18 files changed, 254 insertions, 264 deletions

diff --git a/src/gallium/drivers/cell/common.h b/src/gallium/drivers/cell/common.h
index d5f5c7bbba8..aa29dcb3947 100644
--- a/src/gallium/drivers/cell/common.h
+++ b/src/gallium/drivers/cell/common.h
@@ -358,6 +358,7 @@ struct cell_spu_function_info
 
 
 /** This is the object passed to spe_create_thread() */
+PIPE_ALIGN_TYPE(16,
 struct cell_init_info
 {
    unsigned id;
@@ -370,7 +371,7 @@ struct cell_init_info
    uint *buffer_status;  /**< points at cell_context->buffer_status */
 
    struct cell_spu_function_info *spu_functions;
-} ALIGN16_ATTRIB;
+});
 
 
 #endif /* CELL_COMMON_H */
diff --git a/src/gallium/drivers/cell/ppu/cell_clear.c b/src/gallium/drivers/cell/ppu/cell_clear.c
index 79ad687ea94..3a3f968a492 100644
--- a/src/gallium/drivers/cell/ppu/cell_clear.c
+++ b/src/gallium/drivers/cell/ppu/cell_clear.c
@@ -59,9 +59,9 @@ cell_clear(struct pipe_context *pipe, unsigned buffers, const float *rgba,
 
    if (buffers & PIPE_CLEAR_COLOR) {
       uint surfIndex = 0;
-      uint clearValue;
+      union util_color uc;
 
-      util_pack_color(rgba, cell->framebuffer.cbufs[0]->format, &clearValue);
+      util_pack_color(rgba, cell->framebuffer.cbufs[0]->format, &uc);
 
       /* Build a CLEAR command and place it in the current batch buffer */
       STATIC_ASSERT(sizeof(struct cell_command_clear_surface) % 16 == 0);
@@ -70,7 +70,7 @@ cell_clear(struct pipe_context *pipe, unsigned buffers, const float *rgba,
          cell_batch_alloc16(cell, sizeof(*clr));
       clr->opcode[0] = CELL_CMD_CLEAR_SURFACE;
       clr->surface = surfIndex;
-      clr->value = clearValue;
+      clr->value = uc.ui;
    }
 
    if (buffers & PIPE_CLEAR_DEPTHSTENCIL) {
diff --git a/src/gallium/drivers/cell/ppu/cell_context.h b/src/gallium/drivers/cell/ppu/cell_context.h
index 5c3188e7f9d..3fb6a3227c7 100644
--- a/src/gallium/drivers/cell/ppu/cell_context.h
+++ b/src/gallium/drivers/cell/ppu/cell_context.h
@@ -89,7 +89,7 @@ struct cell_buffer_node;
  */
 struct cell_buffer_list
 {
-   struct cell_fence fence ALIGN16_ATTRIB;
+   PIPE_ALIGN_VAR(16) struct cell_fence fence;
    struct cell_buffer_node *head;
 };
 
@@ -150,18 +150,18 @@ struct cell_context
    /** Mapped constant buffers */
    void *mapped_constants[PIPE_SHADER_TYPES];
 
-   struct cell_spu_function_info spu_functions ALIGN16_ATTRIB;
+   PIPE_ALIGN_VAR(16) struct cell_spu_function_info spu_functions;
 
    uint num_cells, num_spus;
 
    /** Buffers for command batches, vertex/index data */
    uint buffer_size[CELL_NUM_BUFFERS];
-   ubyte buffer[CELL_NUM_BUFFERS][CELL_BUFFER_SIZE] ALIGN16_ATTRIB;
+   PIPE_ALIGN_VAR(16) ubyte buffer[CELL_NUM_BUFFERS][CELL_BUFFER_SIZE];
 
    int cur_batch;  /**< which buffer is being filled w/ commands */
 
    /** [4] to ensure 16-byte alignment for each status word */
-   uint buffer_status[CELL_MAX_SPUS][CELL_NUM_BUFFERS][4] ALIGN16_ATTRIB;
+   PIPE_ALIGN_VAR(16) uint buffer_status[CELL_MAX_SPUS][CELL_NUM_BUFFERS][4];
 
 
    /** Associated with each command/batch buffer is a list of pipe_buffers
diff --git a/src/gallium/drivers/cell/ppu/cell_draw_arrays.c b/src/gallium/drivers/cell/ppu/cell_draw_arrays.c
index 644496db40c..3fa8b975d39 100644
--- a/src/gallium/drivers/cell/ppu/cell_draw_arrays.c
+++ b/src/gallium/drivers/cell/ppu/cell_draw_arrays.c
@@ -59,7 +59,7 @@ cell_map_constant_buffers(struct cell_context *sp)
       }
    }
 
-   draw_set_mapped_constant_buffer(sp->draw,
+   draw_set_mapped_constant_buffer(sp->draw, PIPE_SHADER_VERTEX,
                                    sp->mapped_constants[PIPE_SHADER_VERTEX],
                                    sp->constants[PIPE_SHADER_VERTEX].buffer->size);
 }
@@ -85,7 +85,7 @@ cell_unmap_constant_buffers(struct cell_context *sp)
  *
  * XXX should the element buffer be specified/bound with a separate function?
  */
-static boolean
+static void
 cell_draw_range_elements(struct pipe_context *pipe,
                          struct pipe_buffer *indexBuffer,
                          unsigned indexSize,
@@ -145,47 +145,35 @@ cell_draw_range_elements(struct pipe_context *pipe,
 
    /* Note: leave drawing surfaces mapped */
    cell_unmap_constant_buffers(sp);
-
-   return TRUE;
 }
 
 
-static boolean
+static void
 cell_draw_elements(struct pipe_context *pipe,
                    struct pipe_buffer *indexBuffer,
                    unsigned indexSize,
                    unsigned mode, unsigned start, unsigned count)
 {
-   return cell_draw_range_elements( pipe, indexBuffer,
-                                    indexSize,
-                                    0, 0xffffffff,
-                                    mode, start, count );
+   cell_draw_range_elements( pipe, indexBuffer,
+                             indexSize,
+                             0, 0xffffffff,
+                             mode, start, count );
 }
 
 
-static boolean
+static void
 cell_draw_arrays(struct pipe_context *pipe, unsigned mode,
                      unsigned start, unsigned count)
 {
-   return cell_draw_elements(pipe, NULL, 0, mode, start, count);
-}
-
-
-static void
-cell_set_edgeflags(struct pipe_context *pipe, const unsigned *edgeflags)
-{
-   struct cell_context *cell = cell_context(pipe);
-   draw_set_edgeflags(cell->draw, edgeflags);
+   cell_draw_elements(pipe, NULL, 0, mode, start, count);
 }
 
 
-
 void
 cell_init_draw_functions(struct cell_context *cell)
 {
    cell->pipe.draw_arrays = cell_draw_arrays;
    cell->pipe.draw_elements = cell_draw_elements;
    cell->pipe.draw_range_elements = cell_draw_range_elements;
-   cell->pipe.set_edgeflags = cell_set_edgeflags;
 }
 
diff --git a/src/gallium/drivers/cell/ppu/cell_gen_fp.c b/src/gallium/drivers/cell/ppu/cell_gen_fp.c
index 19e3ab08440..1d8a11a4ac9 100644
--- a/src/gallium/drivers/cell/ppu/cell_gen_fp.c
+++ b/src/gallium/drivers/cell/ppu/cell_gen_fp.c
@@ -237,8 +237,8 @@ is_register_src(struct codegen *gen, int channel,
    if (swizzle > TGSI_SWIZZLE_W || sign_op != TGSI_UTIL_SIGN_KEEP) {
       return FALSE;
    }
-   if (src->SrcRegister.File == TGSI_FILE_TEMPORARY ||
-       src->SrcRegister.File == TGSI_FILE_IMMEDIATE) {
+   if (src->Register.File == TGSI_FILE_TEMPORARY ||
+       src->Register.File == TGSI_FILE_IMMEDIATE) {
       return TRUE;
    }
    return FALSE;
@@ -249,7 +249,7 @@ static boolean
 is_memory_dst(struct codegen *gen, int channel,
               const struct tgsi_full_dst_register *dst)
 {
-   if (dst->DstRegister.File == TGSI_FILE_OUTPUT) {
+   if (dst->Register.File == TGSI_FILE_OUTPUT) {
       return TRUE;
    }
    else {
@@ -279,15 +279,15 @@ get_src_reg(struct codegen *gen,
    assert(swizzle <= TGSI_SWIZZLE_W);
 
    {
-      int index = src->SrcRegister.Index;
+      int index = src->Register.Index;
 
       assert(swizzle < 4);
 
-      if (src->SrcRegister.Indirect) {
+      if (src->Register.Indirect) {
          /* XXX unfinished */
       }
 
