diff options
Diffstat (limited to 'src/gallium/drivers')
97 files changed, 2688 insertions, 1593 deletions
diff --git a/src/gallium/drivers/cell/ppu/cell_clear.c b/src/gallium/drivers/cell/ppu/cell_clear.c index 79ad687ea9..3a3f968a49 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_gen_fp.c b/src/gallium/drivers/cell/ppu/cell_gen_fp.c index 1895a7940c..1d8a11a4ac 100644 --- a/src/gallium/drivers/cell/ppu/cell_gen_fp.c +++ b/src/gallium/drivers/cell/ppu/cell_gen_fp.c @@ -995,7 +995,7 @@ 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); diff --git a/src/gallium/drivers/cell/ppu/cell_texture.c b/src/gallium/drivers/cell/ppu/cell_texture.c index 605d53a948..bc2e625f54 100644 --- a/src/gallium/drivers/cell/ppu/cell_texture.c +++ b/src/gallium/drivers/cell/ppu/cell_texture.c @@ -67,14 +67,11 @@ cell_texture_layout(struct cell_texture *ct) w_tile = align(width, TILE_SIZE); h_tile = align(height, TILE_SIZE); - 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] = pf_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] * pf_get_nblocksy(pt->format, h_tile); if (pt->target == PIPE_TEXTURE_CUBE) size *= 6; else @@ -285,10 +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 * pf_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 * pf_get_nblocksy(ps->format, h_tile) * ct->stride[level]; } else { assert(face == 0); @@ -329,14 +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; @@ -346,10 +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 * pf_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 * pf_get_nblocksy(texture->format, h_tile) * pt->stride; } else { assert(face == 0); @@ -402,7 +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 = pf_get_stride(pt->format, align(texWidth, TILE_SIZE)) * + pf_get_nblocksy(pt->format, align(texHeight, TILE_SIZE)); ctrans->map = align_malloc(size, 16); if (!ctrans->map) diff --git a/src/gallium/drivers/i915/i915_surface.c b/src/gallium/drivers/i915/i915_surface.c index ab8331f3e6..24e1024aaa 100644 --- a/src/gallium/drivers/i915/i915_surface.c +++ b/src/gallium/drivers/i915/i915_surface.c @@ -48,17 +48,19 @@ i915_surface_copy(struct pipe_context *pipe, { struct i915_texture *dst_tex = (struct i915_texture *)dst->texture; struct i915_texture *src_tex = (struct i915_texture *)src->texture; + struct pipe_texture *dpt = &dst_tex->base; + struct pipe_texture *spt = &src_tex->base; assert( dst != src ); - assert( dst_tex->base.block.size == src_tex->base.block.size ); - assert( dst_tex->base.block.width == src_tex->base.block.height ); - assert( dst_tex->base.block.height == src_tex->base.block.height ); - assert( dst_tex->base.block.width == 1 ); - assert( dst_tex->base.block.height == 1 ); + assert( pf_get_blocksize(dpt->format) == pf_get_blocksize(spt->format) ); + assert( pf_get_blockwidth(dpt->format) == pf_get_blockwidth(spt->format) ); + assert( pf_get_blockheight(dpt->format) == pf_get_blockheight(spt->format) ); + assert( pf_get_blockwidth(dpt->format) == 1 ); + assert( pf_get_blockheight(dpt->format) == 1 ); i915_copy_blit( i915_context(pipe), FALSE, - dst_tex->base.block.size, + pf_get_blocksize(dpt->format), (unsigned short) src_tex->stride, src_tex->buffer, src->offset, (unsigned short) dst_tex->stride, dst_tex->buffer, dst->offset, (short) srcx, (short) srcy, (short) dstx, (short) dsty, (short) width, (short) height ); @@ -72,12 +74,13 @@ i915_surface_fill(struct pipe_context *pipe, unsigned width, unsigned height, unsigned value) { struct i915_texture *tex = (struct i915_texture *)dst->texture; + struct pipe_texture *pt = &tex->base; - assert(tex->base.block.width == 1); - assert(tex->base.block.height == 1); + assert(pf_get_blockwidth(pt->format) == 1); + assert(pf_get_blockheight(pt->format) == 1); i915_fill_blit( i915_context(pipe), - tex->base.block.size, + pf_get_blocksize(pt->format), (unsigned short) tex->stride, tex->buffer, dst->offset, (short) dstx, (short) dsty, diff --git a/src/gallium/drivers/i915/i915_texture.c b/src/gallium/drivers/i915/i915_texture.c index c7b86dd4c5..b28b413771 100644 --- a/src/gallium/drivers/i915/i915_texture.c +++ b/src/gallium/drivers/i915/i915_texture.c @@ -74,6 +74,9 @@ static const int step_offsets[6][2] = { {-1, 1} }; +/* XXX really need twice the size if x is already pot? + Otherwise just use