127f8b452acaa0e10ac1d9ebc87fac3e;p=mesa.git

cell: checkpoint commit of new per-fragment processing

Do code generation for alpha test, z test, stencil, blend, colormask
and framebuffer/tile read/write as a single code block.
Ian's previous blend/z/stencil test code is still there but mostly disabled
and will be removed soon.
---

diff --git a/src/gallium/drivers/cell/common.h b/src/gallium/drivers/cell/common.h
index c0ca201e1d1..a62530c64da 100644
--- a/src/gallium/drivers/cell/common.h
+++ b/src/gallium/drivers/cell/common.h
@@ -97,6 +97,7 @@
 #define CELL_CMD_STATE_LOGICOP       21
 #define CELL_CMD_VS_EXECUTE          22
 #define CELL_CMD_FLUSH_BUFFER_RANGE  23
+#define CELL_CMD_STATE_FRAGMENT_OPS  24
 
 
 #define CELL_NUM_BUFFERS 4
@@ -112,30 +113,43 @@
 
 /**
  */
-struct cell_command_depth_stencil_alpha_test {
+struct cell_command_depth_stencil_alpha_test
+{
    uint64_t base;               /**< Effective address of code start. */
    unsigned size;               /**< Size in bytes of SPE code. */
    unsigned read_depth;         /**< Flag: should depth be read? */
    unsigned read_stencil;       /**< Flag: should stencil be read? */
+   struct pipe_depth_stencil_alpha_state state;
 };
 
 
 /**
  * Upload code to perform framebuffer blend operation
  */
-struct cell_command_blend {
+struct cell_command_blend
+{
    uint64_t base;               /**< Effective address of code start. */
    unsigned size;               /**< Size in bytes of SPE code. */
    unsigned read_fb;            /**< Flag: should framebuffer be read? */
 };
 
 
-struct cell_command_logicop {
+struct cell_command_logicop
+{
    uint64_t base;               /**< Effective address of code start. */
    unsigned size;               /**< Size in bytes of SPE code. */
 };
 
