Merge commit 'origin/mesa_7_7_branch'

13 files changed, 97 insertions, 56 deletions

diff --git a/src/mesa/drivers/dri/i965/brw_state.h b/src/mesa/drivers/dri/i965/brw_state.h
index ab6f158080..b129b1f1c3 100644
--- a/src/mesa/drivers/dri/i965/brw_state.h
+++ b/src/mesa/drivers/dri/i965/brw_state.h
@@ -112,6 +112,7 @@ void brw_validate_state(struct brw_context *brw);
 void brw_upload_state(struct brw_context *brw);
 void brw_init_state(struct brw_context *brw);
 void brw_destroy_state(struct brw_context *brw);
+void brw_clear_validated_bos(struct brw_context *brw);
 
 /***********************************************************************
  * brw_state_cache.c
diff --git a/src/mesa/drivers/dri/i965/brw_state_upload.c b/src/mesa/drivers/dri/i965/brw_state_upload.c
index f4283bda1b..af8dfb4c15 100644
--- a/src/mesa/drivers/dri/i965/brw_state_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_state_upload.c
@@ -143,7 +143,7 @@ static void xor_states( struct brw_state_flags *result,
    result->cache = a->cache ^ b->cache;
 }
 
-static void
+void
 brw_clear_validated_bos(struct brw_context *brw)
 {
    int i;
diff --git a/src/mesa/drivers/dri/i965/brw_vtbl.c b/src/mesa/drivers/dri/i965/brw_vtbl.c
index dc47f08dd4..72749b3859 100644
--- a/src/mesa/drivers/dri/i965/brw_vtbl.c
+++ b/src/mesa/drivers/dri/i965/brw_vtbl.c
@@ -46,7 +46,7 @@
 #include "brw_state.h"
 #include "brw_fallback.h"
 #include "brw_vs.h"
-
+#include "brw_wm.h"
 
 static void
 dri_bo_release(dri_bo **bo)
@@ -66,8 +66,14 @@ static void brw_destroy_context( struct intel_context *intel )
 
    brw_destroy_state(brw);
    brw_draw_destroy( brw );
-
-   _mesa_free(brw->wm.compile_data);
+   brw_clear_validated_bos(brw);
+   if (brw->wm.compile_data) {
+      _mesa_free(brw->wm.compile_data->instruction);
+      _mesa_free(brw->wm.compile_data->vreg);
+      _mesa_free(brw->wm.compile_data->refs);
+      _mesa_free(brw->wm.compile_data->prog_instructions);
+      _mesa_free(brw->wm.compile_data);
+   }
 
    for (i = 0; i < brw->state.nr_color_regions; i++)
       intel_region_release(&brw->state.color_regions[i]);
diff --git a/src/mesa/drivers/dri/i965/brw_wm.c b/src/mesa/drivers/dri/i965/brw_wm.c
index 77e3b2c32a..6895f64410 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.c
+++ b/src/mesa/drivers/dri/i965/brw_wm.c
@@ -157,7 +157,6 @@ static void do_wm_prog( struct brw_context *brw,
 					  sizeof(*c->prog_instructions));
       c->vreg = _mesa_calloc(BRW_WM_MAX_VREG * sizeof(*c->vreg));
       c->refs = _mesa_calloc(BRW_WM_MAX_REF * sizeof(*c->refs));
-      c->vreg = _mesa_calloc(BRW_WM_MAX_VREG * sizeof(*c->vreg));
    } else {
       void *instruction = c->instruction;
       void *prog_instructions = c->prog_instructions;
diff --git a/src/mesa/drivers/dri/intel/intel_tex_copy.c b/src/mesa/drivers/dri/intel/intel_tex_copy.c
index 4b5fe7be9f..767d04d2f4 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_copy.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_copy.c
@@ -221,6 +221,8 @@ intelCopyTexImage1D(GLcontext * ctx, GLenum target, GLint level,
    return;
 