util_next_power_of_two? +*/ static unsigned power_of_two(unsigned x) { @@ -83,13 +86,6 @@ power_of_two(unsigned x) return value; } -static unsigned -round_up(unsigned n, unsigned multiple) -{ - return (n + multiple - 1) & ~(multiple - 1); -} - - /* * More advanced helper funcs */ @@ -101,13 +97,8 @@ i915_miptree_set_level_info(struct i915_texture *tex, unsigned nr_images, unsigned w, unsigned h, unsigned d) { - struct pipe_texture *pt = &tex->base; - assert(level < PIPE_MAX_TEXTURE_LEVELS); - pt->nblocksx[level] = pf_get_nblocksx(&pt->block, w); - pt->nblocksy[level] = pf_get_nblocksy(&pt->block, h); - tex->nr_images[level] = nr_images; /* @@ -138,7 +129,7 @@ i915_miptree_set_image_offset(struct i915_texture *tex, assert(img < tex->nr_images[level]); - tex->image_offset[level][img] = y * tex->stride + x * tex->base.block.size; + tex->image_offset[level][img] = y * tex->stride + x * pf_get_blocksize(tex->base.format); /* printf("%s level %d img %d pos %d,%d image_offset %x\n", @@ -160,28 +151,28 @@ i915_scanout_layout(struct i915_texture *tex) { struct pipe_texture *pt = &tex->base; - if (pt->last_level > 0 || pt->block.size != 4) + if (pt->last_level > 0 || pf_get_blocksize(pt->format) != 4) return FALSE; i915_miptree_set_level_info(tex, 0, 1, - tex->base.width0, - tex->base.height0, + pt->width0, + pt->height0, 1); i915_miptree_set_image_offset(tex, 0, 0, 0, 0); - if (tex->base.width0 >= 240) { - tex->stride = power_of_two(tex->base.nblocksx[0] * pt->block.size); - tex->total_nblocksy = round_up(tex->base.nblocksy[0], 8); + if (pt->width0 >= 240) { + tex->stride = power_of_two(pf_get_stride(pt->format, pt->width0)); + tex->total_nblocksy = align(pf_get_nblocksy(pt->format, pt->height0), 8); tex->hw_tiled = INTEL_TILE_X; - } else if (tex->base.width0 == 64 && tex->base.height0 == 64) { - tex->stride = power_of_two(tex->base.nblocksx[0] * pt->block.size); - tex->total_nblocksy = round_up(tex->base.nblocksy[0], 8); + } else if (pt->width0 == 64 && pt->height0 == 64) { + tex->stride = power_of_two(pf_get_stride(pt->format, pt->width0)); + tex->total_nblocksy = align(pf_get_nblocksy(pt->format, pt->height0), 8); } else { return FALSE; } debug_printf("%s size: %d,%d,%d offset %d,%d (0x%x)\n", __FUNCTION__, - tex->base.width0, tex->base.height0, pt->block.size, + pt->width0, pt->height0, pf_get_blocksize(pt->format), tex->stride, tex->total_nblocksy, tex->stride * tex->total_nblocksy); return TRUE; @@ -195,25 +186,25 @@ i915_display_target_layout(struct i915_texture *tex) { struct pipe_texture *pt = &tex->base; - if (pt->last_level > 0 || pt->block.size != 4) + if (pt->last_level > 0 || pf_get_blocksize(pt->format) != 4) return FALSE; /* fallback to normal textures for small textures */ - if (tex->base.width0 < 240) + if (pt->width0 < 240) return FALSE; i915_miptree_set_level_info(tex, 0, 1, - tex->base.width0, - tex->base.height0, + pt->width0, + pt->height0, 1); i915_miptree_set_image_offset(tex, 0, 0, 0, 0); - tex->stride = power_of_two(tex->base.nblocksx[0] * pt->block.size); - tex->total_nblocksy = round_up(tex->base.nblocksy[0], 8); + tex->stride = power_of_two(pf_get_stride(pt->format, pt->width0)); + tex->total_nblocksy = align(pf_get_nblocksy(pt->format, pt->height0), 8); tex->hw_tiled = INTEL_TILE_X; debug_printf("%s size: %d,%d,%d offset %d,%d (0x%x)\n", __FUNCTION__, - tex->base.width0, tex->base.height0, pt->block.size, + pt->width0, pt->height0, pf_get_blocksize(pt->format), tex->stride, tex->total_nblocksy, tex->stride * tex->total_nblocksy); return TRUE; @@ -226,34 +217,32 @@ i915_miptree_layout_2d(struct i915_texture *tex) unsigned level; unsigned width = pt->width0; unsigned height = pt->height0; - unsigned nblocksx = pt->nblocksx[0]; - unsigned nblocksy = pt->nblocksy[0]; + unsigned nblocksy = pf_get_nblocksy(pt->format, pt->width0); /* used for scanouts that need special layouts */ - if (tex->base.tex_usage & PIPE_TEXTURE_USAGE_PRIMARY) + if (pt->tex_usage & PIPE_TEXTURE_USAGE_PRIMARY) if (i915_scanout_layout(tex)) return; /* for shared buffers we use some very like scanout */ - if (tex->base.tex_usage & PIPE_TEXTURE_USAGE_DISPLAY_TARGET) + if (pt->tex_usage & PIPE_TEXTURE_USAGE_DISPLAY_TARGET) if (i915_display_target_layout(tex)) return; - tex->stride = round_up(pt->nblocksx[0] * pt->block.size, 