 
+#define SPU_MAX_FRAGMENT_OPS_INSTS 64
+
+struct cell_command_fragment_ops
+{
+   uint64_t opcode;      /**< CELL_CMD_STATE_FRAGMENT_OPS */
+   unsigned code[SPU_MAX_FRAGMENT_OPS_INSTS];
+};
+
+
 /**
  * Tell SPUs about the framebuffer size, location
  */
diff --git a/src/gallium/drivers/cell/ppu/Makefile b/src/gallium/drivers/cell/ppu/Makefile
index 25473e200cf..b5a6fcb8deb 100644
--- a/src/gallium/drivers/cell/ppu/Makefile
+++ b/src/gallium/drivers/cell/ppu/Makefile
@@ -25,6 +25,7 @@ SOURCES = \
 	cell_context.c \
 	cell_draw_arrays.c \
 	cell_flush.c \
+	cell_gen_fragment.c \
 	cell_state_derived.c \
 	cell_state_emit.c \
 	cell_state_per_fragment.c \
diff --git a/src/gallium/drivers/cell/ppu/cell_gen_fragment.c b/src/gallium/drivers/cell/ppu/cell_gen_fragment.c
new file mode 100644
index 00000000000..df29476be66
--- /dev/null
+++ b/src/gallium/drivers/cell/ppu/cell_gen_fragment.c
@@ -0,0 +1,530 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+
+/**
+ * Generate SPU per-fragment code (actually per-quad code).
+ * \author Brian Paul
+ */
+
+
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+#include "rtasm/rtasm_ppc_spe.h"
+#include "cell_context.h"
+#include "cell_gen_fragment.h"
+
+
+
+/** Do extra optimizations? */
+#define OPTIMIZATIONS 1
+
+
+/**
+ * Generate SPE code to perform Z/depth testing.
+ *
+ * \param dsa         Gallium depth/stencil/alpha state to gen code for
+ * \param f           SPE function to append instruction onto.
+ * \param mask_reg    register containing quad/pixel "alive" mask (in/out)
+ * \param ifragZ_reg  register containing integer fragment Z values (in)
+ * \param ifbZ_reg    register containing integer frame buffer Z values (in/out)
+ * \param zmask_reg   register containing result of Z test/comparison (out)
+ */
+static void
+gen_depth_test(const struct pipe_depth_stencil_alpha_state *dsa,
+               struct spe_function *f,
+               int mask_reg, int ifragZ_reg, int ifbZ_reg, int zmask_reg)
+{
+   ASSERT(dsa->depth.enabled);
+
+   switch (dsa->depth.func) {
+   case PIPE_FUNC_EQUAL:
+      /* zmask = (ifragZ == ref) */
+      spe_ceq(f, zmask_reg, ifragZ_reg, ifbZ_reg);
+      /* mask = (mask & zmask) */
+      spe_and(f, mask_reg, mask_reg, zmask_reg);
+      break;
+
+   case PIPE_FUNC_NOTEQUAL:
+      /* zmask = (ifragZ == ref) */
+      spe_ceq(f, zmask_reg, ifragZ_reg, ifbZ_reg);
+      /* mask = (mask & ~zmask) */
+      spe_andc(f, mask_reg, mask_reg, zmask_reg);
+      break;
+
+   case PIPE_FUNC_GREATER:
+      /* zmask = (ifragZ > ref) */
+      spe_cgt(f, zmask_reg, ifragZ_reg, ifbZ_reg);
+      /* mask = (mask & zmask) */
+      spe_and(f, mask_reg, mask_reg, zmask_reg);
+      break;
+
+   case PIPE_FUNC_LESS:
+      /* zmask = (ref > ifragZ) */
+      spe_cgt(f, zmask_reg, ifbZ_reg, ifragZ_reg);
+      /* mask = (mask & zmask) */
+      spe_and(f, mask_reg, mask_reg, zmask_reg);
+      break;
+
+   case PIPE_FUNC_LEQUAL:
+      /* zmask = (ifragZ > ref) */
+      spe_cgt(f, zmask_reg, ifragZ_reg, ifbZ_reg);
+      /* mask = (mask & ~zmask) */
+      spe_andc(f, mask_reg, mask_reg, zmask_reg);
+      break;
+
+   case PIPE_FUNC_GEQUAL:
+      /* zmask = (ref > ifragZ) */
+      spe_cgt(f, zmask_reg, ifbZ_reg, ifragZ_reg);
+      /* mask = (mask & ~zmask) */
+      spe_andc(f, mask_reg, mask_reg, zmask_reg);
+      break;
+
+   case PIPE_FUNC_NEVER:
+      spe_il(f, mask_reg, 0);  /* mask = {0,0,0,0} */
+      spe_move(f, zmask_reg, mask_reg);  /* zmask = mask */
+      break;
+
+   case PIPE_FUNC_ALWAYS:
+      /* mask unchanged */
+      spe_il(f, zmask_reg, ~0);  /* zmask = {~0,~0,~0,~0} */
+      break;
+
+   default:
+      ASSERT(0);
+      break;
+   }
+
+   if (dsa->depth.writemask) {
+      /*
+       * If (ztest passed) {
+       *    framebufferZ = fragmentZ;
+       * }
+       * OR,
+       * framebufferZ = (ztest_passed ? fragmentZ : framebufferZ;
+       */
+      spe_selb(f, ifbZ_reg, ifbZ_reg, ifragZ_reg, mask_reg);
+   }
+}
+
+
+/**
+ * Generate SPE code to perform alpha testing.
+ *
+ * \param dsa        Gallium depth/stencil/alpha state to gen code for
+ * \param f          SPE function to append instruction onto.
+ * \param mask_reg   register containing quad/pixel "alive" mask (in/out)
+ * \param fragA_reg  register containing four fragment alpha values (in)
+ */
+static void
+gen_alpha_test(const struct pipe_depth_stencil_alpha_state *dsa,
+               struct spe_function *f, int mask_reg, int fragA_reg)
+{
+   int ref_reg = spe_allocate_available_register(f);
+   int amask_reg = spe_allocate_available_register(f);
+
+   ASSERT(dsa->alpha.enabled);
+
+   if ((dsa->alpha.func != PIPE_FUNC_NEVER) &&
+       (dsa->alpha.func != PIPE_FUNC_ALWAYS)) {
+      /* load/splat the alpha reference float value */
+      spe_load_float(f, ref_reg, dsa->alpha.ref);
+   }
+
+   /* emit code to do the alpha comparison, updating 'mask' */
+   switch (dsa->alpha.func) {
+   case PIPE_FUNC_EQUAL:
+      /* amask = (fragA == ref) */
+      spe_fceq(f, amask_reg, fragA_reg, ref_reg);
+      /* mask = (mask & amask) */
+      spe_and(f, mask_reg, mask_reg, amask_reg);
+      break;
+
+   case PIPE_FUNC_NOTEQUAL:
+      /* amask = (fragA == ref) */
+      spe_fceq(f, amask_reg, fragA_reg, ref_reg);
+      /* mask = (mask & ~amask) */
+      spe_andc(f, mask_reg, mask_reg, amask_reg);
+      break;
+
+   case PIPE_FUNC_GREATER:
+      /* amask = (fragA > ref) */
+      spe_fcgt(f, amask_reg, fragA_reg, ref_reg);
+      /* mask = (mask & amask) */
+      spe_and(f, mask_reg, mask_reg, amask_reg);
+      break;
+
+   case PIPE_FUNC_LESS:
+      /* amask = (ref > fragA) */
+      spe_fcgt(f, amask_reg, ref_reg, fragA_reg);
+      /* mask = (mask & amask) */
+      spe_and(f, mask_reg, mask_reg, amask_reg);
+      break;
+
+   case PIPE_FUNC_LEQUAL:
+      /* amask = (fragA > ref) */
+      spe_fcgt(f, amask_reg, fragA_reg, ref_reg);
+      /* mask = (mask & ~amask) */
+      spe_andc(f, mask_reg, mask_reg, amask_reg);
+      break;
+
+   case PIPE_FUNC_GEQUAL:
+      /* amask = (ref > fragA) */
+      spe_fcgt(f, amask_reg, ref_reg, fragA_reg);
+      /* mask = (mask & ~amask) */
+      spe_andc(f, mask_reg, mask_reg, amask_reg);
+      break;
+
+   case PIPE_FUNC_NEVER:
+      spe_il(f, mask_reg, 0);  /* mask = [0,0,0,0] */
+      break;
+
+   case PIPE_FUNC_ALWAYS:
+      /* no-op, mask unchanged */
+      break;
+
+   default:
+      ASSERT(0);
+      break;
+   }
+
+#if OPTIMIZATIONS
+   /* if mask == {0,0,0,0} we're all done, return */
+   {
+      /* re-use amask reg here */
+      int tmp_reg = amask_reg;
+      /* tmp[0] = (mask[0] | mask[1] | mask[2] | mask[3]) */
+      spe_orx(f, tmp_reg, mask_reg);
+      /* if tmp[0] == 0 then return from function call */
+      spe_biz(f, tmp_reg, SPE_REG_RA, 0, 0);
+   }
+#endif
+
+   spe_release_register(f, ref_reg);
+   spe_release_register(f, amask_reg);
+}
+
+
+
+/**
+ * Generate SPE code to implement the fragment operations (alpha test,
+ * depth test, stencil test, blending, colormask, and final
+ * framebuffer write) as specified by the current context state.
+ *
+ * Logically, this code will be called after running the fragment
+ * shader.  But under some circumstances we could run some of this
+ * code before the fragment shader to cull fragments/quads that are
+ * totally occluded/discarded.
+ *
+ * XXX we only support PIPE_FORMAT_Z24S8_UNORM z/stencil buffer right now.
+ *
+ * See the spu_default_fragment_ops() function to see how the per-fragment
+ * operations would be done with ordinary C code.
+ * The code we generate here though has no branches, is SIMD, etc and
+ * should be much faster.
+ *
+ * \param cell  the rendering context (in)
+ * \param f     the generated function (out)
+ */
+void
+gen_fragment_function(struct cell_context *cell, struct spe_function *f)
+{
+   const struct pipe_depth_stencil_alpha_state *dsa =
+      &cell->depth_stencil->base;
+   const struct pipe_blend_state *blend = &cell->blend->base;
+
+   /* For SPE function calls: reg $3 = first param, $4 = second param, etc. */
+   const int x_reg = 3;  /* uint */
+   const int y_reg = 4;  /* uint */
+   const int color_tile_reg = 5;  /* tile_t * */
+   const int depth_tile_reg = 6;  /* tile_t * */
+   const int fragZ_reg = 7;   /* vector float */
+   const int fragR_reg = 8;   /* vector float */
+   const int fragG_reg = 9;   /* vector float */
+   const int fragB_reg = 10;  /* vector float */
+   const int fragA_reg = 11;  /* vector float */
+   const int mask_reg = 12;   /* vector uint */
+
+   /* offset of quad from start of tile
+    * XXX assuming 4-byte pixels for color AND Z/stencil!!!!
+    */
+   int quad_offset_reg;
+
+   int fbRGBA_reg;  /**< framebuffer's RGBA colors for quad */
+   int fbZS_reg;    /**< framebuffer's combined z/stencil values for quad */
+
+   spe_init_func(f, SPU_MAX_FRAGMENT_OPS_INSTS * SPE_INST_SIZE);
+   spe_allocate_register(f, x_reg);
+   spe_allocate_register(f, y_reg);
+   spe_allocate_register(f, color_tile_reg);
+   spe_allocate_register(f, depth_tile_reg);
+   spe_allocate_register(f, fragZ_reg);
+   spe_allocate_register(f, fragR_reg);
+   spe_allocate_register(f, fragG_reg);
+   spe_allocate_register(f, fragB_reg);
+   spe_allocate_register(f, fragA_reg);
+   spe_allocate_register(f, mask_reg);
+
+   quad_offset_reg = spe_allocate_available_register(f);
+   fbRGBA_reg = spe_allocate_available_register(f);
+   fbZS_reg = spe_allocate_available_register(f);
+