-      switch (src->SrcRegister.File) {
+      switch (src->Register.File) {
       case TGSI_FILE_TEMPORARY:
          reg = gen->temp_regs[index][swizzle];
          break;
@@ -374,12 +374,12 @@ get_dst_reg(struct codegen *gen,
 {
    int reg = -1;
 
-   switch (dest->DstRegister.File) {
+   switch (dest->Register.File) {
    case TGSI_FILE_TEMPORARY:
       if (gen->if_nesting > 0 || gen->loop_nesting > 0)
          reg = get_itemp(gen);
       else
-         reg = gen->temp_regs[dest->DstRegister.Index][channel];
+         reg = gen->temp_regs[dest->Register.Index][channel];
       break;
    case TGSI_FILE_OUTPUT:
       reg = get_itemp(gen);
@@ -419,10 +419,10 @@ store_dest_reg(struct codegen *gen,
    }
 #endif
 
-   switch (dest->DstRegister.File) {
+   switch (dest->Register.File) {
    case TGSI_FILE_TEMPORARY:
       if (gen->if_nesting > 0 || gen->loop_nesting > 0) {
-         int d_reg = gen->temp_regs[dest->DstRegister.Index][channel];
+         int d_reg = gen->temp_regs[dest->Register.Index][channel];
          int exec_reg = get_exec_mask_reg(gen);
          /* Mix d with new value according to exec mask:
           * d[i] = mask_reg[i] ? value_reg : d_reg
@@ -437,7 +437,7 @@ store_dest_reg(struct codegen *gen,
    case TGSI_FILE_OUTPUT:
       {
          /* offset is measured in quadwords, not bytes */
-         int offset = dest->DstRegister.Index * 4 + channel;
+         int offset = dest->Register.Index * 4 + channel;
          if (gen->if_nesting > 0 || gen->loop_nesting > 0) {
             int exec_reg = get_exec_mask_reg(gen);
             int curval_reg = get_itemp(gen);
@@ -544,7 +544,7 @@ emit_epilogue(struct codegen *gen)
 
 #define FOR_EACH_ENABLED_CHANNEL(inst, ch) \
    for (ch = 0; ch < 4; ch++) \
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch))
+      if (inst->Dst[0].Register.WriteMask & (1 << ch))
 
 
 static boolean
@@ -552,7 +552,7 @@ emit_ARL(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
    int ch = 0, src_reg, addr_reg;
 
-   src_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+   src_reg = get_src_reg(gen, ch, &inst->Src[0]);
    addr_reg = get_address_reg(gen);
 
    /* convert float to int */
@@ -570,19 +570,19 @@ emit_MOV(struct codegen *gen, const struct tgsi_full_instruction *inst)
    int ch, src_reg[4], dst_reg[4];
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      src_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-      dst_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      src_reg[ch] = get_src_reg(gen, ch, &inst->Src[0]);
+      dst_reg[ch] = get_dst_reg(gen, ch, &inst->Dst[0]);
    }
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      if (is_register_src(gen, ch, &inst->FullSrcRegisters[0]) &&
-          is_memory_dst(gen, ch, &inst->FullDstRegisters[0])) {
+      if (is_register_src(gen, ch, &inst->Src[0]) &&
+          is_memory_dst(gen, ch, &inst->Dst[0])) {
          /* special-case: register to memory store */
-         store_dest_reg(gen, src_reg[ch], ch, &inst->FullDstRegisters[0]);
+         store_dest_reg(gen, src_reg[ch], ch, &inst->Dst[0]);
       }
       else {
          spe_move(gen->f, dst_reg[ch], src_reg[ch]);
-         store_dest_reg(gen, dst_reg[ch], ch, &inst->FullDstRegisters[0]);
+         store_dest_reg(gen, dst_reg[ch], ch, &inst->Dst[0]);
       }
    }
 
@@ -601,9 +601,9 @@ emit_binop(struct codegen *gen, const struct tgsi_full_instruction *inst)
 
    /* Loop over Red/Green/Blue/Alpha channels, fetch src operands */
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-      s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->Src[0]);
+      s2_reg[ch] = get_src_reg(gen, ch, &inst->Src[1]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->Dst[0]);
    }
 
    /* Loop over Red/Green/Blue/Alpha channels, do the op, store results */
@@ -626,7 +626,7 @@ emit_binop(struct codegen *gen, const struct tgsi_full_instruction *inst)
 
    /* Store the result (a no-op for TGSI_FILE_TEMPORARY dests) */
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_reg[ch], ch, &inst->Dst[0]);
    }
 
    /* Free any intermediate temps we allocated */
@@ -645,16 +645,16 @@ emit_MAD(struct codegen *gen, const struct tgsi_full_instruction *inst)
    int ch, s1_reg[4], s2_reg[4], s3_reg[4], d_reg[4];
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-      s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-      s3_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[2]);
-      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->Src[0]);
+      s2_reg[ch] = get_src_reg(gen, ch, &inst->Src[1]);
+      s3_reg[ch] = get_src_reg(gen, ch, &inst->Src[2]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->Dst[0]);
    }
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
       spe_fma(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch], s3_reg[ch]);
    }
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_reg[ch], ch, &inst->Dst[0]);
    }
    free_itemps(gen);
    return TRUE;
@@ -671,10 +671,10 @@ emit_LRP(struct codegen *gen, const struct tgsi_full_instruction *inst)
 
    /* setup/get src/dst/temp regs */
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-      s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-      s3_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[2]);
-      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->Src[0]);
+      s2_reg[ch] = get_src_reg(gen, ch, &inst->Src[1]);
+      s3_reg[ch] = get_src_reg(gen, ch, &inst->Src[2]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->Dst[0]);
       tmp_reg[ch] = get_itemp(gen);
    }
 
@@ -687,7 +687,7 @@ emit_LRP(struct codegen *gen, const struct tgsi_full_instruction *inst)
       spe_fma(gen->f, d_reg[ch], tmp_reg[ch], s1_reg[ch], s3_reg[ch]);
    }
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_reg[ch], ch, &inst->Dst[0]);
    }
    free_itemps(gen);
    return TRUE;
@@ -704,8 +704,8 @@ emit_RCP_RSQ(struct codegen *gen, const struct tgsi_full_instruction *inst)
    int ch, s1_reg[4], d_reg[4], tmp_reg[4];
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->Src[0]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->Dst[0]);
       tmp_reg[ch] = get_itemp(gen);
    }
 
@@ -726,7 +726,7 @@ emit_RCP_RSQ(struct codegen *gen, const struct tgsi_full_instruction *inst)
    }
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_reg[ch], ch, &inst->Dst[0]);
    }
 
    free_itemps(gen);
@@ -747,8 +747,8 @@ emit_ABS(struct codegen *gen, const struct tgsi_full_instruction *inst)
    spe_load_uint(gen->f, bit31mask_reg, (1 << 31));
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->Src[0]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->Dst[0]);
    }
 
    /* d = sign bit cleared in s1 */
@@ -757,7 +757,7 @@ emit_ABS(struct codegen *gen, const struct tgsi_full_instruction *inst)
    }
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_reg[ch], ch, &inst->Dst[0]);
    }
 
    free_itemps(gen);
@@ -775,12 +775,12 @@ emit_DP3(struct codegen *gen, const struct tgsi_full_instruction *inst)
    int s2x_reg, s2y_reg, s2z_reg;
    int t0_reg = get_itemp(gen), t1_reg = get_itemp(gen);
 
-   s1x_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[0]);
-   s2x_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[1]);
-   s1y_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[0]);
-   s2y_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[1]);
-   s1z_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[0]);
-   s2z_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[1]);
+   s1x_reg = get_src_reg(gen, CHAN_X, &inst->Src[0]);
+   s2x_reg = get_src_reg(gen, CHAN_X, &inst->Src[1]);
+   s1y_reg = get_src_reg(gen, CHAN_Y, &inst->Src[0]);
+   s2y_reg = get_src_reg(gen, CHAN_Y, &inst->Src[1]);
+   s1z_reg = get_src_reg(gen, CHAN_Z, &inst->Src[0]);
+   s2z_reg = get_src_reg(gen, CHAN_Z, &inst->Src[1]);
 
    /* t0 = x0 * x1 */
    spe_fm(gen->f, t0_reg, s1x_reg, s2x_reg);
@@ -795,9 +795,9 @@ emit_DP3(struct codegen *gen, const struct tgsi_full_instruction *inst)
    spe_fa(gen->f, t0_reg, t0_reg, t1_reg);
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      int d_reg = get_dst_reg(gen, ch, &inst->Dst[0]);
       spe_move(gen->f, d_reg, t0_reg);
-      store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_reg, ch, &inst->Dst[0]);
    }
 
    free_itemps(gen);
@@ -815,14 +815,14 @@ emit_DP4(struct codegen *gen, const struct tgsi_full_instruction *inst)
    int s1x_reg, s1y_reg, s1z_reg, s1w_reg;
    int t0_reg = get_itemp(gen), t1_reg = get_itemp(gen);
 