4); + tex->stride = align(pf_get_stride(pt->format, pt->width0), 4); tex->total_nblocksy = 0; for (level = 0; level <= pt->last_level; level++) { i915_miptree_set_level_info(tex, level, 1, width, height, 1); i915_miptree_set_image_offset(tex, level, 0, 0, tex->total_nblocksy); - nblocksy = round_up(MAX2(2, nblocksy), 2); + nblocksy = align(MAX2(2, nblocksy), 2); tex->total_nblocksy += nblocksy; width = u_minify(width, 1); height = u_minify(height, 1); - nblocksx = pf_get_nblocksx(&pt->block, width); - nblocksy = pf_get_nblocksy(&pt->block, height); + nblocksy = pf_get_nblocksy(pt->format, height); } } @@ -266,13 +255,12 @@ i915_miptree_layout_3d(struct i915_texture *tex) unsigned width = pt->width0; unsigned height = pt->height0; unsigned depth = pt->depth0; - unsigned nblocksx = pt->nblocksx[0]; - unsigned nblocksy = pt->nblocksy[0]; + unsigned nblocksy = pf_get_nblocksy(pt->format, pt->height0); unsigned stack_nblocksy = 0; /* Calculate the size of a single slice. */ - tex->stride = round_up(pt->nblocksx[0] * pt->block.size, 4); + tex->stride = align(pf_get_stride(pt->format, pt->width0), 4); /* XXX: hardware expects/requires 9 levels at minimum. */ @@ -283,8 +271,7 @@ i915_miptree_layout_3d(struct i915_texture *tex) width = u_minify(width, 1); height = u_minify(height, 1); - nblocksx = pf_get_nblocksx(&pt->block, width); - nblocksy = pf_get_nblocksy(&pt->block, height); + nblocksy = pf_get_nblocksy(pt->format, height); } /* Fixup depth image_offsets: @@ -309,14 +296,14 @@ i915_miptree_layout_cube(struct i915_texture *tex) { struct pipe_texture *pt = &tex->base; unsigned width = pt->width0, height = pt->height0; - const unsigned nblocks = pt->nblocksx[0]; + const unsigned nblocks = pf_get_nblocksx(pt->format, pt->width0); unsigned level; unsigned face; assert(width == height); /* cubemap images are square */ /* double pitch for cube layouts */ - tex->stride = round_up(nblocks * pt->block.size * 2, 4); + tex->stride = align(nblocks * pf_get_blocksize(pt->format) * 2, 4); tex->total_nblocksy = nblocks * 4; for (level = 0; level <= pt->last_level; level++) { @@ -379,8 +366,8 @@ i945_miptree_layout_2d(struct i915_texture *tex) unsigned y = 0; unsigned width = pt->width0; unsigned height = pt->height0; - unsigned nblocksx = pt->nblocksx[0]; - unsigned nblocksy = pt->nblocksy[0]; + unsigned nblocksx = pf_get_nblocksx(pt->format, pt->width0); + unsigned nblocksy = pf_get_nblocksy(pt->format, pt->height0); /* used for scanouts that need special layouts */ if (tex->base.tex_usage & PIPE_TEXTURE_USAGE_PRIMARY) @@ -392,7 +379,7 @@ i945_miptree_layout_2d(struct i915_texture *tex) if (i915_display_target_layout(tex)) return; - tex->stride = round_up(pt->nblocksx[0] * pt->block.size, 4); + tex->stride = align(pf_get_stride(pt->format, pt->width0), 4); /* May need to adjust pitch to accomodate the placement of * the 2nd mipmap level. This occurs when the alignment @@ -401,11 +388,11 @@ i945_miptree_layout_2d(struct i915_texture *tex) */ if (pt->last_level > 0) { unsigned mip1_nblocksx - = align(pf_get_nblocksx(&pt->block, u_minify(width, 1)), align_x) - + pf_get_nblocksx(&pt->block, u_minify(width, 2)); + = align(pf_get_nblocksx(pt->format, u_minify(width, 1)), align_x) + + pf_get_nblocksx(pt->format, u_minify(width, 2)); if (mip1_nblocksx > nblocksx) - tex->stride = mip1_nblocksx * pt->block.size; + tex->stride = mip1_nblocksx * pf_get_blocksize(pt->format); } /* Pitch must be a whole number of dwords @@ -435,8 +422,8 @@ i945_miptree_layout_2d(struct i915_texture *tex) width = u_minify(width, 1); height = u_minify(height, 1); - nblocksx = pf_get_nblocksx(&pt->block, width); - nblocksy = pf_get_nblocksy(&pt->block, height); + nblocksx = pf_get_nblocksx(pt->format, width); + nblocksy = pf_get_nblocksy(pt->format, height); } } @@ -447,17 +434,16 @@ i945_miptree_layout_3d(struct i915_texture *tex) unsigned width = pt->width0; unsigned height = pt->height0; unsigned depth = pt->depth0; - unsigned nblocksx = pt->nblocksx[0]; - unsigned nblocksy = pt->nblocksy[0]; + unsigned nblocksy = pf_get_nblocksy(pt->format, pt->width0); unsigned pack_x_pitch, pack_x_nr; unsigned pack_y_pitch; unsigned level; - tex->stride = round_up(pt->nblocksx[0] * pt->block.size, 4); + tex->stride = align(pf_get_stride(pt->format, pt->width0), 