+   /* compute offset of quad from start of tile, in bytes */
+   {
+      int x2_reg = spe_allocate_available_register(f);
+      int y2_reg = spe_allocate_available_register(f);
+
+      ASSERT(TILE_SIZE == 32);
+
+      spe_rotmi(f, x2_reg, x_reg, -1);  /* x2 = x / 2 */
+      spe_rotmi(f, y2_reg, y_reg, -1);  /* y2 = y / 2 */
+      spe_shli(f, y2_reg, y2_reg, 4);   /* y2 *= 16 */
+      spe_a(f, quad_offset_reg, y2_reg, x2_reg);  /* offset = y2 + x2 */
+      spe_shli(f, quad_offset_reg, quad_offset_reg, 4);   /* offset *= 16 */
+
+      spe_release_register(f, x2_reg);
+      spe_release_register(f, y2_reg);
+   }
+
+
+   if (dsa->alpha.enabled) {
+      gen_alpha_test(dsa, f, mask_reg, fragA_reg);
+   }
+
+   if (dsa->depth.enabled || dsa->stencil[0].enabled) {
+      const enum pipe_format zs_format = cell->framebuffer.zsbuf->format;
+      boolean write_depth_stencil;
+
+      int fbZ_reg = spe_allocate_available_register(f); /* Z values */
+      int fbS_reg = spe_allocate_available_register(f); /* Stencil values */
+
+      /* fetch quad of depth/stencil values from tile at (x,y) */
+      /* Load: fbZS_reg = memory[depth_tile_reg + offset_reg] */
+      spe_lqx(f, fbZS_reg, depth_tile_reg, quad_offset_reg);
+
+      if (dsa->depth.enabled) {
+         /* Extract Z bits from fbZS_reg into fbZ_reg */
+         if (zs_format == PIPE_FORMAT_S8Z24_UNORM ||
+             zs_format == PIPE_FORMAT_X8Z24_UNORM) {
+            int mask_reg = spe_allocate_available_register(f);
+            spe_fsmbi(f, mask_reg, 0x7777);  /* mask[0,1,2,3] = 0x00ffffff */
+            spe_and(f, fbZ_reg, fbZS_reg, mask_reg);  /* fbZ = fbZS & mask */
+            spe_release_register(f, mask_reg);
+            /* OK, fbZ_reg has four 24-bit Z values now */
+         }
+         else {
+            /* XXX handle other z/stencil formats */
+            ASSERT(0);
+         }
+
+         /* Convert fragZ values from float[4] to uint[4] */
+         if (zs_format == PIPE_FORMAT_S8Z24_UNORM ||
+             zs_format == PIPE_FORMAT_X8Z24_UNORM ||
+             zs_format == PIPE_FORMAT_Z24S8_UNORM ||
+             zs_format == PIPE_FORMAT_Z24X8_UNORM) {
+            /* 24-bit Z values */
+            int scale_reg = spe_allocate_available_register(f);
+
+            /* scale_reg[0,1,2,3] = float(2^24-1) */
+            spe_load_float(f, scale_reg, (float) 0xffffff);
+
+            /* XXX these two instructions might be combined */
+            spe_fm(f, fragZ_reg, fragZ_reg, scale_reg); /* fragZ *= scale */
+            spe_cfltu(f, fragZ_reg, fragZ_reg, 0);  /* fragZ = (int) fragZ */
+
+            spe_release_register(f, scale_reg);
+         }
+         else {
+            /* XXX handle 16-bit Z format */
+            ASSERT(0);
+         }
+      }
+
+      if (dsa->stencil[0].enabled) {
+         /* Extract Stencil bit sfrom fbZS_reg into fbS_reg */
+         if (zs_format == PIPE_FORMAT_S8Z24_UNORM ||
+             zs_format == PIPE_FORMAT_X8Z24_UNORM) {
+            /* XXX extract with a shift */
+            ASSERT(0);
+         }
+         else if (zs_format == PIPE_FORMAT_Z24S8_UNORM ||
+                  zs_format == PIPE_FORMAT_Z24X8_UNORM) {
+            /* XXX extract with a mask */
+            ASSERT(0);
+         }
+      }
+
+
+      if (dsa->stencil[0].enabled) {
+         /* XXX this may involve depth testing too */
+         // gen_stencil_test(dsa, f, ... );
+         ASSERT(0);
+      }
+      else if (dsa->depth.enabled) {
+         int zmask_reg = spe_allocate_available_register(f);
+         gen_depth_test(dsa, f, mask_reg, fragZ_reg, fbZ_reg, zmask_reg);
+         spe_release_register(f, zmask_reg);
+      }
+
+      /* do we need to write Z and/or Stencil back into framebuffer? */
+      write_depth_stencil = (dsa->depth.writemask |
+                             dsa->stencil[0].write_mask |
+                             dsa->stencil[1].write_mask);
+
+      if (write_depth_stencil) {
+         /* Merge latest Z and Stencil values into fbZS_reg.
+          * fbZ_reg has four Z vals in bits [23..0] or bits [15..0].
+          * fbS_reg has four 8-bit Z values in bits [7..0].
+          */
+         if (zs_format == PIPE_FORMAT_S8Z24_UNORM ||
+             zs_format == PIPE_FORMAT_X8Z24_UNORM) {
+            spe_shli(f, fbS_reg, fbS_reg, 24); /* fbS = fbS << 24 */
+            spe_or(f, fbZS_reg, fbS_reg, fbZ_reg); /* fbZS = fbS | fbZ */
+         }
+         else if (zs_format == PIPE_FORMAT_S8Z24_UNORM ||
+                  zs_format == PIPE_FORMAT_X8Z24_UNORM) {
+            /* XXX to do */
+            ASSERT(0);
+         }
+         else if (zs_format == PIPE_FORMAT_Z16_UNORM) {
+            /* XXX to do */
+            ASSERT(0);
+         }
+         else if (zs_format == PIPE_FORMAT_S8_UNORM) {
+            /* XXX to do */
+            ASSERT(0);
+         }
+         else {
+            /* bad zs_format */
+            ASSERT(0);
+         }
+
+         /* Store: memory[depth_tile_reg + quad_offset_reg] = fbZS */
+         spe_stqx(f, fbZS_reg, depth_tile_reg, quad_offset_reg);
+      }
+
+      spe_release_register(f, fbZ_reg);
+      spe_release_register(f, fbS_reg);
+   }
+
+
+   /* Get framebuffer quad/colors.  We'll need these for blending,
+    * color masking, and to obey the quad/pixel mask.
+    * Load: fbRGBA_reg = memory[color_tile + quad_offset]
+    * Note: if mask={~0,~0,~0,~0} and we're not blending or colormasking
+    * we could skip this load.
+    */
+   spe_lqx(f, fbRGBA_reg, color_tile_reg, quad_offset_reg);
+
+
+   if (blend->blend_enable) {
+      /* convert packed tile colors in fbRGBA_reg to float[4] vectors */
+
+      // gen_blend_code(blend, f, mask_reg, ... );
+
+   }
+
+
+
+   /*
+    * Write fragment colors to framebuffer/tile.
+    * This involves converting the fragment colors from float[4] to the
+    * tile's specific format and obeying the quad/pixel mask.
+    */
+   {
+      const enum pipe_format color_format = cell->framebuffer.cbufs[0]->format;
+      int rgba_reg = spe_allocate_available_register(f);
+
+      /* Convert float[4] in [0.0,1.0] to int[4] in [0,~0], with clamping */
+      spe_cfltu(f, fragR_reg, fragR_reg, 32);
+      spe_cfltu(f, fragG_reg, fragG_reg, 32);
+      spe_cfltu(f, fragB_reg, fragB_reg, 32);
+      spe_cfltu(f, fragA_reg, fragA_reg, 32);
+
+      /* Shift most the significant bytes to least the significant positions.
+       * I.e.: reg = reg >> 24
+       */
+      spe_rotmi(f, fragR_reg, fragR_reg, -24);
+      spe_rotmi(f, fragG_reg, fragG_reg, -24);
+      spe_rotmi(f, fragB_reg, fragB_reg, -24);
+      spe_rotmi(f, fragA_reg, fragA_reg, -24);
+
+      /* Shift the color bytes according to the surface format */
+      if (color_format == PIPE_FORMAT_A8R8G8B8_UNORM) {
+         spe_roti(f, fragG_reg, fragG_reg, 8);   /* green <<= 8 */
+         spe_roti(f, fragR_reg, fragR_reg, 16);  /* red <<= 16 */
+         spe_roti(f, fragA_reg, fragA_reg, 24);  /* alpha <<= 24 */
+      }
+      else if (color_format == PIPE_FORMAT_B8G8R8A8_UNORM) {
+         spe_roti(f, fragR_reg, fragR_reg, 8);   /* red <<= 8 */
+         spe_roti(f, fragG_reg, fragG_reg, 16);  /* green <<= 16 */
+         spe_roti(f, fragB_reg, fragB_reg, 24);  /* blue <<= 24 */
+      }
+      else {
+         ASSERT(0);
+      }
+
+      /* Merge red, green, blue, alpha registers to make packed RGBA colors.
+       * Eg: after shifting according to color_format we might have:
+       *     R = {0x00ff0000, 0x00110000, 0x00220000, 0x00330000}
+       *     G = {0x0000ff00, 0x00004400, 0x00005500, 0x00006600}
+       *     B = {0x000000ff, 0x00000077, 0x00000088, 0x00000099}
+       *     A = {0xff000000, 0xaa000000, 0xbb000000, 0xcc000000}
+       * OR-ing all those together gives us four packed colors:
+       *  RGBA = {0xffffffff, 0xaa114477, 0xbb225588, 0xcc336699}
+       */
+      spe_or(f, rgba_reg, fragR_reg, fragG_reg);
+      spe_or(f, rgba_reg, rgba_reg, fragB_reg);
+      spe_or(f, rgba_reg, rgba_reg, fragA_reg);
+
+      /* Mix fragment colors with framebuffer colors using the quad/pixel mask:
+       * if (mask[i])
+       *    rgba[i] = rgba[i];
+       * else
+       *    rgba[i] = framebuffer[i];
+       */
+      spe_selb(f, rgba_reg, fbRGBA_reg, rgba_reg, mask_reg);
+
+      /* Store updated quad in tile:
+       * memory[color_tile + quad_offset] = rgba_reg;
+       */
+      spe_stqx(f, rgba_reg, color_tile_reg, quad_offset_reg);
+
+      spe_release_register(f, rgba_reg);
+   }
+
+   printf("gen_fragment_ops nr instructions: %u\n", f->num_inst);
+
+   spe_bi(f, SPE_REG_RA, 0, 0);  /* return from function call */
+
+
+   spe_release_register(f, fbRGBA_reg);
+   spe_release_register(f, fbZS_reg);
+   spe_release_register(f, quad_offset_reg);
+}
+
diff --git a/src/gallium/drivers/cell/ppu/cell_gen_fragment.h b/src/gallium/drivers/cell/ppu/cell_gen_fragment.h
new file mode 100644
index 00000000000..0ea0fc690c8
--- /dev/null
+++ b/src/gallium/drivers/cell/ppu/cell_gen_fragment.h
@@ -0,0 +1,38 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+#ifndef CELL_GEN_FRAGMENT_H
+#define CELL_GEN_FRAGMENT_H
+
+
+extern void
+gen_fragment_function(struct cell_context *cell, struct spe_function *f);
+
+
+#endif /* CELL_GEN_FRAGMENT_H */
+
diff --git a/src/gallium/drivers/cell/ppu/cell_state_emit.c b/src/gallium/drivers/cell/ppu/cell_state_emit.c
index f2feaa329ab..06777aac14c 100644
--- a/src/gallium/drivers/cell/ppu/cell_state_emit.c
+++ b/src/gallium/drivers/cell/ppu/cell_state_emit.c
@@ -27,6 +27,7 @@
 
