[965] Replace various alignment code with a shared ALIGN() macro.

In the process, fix some alignment issues:
- Scratch space allocation was aligned into units of 1KB, while the allocation
  wanted units of bytes, so we never allocated enough space for scratch.
- GRF register count was programmed as ALIGN(val - 1, 16) / 16 instead of
  ALIGN(val, 16) / 16 - 1, which overcounted for val != 16n+1.

18 files changed, 33 insertions, 41 deletions

diff --git a/src/mesa/drivers/dri/i915/intel_context.h b/src/mesa/drivers/dri/i915/intel_context.h
index c8298dd9c4..ce9a362944 100644
--- a/src/mesa/drivers/dri/i915/intel_context.h
+++ b/src/mesa/drivers/dri/i915/intel_context.h
@@ -292,6 +292,8 @@ extern char *__progname;
 #define SUBPIXEL_X 0.125
 #define SUBPIXEL_Y 0.125
 
+#define ALIGN(value, alignment)  ((value + alignment - 1) & ~(alignment - 1))
+
 #define INTEL_FIREVERTICES(intel)		\
 do {						\
    if ((intel)->prim.flush)			\
diff --git a/src/mesa/drivers/dri/i965/brw_clip_state.c b/src/mesa/drivers/dri/i965/brw_clip_state.c
index ae46d7a86e..ba2f0edf51 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_state.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_state.c
@@ -43,7 +43,8 @@ static void upload_clip_unit( struct brw_context *brw )
    memset(&clip, 0, sizeof(clip));
 
    /* CACHE_NEW_CLIP_PROG */
-   clip.thread0.grf_reg_count = ((brw->clip.prog_data->total_grf-1) & ~15) / 16;
+   clip.thread0.grf_reg_count =
+      ALIGN(brw->clip.prog_data->total_grf, 16) / 16 - 1;
    clip.thread0.kernel_start_pointer = brw->clip.prog_gs_offset >> 6;
    clip.thread3.urb_entry_read_length = brw->clip.prog_data->urb_read_length;
    clip.thread3.const_urb_entry_read_length = brw->clip.prog_data->curb_read_length;
diff --git a/src/mesa/drivers/dri/i965/brw_curbe.c b/src/mesa/drivers/dri/i965/brw_curbe.c
index d3c88c1dca..fa4ea42aa6 100644
--- a/src/mesa/drivers/dri/i965/brw_curbe.c
+++ b/src/mesa/drivers/dri/i965/brw_curbe.c
@@ -304,7 +304,7 @@ static void upload_constant_buffer(struct brw_context *brw)
       
       if (!brw_pool_alloc(pool, 
 			  bufsz,
-			  6,
+			  1 << 6,
 			  &brw->curbe.gs_offset)) {
 	 _mesa_printf("out of GS memory for curbe\n");
 	 assert(0);
diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c
index 89cd063d46..b7795703fd 100644
--- a/src/mesa/drivers/dri/i965/brw_draw_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c
@@ -290,7 +290,7 @@ static void get_space( struct brw_context *brw,
 		       struct gl_buffer_object **vbo_return,
 		       GLuint *offset_return )
 {
-   size = (size + 63) & ~63;
+   size = ALIGN(size, 64);
    
    if (brw->vb.upload.offset + size > BRW_UPLOAD_INIT_SIZE)
       wrap_buffers(brw, size);
diff --git a/src/mesa/drivers/dri/i965/brw_gs_state.c b/src/mesa/drivers/dri/i965/brw_gs_state.c
index 5826c01d4f..5db4dd4603 100644
--- a/src/mesa/drivers/dri/i965/brw_gs_state.c
+++ b/src/mesa/drivers/dri/i965/brw_gs_state.c
@@ -46,7 +46,8 @@ static void upload_gs_unit( struct brw_context *brw )
 
    /* CACHE_NEW_GS_PROG */
    if (brw->gs.prog_active) {
-      gs.thread0.grf_reg_count = ((brw->gs.prog_data->total_grf-1) & ~15) / 16;
+      gs.thread0.grf_reg_count =
+	 ALIGN(brw->gs.prog_data->total_grf, 16) / 16 - 1;
       gs.thread0.kernel_start_pointer = brw->gs.prog_gs_offset >> 6;
       gs.thread3.urb_entry_read_length = brw->gs.prog_data->urb_read_length;
    }
diff --git a/src/mesa/drivers/dri/i965/brw_sf_state.c b/src/mesa/drivers/dri/i965/brw_sf_state.c
index 236c6fd42a..2257916aae 100644
--- a/src/mesa/drivers/dri/i965/brw_sf_state.c
+++ b/src/mesa/drivers/dri/i965/brw_sf_state.c
@@ -118,7 +118,7 @@ static void upload_sf_unit( struct brw_context *brw )
    memset(&sf, 0, sizeof(sf));
 