4); tex->total_nblocksy = 0; - pack_y_pitch = MAX2(pt->nblocksy[0], 2); - pack_x_pitch = tex->stride / pt->block.size; + pack_y_pitch = MAX2(nblocksy, 2); + pack_x_pitch = tex->stride / pf_get_blocksize(pt->format); pack_x_nr = 1; for (level = 0; level <= pt->last_level; level++) { @@ -482,7 +468,7 @@ i945_miptree_layout_3d(struct i915_texture *tex) if (pack_x_pitch > 4) { pack_x_pitch >>= 1; pack_x_nr <<= 1; - assert(pack_x_pitch * pack_x_nr * pt->block.size <= tex->stride); + assert(pack_x_pitch * pack_x_nr * pf_get_blocksize(pt->format) <= tex->stride); } if (pack_y_pitch > 2) { @@ -492,8 +478,7 @@ i945_miptree_layout_3d(struct i915_texture *tex) width = u_minify(width, 1); height = u_minify(height, 1); depth = u_minify(depth, 1); - nblocksx = pf_get_nblocksx(&pt->block, width); - nblocksy = pf_get_nblocksy(&pt->block, height); + nblocksy = pf_get_nblocksy(pt->format, height); } } @@ -503,7 +488,7 @@ i945_miptree_layout_cube(struct i915_texture *tex) struct pipe_texture *pt = &tex->base; unsigned level; - const unsigned nblocks = pt->nblocksx[0]; + const unsigned nblocks = pf_get_nblocksx(pt->format, pt->width0); unsigned face; unsigned width = pt->width0; unsigned height = pt->height0; @@ -523,9 +508,9 @@ i945_miptree_layout_cube(struct i915_texture *tex) * or the final row of 4x4, 2x2 and 1x1 faces below this. */ if (nblocks > 32) - tex->stride = round_up(nblocks * pt->block.size * 2, 4); + tex->stride = align(nblocks * pf_get_blocksize(pt->format) * 2, 4); else - tex->stride = 14 * 8 * pt->block.size; + tex->stride = 14 * 8 * pf_get_blocksize(pt->format); tex->total_nblocksy = nblocks * 4; @@ -645,9 +630,6 @@ i915_texture_create(struct pipe_screen *screen, pipe_reference_init(&tex->base.reference, 1); tex->base.screen = screen; - tex->base.nblocksx[0] = pf_get_nblocksx(&tex->base.block, tex->base.width0); - tex->base.nblocksy[0] = pf_get_nblocksy(&tex->base.block, tex->base.height0); - if (is->is_i945) { if (!i945_miptree_layout(tex)) goto fail; @@ -829,14 +811,10 @@ i915_get_tex_transfer(struct pipe_screen *screen, trans = CALLOC_STRUCT(i915_transfer); if (trans) { pipe_texture_reference(&trans->base.texture, texture); - trans->base.format = trans->base.format; trans->base.x = x; trans->base.y = y; trans->base.width = w; trans->base.height = h; - trans->base.block = texture->block; - trans->base.nblocksx = texture->nblocksx[level]; - trans->base.nblocksy = texture->nblocksy[level]; trans->base.stride = tex->stride; trans->offset = offset; trans->base.usage = usage; @@ -852,6 +830,7 @@ i915_transfer_map(struct pipe_screen *screen, struct intel_winsys *iws = i915_screen(tex->base.screen)->iws; char *map; boolean write = FALSE; + enum pipe_format format = tex->base.format; if (transfer->usage & PIPE_TRANSFER_WRITE) write = TRUE; @@ -861,8 +840,8 @@ i915_transfer_map(struct pipe_screen *screen, return NULL; return map + i915_transfer(transfer)->offset + - transfer->y / transfer->block.height * transfer->stride + - transfer->x / transfer->block.width * transfer->block.size; + transfer->y / pf_get_blockheight(format) * transfer->stride + + transfer->x / pf_get_blockwidth(format) * pf_get_blocksize(format); } static void diff --git a/src/gallium/drivers/identity/id_context.c b/src/gallium/drivers/identity/id_context.c index 4509c7b1e5..bedab56f59 100644 --- a/src/gallium/drivers/identity/id_context.c +++ b/src/gallium/drivers/identity/id_context.c @@ -742,7 +742,7 @@ identity_context_create(struct pipe_screen *_screen, struct pipe_context *pipe) id_pipe->base.set_polygon_stipple = identity_set_polygon_stipple; id_pipe->base.set_scissor_state = identity_set_scissor_state; id_pipe->base.set_viewport_state = identity_set_viewport_state; - id_pipe->base.set_fragment_sampler_textures = identity_set_vertex_sampler_textures; + id_pipe->base.set_fragment_sampler_textures = identity_set_fragment_sampler_textures; id_pipe->base.set_vertex_sampler_textures = identity_set_vertex_sampler_textures; id_pipe->base.set_vertex_buffers = identity_set_vertex_buffers; id_pipe->base.set_vertex_elements = identity_set_vertex_elements; diff --git a/src/gallium/drivers/llvmpipe/lp_bld_arit.c b/src/gallium/drivers/llvmpipe/lp_bld_arit.c index 9c59677a74..eea6b5d6a5 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_arit.