  fail:
+   if (INTEL_DEBUG & DEBUG_FALLBACKS)
+      fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
    _mesa_meta_CopyTexImage1D(ctx, target, level, internalFormat, x, y,
                              width, border);
 }
@@ -268,6 +270,8 @@ intelCopyTexImage2D(GLcontext * ctx, GLenum target, GLint level,
    return;
 
  fail:
+   if (INTEL_DEBUG & DEBUG_FALLBACKS)
+      fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
    _mesa_meta_CopyTexImage2D(ctx, target, level, internalFormat, x, y,
                              width, height, border);
 }
@@ -292,6 +296,8 @@ intelCopyTexSubImage1D(GLcontext * ctx, GLenum target, GLint level,
    if (!do_copy_texsubimage(intel_context(ctx), target,
                             intel_texture_image(texImage),
                             internalFormat, xoffset, 0, x, y, width, 1)) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+         fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
       _mesa_meta_CopyTexSubImage1D(ctx, target, level, xoffset, x, y, width);
    }
 }
@@ -317,8 +323,8 @@ intelCopyTexSubImage2D(GLcontext * ctx, GLenum target, GLint level,
                             internalFormat,
                             xoffset, yoffset, x, y, width, height)) {
 
-      DBG("%s - fallback to _mesa_meta_CopyTexSubImage2D\n", __FUNCTION__);
-
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+         fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
       _mesa_meta_CopyTexSubImage2D(ctx, target, level,
                                    xoffset, yoffset, x, y, width, height);
    }
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program.h b/src/mesa/drivers/dri/r300/compiler/radeon_program.h
index 33db3ea0ff..03592884eb 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_program.h
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program.h
@@ -191,7 +191,7 @@ struct rc_src_register lmul_swizzle(unsigned int swizzle, struct rc_src_register
 
 static inline void reset_srcreg(struct rc_src_register* reg)
 {
-	memset(reg, 0, sizeof(reg));
+	memset(reg, 0, sizeof(struct rc_src_register));
 	reg->Swizzle = RC_SWIZZLE_XYZW;
 }
 
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c
index 0326d25233..ced66af1eb 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c
@@ -560,23 +560,23 @@ static void sincos_constants(struct radeon_compiler* c, unsigned int *constants)
  * MAD dest, tmp.y, weight, tmp.x
  */
 static void sin_approx(
-	struct radeon_compiler* c, struct rc_instruction * before,
+	struct radeon_compiler* c, struct rc_instruction * inst,
 	struct rc_dst_register dst, struct rc_src_register src, const unsigned int* constants)
 {
 	unsigned int tempreg = rc_find_free_temporary(c);
 
-	emit2(c, before, RC_OPCODE_MUL, 0, dstregtmpmask(tempreg, RC_MASK_XY),
+	emit2(c, inst->Prev, RC_OPCODE_MUL, 0, dstregtmpmask(tempreg, RC_MASK_XY),
 		swizzle(src, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X),
 		srcreg(RC_FILE_CONSTANT, constants[0]));
-	emit3(c, before, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_X),
+	emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_X),
 		swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y),
 		absolute(swizzle(src, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X)),
 		swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X));
-	emit3(c, before, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_Y),
+	emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_Y),
 		swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X),
 		absolute(swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X)),
 		negate(swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X)));
-	emit3(c, before, RC_OPCODE_MAD, 0, dst,
+	emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dst,
 		swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y),
 		swizzle(srcreg(RC_FILE_CONSTANT, constants[0]), RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W),
 		swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X));
diff --git a/src/mesa/drivers/dri/radeon/radeon_common_context.c b/src/mesa/drivers/dri/radeon/radeon_common_context.c
index 2a38c4599c..71f70d724b 100644
--- a/src/mesa/drivers/dri/radeon/radeon_common_context.c
+++ b/src/mesa/drivers/dri/radeon/radeon_common_context.c
@@ -262,7 +262,7 @@ GLboolean radeonInitContext(radeonContextPtr radeon,
 		else
 			radeon->texture_row_align = 32;
 		radeon->texture_rect_row_align = 64;
-		radeon->texture_compressed_row_align = 64;
+		radeon->texture_compressed_row_align = 32;
 	}
 
 	radeon_init_dma(radeon);
diff --git a/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c b/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c
index 0497fa7db5..46603de2e7 100644
--- a/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c
+++ b/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c
@@ -37,26 +37,35 @@
 #include "main/texobj.h"
 #include "radeon_texture.h"
 