    /* CACHE_NEW_SF_PROG */
-   sf.thread0.grf_reg_count = ((brw->sf.prog_data->total_grf-1) & ~15) / 16;
+   sf.thread0.grf_reg_count = ALIGN(brw->sf.prog_data->total_grf, 16) / 16 - 1;
    sf.thread0.kernel_start_pointer = brw->sf.prog_gs_offset >> 6;
    sf.thread3.urb_entry_read_length = brw->sf.prog_data->urb_read_length;
 
diff --git a/src/mesa/drivers/dri/i965/brw_state_cache.c b/src/mesa/drivers/dri/i965/brw_state_cache.c
index 98d765ac0e..0e73ff8390 100644
--- a/src/mesa/drivers/dri/i965/brw_state_cache.c
+++ b/src/mesa/drivers/dri/i965/brw_state_cache.c
@@ -148,7 +148,7 @@ GLuint brw_upload_cache( struct brw_cache *cache,
    GLuint hash = hash_key(key, key_size);
    void *tmp = _mesa_malloc(key_size + cache->aux_size);
    
-   if (!brw_pool_alloc(cache->pool, data_size, 6, &offset)) {
+   if (!brw_pool_alloc(cache->pool, data_size, 1 << 6, &offset)) {
       /* Should not be possible: 
        */
       _mesa_printf("brw_pool_alloc failed\n");
diff --git a/src/mesa/drivers/dri/i965/brw_state_pool.c b/src/mesa/drivers/dri/i965/brw_state_pool.c
index 708ae857ab..eda92a2fa8 100644
--- a/src/mesa/drivers/dri/i965/brw_state_pool.c
+++ b/src/mesa/drivers/dri/i965/brw_state_pool.c
@@ -41,10 +41,9 @@ GLboolean brw_pool_alloc( struct brw_mem_pool *pool,
 			  GLuint align,
 			  GLuint *offset_return)
 {
-   GLuint align_mask = (1<<align)-1;
-   GLuint fixup = ((pool->offset + align_mask) & ~align_mask) - pool->offset;
+   GLuint fixup = ALIGN(pool->offset, align) - pool->offset;
 
-   size = (size + 3) & ~3;
+   size = ALIGN(size, 4);
 
    if (pool->offset + fixup + size >= pool->size) {
       _mesa_printf("%s failed\n", __FUNCTION__);
diff --git a/src/mesa/drivers/dri/i965/brw_tex_layout.c b/src/mesa/drivers/dri/i965/brw_tex_layout.c
index 2094a1c8ad..e306c9cf10 100644
--- a/src/mesa/drivers/dri/i965/brw_tex_layout.c
+++ b/src/mesa/drivers/dri/i965/brw_tex_layout.c
@@ -37,8 +37,6 @@
 #include "intel_tex_layout.h"
 #include "macros.h"
 
-#define ALIGN(value, alignment)  ((value + alignment - 1) & ~(alignment - 1))
-
 GLboolean brw_miptree_layout( struct intel_mipmap_tree *mt )
 {
    /* XXX: these vary depending on image format: 
@@ -64,7 +62,7 @@ GLboolean brw_miptree_layout( struct intel_mipmap_tree *mt )
           mt->pitch = ALIGN(width, align_w);
           pack_y_pitch = (height + 3) / 4;
       } else {
-          mt->pitch = ((mt->width0 * mt->cpp + 3) & ~3) / mt->cpp;
+          mt->pitch = ALIGN(mt->width0 * mt->cpp, 4) / mt->cpp;
           pack_y_pitch = ALIGN(mt->height0, align_h);
       }
 
diff --git a/src/mesa/drivers/dri/i965/brw_vs_state.c b/src/mesa/drivers/dri/i965/brw_vs_state.c
index c225bf8f5c..f561979138 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_state.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_state.c
@@ -44,7 +44,7 @@ static void upload_vs_unit( struct brw_context *brw )
 
    /* CACHE_NEW_VS_PROG */
    vs.thread0.kernel_start_pointer = brw->vs.prog_gs_offset >> 6;
-   vs.thread0.grf_reg_count = ((brw->vs.prog_data->total_grf-1) & ~15) / 16;
+   vs.thread0.grf_reg_count = ALIGN(brw->vs.prog_data->total_grf, 16) / 16 - 1;
    vs.thread3.urb_entry_read_length = brw->vs.prog_data->urb_read_length;
    vs.thread3.const_urb_entry_read_length = brw->vs.prog_data->curb_read_length;
    vs.thread3.dispatch_grf_start_reg = 1;
diff --git a/src/mesa/drivers/dri/i965/brw_wm_state.c b/src/mesa/drivers/dri/i965/brw_wm_state.c
index 5b4f2abd0e..351de6d90e 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_state.c
@@ -62,7 +62,7 @@ static void upload_wm_unit(struct brw_context *brw )
    memset(&wm, 0, sizeof(wm));
 