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_arit.c @@ -629,7 +629,8 @@ lp_build_abs(struct lp_build_context *bld, if(type.floating) { /* Mask out the sign bit */ LLVMTypeRef int_vec_type = lp_build_int_vec_type(type); - LLVMValueRef mask = lp_build_int_const_scalar(type, ((unsigned long long)1 << type.width) - 1); + unsigned long absMask = ~(1 << (type.width - 1)); + LLVMValueRef mask = lp_build_int_const_scalar(type, ((unsigned long long) absMask)); a = LLVMBuildBitCast(bld->builder, a, int_vec_type, ""); a = LLVMBuildAnd(bld->builder, a, mask, ""); a = LLVMBuildBitCast(bld->builder, a, vec_type, ""); @@ -1083,7 +1084,7 @@ lp_build_log(struct lp_build_context *bld, LLVMValueRef x) { /* log(2) */ - LLVMValueRef log2 = lp_build_const_scalar(bld->type, 1.4426950408889634); + LLVMValueRef log2 = lp_build_const_scalar(bld->type, 0.69314718055994529); return lp_build_mul(bld, log2, lp_build_exp2(bld, x)); } @@ -1095,7 +1096,7 @@ lp_build_log(struct lp_build_context *bld, /** * Generate polynomial. - * Ex: x^2 * coeffs[0] + x * coeffs[1] + coeffs[2]. + * Ex: coeffs[0] + x * coeffs[1] + x^2 * coeffs[2]. */ static LLVMValueRef lp_build_polynomial(struct lp_build_context *bld, @@ -1285,13 +1286,13 @@ lp_build_log2_approx(struct lp_build_context *bld, /* mant = (float) mantissa(x) */ mant = LLVMBuildAnd(bld->builder, i, mantmask, ""); mant = LLVMBuildOr(bld->builder, mant, one, ""); - mant = LLVMBuildSIToFP(bld->builder, mant, vec_type, ""); + mant = LLVMBuildBitCast(bld->builder, mant, vec_type, ""); logmant = lp_build_polynomial(bld, mant, lp_build_log2_polynomial, Elements(lp_build_log2_polynomial)); /* This effectively increases the polynomial degree by one, but ensures that log2(1) == 0*/ - logmant = LLVMBuildMul(bld->builder, logmant, LLVMBuildMul(bld->builder, mant, bld->one, ""), ""); + logmant = LLVMBuildMul(bld->builder, logmant, LLVMBuildSub(bld->builder, mant, bld->one, ""), ""); res = LLVMBuildAdd(bld->builder, logmant, logexp, ""); } diff --git a/src/gallium/drivers/llvmpipe/lp_bld_debug.c b/src/gallium/drivers/llvmpipe/lp_bld_debug.c index 59d8f492e6..39dfc51e50 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_debug.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_debug.c @@ -77,10 +77,10 @@ lp_disassemble(const void* func) while (ud_disassemble(&ud_obj)) { #ifdef PIPE_ARCH_X86 - debug_printf("%08lx: ", (unsigned long)ud_insn_off(&ud_obj)); + debug_printf("0x%08lx:\t", (unsigned long)ud_insn_off(&ud_obj)); #endif #ifdef PIPE_ARCH_X86_64 - debug_printf("%016llx: ", (unsigned long long)ud_insn_off(&ud_obj)); + debug_printf("0x%016llx:\t", (unsigned long long)ud_insn_off(&ud_obj)); #endif #if 0 @@ -115,9 +115,16 @@ lp_disassemble(const void* func) } } - if (ud_insn_off(&ud_obj) >= max_jmp_pc && ud_obj.mnemonic == UD_Iret) + if ((ud_insn_off(&ud_obj) >= max_jmp_pc && ud_obj.mnemonic == UD_Iret) || + ud_obj.mnemonic == UD_Iinvalid) break; } + +#if 0 + /* Print GDB command, useful to verify udis86 output */ + debug_printf("disassemble %p %p\n", func, (void*)(uintptr_t)ud_obj.pc); +#endif + debug_printf("\n"); #else (void)func; diff --git a/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c b/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c index fe2db04d8f..a67c70ff25 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c @@ -560,9 +560,9 @@ emit_instruction( if (IS_DST0_CHANNEL_ENABLED( inst, CHAN_X ) || IS_DST0_CHANNEL_ENABLED( inst, CHAN_Y ) || IS_DST0_CHANNEL_ENABLED( inst, CHAN_Z )) { - LLVMValueRef *p_floor_log2; - LLVMValueRef *p_exp; - LLVMValueRef *p_log2; + LLVMValueRef *p_floor_log2 = NULL; + LLVMValueRef *p_exp = NULL; + LLVMValueRef *p_log2 = NULL; src0 = emit_fetch( bld, inst, 0, CHAN_X ); src0 = lp_build_abs( &bld->base, src0 ); @@ -763,7 +763,7 @@ emit_instruction( FOR_EACH_DST0_ENABLED_CHANNEL( inst, chan_index ) { src0 = emit_fetch( bld, inst, 0, chan_index ); tmp0 = lp_build_floor(&bld->base, src0); - tmp0 = lp_build_sub(&bld->base, tmp0, src0); + tmp0 = lp_build_sub(&bld->base, src0, tmp0); dst0[chan_index] = tmp0; } break; diff --git a/src/gallium/drivers/llvmpipe/lp_clear.c b/src/gallium/drivers/llvmpipe/lp_clear.c index bdcff94b9b..08d9f2e273 100644 --- a/src/gallium/drivers/llvmpipe/lp_clear.c +++ b/src/gallium/drivers/llvmpipe/lp_clear.c @@ -50,6 +50,7 @@ llvmpipe_clear(struct