-static GLuint radeon_compressed_texture_size(GLcontext *ctx,
-		GLsizei width, GLsizei height, GLsizei depth,
-		gl_format mesaFormat)
+static unsigned get_aligned_compressed_row_stride(
+		gl_format format,
+		unsigned width,
+		unsigned minStride)
 {
-	GLuint size = _mesa_format_image_size(mesaFormat, width, height, depth);
-
-	if (mesaFormat == MESA_FORMAT_RGB_DXT1 ||
-	    mesaFormat == MESA_FORMAT_RGBA_DXT1) {
-		if (width + 3 < 8)	/* width one block */
-			size = size * 4;
-		else if (width + 3 < 16)
-			size = size * 2;
-	} else {
-		/* DXT3/5, 16 bytes per block */
-	  //		WARN_ONCE("DXT 3/5 suffers from multitexturing problems!\n");
-		if (width + 3 < 8)
-			size = size * 2;
+	const unsigned blockSize = _mesa_get_format_bytes(format);
+	unsigned blockWidth, blockHeight, numXBlocks;
+
+	_mesa_get_format_block_size(format, &blockWidth, &blockHeight);
+	numXBlocks = (width + blockWidth - 1) / blockWidth;
+
+	while (numXBlocks * blockSize < minStride)
+	{
+		++numXBlocks;
 	}
 
-	return size;
+	return numXBlocks * blockSize;
+}
+
+static unsigned get_compressed_image_size(
+		gl_format format,
+		unsigned rowStride,
+		unsigned height)
+{
+	unsigned blockWidth, blockHeight;
+
+	_mesa_get_format_block_size(format, &blockWidth, &blockHeight);
+
+	return rowStride * ((height + blockHeight - 1) / blockHeight);
 }
 
 /**
@@ -74,10 +83,8 @@ static void compute_tex_image_offset(radeonContextPtr rmesa, radeon_mipmap_tree
 
 	/* Find image size in bytes */
 	if (_mesa_is_format_compressed(mt->mesaFormat)) {
-		/* TODO: Is this correct? Need test cases for compressed textures! */
-		row_align = rmesa->texture_compressed_row_align - 1;
-		lvl->rowstride = (_mesa_format_row_stride(mt->mesaFormat, lvl->width) + row_align) & ~row_align;
-		lvl->size = radeon_compressed_texture_size(rmesa->glCtx, lvl->width, lvl->height, lvl->depth, mt->mesaFormat);
+		lvl->rowstride = get_aligned_compressed_row_stride(mt->mesaFormat, lvl->width, rmesa->texture_compressed_row_align);
+		lvl->size = get_compressed_image_size(mt->mesaFormat, lvl->rowstride, lvl->height);
 	} else if (mt->target == GL_TEXTURE_RECTANGLE_NV) {
 		row_align = rmesa->texture_rect_row_align - 1;
 		lvl->rowstride = (_mesa_format_row_stride(mt->mesaFormat, lvl->width) + row_align) & ~row_align;
@@ -485,11 +492,12 @@ static radeon_mipmap_tree * get_biggest_matching_miptree(radeonTexObj *texObj,
 														 unsigned firstLevel,
 														 unsigned lastLevel)
 {
-	const unsigned numLevels = lastLevel - firstLevel;
+	const unsigned numLevels = lastLevel - firstLevel + 1;
 	unsigned *mtSizes = calloc(numLevels, sizeof(unsigned));
 	radeon_mipmap_tree **mts = calloc(numLevels, sizeof(radeon_mipmap_tree *));
 	unsigned mtCount = 0;
 	unsigned maxMtIndex = 0;
+	radeon_mipmap_tree *tmp;
 