-   s0x_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[0]);
-   s1x_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[1]);
-   s0y_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[0]);
-   s1y_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[1]);
-   s0z_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[0]);
-   s1z_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[1]);
-   s0w_reg = get_src_reg(gen, CHAN_W, &inst->FullSrcRegisters[0]);
-   s1w_reg = get_src_reg(gen, CHAN_W, &inst->FullSrcRegisters[1]);
+   s0x_reg = get_src_reg(gen, CHAN_X, &inst->Src[0]);
+   s1x_reg = get_src_reg(gen, CHAN_X, &inst->Src[1]);
+   s0y_reg = get_src_reg(gen, CHAN_Y, &inst->Src[0]);
+   s1y_reg = get_src_reg(gen, CHAN_Y, &inst->Src[1]);
+   s0z_reg = get_src_reg(gen, CHAN_Z, &inst->Src[0]);
+   s1z_reg = get_src_reg(gen, CHAN_Z, &inst->Src[1]);
+   s0w_reg = get_src_reg(gen, CHAN_W, &inst->Src[0]);
+   s1w_reg = get_src_reg(gen, CHAN_W, &inst->Src[1]);
 
    /* t0 = x0 * x1 */
    spe_fm(gen->f, t0_reg, s0x_reg, s1x_reg);
@@ -840,9 +840,9 @@ emit_DP4(struct codegen *gen, const struct tgsi_full_instruction *inst)
    spe_fa(gen->f, t0_reg, t0_reg, t1_reg);
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      int d_reg = get_dst_reg(gen, ch, &inst->Dst[0]);
       spe_move(gen->f, d_reg, t0_reg);
-      store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_reg, ch, &inst->Dst[0]);
    }
 
    free_itemps(gen);
@@ -857,31 +857,31 @@ emit_DPH(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
    /* XXX rewrite this function to look more like DP3/DP4 */
    int ch;
-   int s1_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[0]);
-   int s2_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[1]);
+   int s1_reg = get_src_reg(gen, CHAN_X, &inst->Src[0]);
+   int s2_reg = get_src_reg(gen, CHAN_X, &inst->Src[1]);
    int tmp_reg = get_itemp(gen);
 
    /* t = x0 * x1 */
    spe_fm(gen->f, tmp_reg, s1_reg, s2_reg);
 
-   s1_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[0]);
-   s2_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[1]);
+   s1_reg = get_src_reg(gen, CHAN_Y, &inst->Src[0]);
+   s2_reg = get_src_reg(gen, CHAN_Y, &inst->Src[1]);
    /* t = y0 * y1 + t */
    spe_fma(gen->f, tmp_reg, s1_reg, s2_reg, tmp_reg);
 
-   s1_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[0]);
-   s2_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[1]);
+   s1_reg = get_src_reg(gen, CHAN_Z, &inst->Src[0]);
+   s2_reg = get_src_reg(gen, CHAN_Z, &inst->Src[1]);
    /* t = z0 * z1 + t */
    spe_fma(gen->f, tmp_reg, s1_reg, s2_reg, tmp_reg);
 
-   s2_reg = get_src_reg(gen, CHAN_W, &inst->FullSrcRegisters[1]);
+   s2_reg = get_src_reg(gen, CHAN_W, &inst->Src[1]);
    /* t = w1 + t */
    spe_fa(gen->f, tmp_reg, s2_reg, tmp_reg);
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      int d_reg = get_dst_reg(gen, ch, &inst->Dst[0]);
       spe_move(gen->f, d_reg, tmp_reg);
-      store_dest_reg(gen, tmp_reg, ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, tmp_reg, ch, &inst->Dst[0]);
    }
 
    free_itemps(gen);
@@ -898,9 +898,9 @@ emit_NRM3(struct codegen *gen, const struct tgsi_full_instruction *inst)
    int src_reg[3];
    int t0_reg = get_itemp(gen), t1_reg = get_itemp(gen);
 
-   src_reg[0] = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[0]);
-   src_reg[1] = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[0]);
-   src_reg[2] = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[0]);
+   src_reg[0] = get_src_reg(gen, CHAN_X, &inst->Src[0]);
+   src_reg[1] = get_src_reg(gen, CHAN_Y, &inst->Src[0]);
+   src_reg[2] = get_src_reg(gen, CHAN_Z, &inst->Src[0]);
 
    /* t0 = x * x */
    spe_fm(gen->f, t0_reg, src_reg[0], src_reg[0]);
@@ -919,10 +919,10 @@ emit_NRM3(struct codegen *gen, const struct tgsi_full_instruction *inst)
    spe_fi(gen->f, t1_reg, t0_reg, t1_reg);
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      int d_reg = get_dst_reg(gen, ch, &inst->Dst[0]);
       /* dst = src[ch] * t1 */
       spe_fm(gen->f, d_reg, src_reg[ch], t1_reg);
-      store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_reg, ch, &inst->Dst[0]);
    }
 
    free_itemps(gen);
@@ -936,48 +936,48 @@ emit_NRM3(struct codegen *gen, const struct tgsi_full_instruction *inst)
 static boolean
 emit_XPD(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
-   int s1_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[0]);
-   int s2_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[1]);
+   int s1_reg = get_src_reg(gen, CHAN_Z, &inst->Src[0]);
+   int s2_reg = get_src_reg(gen, CHAN_Y, &inst->Src[1]);
    int tmp_reg = get_itemp(gen);
 
    /* t = z0 * y1 */
    spe_fm(gen->f, tmp_reg, s1_reg, s2_reg);
 
-   s1_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[0]);
-   s2_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[1]);
+   s1_reg = get_src_reg(gen, CHAN_Y, &inst->Src[0]);
+   s2_reg = get_src_reg(gen, CHAN_Z, &inst->Src[1]);
    /* t = y0 * z1 - t */
    spe_fms(gen->f, tmp_reg, s1_reg, s2_reg, tmp_reg);
 
-   if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << CHAN_X)) {
-      store_dest_reg(gen, tmp_reg, CHAN_X, &inst->FullDstRegisters[0]);
+   if (inst->Dst[0].Register.WriteMask & (1 << CHAN_X)) {
+      store_dest_reg(gen, tmp_reg, CHAN_X, &inst->Dst[0]);
    }
 
-   s1_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[0]);
-   s2_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[1]);
+   s1_reg = get_src_reg(gen, CHAN_X, &inst->Src[0]);
+   s2_reg = get_src_reg(gen, CHAN_Z, &inst->Src[1]);
    /* t = x0 * z1 */
    spe_fm(gen->f, tmp_reg, s1_reg, s2_reg);
 
-   s1_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[0]);
-   s2_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[1]);
+   s1_reg = get_src_reg(gen, CHAN_Z, &inst->Src[0]);
+   s2_reg = get_src_reg(gen, CHAN_X, &inst->Src[1]);
    /* t = z0 * x1 - t */
    spe_fms(gen->f, tmp_reg, s1_reg, s2_reg, tmp_reg);
 
-   if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << CHAN_Y)) {
-      store_dest_reg(gen, tmp_reg, CHAN_Y, &inst->FullDstRegisters[0]);
+   if (inst->Dst[0].Register.WriteMask & (1 << CHAN_Y)) {
+      store_dest_reg(gen, tmp_reg, CHAN_Y, &inst->Dst[0]);
    }
 
-   s1_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[0]);
-   s2_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[1]);
+   s1_reg = get_src_reg(gen, CHAN_Y, &inst->Src[0]);
+   s2_reg = get_src_reg(gen, CHAN_X, &inst->Src[1]);
    /* t = y0 * x1 */
    spe_fm(gen->f, tmp_reg, s1_reg, s2_reg);
 
-   s1_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[0]);
-   s2_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[1]);
+   s1_reg = get_src_reg(gen, CHAN_X, &inst->Src[0]);
+   s2_reg = get_src_reg(gen, CHAN_Y, &inst->Src[1]);
    /* t = x0 * y1 - t */
    spe_fms(gen->f, tmp_reg, s1_reg, s2_reg, tmp_reg);
 