pipe_context *pipe, unsigned buffers, const float *rgba, double depth, unsigned stencil) { struct llvmpipe_context *llvmpipe = llvmpipe_context(pipe); + union util_color uc; unsigned cv; uint i; @@ -64,8 +65,8 @@ llvmpipe_clear(struct pipe_context *pipe, unsigned buffers, const float *rgba, for (i = 0; i < llvmpipe->framebuffer.nr_cbufs; i++) { struct pipe_surface *ps = llvmpipe->framebuffer.cbufs[i]; - util_pack_color(rgba, ps->format, &cv); - lp_tile_cache_clear(llvmpipe->cbuf_cache[i], rgba, cv); + util_pack_color(rgba, ps->format, &uc); + lp_tile_cache_clear(llvmpipe->cbuf_cache[i], rgba, uc.ui); } llvmpipe->dirty_render_cache = TRUE; } diff --git a/src/gallium/drivers/llvmpipe/lp_context.c b/src/gallium/drivers/llvmpipe/lp_context.c index c081f6de03..679e244274 100644 --- a/src/gallium/drivers/llvmpipe/lp_context.c +++ b/src/gallium/drivers/llvmpipe/lp_context.c @@ -118,6 +118,11 @@ static void llvmpipe_destroy( struct pipe_context *pipe ) pipe_texture_reference(&llvmpipe->texture[i], NULL); } + for (i = 0; i < PIPE_MAX_VERTEX_SAMPLERS; i++) { + lp_destroy_tex_tile_cache(llvmpipe->vertex_tex_cache[i]); + pipe_texture_reference(&llvmpipe->vertex_textures[i], NULL); + } + for (i = 0; i < Elements(llvmpipe->constants); i++) { if (llvmpipe->constants[i].buffer) { pipe_buffer_reference(&llvmpipe->constants[i].buffer, NULL); @@ -145,6 +150,11 @@ llvmpipe_is_texture_referenced( struct pipe_context *pipe, llvmpipe->framebuffer.zsbuf->texture == texture) return PIPE_REFERENCED_FOR_WRITE; } + for (i = 0; i < PIPE_MAX_VERTEX_SAMPLERS; i++) { + if (llvmpipe->vertex_tex_cache[i] && + llvmpipe->vertex_tex_cache[i]->texture == texture) + return PIPE_REFERENCED_FOR_READ; + } return PIPE_UNREFERENCED; } @@ -181,6 +191,7 @@ llvmpipe_create( struct pipe_screen *screen ) llvmpipe->pipe.create_sampler_state = llvmpipe_create_sampler_state; llvmpipe->pipe.bind_fragment_sampler_states = llvmpipe_bind_sampler_states; + llvmpipe->pipe.bind_vertex_sampler_states = llvmpipe_bind_vertex_sampler_states; llvmpipe->pipe.delete_sampler_state = llvmpipe_delete_sampler_state; llvmpipe->pipe.create_depth_stencil_alpha_state = llvmpipe_create_depth_stencil_state; @@ -206,6 +217,7 @@ llvmpipe_create( struct pipe_screen *screen ) llvmpipe->pipe.set_polygon_stipple = llvmpipe_set_polygon_stipple; llvmpipe->pipe.set_scissor_state = llvmpipe_set_scissor_state; llvmpipe->pipe.set_fragment_sampler_textures = llvmpipe_set_sampler_textures; + llvmpipe->pipe.set_vertex_sampler_textures = llvmpipe_set_vertex_sampler_textures; llvmpipe->pipe.set_viewport_state = llvmpipe_set_viewport_state; llvmpipe->pipe.set_vertex_buffers = llvmpipe_set_vertex_buffers; @@ -234,13 +246,15 @@ llvmpipe_create( struct pipe_screen *screen ) for (i = 0; i < PIPE_MAX_SAMPLERS; i++) llvmpipe->tex_cache[i] = lp_create_tex_tile_cache( screen ); + for (i = 0; i < PIPE_MAX_VERTEX_SAMPLERS; i++) + llvmpipe->vertex_tex_cache[i] = lp_create_tex_tile_cache(screen); /* vertex shader samplers */ - for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { + for (i = 0; i < PIPE_MAX_VERTEX_SAMPLERS; i++) { llvmpipe->tgsi.vert_samplers[i].base.get_samples = lp_get_samples; llvmpipe->tgsi.vert_samplers[i].processor = TGSI_PROCESSOR_VERTEX; - llvmpipe->tgsi.vert_samplers[i].cache = llvmpipe->tex_cache[i]; + llvmpipe->tgsi.vert_samplers[i].cache = llvmpipe->vertex_tex_cache[i]; llvmpipe->tgsi.vert_samplers_list[i] = &llvmpipe->tgsi.vert_samplers[i]; } @@ -260,7 +274,7 @@ llvmpipe_create( struct pipe_screen *screen ) goto fail; draw_texture_samplers(llvmpipe->draw, - PIPE_MAX_SAMPLERS, + PIPE_MAX_VERTEX_SAMPLERS, (struct tgsi_sampler **) llvmpipe->tgsi.vert_samplers_list); diff --git a/src/gallium/drivers/llvmpipe/lp_context.h b/src/gallium/drivers/llvmpipe/lp_context.h index 3ad95d0bfc..cc4d5ad5fd 100644 --- a/src/gallium/drivers/llvmpipe/lp_context.h +++ b/src/gallium/drivers/llvmpipe/lp_context.h @@ -55,6 +55,7 @@ struct llvmpipe_context { /** Constant state objects */ const struct pipe_blend_state *blend; const struct pipe_sampler_state *sampler[PIPE_MAX_SAMPLERS]; + struct pipe_sampler_state *vertex_samplers[PIPE_MAX_VERTEX_SAMPLERS]; const struct pipe_depth_stencil_alpha_state *depth_stencil; const struct pipe_rasterizer_state *rasterizer; struct lp_fragment_shader *fs; @@ -68,12 +69,15 @@ struct llvmpipe_context { struct pipe_poly_stipple poly_stipple; struct pipe_scissor_state scissor; struct pipe_texture *texture[PIPE_MAX_SAMPLERS]; + struct pipe_texture *vertex_textures[PIPE_MAX_VERTEX_SAMPLERS]; struct pipe_viewport_state viewport; struct pipe_vertex_buffer vertex_buffer[PIPE_MAX_ATTRIBS]; struct pipe_vertex_element vertex_element[PIPE_MAX_ATTRIBS]; unsigned num_samplers; unsigned num_textures; + unsigned num_vertex_samplers; + unsigned num_vertex_textures; unsigned num_vertex_elements; unsigned num_vertex_buffers; @@ -136,6 +140,7 @@ struct llvmpipe_context { unsigned tex_timestamp; struct llvmpipe_tex_tile_cache *tex_cache[PIPE_MAX_SAMPLERS]; + struct llvmpipe_tex_tile_cache *vertex_tex_cache[PIPE_MAX_VERTEX_SAMPLERS]; unsigned no_rast : 1; diff --git a/src/gallium/drivers/llvmpipe/lp_debug.h b/src/gallium/drivers/llvmpipe/lp_debug.h new file mode 100644 index 0000000000..74b2757494 --- /dev/null +++ b/src/gallium/drivers/llvmpipe/lp_debug.h @@ -0,0 +1,71 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + + +#ifndef LP_DEBUG_H +#define LP_DEBUG_H + +#include "pipe/p_compiler.h" +#include "util/u_debug.h" + +extern void +st_print_current(void); + + +#define DEBUG_PIPE 0x1 +#define DEBUG_TGSI 0x2 +#define DEBUG_TEX 0x4 +#define DEBUG_ASM 0x8 +#define DEBUG_SETUP 0x10 +#define DEBUG_RAST 0x20 +#define DEBUG_QUERY 0x40 +#define DEBUG_SCREEN 0x80 +#define DEBUG_JIT 0x100 + +#ifdef DEBUG +extern int LP_DEBUG; +#else +#define LP_DEBUG 0 +#endif + +void st_debug_init( void ); + +static INLINE void +LP_DBG( unsigned flag, const char *fmt, ... ) +{ + if (LP_DEBUG & flag) + { + va_list args; + + va_start( args, fmt ); + debug_vprintf( fmt, args ); + va_end( args ); + } +} + + +#endif /* LP_DEBUG_H */ diff --git a/src/gallium/drivers/llvmpipe/lp_screen.c b/src/gallium/drivers/llvmpipe/lp_screen.c index a6ecaa0b2b..9b47415f00 100644 --- a/src/gallium/drivers/llvmpipe/lp_screen.c +++ b/src/gallium/drivers/llvmpipe/lp_screen.c @@ -36,6 +36,24 @@ #include "lp_winsys.h" #include "lp_jit.h" #include "lp_screen.h" +#include "lp_debug.h" + +#ifdef DEBUG +int LP_DEBUG = 0; + +static const struct debug_named_value lp_debug_flags[] = { + { "pipe", DEBUG_PIPE }, + { "tgsi", DEBUG_TGSI }, + { "tex", DEBUG_TEX }, + { "asm", DEBUG_ASM }, + { "setup", DEBUG_SETUP }, + { "rast", DEBUG_RAST }, + { "query", DEBUG_QUERY }, + { "screen", DEBUG_SCREEN }, + { "jit", DEBUG_JIT }, + {NULL, 0} +}; +#endif static const char * @@ -59,7 +77,9 @@ llvmpipe_get_param(struct pipe_screen *screen, int param) case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: return PIPE_MAX_SAMPLERS; case PIPE_CAP_MAX_VERTEX_TEXTURE_UNITS: - return 0; + return PIPE_MAX_VERTEX_SAMPLERS; + case PIPE_CAP_MAX_COMBINED_SAMPLERS: + return PIPE_MAX_SAMPLERS + PIPE_MAX_VERTEX_SAMPLERS; case PIPE_CAP_NPOT_TEXTURES: return 1; case PIPE_CAP_TWO_SIDED_STENCIL: @@ -257,6 +277,10 @@ llvmpipe_create_screen(struct llvmpipe_winsys *winsys) { struct llvmpipe_screen *screen = CALLOC_STRUCT(llvmpipe_screen); +#ifdef DEBUG + LP_DEBUG = debug_get_flags_option("LP_DEBUG", lp_debug_flags, 0 ); +#endif + if (!screen) return NULL; diff --git a/src/gallium/drivers/llvmpipe/lp_setup.c b/src/gallium/drivers/llvmpipe/lp_setup.c index ffcbc9a379..b4aabd4d7c 100644 --- a/src/gallium/drivers/llvmpipe/lp_setup.c +++ b/src/gallium/drivers/llvmpipe/lp_setup.c @@ -166,7 +166,7 @@ shade_quads(struct llvmpipe_context *llvmpipe, assert((y % 2) == 0); depth = llvmpipe->zsbuf_map + y*llvmpipe->zsbuf_transfer->stride + - 2*x*llvmpipe->zsbuf_transfer->block.size; + 2*x*pf_get_blocksize(llvmpipe->zsbuf_transfer->texture->format); } else depth = NULL; diff --git a/src/gallium/drivers/llvmpipe/lp_state.h b/src/gallium/drivers/llvmpipe/lp_state.h index 7b26ce61a3..d1c74ab07b 100644 --- a/src/gallium/drivers/llvmpipe/lp_state.h +++ b/src/gallium/drivers/llvmpipe/lp_state.h @@ -126,6 +126,10 @@ void * llvmpipe_create_sampler_state(struct pipe_context *, const struct pipe_sampler_state *); void llvmpipe_bind_sampler_states(struct pipe_context *, unsigned, void **); +void +llvmpipe_bind_vertex_sampler_states(struct pipe_context *, + unsigned num_samplers, + void **samplers); void llvmpipe_delete_sampler_state(struct pipe_context *, void *); void * @@ -172,6 +176,11 @@ void