 #include "util/u_memory.h"
 #include "cell_context.h"
+#include "cell_gen_fragment.h"
 #include "cell_state.h"
 #include "cell_state_emit.h"
 #include "cell_state_per_fragment.h"
@@ -83,6 +84,29 @@ cell_emit_state(struct cell_context *cell)
       fb->depth_format = zbuf ? zbuf->format : PIPE_FORMAT_NONE;
       fb->width = cell->framebuffer.width;
       fb->height = cell->framebuffer.height;
+#if 0
+      printf("EMIT color format %s\n", pf_name(fb->color_format));
+      printf("EMIT depth format %s\n", pf_name(fb->depth_format));
+#endif
+   }
+
+
+   if (cell->dirty & (CELL_NEW_FRAMEBUFFER | CELL_NEW_DEPTH_STENCIL)) {
+      /* XXX we don't want to always do codegen here.  We should have
+       * a hash/lookup table to cache previous results...
+       */
+      struct cell_command_fragment_ops *fops
+            = cell_batch_alloc(cell, sizeof(*fops));
+      struct spe_function spe_code;
+
+      /* generate new code */
+      gen_fragment_function(cell, &spe_code);
+      /* put the new code into the batch buffer */
+      fops->opcode = CELL_CMD_STATE_FRAGMENT_OPS;
+      memcpy(&fops->code, spe_code.store,
+             SPU_MAX_FRAGMENT_OPS_INSTS * SPE_INST_SIZE);
+      /* free codegen buffer */
+      spe_release_func(&spe_code);
    }
 