-   if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << CHAN_Z)) {
-      store_dest_reg(gen, tmp_reg, CHAN_Z, &inst->FullDstRegisters[0]);
+   if (inst->Dst[0].Register.WriteMask & (1 << CHAN_Z)) {
+      store_dest_reg(gen, tmp_reg, CHAN_Z, &inst->Dst[0]);
    }
 
    free_itemps(gen);
@@ -995,14 +995,14 @@ static boolean
 emit_inequality(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
    int ch, s1_reg[4], s2_reg[4], d_reg[4], one_reg;
-   bool complement = FALSE;
+   boolean complement = FALSE;
 
    one_reg = get_const_one_reg(gen);
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-      s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->Src[0]);
+      s2_reg[ch] = get_src_reg(gen, ch, &inst->Src[1]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->Dst[0]);
    }
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
@@ -1043,7 +1043,7 @@ emit_inequality(struct codegen *gen, const struct tgsi_full_instruction *inst)
    }
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_reg[ch], ch, &inst->Dst[0]);
    }
 
    free_itemps(gen);
@@ -1060,10 +1060,10 @@ emit_CMP(struct codegen *gen, const struct tgsi_full_instruction *inst)
    int ch;
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-      int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-      int s3_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[2]);
-      int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      int s1_reg = get_src_reg(gen, ch, &inst->Src[0]);
+      int s2_reg = get_src_reg(gen, ch, &inst->Src[1]);
+      int s3_reg = get_src_reg(gen, ch, &inst->Src[2]);
+      int d_reg = get_dst_reg(gen, ch, &inst->Dst[0]);
       int zero_reg = get_itemp(gen);
    
       spe_zero(gen->f, zero_reg);
@@ -1072,7 +1072,7 @@ emit_CMP(struct codegen *gen, const struct tgsi_full_instruction *inst)
       spe_fcgt(gen->f, d_reg, zero_reg, s1_reg);
       spe_selb(gen->f, d_reg, s3_reg, s2_reg, d_reg);
 
-      store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_reg, ch, &inst->Dst[0]);
       free_itemps(gen);
    }
 
@@ -1090,8 +1090,8 @@ emit_TRUNC(struct codegen *gen, const struct tgsi_full_instruction *inst)
    int ch, s1_reg[4], d_reg[4];
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->Src[0]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->Dst[0]);
    }
 
    /* Convert float to int */
@@ -1105,7 +1105,7 @@ emit_TRUNC(struct codegen *gen, const struct tgsi_full_instruction *inst)
    }
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_reg[ch], ch, &inst->Dst[0]);
    }
 
    free_itemps(gen);
@@ -1129,8 +1129,8 @@ emit_FLR(struct codegen *gen, const struct tgsi_full_instruction *inst)
    one_reg = get_const_one_reg(gen);
    
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->Src[0]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->Dst[0]);
       tmp_reg[ch] = get_itemp(gen);
    }
 
@@ -1156,7 +1156,7 @@ emit_FLR(struct codegen *gen, const struct tgsi_full_instruction *inst)
    }
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_reg[ch], ch, &inst->Dst[0]);
    }
 
    free_itemps(gen);
@@ -1177,8 +1177,8 @@ emit_FRC(struct codegen *gen, const struct tgsi_full_instruction *inst)
    one_reg = get_const_one_reg(gen);
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->Src[0]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->Dst[0]);
       tmp_reg[ch] = get_itemp(gen);
    }
 
@@ -1210,7 +1210,7 @@ emit_FRC(struct codegen *gen, const struct tgsi_full_instruction *inst)
 
    /* store result */
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_reg[ch], ch, &inst->Dst[0]);
    }
 
    free_itemps(gen);
@@ -1272,7 +1272,7 @@ emit_function_call(struct codegen *gen,
 
    if (scalar) {
       for (a = 0; a < num_args; a++) {
-         s_regs[a] = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[a]);
+         s_regs[a] = get_src_reg(gen, CHAN_X, &inst->Src[a]);
       }
       /* we'll call the function, put the return value in this register,
        * then replicate it across all write-enabled components in d_reg.
@@ -1287,11 +1287,11 @@ emit_function_call(struct codegen *gen,
 
       if (!scalar) {
          for (a = 0; a < num_args; a++) {
-            s_regs[a] = get_src_reg(gen, ch, &inst->FullSrcRegisters[a]);
+            s_regs[a] = get_src_reg(gen, ch, &inst->Src[a]);
          }
       }
 
-      d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      d_reg = get_dst_reg(gen, ch, &inst->Dst[0]);
 
       if (!scalar || !func_called) {
          /* for a scalar function, we'll really only call the function once */
@@ -1336,7 +1336,7 @@ emit_function_call(struct codegen *gen,
          spe_move(gen->f, d_reg, retval_reg);
       }
 
-      store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_reg, ch, &inst->Dst[0]);
       free_itemps(gen);
    }
 
@@ -1351,8 +1351,8 @@ emit_function_call(struct codegen *gen,
 static boolean
 emit_TEX(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
-   const uint target = inst->InstructionExtTexture.Texture;
-   const uint unit = inst->FullSrcRegisters[1].SrcRegister.Index;
+   const uint target = inst->Texture.Texture;
+   const uint unit = inst->Src[1].Register.Index;
    uint addr;
    int ch;
    int coord_regs[4], d_regs[4];
@@ -1373,14 +1373,14 @@ emit_TEX(struct codegen *gen, const struct tgsi_full_instruction *inst)
       return FALSE;
    }
 
-   assert(inst->FullSrcRegisters[1].SrcRegister.File == TGSI_FILE_SAMPLER);
+   assert(inst->Src[1].Register.File == TGSI_FILE_SAMPLER);
 
    spe_comment(gen->f, -4, "CALL tex:");
 
    /* get src/dst reg info */
    for (ch = 0; ch < 4; ch++) {
-      coord_regs[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-      d_regs[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      coord_regs[ch] = get_src_reg(gen, ch, &inst->Src[0]);
+      d_regs[ch] = get_dst_reg(gen, ch, &inst->Dst[0]);
    }
 
    {
@@ -1425,7 +1425,7 @@ emit_TEX(struct codegen *gen, const struct tgsi_full_instruction *inst)
    }
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      store_dest_reg(gen, d_regs[ch], ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_regs[ch], ch, &inst->Dst[0]);
       free_itemps(gen);
    }
 
@@ -1452,7 +1452,7 @@ emit_KIL(struct codegen *gen, const struct tgsi_full_instruction *inst)
 
    /* get src regs */
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      s_regs[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      s_regs[ch] = get_src_reg(gen, ch, &inst->Src[0]);
    }
 
    /* test if any src regs are < 0 */
@@ -1500,9 +1500,9 @@ emit_MIN_MAX(struct codegen *gen, const struct tgsi_full_instruction *inst)
    int ch, s0_reg[4], s1_reg[4], d_reg[4], tmp_reg[4];
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      s0_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      s0_reg[ch] = get_src_reg(gen, ch, &inst->Src[0]);
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->Src[1]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->Dst[0]);
       tmp_reg[ch] = get_itemp(gen);         
    }
 
@@ -1518,7 +1518,7 @@ emit_MIN_MAX(struct codegen *gen, const struct tgsi_full_instruction *inst)
    }
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+      store_dest_reg(gen, d_reg[ch], ch, &inst->Dst[0]);
    }
 
    free_itemps(gen);
@@ -1575,7 +1575,7 @@ emit_IF(struct codegen *gen, const struct tgsi_full_instruction *inst)
 
    /* update conditional execution mask with the predicate register */
    int tmp_reg = get_itemp(gen);
-   int s1_reg = get_src_reg(gen, channel, &inst->FullSrcRegisters[0]);
+   int s1_reg = get_src_reg(gen, channel, &inst->Src[0]);
 
    /* tmp = (s1_reg == 0) */
    spe_ceqi(gen->f, tmp_reg, s1_reg, 0);
@@ -1699,8 +1699,8 @@ emit_DDX_DDY(struct codegen *gen, const struct tgsi_full_instruction *inst,
    int ch;
 
    FOR_EACH_ENABLED_CHANNEL(inst, ch) {
-      int s_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-      int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      int s_reg = get_src_reg(gen, ch, &inst->Src[0]);
+      int d_reg = get_dst_reg(gen, ch, &inst->Dst[0]);
 
       int t1_reg = get_itemp(gen);
       int t2_reg = get_itemp(gen);
@@ -1909,8 +1909,8 @@ emit_declaration(struct cell_context *cell,
 
    switch (decl->Declaration.File) {
    case TGSI_FILE_TEMPORARY:
-      for (i = decl->DeclarationRange.First;
-           i <= decl->DeclarationRange.Last;
+      for (i = decl->Range.First;
+           i <= decl->Range.Last;
            i++) {
          assert(i < MAX_TEMPS);
          for (ch = 0; ch < 4; ch++) {
diff --git a/src/gallium/drivers/cell/ppu/cell_pipe_state.c b/src/gallium/drivers/cell/ppu/cell_pipe_state.c
index ccd0fef6e84..c18a5d0635e 100644
--- a/src/gallium/drivers/cell/ppu/cell_pipe_state.c
+++ b/src/gallium/drivers/cell/ppu/cell_pipe_state.c
@@ -383,10 +383,10 @@ cell_init_state_functions(struct cell_context *cell)
    cell->pipe.delete_blend_state = cell_delete_blend_state;
 
    cell->pipe.create_sampler_state = cell_create_sampler_state;
-   cell->pipe.bind_sampler_states = cell_bind_sampler_states;
+   cell->pipe.bind_fragment_sampler_states = cell_bind_sampler_states;
    cell->pipe.delete_sampler_state = cell_delete_sampler_state;
 