    /* CACHE_NEW_WM_PROG */
-   wm.thread0.grf_reg_count = ((brw->wm.prog_data->total_grf-1) & ~15) / 16;
+   wm.thread0.grf_reg_count = ALIGN(brw->wm.prog_data->total_grf, 16) / 16 - 1;
    wm.thread0.kernel_start_pointer = brw->wm.prog_gs_offset >> 6;
    wm.thread3.dispatch_grf_start_reg = brw->wm.prog_data->first_curbe_grf;
    wm.thread3.urb_entry_read_length = brw->wm.prog_data->urb_read_length;
@@ -71,7 +71,7 @@ static void upload_wm_unit(struct brw_context *brw )
    wm.wm5.max_threads = max_threads;      
 
    if (brw->wm.prog_data->total_scratch) {
-      GLuint per_thread = (brw->wm.prog_data->total_scratch + 1023) / 1024;
+      GLuint per_thread = ALIGN(brw->wm.prog_data->total_scratch, 1024);
       GLuint total = per_thread * (max_threads + 1);
 
       /* Scratch space -- just have to make sure there is sufficient
diff --git a/src/mesa/drivers/dri/i965/bufmgr_fake.c b/src/mesa/drivers/dri/i965/bufmgr_fake.c
index a85121122f..65760c40d4 100644
--- a/src/mesa/drivers/dri/i965/bufmgr_fake.c
+++ b/src/mesa/drivers/dri/i965/bufmgr_fake.c
@@ -168,7 +168,7 @@ static GLboolean alloc_from_pool( struct intel_context *intel,
    if (!block)
       return GL_FALSE;
 
-   sz = (buf->size + align-1) & ~(align-1);
+   sz = ALIGN(buf->size, align);
 
    block->mem = mmAllocMem(pool->heap, 
 			   sz, 
diff --git a/src/mesa/drivers/dri/i965/intel_batchbuffer.c b/src/mesa/drivers/dri/i965/intel_batchbuffer.c
index fb58c0e708..7a6293b557 100644
--- a/src/mesa/drivers/dri/i965/intel_batchbuffer.c
+++ b/src/mesa/drivers/dri/i965/intel_batchbuffer.c
@@ -37,7 +37,7 @@ static void intel_batchbuffer_reset( struct intel_batchbuffer *batch )
    assert(batch->map == NULL);
 
    batch->offset = (unsigned long)batch->ptr;
-   batch->offset = (batch->offset + 63) & ~63;
+   batch->offset = ALIGN(batch->offset, 64);
    batch->ptr = (unsigned char *) batch->offset;
 
    if (BATCH_SZ - batch->offset < BATCH_REFILL) {
@@ -208,7 +208,7 @@ void intel_batchbuffer_align( struct intel_batchbuffer *batch,
 			      GLuint sz )
 {
    unsigned long ptr = (unsigned long) batch->ptr;
-   unsigned long aptr = (ptr + align) & ~((unsigned long)align-1);
+   unsigned long aptr = ALIGN(ptr, align);
    GLuint fixup = aptr - ptr;
 
    if (intel_batchbuffer_space(batch) < fixup + sz)
diff --git a/src/mesa/drivers/dri/i965/intel_blit.c b/src/mesa/drivers/dri/i965/intel_blit.c
index 7a9e1a2a3f..d1c1c8afb6 100644
--- a/src/mesa/drivers/dri/i965/intel_blit.c
+++ b/src/mesa/drivers/dri/i965/intel_blit.c
@@ -533,7 +533,7 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel,
 				  GLenum logic_op)
 {
    struct xy_text_immediate_blit text;
-   int dwords = ((src_size + 7) & ~7) / 4;
+   int dwords = ALIGN(src_size, 8) / 4;
    uint32_t opcode, br13;
 
    assert( logic_op - GL_CLEAR >= 0 );
diff --git a/src/mesa/drivers/dri/i965/intel_context.h b/src/mesa/drivers/dri/i965/intel_context.h
index f63c2f613d..65898caaa7 100644
--- a/src/mesa/drivers/dri/i965/intel_context.h
+++ b/src/mesa/drivers/dri/i965/intel_context.h
@@ -252,6 +252,8 @@ void UNLOCK_HARDWARE( struct intel_context *intel );
 #define SUBPIXEL_X 0.125
 #define SUBPIXEL_Y 0.125
 