    if (cell->dirty & CELL_NEW_BLEND) {
@@ -90,8 +114,7 @@ cell_emit_state(struct cell_context *cell)
 
       if (cell->blend != NULL) {
          blend.base = (intptr_t) cell->blend->code.store;
-         blend.size = (char *) cell->blend->code.csr
-             - (char *) cell->blend->code.store;
+         blend.size = cell->blend->code.num_inst * SPE_INST_SIZE;
          blend.read_fb = TRUE;
       }
       else {
@@ -108,10 +131,10 @@ cell_emit_state(struct cell_context *cell)
 
       if (cell->depth_stencil != NULL) {
 	 dsat.base = (intptr_t) cell->depth_stencil->code.store;
-	 dsat.size = (char *) cell->depth_stencil->code.csr
-	     - (char *) cell->depth_stencil->code.store;
+	 dsat.size = cell->depth_stencil->code.num_inst * SPE_INST_SIZE;
 	 dsat.read_depth = TRUE;
 	 dsat.read_stencil = FALSE;
+         dsat.state = cell->depth_stencil->base;
       }
       else {
 	 dsat.base = 0;
diff --git a/src/gallium/drivers/cell/ppu/cell_state_per_fragment.c b/src/gallium/drivers/cell/ppu/cell_state_per_fragment.c
index 705867107bf..78cb446c14a 100644
--- a/src/gallium/drivers/cell/ppu/cell_state_per_fragment.c
+++ b/src/gallium/drivers/cell/ppu/cell_state_per_fragment.c
@@ -1158,7 +1158,7 @@ cell_generate_alpha_blend(struct cell_blend_state *cb)
 static int
 PC_OFFSET(const struct spe_function *f, const void *d)
 {
-   const intptr_t pc = (intptr_t) f->csr;
+   const intptr_t pc = (intptr_t) &f->store[f->num_inst];
    const intptr_t ea = ~0x0f & (intptr_t) d;
 
    return (ea - pc) >> 2;
diff --git a/src/gallium/drivers/cell/spu/Makefile b/src/gallium/drivers/cell/spu/Makefile
index d49abb2e825..e285ae9fdba 100644
--- a/src/gallium/drivers/cell/spu/Makefile
+++ b/src/gallium/drivers/cell/spu/Makefile
@@ -43,7 +43,7 @@ INCLUDE_DIRS = \
 	$(SPU_CC) $(SPU_CFLAGS) -c $<
 
 .c.s:
-	$(SPU_CC) $(SPU_CFLAGS) -S $<
+	$(SPU_CC) $(SPU_CFLAGS) -O3 -S $<
 
 
 # The .a file will be linked into the main/PPU executable
diff --git a/src/gallium/drivers/cell/spu/spu_main.c b/src/gallium/drivers/cell/spu/spu_main.c
index c4236817a9a..4e0ec159250 100644
--- a/src/gallium/drivers/cell/spu/spu_main.c
+++ b/src/gallium/drivers/cell/spu/spu_main.c
@@ -34,6 +34,7 @@
 
 #include "spu_main.h"
 #include "spu_render.h"
+#include "spu_per_fragment_op.h"
 #include "spu_texture.h"
 #include "spu_tile.h"
 //#include "spu_test.h"
@@ -46,7 +47,7 @@
 /*
 helpful headers:
 /usr/lib/gcc/spu/4.1.1/include/spu_mfcio.h
-/opt/ibm/cell-sdk/prototype/sysroot/usr/include/libmisc.h
+/opt/cell/sdk/usr/include/libmisc.h
 */
 
 boolean Debug = FALSE;
@@ -226,6 +227,24 @@ cmd_release_verts(const struct cell_command_release_verts *release)
 }
 
 
+/**
+ * Process a CELL_CMD_STATE_FRAGMENT_OPS command.
+ * This involves installing new fragment ops SPU code.
+ * If this function is never called, we'll use a regular C fallback function
+ * for fragment processing.
+ */
+static void
+cmd_state_fragment_ops(const struct cell_command_fragment_ops *fops)
+{
+   if (Debug)
+      printf("SPU %u: CMD_STATE_FRAGMENT_OPS\n", spu.init.id);
+   /* Copy SPU code from batch buffer to spu buffer */
+   memcpy(spu.fragment_ops.code, fops->code, SPU_MAX_FRAGMENT_OPS_INSTS * 4);
+   /* Point function pointer at new code */
+   spu.fragment_ops.func = (spu_fragment_ops_func) spu.fragment_ops.code;
+}
+
+
 static void
 cmd_state_framebuffer(const struct cell_command_framebuffer *cmd)
 {
@@ -257,6 +276,8 @@ cmd_state_framebuffer(const struct cell_command_framebuffer *cmd)
       break;
    case PIPE_FORMAT_Z24S8_UNORM:
    case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_Z24X8_UNORM:
+   case PIPE_FORMAT_X8Z24_UNORM:
       spu.fb.zsize = 4;
       spu.fb.zscale = (float) 0x00ffffffu;
       break;
@@ -282,6 +303,8 @@ cmd_state_framebuffer(const struct cell_command_framebuffer *cmd)
 }
 
 
+#define NEW_FRAGMENT_FUNCTION 01
+
 static void
 cmd_state_blend(const struct cell_command_blend *state)
 {
@@ -302,7 +325,9 @@ cmd_state_blend(const struct cell_command_blend *state)
       wait_on_mask(1 << TAG_BATCH_BUFFER);
       spu.blend = (blend_func) fb_blend_code_buffer;
       spu.read_fb = state->read_fb;
-   } else {
+   }
+   else
+   {
       spu.read_fb = FALSE;
    }
 }
@@ -326,7 +351,9 @@ cmd_state_depth_stencil(const struct cell_command_depth_stencil_alpha_test *stat
 	      0, /* tid */
 	      0  /* rid */);
       wait_on_mask(1 << TAG_BATCH_BUFFER);
-   } else {
+   }
+   else
+   {
       /* If there is no code, emit a return instruction.
        */
       depth_stencil_code_buffer[0] = 0x35;
@@ -338,12 +365,14 @@ cmd_state_depth_stencil(const struct cell_command_depth_stencil_alpha_test *stat
    spu.frag_test = (frag_test_func) depth_stencil_code_buffer;
    spu.read_depth = state->read_depth;
    spu.read_stencil = state->read_stencil;
+   spu.depth_stencil_alpha = state->state;
 }
 
 
 static void
 cmd_state_logicop(const struct cell_command_logicop * code)
 {
+#if !NEW_FRAGMENT_FUNCTION
    mfc_get(logicop_code_buffer,
            (unsigned int) code->base,  /* src */
            code->size,
@@ -353,6 +382,7 @@ cmd_state_logicop(const struct cell_command_logicop * code)
    wait_on_mask(1 << TAG_BATCH_BUFFER);
 
    spu.logicop = (logicop_func) logicop_code_buffer;
+#endif
 }
 
 
@@ -455,7 +485,9 @@ cmd_finish(void)
 