-   cell->pipe.set_sampler_textures = cell_set_sampler_textures;
+   cell->pipe.set_fragment_sampler_textures = cell_set_sampler_textures;
 
    cell->pipe.create_depth_stencil_alpha_state = cell_create_depth_stencil_alpha_state;
    cell->pipe.bind_depth_stencil_alpha_state   = cell_bind_depth_stencil_alpha_state;
diff --git a/src/gallium/drivers/cell/ppu/cell_state_derived.c b/src/gallium/drivers/cell/ppu/cell_state_derived.c
index efc4f78364b..b723e794e71 100644
--- a/src/gallium/drivers/cell/ppu/cell_state_derived.c
+++ b/src/gallium/drivers/cell/ppu/cell_state_derived.c
@@ -66,7 +66,7 @@ calculate_vertex_layout( struct cell_context *cell )
    vinfo->num_attribs = 0;
 
    /* we always want to emit vertex pos */
-   src = draw_find_vs_output(cell->draw, TGSI_SEMANTIC_POSITION, 0);
+   src = draw_find_shader_output(cell->draw, TGSI_SEMANTIC_POSITION, 0);
    assert(src >= 0);
    draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_POS, src);
 
@@ -82,14 +82,14 @@ calculate_vertex_layout( struct cell_context *cell )
          break;
 
       case TGSI_SEMANTIC_COLOR:
-         src = draw_find_vs_output(cell->draw, TGSI_SEMANTIC_COLOR, 
+         src = draw_find_shader_output(cell->draw, TGSI_SEMANTIC_COLOR, 
                                    fs->info.input_semantic_index[i]);
          assert(src >= 0);
          draw_emit_vertex_attr(vinfo, EMIT_4F, colorInterp, src);
          break;
 
       case TGSI_SEMANTIC_FOG:
-         src = draw_find_vs_output(cell->draw, TGSI_SEMANTIC_FOG, 0);
+         src = draw_find_shader_output(cell->draw, TGSI_SEMANTIC_FOG, 0);
 #if 1
          if (src < 0) /* XXX temp hack, try demos/fogcoord.c with this */
             src = 0;
@@ -100,7 +100,7 @@ calculate_vertex_layout( struct cell_context *cell )
 
       case TGSI_SEMANTIC_GENERIC:
          /* this includes texcoords and varying vars */
-         src = draw_find_vs_output(cell->draw, TGSI_SEMANTIC_GENERIC,
+         src = draw_find_shader_output(cell->draw, TGSI_SEMANTIC_GENERIC,
                               fs->info.input_semantic_index[i]);
          assert(src >= 0);
          draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src);
diff --git a/src/gallium/drivers/cell/ppu/cell_state_emit.c b/src/gallium/drivers/cell/ppu/cell_state_emit.c
index 9479c0898fd..5b87286d4c5 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 "pipe/p_inlines.h"
 #include "util/u_memory.h"
+#include "util/u_math.h"
 #include "cell_context.h"
 #include "cell_gen_fragment.h"
 #include "cell_state.h"
@@ -299,9 +300,9 @@ cell_emit_state(struct cell_context *cell)
                for (level = 0; level < CELL_MAX_TEXTURE_LEVELS; level++) {
                   texture->start[level] = (ct->mapped +
                                            ct->level_offset[level]);
-                  texture->width[level] = ct->base.width[level];
-                  texture->height[level] = ct->base.height[level];
-                  texture->depth[level] = ct->base.depth[level];
+                  texture->width[level] = u_minify(ct->base.width0, level);
+                  texture->height[level] = u_minify(ct->base.height0, level);
+                  texture->depth[level] = u_minify(ct->base.depth0, level);
                }
                texture->target = ct->base.target;
             }
@@ -330,7 +331,7 @@ cell_emit_state(struct cell_context *cell)
       const struct draw_context *const draw = cell->draw;
       struct cell_shader_info info;
 
-      info.num_outputs = draw_num_vs_outputs(draw);
+      info.num_outputs = draw_num_shader_outputs(draw);
       info.declarations = (uintptr_t) draw->vs.machine.Declarations;
       info.num_declarations = draw->vs.machine.NumDeclarations;
       info.instructions = (uintptr_t) draw->vs.machine.Instructions;
diff --git a/src/gallium/drivers/cell/ppu/cell_texture.c b/src/gallium/drivers/cell/ppu/cell_texture.c
index ae4c61efb3b..998944f77a3 100644
--- a/src/gallium/drivers/cell/ppu/cell_texture.c
+++ b/src/gallium/drivers/cell/ppu/cell_texture.c
@@ -35,6 +35,8 @@
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
 #include "pipe/internal/p_winsys_screen.h"
+
+#include "util/u_format.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
 
@@ -49,9 +51,9 @@ cell_texture_layout(struct cell_texture *ct)
 {
    struct pipe_texture *pt = &ct->base;
    unsigned level;
-   unsigned width = pt->width[0];
-   unsigned height = pt->height[0];
-   unsigned depth = pt->depth[0];
+   unsigned width = pt->width0;
+   unsigned height = pt->height0;
+   unsigned depth = pt->depth0;
 
    ct->buffer_size = 0;
 
@@ -65,17 +67,11 @@ cell_texture_layout(struct cell_texture *ct)
       w_tile = align(width, TILE_SIZE);
       h_tile = align(height, TILE_SIZE);
 
-      pt->width[level] = width;
-      pt->height[level] = height;
-      pt->depth[level] = depth;
-      pt->nblocksx[level] = pf_get_nblocksx(&pt->block, w_tile);  
-      pt->nblocksy[level] = pf_get_nblocksy(&pt->block, h_tile);  
-
-      ct->stride[level] = pt->nblocksx[level] * pt->block.size;
+      ct->stride[level] = util_format_get_stride(pt->format, w_tile);
 
       ct->level_offset[level] = ct->buffer_size;
 
-      size = pt->nblocksx[level] * pt->nblocksy[level] * pt->block.size;
+      size = ct->stride[level] * util_format_get_nblocksy(pt->format, h_tile);
       if (pt->target == PIPE_TEXTURE_CUBE)
          size *= 6;
       else
@@ -83,9 +79,9 @@ cell_texture_layout(struct cell_texture *ct)
 
       ct->buffer_size += size;
 
-      width = minify(width);
-      height = minify(height);
-      depth = minify(depth);
+      width = u_minify(width, 1);
+      height = u_minify(height, 1);
+      depth = u_minify(depth, 1);
    }
 }
 
@@ -276,8 +272,8 @@ cell_get_tex_surface(struct pipe_screen *screen,
       pipe_reference_init(&ps->reference, 1);
       pipe_texture_reference(&ps->texture, pt);
       ps->format = pt->format;
-      ps->width = pt->width[level];
-      ps->height = pt->height[level];
+      ps->width = u_minify(pt->width0, level);
+      ps->height = u_minify(pt->height0, level);
       ps->offset = ct->level_offset[level];
       /* XXX may need to override usage flags (see sp_texture.c) */
       ps->usage = usage;
@@ -286,10 +282,12 @@ cell_get_tex_surface(struct pipe_screen *screen,
       ps->zslice = zslice;
 
       if (pt->target == PIPE_TEXTURE_CUBE) {
-         ps->offset += face * pt->nblocksy[level] * ct->stride[level];
+         unsigned h_tile = align(ps->height, TILE_SIZE);
+         ps->offset += face * util_format_get_nblocksy(ps->format, h_tile) * ct->stride[level];
       }
       else if (pt->target == PIPE_TEXTURE_3D) {
-         ps->offset += zslice * pt->nblocksy[level] * ct->stride[level];
+         unsigned h_tile = align(ps->height, TILE_SIZE);
+         ps->offset += zslice * util_format_get_nblocksy(ps->format, h_tile) * ct->stride[level];
       }
       else {
          assert(face == 0);
@@ -330,14 +328,10 @@ cell_get_tex_transfer(struct pipe_screen *screen,
    if (ctrans) {
       struct pipe_transfer *pt = &ctrans->base;
       pipe_texture_reference(&pt->texture, texture);
-      pt->format = texture->format;
-      pt->block = texture->block;
       pt->x = x;
       pt->y = y;
       pt->width = w;
       pt->height = h;
-      pt->nblocksx = texture->nblocksx[level];
-      pt->nblocksy = texture->nblocksy[level];
       pt->stride = ct->stride[level];
       pt->usage = usage;
       pt->face = face;
@@ -347,10 +341,12 @@ cell_get_tex_transfer(struct pipe_screen *screen,
       ctrans->offset = ct->level_offset[level];
 