+#define ALIGN(value, alignment)  ((value + alignment - 1) & ~(alignment - 1))
+
 /* ================================================================
  * Color packing:
  */
diff --git a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
index 0fb33e27f4..268a982a97 100644
--- a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
+++ b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
@@ -233,8 +233,8 @@ GLboolean intel_miptree_image_data(struct intel_context *intel,
 
    if (dst->compressed) {
        alignment = intel_compressed_alignment(dst->internal_format);
-       src_row_pitch = ((src_row_pitch + alignment - 1) & ~(alignment - 1));
-       width = ((width + alignment - 1) & ~(alignment - 1));
+       src_row_pitch = ALIGN(src_row_pitch, alignment);
+       width = ALIGN(width, alignment);
        height = (height + 3) / 4;
    }
 
diff --git a/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c b/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c
index 79c1fee9c0..3777422619 100644
--- a/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c
+++ b/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c
@@ -91,11 +91,6 @@ static void set_bit( GLubyte *dest,
    dest[bit/8] |= 1 << (bit % 8);
 }
 
-static int align(int x, int align)
-{
-   return (x + align - 1) & ~(align - 1);
-}
-
 /* Extract a rectangle's worth of data from the bitmap.  Called
  * per-cliprect.
  */
@@ -147,7 +142,7 @@ static GLuint get_bitmap_rect(GLsizei width, GLsizei height,
       }
 
       if (row_align)
-	 bit = (bit + row_align - 1) & ~(row_align - 1);
+	 bit = ALIGN(bit, row_align);
    }
 
    return count;
@@ -268,7 +263,7 @@ do_blit_bitmap( GLcontext *ctx,
 	    for (px = 0; px < box_w; px += DX) { 
 	       int h = MIN2(DY, box_h - py);
 	       int w = MIN2(DX, box_w - px); 
-	       GLuint sz = align(align(w,8) * h, 64)/8;
+	       GLuint sz = ALIGN(ALIGN(w,8) * h, 64)/8;
 	       GLenum logic_op = ctx->Color.ColorLogicOpEnabled ?
 		  ctx->Color.LogicOp : GL_COPY;
 
diff --git a/src/mesa/drivers/dri/intel/intel_tex_layout.c b/src/mesa/drivers/dri/intel/intel_tex_layout.c
index fdecd3e186..e3c6e1c17c 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_layout.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_layout.c
@@ -34,12 +34,6 @@
 #include "intel_tex_layout.h"
 #include "macros.h"
 
-
-static int align(int value, int alignment)
-{
-   return (value + alignment - 1) & ~(alignment - 1);
-}
-
 GLuint intel_compressed_alignment(GLenum internalFormat)
 {
     GLuint alignment = 4;
@@ -70,7 +64,7 @@ void i945_miptree_layout_2d( struct intel_mipmap_tree *mt )
 
    if (mt->compressed) {
        align_w = intel_compressed_alignment(mt->internal_format);
-       mt->pitch = align(mt->width0, align_w);
+       mt->pitch = ALIGN(mt->width0, align_w);
    }
 
    /* May need to adjust pitch to accomodate the placement of
@@ -82,10 +76,10 @@ void i945_miptree_layout_2d( struct intel_mipmap_tree *mt )
        GLuint mip1_width;
 
        if (mt->compressed) {
-           mip1_width = align(minify(mt->width0), align_w)
-               + align(minify(minify(mt->width0)), align_w);
+           mip1_width = ALIGN(minify(mt->width0), align_w)
+               + ALIGN(minify(minify(mt->width0)), align_w);
        } else {
-           mip1_width = align(minify(mt->width0), align_w)
+           mip1_width = ALIGN(minify(mt->width0), align_w)
                + minify(minify(mt->width0));
        }
 
@@ -97,7 +91,7 @@ void i945_miptree_layout_2d( struct intel_mipmap_tree *mt )
    /* Pitch must be a whole number of dwords, even though we
     * express it in texels.
     */
-   mt->pitch = align(mt->pitch * mt->cpp, 4) / mt->cpp;
+   mt->pitch = ALIGN(mt->pitch * mt->cpp, 4) / mt->cpp;
    mt->total_height = 0;
 
    for ( level = mt->first_level ; level <= mt->last_level ; level++ ) {
@@ -109,7 +103,7 @@ void i945_miptree_layout_2d( struct intel_mipmap_tree *mt )
       if (mt->compressed)
 	 img_height = MAX2(1, height/4);
       else
-	 img_height = align(height, align_h);
+	 img_height = ALIGN(height, align_h);
 
 
       /* Because the images are packed better, the final offset
@@ -120,7 +114,7 @@ void i945_miptree_layout_2d( struct intel_mipmap_tree *mt )
       /* Layout_below: step right after second mipmap.
        */
       if (level == mt->first_level + 1) {
-	 x += align(width, align_w);
+	 x += ALIGN(width, align_w);
       }
       else {
 	 y += img_height;