 
 /**
- * Execute a batch of commands
+ * Execute a batch of commands which was sent to us by the PPU.
+ * See the cell_emit_state.c code to see where the commands come from.
+ *
  * The opcode param encodes the location of the buffer and its size.
  */
 static void
@@ -519,6 +551,14 @@ cmd_batch(uint opcode)
             pos += pos_incr;
          }
          break;
+      case CELL_CMD_STATE_FRAGMENT_OPS:
+         {
+            struct cell_command_fragment_ops *fops
+               = (struct cell_command_fragment_ops *) &buffer[pos];
+            cmd_state_fragment_ops(fops);
+            pos += sizeof(*fops) / 8;
+         }
+         break;
       case CELL_CMD_RELEASE_VERTS:
          {
             struct cell_command_release_verts *release
@@ -680,6 +720,11 @@ one_time_init(void)
    memset(spu.ctile_status, TILE_STATUS_DEFINED, sizeof(spu.ctile_status));
    memset(spu.ztile_status, TILE_STATUS_DEFINED, sizeof(spu.ztile_status));
    invalidate_tex_cache();
+
+   /* Install default/fallback fragment processing function.
+    * This will normally be overriden by a code-gen'd function.
+    */
+   spu.fragment_ops.func = spu_fallback_fragment_ops;
 }
 
 
diff --git a/src/gallium/drivers/cell/spu/spu_main.h b/src/gallium/drivers/cell/spu/spu_main.h
index c2a53c9dcf6..7ab34f52229 100644
--- a/src/gallium/drivers/cell/spu/spu_main.h
+++ b/src/gallium/drivers/cell/spu/spu_main.h
@@ -91,6 +91,24 @@ typedef struct spu_blend_results (*logicop_func)(
 
 typedef vector float (*sample_texture_func)(uint unit, vector float texcoord);
 
+
+typedef void (*spu_fragment_ops_func)(uint x, uint y,
+                                      tile_t *colorTile,
+                                      tile_t *depthStencilTile,
+                                      vector float fragZ,
+                                      vector float fragRed,
+                                      vector float fragGreen,
+                                      vector float fragBlue,
+                                      vector float fragAlpha,
+                                      vector unsigned int mask);
+
+struct spu_fragment_ops
+{
+   uint code[SPU_MAX_FRAGMENT_OPS_INSTS];
+   spu_fragment_ops_func func;  /**< Current fragment ops function */
+} ALIGN16_ATTRIB;
+
+
 struct spu_framebuffer {
    void *color_start;              /**< addr of color surface in main memory */
    void *depth_start;              /**< addr of depth surface in main memory */
@@ -127,6 +145,9 @@ struct spu_global
    struct cell_init_info init;
 
    struct spu_framebuffer fb;
+
+   struct pipe_depth_stencil_alpha_state depth_stencil_alpha;
+
    boolean read_depth;
    boolean read_stencil;
    frag_test_func frag_test;  /**< Current depth/stencil test code */
@@ -142,6 +163,8 @@ struct spu_global
 
    struct vertex_info vertex_info;
 
+   struct spu_fragment_ops fragment_ops;
+
    /* XXX more state to come */
 
 
diff --git a/src/gallium/drivers/cell/spu/spu_per_fragment_op.c b/src/gallium/drivers/cell/spu/spu_per_fragment_op.c
index 29dc07a2e8d..ffc596aa622 100644
--- a/src/gallium/drivers/cell/spu/spu_per_fragment_op.c
+++ b/src/gallium/drivers/cell/spu/spu_per_fragment_op.c
@@ -29,8 +29,11 @@
  * \author Ian Romanick <idr@us.ibm.com>
  */
 
+
+#include <transpose_matrix4x4.h>
 #include "pipe/p_format.h"
 #include "spu_main.h"
+#include "spu_colorpack.h"
 #include "spu_per_fragment_op.h"
 
 #define ZERO 0x80
@@ -90,7 +93,8 @@ read_ds_quad(tile_t *tile, unsigned x, unsigned y,
       break;
    }
 
-   case PIPE_FORMAT_S8Z24_UNORM: {
+   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_X8Z24_UNORM: {
       qword *ptr = (qword *) &tile->ui4[iy][ix];
 
       *depth = si_and(*ptr, si_fsmbi(0x7777));
@@ -153,7 +157,8 @@ write_ds_quad(tile_t *buffer, unsigned x, unsigned y,
       break;
    }
 
-   case PIPE_FORMAT_S8Z24_UNORM: {
+   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_X8Z24_UNORM: {
       qword *ptr = (qword *) &buffer->ui4[iy][ix];
       /* form select mask = 0111,0111,0111,0111 */
       qword mask = si_fsmbi(0x7777);
@@ -217,3 +222,225 @@ spu_do_depth_stencil(int x, int y,
 