       if (texture->target == PIPE_TEXTURE_CUBE) {
-         ctrans->offset += face * pt->nblocksy * pt->stride;
+         unsigned h_tile = align(u_minify(texture->height0, level), TILE_SIZE);
+         ctrans->offset += face * util_format_get_nblocksy(texture->format, h_tile) * pt->stride;
       }
       else if (texture->target == PIPE_TEXTURE_3D) {
-         ctrans->offset += zslice * pt->nblocksy * pt->stride;
+         unsigned h_tile = align(u_minify(texture->height0, level), TILE_SIZE);
+         ctrans->offset += zslice * util_format_get_nblocksy(texture->format, h_tile) * pt->stride;
       }
       else {
          assert(face == 0);
@@ -386,8 +382,8 @@ cell_transfer_map(struct pipe_screen *screen, struct pipe_transfer *transfer)
    struct pipe_texture *pt = transfer->texture;
    struct cell_texture *ct = cell_texture(pt);
    const uint level = ctrans->base.level;
-   const uint texWidth = pt->width[level];
-   const uint texHeight = pt->height[level];
+   const uint texWidth = u_minify(pt->width0, level);
+   const uint texHeight = u_minify(pt->height0, level);
    const uint stride = ct->stride[level];
    unsigned size;
 
@@ -403,7 +399,8 @@ cell_transfer_map(struct pipe_screen *screen, struct pipe_transfer *transfer)
     * Create a buffer of ordinary memory for the linear texture.
     * This is the memory that the user will read/write.
     */
-   size = pt->nblocksx[level] * pt->nblocksy[level] * pt->block.size;
+   size = util_format_get_stride(pt->format, align(texWidth, TILE_SIZE)) *
+          util_format_get_nblocksy(pt->format, align(texHeight, TILE_SIZE));
 
    ctrans->map = align_malloc(size, 16);
    if (!ctrans->map)
@@ -411,7 +408,7 @@ cell_transfer_map(struct pipe_screen *screen, struct pipe_transfer *transfer)
 
    if (transfer->usage & PIPE_TRANSFER_READ) {
       /* need to untwiddle the texture to make a linear version */
-      const uint bpp = pf_get_size(ct->base.format);
+      const uint bpp = util_format_get_blocksize(ct->base.format);
       if (bpp == 4) {
          const uint *src = (uint *) (ct->mapped + ctrans->offset);
          uint *dst = ctrans->map;
@@ -440,8 +437,8 @@ cell_transfer_unmap(struct pipe_screen *screen,
    struct pipe_texture *pt = transfer->texture;
    struct cell_texture *ct = cell_texture(pt);
    const uint level = ctrans->base.level;
-   const uint texWidth = pt->width[level];
-   const uint texHeight = pt->height[level];
+   const uint texWidth = u_minify(pt->width0, level);
+   const uint texHeight = u_minify(pt->height0, level);
    const uint stride = ct->stride[level];
 
    if (!ct->mapped) {
@@ -454,7 +451,7 @@ cell_transfer_unmap(struct pipe_screen *screen,
       /* The user wrote new texture data into the mapped buffer.
        * We need to convert the new linear data into the twiddled/tiled format.
        */
-      const uint bpp = pf_get_size(ct->base.format);
+      const uint bpp = util_format_get_blocksize(ct->base.format);
       if (bpp == 4) {
          const uint *src = ctrans->map;
          uint *dst = (uint *) (ct->mapped + ctrans->offset);
diff --git a/src/gallium/drivers/cell/spu/spu_command.c b/src/gallium/drivers/cell/spu/spu_command.c
index 5c0179d9546..55bd85bde2b 100644
--- a/src/gallium/drivers/cell/spu/spu_command.c
+++ b/src/gallium/drivers/cell/spu/spu_command.c
@@ -53,8 +53,7 @@ struct spu_vs_context draw;
 /**
  * Buffers containing dynamically generated SPU code:
  */
-static unsigned char attribute_fetch_code_buffer[136 * PIPE_MAX_ATTRIBS]
-    ALIGN16_ATTRIB;
+PIPE_ALIGN_VAR(16) static unsigned char attribute_fetch_code_buffer[136 * PIPE_MAX_ATTRIBS];
 
 
 
@@ -405,8 +404,6 @@ cmd_state_sampler(const struct cell_command_sampler *sampler)
    case PIPE_TEX_FILTER_LINEAR:
       spu.min_sample_texture_2d[unit] = sample_texture_2d_bilinear;
       break;
-   case PIPE_TEX_FILTER_ANISO:
-      /* fall-through, for now */
    case PIPE_TEX_FILTER_NEAREST:
       spu.min_sample_texture_2d[unit] = sample_texture_2d_nearest;
       break;
@@ -418,8 +415,6 @@ cmd_state_sampler(const struct cell_command_sampler *sampler)
    case PIPE_TEX_FILTER_LINEAR:
       spu.mag_sample_texture_2d[unit] = sample_texture_2d_bilinear;
       break;
-   case PIPE_TEX_FILTER_ANISO:
-      /* fall-through, for now */
    case PIPE_TEX_FILTER_NEAREST:
       spu.mag_sample_texture_2d[unit] = sample_texture_2d_nearest;
       break;
@@ -547,7 +542,7 @@ cmd_batch(uint opcode)
 {
    const uint buf = (opcode >> 8) & 0xff;
    uint size = (opcode >> 16);
-   qword buffer[CELL_BUFFER_SIZE / 16] ALIGN16_ATTRIB;
+   PIPE_ALIGN_VAR(16) qword buffer[CELL_BUFFER_SIZE / 16];
    const unsigned usize = ROUNDUP16(size) / sizeof(buffer[0]);
    uint pos;
 
diff --git a/src/gallium/drivers/cell/spu/spu_exec.c b/src/gallium/drivers/cell/spu/spu_exec.c
index 4c32b2d06d7..d2166a49016 100644
--- a/src/gallium/drivers/cell/spu/spu_exec.c
+++ b/src/gallium/drivers/cell/spu/spu_exec.c
@@ -108,10 +108,10 @@
    for (CHAN = 0; CHAN < 4; CHAN++)
 
 #define IS_CHANNEL_ENABLED(INST, CHAN)\
-   ((INST).FullDstRegisters[0].DstRegister.WriteMask & (1 << (CHAN)))
+   ((INST).Dst[0].Register.WriteMask & (1 << (CHAN)))
 
 #define IS_CHANNEL_ENABLED2(INST, CHAN)\
-   ((INST).FullDstRegisters[1].DstRegister.WriteMask & (1 << (CHAN)))
+   ((INST).Dst[1].Register.WriteMask & (1 << (CHAN)))
 
 #define FOR_EACH_ENABLED_CHANNEL(INST, CHAN)\
    FOR_EACH_CHANNEL( CHAN )\
@@ -431,22 +431,22 @@ fetch_source(
    index.i[0] =
    index.i[1] =
    index.i[2] =
-   index.i[3] = reg->SrcRegister.Index;
+   index.i[3] = reg->Register.Index;
 
-   if (reg->SrcRegister.Indirect) {
+   if (reg->Register.Indirect) {
       union spu_exec_channel index2;
       union spu_exec_channel indir_index;
 
       index2.i[0] =
       index2.i[1] =
       index2.i[2] =
-      index2.i[3] = reg->SrcRegisterInd.Index;
+      index2.i[3] = reg->Indirect.Index;
 
-      swizzle = tgsi_util_get_src_register_swizzle(&reg->SrcRegisterInd,
+      swizzle = tgsi_util_get_src_register_swizzle(&reg->Indirect,
                                                    CHAN_X);
       fetch_src_file_channel(
          mach,
-         reg->SrcRegisterInd.File,
+         reg->Indirect.File,
          swizzle,
          &index2,
          &indir_index );
@@ -454,8 +454,8 @@ fetch_source(
       index.q = si_a(index.q, indir_index.q);
    }
 
-   if( reg->SrcRegister.Dimension ) {
-      switch( reg->SrcRegister.File ) {
+   if( reg->Register.Dimension ) {
+      switch( reg->Register.File ) {
       case TGSI_FILE_INPUT:
          index.q = si_mpyi(index.q, 17);
          break;
@@ -466,24 +466,24 @@ fetch_source(
          ASSERT( 0 );
       }
 