 	for (unsigned level = firstLevel; level <= lastLevel; ++level) {
 		radeon_texture_image *img = get_radeon_texture_image(texObj->base.Image[0][level]);
@@ -511,7 +519,7 @@ static radeon_mipmap_tree * get_biggest_matching_miptree(radeonTexObj *texObj,
 
 		if (!found) {
 			mtSizes[mtCount] += img->mt->levels[img->mtlevel].size;
-			mts[mtCount++] = img->mt;
+			mts[mtCount] = img->mt;
 			mtCount++;
 		}
 	}
@@ -526,7 +534,11 @@ static radeon_mipmap_tree * get_biggest_matching_miptree(radeonTexObj *texObj,
 		}
 	}
 
-	return mts[maxMtIndex];
+	tmp = mts[maxMtIndex];
+	free(mtSizes);
+	free(mts);
+
+	return tmp;
 }
 
 /**
diff --git a/src/mesa/drivers/dri/radeon/radeon_texture.c b/src/mesa/drivers/dri/radeon/radeon_texture.c
index 6f11f1fa4a..1ee9e2792a 100644
--- a/src/mesa/drivers/dri/radeon/radeon_texture.c
+++ b/src/mesa/drivers/dri/radeon/radeon_texture.c
@@ -582,12 +582,12 @@ static void radeon_store_teximage(GLcontext* ctx, int dims,
 		/* TODO */
 		assert(0);
 	} else {
-		dstRowStride = _mesa_format_row_stride(texImage->TexFormat, width);
+		dstRowStride = _mesa_format_row_stride(texImage->TexFormat, texImage->Width);
 	}
 
 	if (dims == 3) {
 		unsigned alignedWidth = dstRowStride/_mesa_get_format_bytes(texImage->TexFormat);
-		dstImageOffsets = allocate_image_offsets(ctx, alignedWidth, height, depth);
+		dstImageOffsets = allocate_image_offsets(ctx, alignedWidth, texImage->Height, texImage->Depth);
 		if (!dstImageOffsets) {
 			return;
 		}
@@ -598,8 +598,11 @@ static void radeon_store_teximage(GLcontext* ctx, int dims,
 	radeon_teximage_map(image, GL_TRUE);
 
 	if (compressed) {
-		uint32_t srcRowStride, bytesPerRow, rows;
+		uint32_t srcRowStride, bytesPerRow, rows, block_width, block_height;
 		GLubyte *img_start;
+
+		_mesa_get_format_block_size(texImage->TexFormat, &block_width, &block_height);
+
 		if (!image->mt) {
 			dstRowStride = _mesa_format_row_stride(texImage->TexFormat, texImage->Width);
 			img_start = _mesa_compressed_image_address(xoffset, yoffset, 0,
@@ -607,17 +610,16 @@ static void radeon_store_teximage(GLcontext* ctx, int dims,
 									texImage->Width, texImage->Data);
 		}
 		else {
-			uint32_t blocks_x, block_width, block_height;
-			_mesa_get_format_block_size(image->mt->mesaFormat, &block_width, &block_height);
-			blocks_x = dstRowStride / block_width;
-			img_start = texImage->Data + _mesa_get_format_bytes(image->mt->mesaFormat) * 4 * (blocks_x * (yoffset / 4) + xoffset / 4);
+			uint32_t offset;
+			offset = dstRowStride / _mesa_get_format_bytes(texImage->TexFormat) * yoffset / block_height + xoffset / block_width;
+			offset *= _mesa_get_format_bytes(texImage->TexFormat);
+			img_start = texImage->Data + offset;
 		}
 		srcRowStride = _mesa_format_row_stride(texImage->TexFormat, width);
 		bytesPerRow = srcRowStride;
-		rows = (height + 3) / 4;
-
-		copy_rows(img_start, dstRowStride,  pixels, srcRowStride, rows,  bytesPerRow);
+		rows = (height + block_height - 1) / block_height;
 