    return result.mask;
 }
+
+
+
+
+/**
+ * Called by rasterizer for each quad after the shader has run.  This
+ * is a fallback/debug function.  In reality we'll use a generated
+ * function produced by the PPU.  But this function is useful for
+ * debug/validation.
+ */
+void
+spu_fallback_fragment_ops(uint x, uint y,
+                          tile_t *colorTile,
+                          tile_t *depthStencilTile,
+                          vector float fragZ,
+                          vector float fragRed,
+                          vector float fragGreen,
+                          vector float fragBlue,
+                          vector float fragAlpha,
+                          vector unsigned int mask)
+{
+   vector float frag_soa[4], frag_aos[4];
+   unsigned int c0, c1, c2, c3;
+
+   /* do alpha test */
+   if (spu.depth_stencil_alpha.alpha.enabled) {
+      vector float ref = spu_splats(spu.depth_stencil_alpha.alpha.ref);
+      vector unsigned int amask;
+
+      switch (spu.depth_stencil_alpha.alpha.func) {
+      case PIPE_FUNC_LESS:
+         amask = spu_cmpgt(ref, fragAlpha);  /* mask = (fragAlpha < ref) */
+         break;
+      case PIPE_FUNC_GREATER:
+         amask = spu_cmpgt(fragAlpha, ref);  /* mask = (fragAlpha > ref) */
+         break;
+      case PIPE_FUNC_GEQUAL:
+         amask = spu_cmpgt(ref, fragAlpha);
+         amask = spu_nor(amask, amask);
+         break;
+      case PIPE_FUNC_LEQUAL:
+         amask = spu_cmpgt(fragAlpha, ref);
+         amask = spu_nor(amask, amask);
+         break;
+      case PIPE_FUNC_EQUAL:
+         amask = spu_cmpeq(ref, fragAlpha);
+         break;
+      case PIPE_FUNC_NOTEQUAL:
+         amask = spu_cmpeq(ref, fragAlpha);
+         amask = spu_nor(amask, amask);
+         break;
+      case PIPE_FUNC_ALWAYS:
+         amask = spu_splats(0xffffffffU);
+         break;
+      case PIPE_FUNC_NEVER:
+         amask = spu_splats( 0x0U);
+         break;
+      default:
+         ;
+      }
+
+      mask = spu_and(mask, amask);
+   }
+
+   /* Z and/or stencil testing... */
+   if (spu.depth_stencil_alpha.depth.enabled ||
+       spu.depth_stencil_alpha.stencil[0].enabled) {
+
+      /* get four Z/Stencil values from tile */
+      vector unsigned int mask24 = spu_splats((unsigned int)0x00ffffffU);
+      vector unsigned int ifbZS = depthStencilTile->ui4[y/2][x/2];
+      vector unsigned int ifbZ = spu_and(ifbZS, mask24);
+      vector unsigned int ifbS = spu_andc(ifbZS, mask24);
+
+      if (spu.depth_stencil_alpha.stencil[0].enabled) {
+         /* do stencil test */
+         ASSERT(spu.fb.depth_format == PIPE_FORMAT_S8Z24_UNORM);
+
+      }
+      else if (spu.depth_stencil_alpha.depth.enabled) {
+         /* do depth test */
+
+         ASSERT(spu.fb.depth_format == PIPE_FORMAT_S8Z24_UNORM ||
+                spu.fb.depth_format == PIPE_FORMAT_X8Z24_UNORM);
+
+         vector unsigned int ifragZ;
+         vector unsigned int zmask;
+
+         /* convert four fragZ from float to uint */
+         fragZ = spu_mul(fragZ, spu_splats((float) 0xffffff));
+         ifragZ = spu_convtu(fragZ, 0);
+
+         /* do depth comparison, setting zmask with results */
+         switch (spu.depth_stencil_alpha.depth.func) {
+         case PIPE_FUNC_LESS:
+            zmask = spu_cmpgt(ifbZ, ifragZ);  /* mask = (ifragZ < ifbZ) */
+            break;
+         case PIPE_FUNC_GREATER:
+            zmask = spu_cmpgt(ifragZ, ifbZ);  /* mask = (ifbZ > ifragZ) */
+            break;
+         case PIPE_FUNC_GEQUAL:
+            zmask = spu_cmpgt(ifbZ, ifragZ);
+            zmask = spu_nor(zmask, zmask);
+            break;
+         case PIPE_FUNC_LEQUAL:
+            zmask = spu_cmpgt(ifragZ, ifbZ);
+            zmask = spu_nor(zmask, zmask);
+            break;
+         case PIPE_FUNC_EQUAL:
+            zmask = spu_cmpeq(ifbZ, ifragZ);
+            break;
+         case PIPE_FUNC_NOTEQUAL:
+            zmask = spu_cmpeq(ifbZ, ifragZ);
+            zmask = spu_nor(zmask, zmask);
+            break;
+         case PIPE_FUNC_ALWAYS:
+            zmask = spu_splats(0xffffffffU);
+            break;
+         case PIPE_FUNC_NEVER:
+            zmask = spu_splats( 0x0U);
+            break;
+         default:
+            ;
+         }
+
+         mask = spu_and(mask, zmask);
+
+         /* merge framebuffer Z and fragment Z according to the mask */
+         ifbZ = spu_or(spu_and(ifragZ, mask),
+                       spu_andc(ifbZ, mask));
+      }
+
+      if (spu_extract(spu_orx(mask), 0)) {
+         /* put new fragment Z/Stencil values back into Z/Stencil tile */
+         depthStencilTile->ui4[y/2][x/2] = spu_or(ifbZ, ifbS);
+
+         spu.cur_ztile_status = TILE_STATUS_DIRTY;
+      }
+   }
+
+   /* XXX do blending here */
+
+   /* XXX do colormask test here */
+
+
+   if (spu_extract(spu_orx(mask), 0)) {
+      spu.cur_ctile_status = TILE_STATUS_DIRTY;
+   }
+   else {
+      return;
+   }
+
+   /* convert RRRR,GGGG,BBBB,AAAA to RGBA,RGBA,RGBA,RGBA */
+#if 0
+   {
+      vector float frag_soa[4];
+      frag_soa[0] = fragRed;
+      frag_soa[1] = fragGreen;
+      frag_soa[2] = fragBlue;
+      frag_soa[3] = fragAlpha;
+      _transpose_matrix4x4(frag_aos, frag_soa);
+   }
+#else
+   /* short-cut relying on function parameter layout: */
+   _transpose_matrix4x4(frag_aos, &fragRed);
+   (void) fragGreen;
+   (void) fragBlue;
+#endif
+
+   switch (spu.fb.color_format) {
+   case PIPE_FORMAT_A8R8G8B8_UNORM:
+      c0 = spu_pack_A8R8G8B8(frag_aos[0]);
+      c1 = spu_pack_A8R8G8B8(frag_aos[1]);
+      c2 = spu_pack_A8R8G8B8(frag_aos[2]);
+      c3 = spu_pack_A8R8G8B8(frag_aos[3]);
+      break;
+
+   case PIPE_FORMAT_B8G8R8A8_UNORM:
+      c0 = spu_pack_B8G8R8A8(frag_aos[0]);
+      c1 = spu_pack_B8G8R8A8(frag_aos[1]);
+      c2 = spu_pack_B8G8R8A8(frag_aos[2]);
+      c3 = spu_pack_B8G8R8A8(frag_aos[3]);
+      break;
+   default:
+      fprintf(stderr, "SPU: Bad pixel format in spu_default_fragment_ops\n");
+      ASSERT(0);
+   }
+
+#if 0
+   /*
+    * Quad layout:
+    *  +--+--+
+    *  |p0|p1|
+    *  +--+--+
+    *  |p2|p3|
+    *  +--+--+
+    */
+   if (spu_extract(mask, 0))
+      colorTile->ui[y+0][x+0] = c0;
+   if (spu_extract(mask, 1))
+      colorTile->ui[y+0][x+1] = c1;
+   if (spu_extract(mask, 2))
+      colorTile->ui[y+1][x+0] = c2;
+   if (spu_extract(mask, 3))
+      colorTile->ui[y+1][x+1] = c3;   
+#else
+   /*
+    * Quad layout:
+    *  +--+--+--+--+
+    *  |p0|p1|p2|p3|
+    *  +--+--+--+--+
+    */
+   if (spu_extract(mask, 0))
+      colorTile->ui[y][x*2] = c0;
+   if (spu_extract(mask, 1))
+      colorTile->ui[y][x*2+1] = c1;
+   if (spu_extract(mask, 2))
+      colorTile->ui[y][x*2+2] = c2;
+   if (spu_extract(mask, 3))
+      colorTile->ui[y][x*2+3] = c3;   
+#endif
+}
diff --git a/src/gallium/drivers/cell/spu/spu_per_fragment_op.h b/src/gallium/drivers/cell/spu/spu_per_fragment_op.h
index 65712586992..ffadf0661ca 100644
--- a/src/gallium/drivers/cell/spu/spu_per_fragment_op.h
+++ b/src/gallium/drivers/cell/spu/spu_per_fragment_op.h
@@ -29,4 +29,15 @@ extern qword
 spu_do_depth_stencil(int x, int y, qword frag_mask, qword frag_depth,
 		     qword frag_alpha, qword facing);
 