-      index.i[0] += reg->SrcRegisterDim.Index;
-      index.i[1] += reg->SrcRegisterDim.Index;
-      index.i[2] += reg->SrcRegisterDim.Index;
-      index.i[3] += reg->SrcRegisterDim.Index;
+      index.i[0] += reg->Dimension.Index;
+      index.i[1] += reg->Dimension.Index;
+      index.i[2] += reg->Dimension.Index;
+      index.i[3] += reg->Dimension.Index;
 
-      if (reg->SrcRegisterDim.Indirect) {
+      if (reg->Dimension.Indirect) {
          union spu_exec_channel index2;
          union spu_exec_channel indir_index;
 
          index2.i[0] =
          index2.i[1] =
          index2.i[2] =
-         index2.i[3] = reg->SrcRegisterDimInd.Index;
+         index2.i[3] = reg->DimIndirect.Index;
 
-         swizzle = tgsi_util_get_src_register_swizzle( &reg->SrcRegisterDimInd, CHAN_X );
+         swizzle = tgsi_util_get_src_register_swizzle( &reg->DimIndirect, CHAN_X );
          fetch_src_file_channel(
             mach,
-            reg->SrcRegisterDimInd.File,
+            reg->DimIndirect.File,
             swizzle,
             &index2,
             &indir_index );
@@ -495,7 +495,7 @@ fetch_source(
    swizzle = tgsi_util_get_full_src_register_swizzle( reg, chan_index );
    fetch_src_file_channel(
       mach,
-      reg->SrcRegister.File,
+      reg->Register.File,
       swizzle,
       &index,
       chan );
@@ -517,7 +517,7 @@ fetch_source(
       break;
    }
 
-   if (reg->SrcRegisterExtMod.Complement) {
+   if (reg->RegisterExtMod.Complement) {
       chan->q = si_fs(mach->Temps[TEMP_1_I].xyzw[TEMP_1_C].q, chan->q);
    }
 }
@@ -532,21 +532,21 @@ store_dest(
 {
    union spu_exec_channel *dst;
 
-   switch( reg->DstRegister.File ) {
+   switch( reg->Register.File ) {
    case TGSI_FILE_NULL:
       return;
 
    case TGSI_FILE_OUTPUT:
       dst = &mach->Outputs[mach->Temps[TEMP_OUTPUT_I].xyzw[TEMP_OUTPUT_C].u[0]
-                           + reg->DstRegister.Index].xyzw[chan_index];
+                           + reg->Register.Index].xyzw[chan_index];
       break;
 
    case TGSI_FILE_TEMPORARY:
-      dst = &mach->Temps[reg->DstRegister.Index].xyzw[chan_index];
+      dst = &mach->Temps[reg->Register.Index].xyzw[chan_index];
       break;
 
    case TGSI_FILE_ADDRESS:
-      dst = &mach->Addrs[reg->DstRegister.Index].xyzw[chan_index];
+      dst = &mach->Addrs[reg->Register.Index].xyzw[chan_index];
       break;
 
    default:
@@ -583,10 +583,10 @@ store_dest(
 }
 
 #define FETCH(VAL,INDEX,CHAN)\
-    fetch_source (mach, VAL, &inst->FullSrcRegisters[INDEX], CHAN)
+    fetch_source (mach, VAL, &inst->Src[INDEX], CHAN)
 
 #define STORE(VAL,INDEX,CHAN)\
-    store_dest (mach, VAL, &inst->FullDstRegisters[INDEX], inst, CHAN )
+    store_dest (mach, VAL, &inst->Dst[INDEX], inst, CHAN )
 
 
 /**
@@ -612,7 +612,7 @@ exec_kil(struct spu_exec_machine *mach,
 
       /* unswizzle channel */
       swizzle = tgsi_util_get_full_src_register_swizzle (
-                        &inst->FullSrcRegisters[0],
+                        &inst->Src[0],
                         chan_index);
 
       /* check if the component has not been already tested */
@@ -677,7 +677,7 @@ exec_tex(struct spu_exec_machine *mach,
          const struct tgsi_full_instruction *inst,
          boolean biasLod, boolean projected)
 {
-   const uint unit = inst->FullSrcRegisters[1].SrcRegister.Index;
+   const uint unit = inst->Src[1].Register.Index;
    union spu_exec_channel r[8];
    uint chan_index;
    float lodBias;
@@ -833,8 +833,8 @@ exec_declaration(struct spu_exec_machine *mach,
          unsigned first, last, mask;
          interpolation_func interp;
 
-         first = decl->DeclarationRange.First;
-         last = decl->DeclarationRange.Last;
+         first = decl->Range.First;
+         last = decl->Range.Last;
          mask = decl->Declaration.UsageMask;
 
          switch( decl->Declaration.Interpolate ) {
@@ -1681,7 +1681,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_SHR:
+   case TGSI_OPCODE_ISHR:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          FETCH( &r[1], 1, chan_index );
@@ -1839,10 +1839,11 @@ spu_exec_machine_run( struct spu_exec_machine *mach )
    /* execute declarations (interpolants) */
    if( mach->Processor == TGSI_PROCESSOR_FRAGMENT ) {
       for (i = 0; i < mach->NumDeclarations; i++) {
+         PIPE_ALIGN_VAR(16)
          union {
             struct tgsi_full_declaration decl;
             qword buffer[ROUNDUP16(sizeof(struct tgsi_full_declaration)) / 16];
-         } d ALIGN16_ATTRIB;
+         } d;
          unsigned ea = (unsigned) (mach->Declarations + pc);
 
          spu_dcache_fetch_unaligned(d.buffer, ea, sizeof(d.decl));
@@ -1853,10 +1854,11 @@ spu_exec_machine_run( struct spu_exec_machine *mach )
 
    /* execute instructions, until pc is set to -1 */
    while (pc != -1) {
+      PIPE_ALIGN_VAR(16)
       union {
          struct tgsi_full_instruction inst;
          qword buffer[ROUNDUP16(sizeof(struct tgsi_full_instruction)) / 16];
-      } i ALIGN16_ATTRIB;
+      } i;
       unsigned ea = (unsigned) (mach->Instructions + pc);
 
       spu_dcache_fetch_unaligned(i.buffer, ea, sizeof(i.inst));
diff --git a/src/gallium/drivers/cell/spu/spu_exec.h b/src/gallium/drivers/cell/spu/spu_exec.h
index 86056799405..0ca92af248d 100644
--- a/src/gallium/drivers/cell/spu/spu_exec.h
+++ b/src/gallium/drivers/cell/spu/spu_exec.h
@@ -98,9 +98,9 @@ struct spu_exec_machine
     * 4  internal temporaries
     * 1  address
     */
+   PIPE_ALIGN_VAR(16)
    struct spu_exec_vector       Temps[TGSI_EXEC_NUM_TEMPS 
-                                      + TGSI_EXEC_NUM_TEMP_EXTRAS + 1]
-       ALIGN16_ATTRIB;
+                                      + TGSI_EXEC_NUM_TEMP_EXTRAS + 1];
 
    struct spu_exec_vector       *Addrs;
 
diff --git a/src/gallium/drivers/cell/spu/spu_funcs.c b/src/gallium/drivers/cell/spu/spu_funcs.c
index ff3d609d258..98919c43ffc 100644
--- a/src/gallium/drivers/cell/spu/spu_funcs.c
+++ b/src/gallium/drivers/cell/spu/spu_funcs.c
@@ -144,7 +144,7 @@ export_func(struct cell_spu_function_info *spu_functions,
 void
 return_function_info(void)
 {
-   struct cell_spu_function_info funcs ALIGN16_ATTRIB;
+   PIPE_ALIGN_VAR(16) struct cell_spu_function_info funcs;
    int tag = TAG_MISC;
 
    ASSERT(sizeof(funcs) == 256); /* must be multiple of 16 bytes */
diff --git a/src/gallium/drivers/cell/spu/spu_main.h b/src/gallium/drivers/cell/spu/spu_main.h
index 33767e7c51d..b18f4c22ef1 100644
--- a/src/gallium/drivers/cell/spu/spu_main.h
+++ b/src/gallium/drivers/cell/spu/spu_main.h
@@ -93,6 +93,7 @@ typedef vector unsigned int (*spu_fragment_program_func)(vector float *inputs,
                                                          vector float *constants);
 
 
+PIPE_ALIGN_TYPE(16,
 struct spu_framebuffer
 {
    void *color_start;              /**< addr of color surface in main memory */
@@ -107,10 +108,11 @@ struct spu_framebuffer
 
    uint zsize;                     /**< 0, 2 or 4 bytes per Z */
    float zscale;                   /**< 65535.0, 2^24-1 or 2^32-1 */
-} ALIGN16_ATTRIB;
+});
 