+		copy_rows(img_start, dstRowStride, pixels, srcRowStride, rows, bytesPerRow);
 	}
 	else {
 		if (!_mesa_texstore(ctx, dims, texImage->_BaseFormat,
diff --git a/src/mesa/main/texfetch.c b/src/mesa/main/texfetch.c
index f4f2be48c3..b37039429f 100644
--- a/src/mesa/main/texfetch.c
+++ b/src/mesa/main/texfetch.c
@@ -570,7 +570,7 @@ texfetch_funcs[MESA_FORMAT_COUNT] =
 static FetchTexelFuncF
 _mesa_get_texel_fetch_func(gl_format format, GLuint dims)
 {
-   FetchTexelFuncF f;
+   FetchTexelFuncF f = NULL;
    GLuint i;
    /* XXX replace loop with direct table lookup */
    for (i = 0; i < MESA_FORMAT_COUNT; i++) {
diff --git a/src/mesa/main/texgetimage.c b/src/mesa/main/texgetimage.c
index 2f88718933..008407210d 100644
--- a/src/mesa/main/texgetimage.c
+++ b/src/mesa/main/texgetimage.c
@@ -521,10 +521,11 @@ _mesa_get_compressed_teximage(GLcontext *ctx, GLenum target, GLint level,
                               struct gl_texture_object *texObj,
                               struct gl_texture_image *texImage)
 {
-   const GLuint size = _mesa_format_image_size(texImage->TexFormat,
-                                               texImage->Width,
-                                               texImage->Height,
-                                               texImage->Depth);
+   const GLuint row_stride = _mesa_format_row_stride(texImage->TexFormat,
+                                                     texImage->Width);
+   const GLuint row_stride_stored = _mesa_format_row_stride(texImage->TexFormat,
+                                                            texImage->RowStride);
+   GLuint i;
 
    if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) {
       /* pack texture image into a PBO */
@@ -540,8 +541,22 @@ _mesa_get_compressed_teximage(GLcontext *ctx, GLenum target, GLint level,
       img = ADD_POINTERS(buf, img);
    }
 
-   /* just memcpy, no pixelstore or pixel transfer */
-   _mesa_memcpy(img, texImage->Data, size);
+   /* no pixelstore or pixel transfer, but respect stride */
+
+   if (row_stride == row_stride_stored) {
+      const GLuint size = _mesa_format_image_size(texImage->TexFormat,
+                                                  texImage->Width,
+                                                  texImage->Height,
+                                                  texImage->Depth);
+      _mesa_memcpy(img, texImage->Data, size);
+   }
+   else {
+      GLuint bw, bh;
+      _mesa_get_format_block_size(texImage->TexFormat, &bw, &bh);
+      for (i = 0; i < (texImage->Height + bh - 1) / bh; i++) {
+         memcpy(img + i * row_stride, texImage->Data + i * row_stride_stored, row_stride);
+      }
+   }
 
    if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) {
       ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_PACK_BUFFER_EXT,
diff --git a/src/mesa/shader/prog_optimize.c b/src/mesa/shader/prog_optimize.c
index 3d28d885a4..4fe351251e 100644
--- a/src/mesa/shader/prog_optimize.c
+++ b/src/mesa/shader/prog_optimize.c
@@ -443,7 +443,7 @@ _mesa_remove_extra_move_use(struct gl_program *prog)
     *    FOO tmpY, arg0, arg1;
     */
 
-   for (i = 0; i < prog->NumInstructions - 1; i++) {
+   for (i = 0; i + 1 < prog->NumInstructions; i++) {
       const struct prog_instruction *mov = prog->Instructions + i;
 
       if (mov->Opcode != OPCODE_MOV ||