+extern void
+spu_fallback_fragment_ops(uint x, uint y,
+                          tile_t *colorTile,
+                          tile_t *depthStencilTile,
+                          vector float fragZ,
+                          vector float fragRed,
+                          vector float fragGreen,
+                          vector float fragBlue,
+                          vector float fragAlpha,
+                          vector unsigned int mask);
+
 #endif /* SPU_PER_FRAGMENT_OP */
diff --git a/src/gallium/drivers/cell/spu/spu_tri.c b/src/gallium/drivers/cell/spu/spu_tri.c
index a3ea0a3e69b..71ef6ca24fb 100644
--- a/src/gallium/drivers/cell/spu/spu_tri.c
+++ b/src/gallium/drivers/cell/spu/spu_tri.c
@@ -297,9 +297,12 @@ emit_quad( int x, int y, mask_t mask )
    sp->quad.first->run(sp->quad.first, &setup.quad);
 #else
 
+#define NEW_FRAGMENT_FUNCTION 01
+#if !NEW_FRAGMENT_FUNCTION
    if (spu.read_depth) {
       mask = do_depth_test(x, y, mask);
    }
+#endif
 
    /* If any bits in mask are set... */
    if (spu_extract(spu_orx(mask), 0)) {
@@ -308,6 +311,7 @@ emit_quad( int x, int y, mask_t mask )
       vector float colors[4];
 
       spu.cur_ctile_status = TILE_STATUS_DIRTY;
+      spu.cur_ztile_status = TILE_STATUS_DIRTY;
 
       if (spu.texture[0].start) {
          /* texture mapping */
@@ -355,6 +359,29 @@ emit_quad( int x, int y, mask_t mask )
       }
 
 
+#if NEW_FRAGMENT_FUNCTION
+      {
+         /* Convert fragment data from AoS to SoA format.
+          * I.e. (RGBA,RGBA,RGBA,RGBA) -> (RRRR,GGGG,BBBB,AAAA)
+          * This is temporary!
+          */
+         vector float soa_frag[4];
+         _transpose_matrix4x4(soa_frag, colors);
+
+         float4 fragZ;
+
+         fragZ.v = eval_z((float) x, (float) y);
+
+         /* Do all per-fragment/quad operations here, including:
+          *  alpha test, z test, stencil test, blend and framebuffer writing.
+          */
+         spu.fragment_ops.func(ix, iy, &spu.ctile, &spu.ztile,
+                               fragZ.v,
+                               soa_frag[0], soa_frag[1],
+                               soa_frag[2], soa_frag[3],
+                               mask);
+      }
+#else
       /* Convert fragment data from AoS to SoA format.
        * I.e. (RGBA,RGBA,RGBA,RGBA) -> (RRRR,GGGG,BBBB,AAAA)
        */
@@ -405,6 +432,9 @@ emit_quad( int x, int y, mask_t mask )
       spu.ctile.ui[iy+0][ix+1] = spu_extract((vec_uint4) result.g, 0);
       spu.ctile.ui[iy+1][ix+0] = spu_extract((vec_uint4) result.b, 0);
       spu.ctile.ui[iy+1][ix+1] = spu_extract((vec_uint4) result.a, 0);
+
+#endif /* NEW_FRAGMENT_FUNCTION */
+
    }
 #endif
 }
diff --git a/src/gallium/winsys/xlib/xm_api.c b/src/gallium/winsys/xlib/xm_api.c
index b0105131071..28bd6ceab42 100644
--- a/src/gallium/winsys/xlib/xm_api.c
+++ b/src/gallium/winsys/xlib/xm_api.c
@@ -349,12 +349,17 @@ create_xmesa_buffer(XMesaDrawable d, BufferType type,
 
    if (vis->mesa_visual.depthBits == 0)
       depthFormat = PIPE_FORMAT_NONE;
+#ifdef GALLIUM_CELL /* XXX temporary for Cell! */
+   else
+      depthFormat = PIPE_FORMAT_S8Z24_UNORM;
+#else
    else if (vis->mesa_visual.depthBits <= 16)
-      depthFormat = PIPE_FORMAT_Z16_UNORM;
+      depthFormat = PIPE_FORMAT_Z16UNORM;
    else if (vis->mesa_visual.depthBits <= 24)
       depthFormat = PIPE_FORMAT_S8Z24_UNORM;
    else
       depthFormat = PIPE_FORMAT_Z32_UNORM;
+#endif
 
    if (vis->mesa_visual.stencilBits == 8) {
       if (depthFormat == PIPE_FORMAT_S8Z24_UNORM)
diff --git a/src/gallium/winsys/xlib/xm_winsys.c b/src/gallium/winsys/xlib/xm_winsys.c
index 5e9a1f92f11..c4a30d37023 100644
--- a/src/gallium/winsys/xlib/xm_winsys.c
+++ b/src/gallium/winsys/xlib/xm_winsys.c
@@ -275,6 +275,39 @@ xm_buffer_destroy(struct pipe_winsys *pws,
 }
 
 
+/**
+ * For Cell.  Basically, rearrange the pixels/quads from this layout:
+ *  +--+--+--+--+
+ *  |p0|p1|p2|p3|....
+ *  +--+--+--+--+
+ *
+ * to this layout:
+ *  +--+--+
+ *  |p0|p1|....
+ *  +--+--+
+ *  |p2|p3|
+ *  +--+--+
+ */
+static void
+twiddle_tile(uint *tile)
+{
+   uint tile2[TILE_SIZE * TILE_SIZE];
+   int y, x;
+
+   for (y = 0; y < TILE_SIZE; y+=2) {
+      for (x = 0; x < TILE_SIZE; x+=2) {
+         int k = 4 * (y/2 * TILE_SIZE/2 + x/2);
+         tile2[y * TILE_SIZE + (x + 0)] = tile[k];
+         tile2[y * TILE_SIZE + (x + 1)] = tile[k+1];
+         tile2[(y + 1) * TILE_SIZE + (x + 0)] = tile[k+2];
+         tile2[(y + 1) * TILE_SIZE + (x + 1)] = tile[k+3];
+      }
+   }
+   memcpy(tile, tile2, sizeof(tile2));
+}
+
+
+
 /**
  * Display a surface that's in a tiled configuration.  That is, all the
  * pixels for a TILE_SIZExTILE_SIZE block are contiguous in memory.
@@ -321,6 +354,8 @@ xmesa_display_surface_tiled(XMesaBuffer b, const struct pipe_surface *surf)
 
          ximage->data = (char *) xm_buf->data + offset;
 
+         twiddle_tile((uint *) ximage->data);
+
          if (XSHM_ENABLED(xm_buf)) {
 #if defined(USE_XSHM) && !defined(XFree86Server)
             XShmPutImage(b->xm_visual->display, b->drawable, b->gc,