 
 /** per-texture level info */
+PIPE_ALIGN_TYPE(16,
 struct spu_texture_level
 {
    void *start;
@@ -123,20 +125,22 @@ struct spu_texture_level
    vector signed int mask_s, mask_t, mask_r;
    /** texcoord clamp limits */
    vector signed int max_s, max_t, max_r;
-} ALIGN16_ATTRIB;
+});
 
 
+PIPE_ALIGN_TYPE(16,
 struct spu_texture
 {
    struct spu_texture_level level[CELL_MAX_TEXTURE_LEVELS];
    uint max_level;
    uint target;  /**< PIPE_TEXTURE_x */
-} ALIGN16_ATTRIB;
+});
 
 
 /**
  * All SPU global/context state will be in a singleton object of this type:
  */
+PIPE_ALIGN_TYPE(16,
 struct spu_global
 {
    /** One-time init/constant info */
@@ -155,8 +159,8 @@ struct spu_global
    struct vertex_info vertex_info;
 
    /** Current color and Z tiles */
-   tile_t ctile ALIGN16_ATTRIB;
-   tile_t ztile ALIGN16_ATTRIB;
+   PIPE_ALIGN_VAR(16) tile_t ctile;
+   PIPE_ALIGN_VAR(16) tile_t ztile;
 
    /** Read depth/stencil tiles? */
    boolean read_depth_stencil;
@@ -165,8 +169,8 @@ struct spu_global
    ubyte cur_ctile_status, cur_ztile_status;
 
    /** Status of all tiles in framebuffer */
-   ubyte ctile_status[CELL_MAX_HEIGHT/TILE_SIZE][CELL_MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB;
-   ubyte ztile_status[CELL_MAX_HEIGHT/TILE_SIZE][CELL_MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB;
+   PIPE_ALIGN_VAR(16) ubyte ctile_status[CELL_MAX_HEIGHT/TILE_SIZE][CELL_MAX_WIDTH/TILE_SIZE];
+   PIPE_ALIGN_VAR(16) ubyte ztile_status[CELL_MAX_HEIGHT/TILE_SIZE][CELL_MAX_WIDTH/TILE_SIZE];
 
    /** Current fragment ops machine code, at 8-byte boundary */
    uint *fragment_ops_code;
@@ -175,7 +179,7 @@ struct spu_global
    spu_fragment_ops_func fragment_ops[2];
 
    /** Current fragment program machine code, at 8-byte boundary */
-   uint fragment_program_code[SPU_MAX_FRAGMENT_PROGRAM_INSTS] ALIGN8_ATTRIB;
+   PIPE_ALIGN_VAR(8) uint fragment_program_code[SPU_MAX_FRAGMENT_PROGRAM_INSTS];
    /** Current fragment ops function */
    spu_fragment_program_func fragment_program;
 
@@ -187,7 +191,7 @@ struct spu_global
    /** Fragment program constants */
    vector float constants[4 * CELL_MAX_CONSTANTS];
 
-} ALIGN16_ATTRIB;
+});
 
 
 extern struct spu_global spu;
diff --git a/src/gallium/drivers/cell/spu/spu_render.c b/src/gallium/drivers/cell/spu/spu_render.c
index 5ffb7073abf..14987e3c3a2 100644
--- a/src/gallium/drivers/cell/spu/spu_render.c
+++ b/src/gallium/drivers/cell/spu/spu_render.c
@@ -169,7 +169,7 @@ void
 cmd_render(const struct cell_command_render *render, uint *pos_incr)
 {
    /* we'll DMA into these buffers */
-   ubyte vertex_data[CELL_BUFFER_SIZE] ALIGN16_ATTRIB;
+   PIPE_ALIGN_VAR(16) ubyte vertex_data[CELL_BUFFER_SIZE];
    const uint vertex_size = render->vertex_size; /* in bytes */
    /*const*/ uint total_vertex_bytes = render->num_verts * vertex_size;
    uint index_bytes;
diff --git a/src/gallium/drivers/cell/spu/spu_util.c b/src/gallium/drivers/cell/spu/spu_util.c
index c2c32b22d5a..24057e29e36 100644
--- a/src/gallium/drivers/cell/spu/spu_util.c
+++ b/src/gallium/drivers/cell/spu/spu_util.c
@@ -33,7 +33,7 @@ tgsi_util_get_full_src_register_swizzle(
    unsigned component )
 {
    return tgsi_util_get_src_register_swizzle(
-      reg->SrcRegister,
+      reg->Register,
       component );
 }
 
@@ -45,10 +45,10 @@ tgsi_util_get_full_src_register_sign_mode(
 {
    unsigned sign_mode;
 
-   if( reg->SrcRegisterExtMod.Absolute ) {
+   if( reg->RegisterExtMod.Absolute ) {
       /* Consider only the post-abs negation. */
 
-      if( reg->SrcRegisterExtMod.Negate ) {
+      if( reg->RegisterExtMod.Negate ) {
          sign_mode = TGSI_UTIL_SIGN_SET;
       }
       else {
@@ -60,8 +60,8 @@ tgsi_util_get_full_src_register_sign_mode(
 
       unsigned negate;
 
-      negate = reg->SrcRegister.Negate;
-      if( reg->SrcRegisterExtMod.Negate ) {
+      negate = reg->Register.Negate;
+      if( reg->RegisterExtMod.Negate ) {
          negate = !negate;
       }
 
diff --git a/src/gallium/drivers/cell/spu/spu_vertex_fetch.c b/src/gallium/drivers/cell/spu/spu_vertex_fetch.c
index 03375d84a57..087963960df 100644
--- a/src/gallium/drivers/cell/spu/spu_vertex_fetch.c
+++ b/src/gallium/drivers/cell/spu/spu_vertex_fetch.c
@@ -43,7 +43,8 @@ typedef void (*spu_fetch_func)(qword *out, const qword *in,
 			       const qword *shuffle_data);
 
 
-static const qword fetch_shuffle_data[5] ALIGN16_ATTRIB = {
+PIPE_ALIGN_VAR(16) static const qword
+fetch_shuffle_data[5] = {
    /* Shuffle used by CVT_64_FLOAT
     */
    {
@@ -110,7 +111,7 @@ 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] ALIGN16_ATTRIB;
+      PIPE_ALIGN_VAR(16) qword in[2 * 4];
 
 
       /* 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 fbe5b34d397..3e9804bf8ee 100644
--- a/src/gallium/drivers/cell/spu/spu_vertex_shader.c
+++ b/src/gallium/drivers/cell/spu/spu_vertex_shader.c
@@ -107,8 +107,8 @@ 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_MAX_ATTRIBS);
-   ALIGN16_DECL(struct spu_exec_vector, outputs, PIPE_MAX_ATTRIBS);
+   PIPE_ALIGN_VAR(16) struct spu_exec_vector inputs[PIPE_MAX_ATTRIBS];
+   PIPE_ALIGN_VAR(16) struct spu_exec_vector outputs[PIPE_MAX_ATTRIBS];
    const float *scale = draw->viewport.scale;
    const float *trans = draw->viewport.translate;
 
@@ -119,8 +119,8 @@ run_vertex_program(struct spu_vs_context *draw,
    ASSERT_ALIGN16(draw->constants);
    machine->Consts = (float (*)[4]) draw->constants;
 
-   machine->Inputs = ALIGN16_ASSIGN(inputs);
-   machine->Outputs = ALIGN16_ASSIGN(outputs);
+   machine->Inputs = inputs;
+   machine->Outputs = outputs;
 
    spu_vertex_fetch( draw, machine, elts, count );
 
@@ -132,8 +132,9 @@ run_vertex_program(struct spu_vs_context *draw,
    for (j = 0; j < count; j++) {
       unsigned slot;
       float x, y, z, w;
+      PIPE_ALIGN_VAR(16)
       unsigned char buffer[sizeof(struct vertex_header)
-          + MAX_VERTEX_SIZE] ALIGN16_ATTRIB;
+          + MAX_VERTEX_SIZE];
       struct vertex_header *const tmpOut =
           (struct vertex_header *) buffer;
       const unsigned vert_size = ROUNDUP16(sizeof(struct vertex_header)
@@ -186,8 +187,8 @@ run_vertex_program(struct spu_vs_context *draw,
 }
 
 
-unsigned char immediates[(sizeof(float) * 4 * TGSI_EXEC_NUM_IMMEDIATES) + 32]
-    ALIGN16_ATTRIB;
+PIPE_ALIGN_VAR(16) unsigned char
+immediates[(sizeof(float) * 4 * TGSI_EXEC_NUM_IMMEDIATES) + 32]);
 
 
 void