llvmpipe_set_sampler_textures( struct pipe_context *, unsigned num, struct pipe_texture ** ); +void +llvmpipe_set_vertex_sampler_textures(struct pipe_context *, + unsigned num_textures, + struct pipe_texture **); + void llvmpipe_set_viewport_state( struct pipe_context *, const struct pipe_viewport_state * ); diff --git a/src/gallium/drivers/llvmpipe/lp_state_derived.c b/src/gallium/drivers/llvmpipe/lp_state_derived.c index c753b183c0..e703964aaa 100644 --- a/src/gallium/drivers/llvmpipe/lp_state_derived.c +++ b/src/gallium/drivers/llvmpipe/lp_state_derived.c @@ -198,10 +198,14 @@ update_tgsi_samplers( struct llvmpipe_context *llvmpipe ) unsigned i; /* vertex shader samplers */ - for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { - llvmpipe->tgsi.vert_samplers[i].sampler = llvmpipe->sampler[i]; - llvmpipe->tgsi.vert_samplers[i].texture = llvmpipe->texture[i]; - llvmpipe->tgsi.frag_samplers[i].base.get_samples = lp_get_samples; + for (i = 0; i < PIPE_MAX_VERTEX_SAMPLERS; i++) { + llvmpipe->tgsi.vert_samplers[i].sampler = llvmpipe->vertex_samplers[i]; + llvmpipe->tgsi.vert_samplers[i].texture = llvmpipe->vertex_textures[i]; + llvmpipe->tgsi.vert_samplers[i].base.get_samples = lp_get_samples; + } + + for (i = 0; i < PIPE_MAX_VERTEX_SAMPLERS; i++) { + lp_tex_tile_cache_validate_texture( llvmpipe->vertex_tex_cache[i] ); } /* fragment shader samplers */ diff --git a/src/gallium/drivers/llvmpipe/lp_state_fs.c b/src/gallium/drivers/llvmpipe/lp_state_fs.c index ee0f69b2af..22683ff8b4 100644 --- a/src/gallium/drivers/llvmpipe/lp_state_fs.c +++ b/src/gallium/drivers/llvmpipe/lp_state_fs.c @@ -87,6 +87,7 @@ #include "lp_state.h" #include "lp_quad.h" #include "lp_tex_sample.h" +#include "lp_debug.h" static const unsigned char quad_offset_x[4] = {0, 1, 0, 1}; @@ -408,59 +409,58 @@ generate_fragment(struct llvmpipe_context *lp, unsigned i; unsigned chan; -#ifdef DEBUG - tgsi_dump(shader->base.tokens, 0); - if(key->depth.enabled) { - debug_printf("depth.format = %s\n", pf_name(key->zsbuf_format)); - debug_printf("depth.func = %s\n", debug_dump_func(key->depth.func, TRUE)); - debug_printf("depth.writemask = %u\n", key->depth.writemask); - } - if(key->alpha.enabled) { - debug_printf("alpha.func = %s\n", debug_dump_func(key->alpha.func, TRUE)); - debug_printf("alpha.ref_value = %f\n", key->alpha.ref_value); - } - if(key->blend.logicop_enable) { - debug_printf("blend.logicop_func = %u\n", key->blend.logicop_func); - } - else if(key->blend.blend_enable) { - debug_printf("blend.rgb_func = %s\n", debug_dump_blend_func (key->blend.rgb_func, TRUE)); - debug_printf("rgb_src_factor = %s\n", debug_dump_blend_factor(key->blend.rgb_src_factor, TRUE)); - debug_printf("rgb_dst_factor = %s\n", debug_dump_blend_factor(key->blend.rgb_dst_factor, TRUE)); - debug_printf("alpha_func = %s\n", debug_dump_blend_func (key->blend.alpha_func, TRUE)); - debug_printf("alpha_src_factor = %s\n", debug_dump_blend_factor(key->blend.alpha_src_factor, TRUE)); - debug_printf("alpha_dst_factor = %s\n", debug_dump_blend_factor(key->blend.alpha_dst_factor, TRUE)); - } - debug_printf("blend.colormask = 0x%x\n", key->blend.colormask); - for(i = 0; i < PIPE_MAX_SAMPLERS; ++i) { - if(key->sampler[i].format) { - debug_printf("sampler[%u] = \n", i); - debug_printf(" .format = %s\n", - pf_name(key->sampler[i].format)); - debug_printf(" .target = %s\n", - debug_dump_tex_target(key->sampler[i].target, TRUE)); - debug_printf(" .pot = %u %u %u\n", - key->sampler[i].pot_width, - key->sampler[i].pot_height, - key->sampler[i].pot_depth); - debug_printf(" .wrap = %s %s %s\n", - debug_dump_tex_wrap(key->sampler[i].wrap_s, TRUE), - debug_dump_tex_wrap(key->sampler[i].wrap_t, TRUE), - debug_dump_tex_wrap(key->sampler[i].wrap_r, TRUE)); - debug_printf(" .min_img_filter = %s\n", - debug_dump_tex_filter(key->sampler[i].min_img_filter, TRUE)); - debug_printf(" .min_mip_filter = %s\n", - debug_dump_tex_mipfilter(key->sampler[i].min_mip_filter, TRUE)); - debug_printf(" .mag_img_filter = %s\n", - debug_dump_tex_filter(key->sampler[i].mag_img_filter, TRUE)); - if(key->sampler[i].compare_mode) - debug_printf(" .compare_mode = %s\n", debug_dump_func(key->sampler[i].compare_func, TRUE)); - debug_printf(" .normalized_coords = %u\n", key->sampler[i].normalized_coords); - debug_printf(" .prefilter = %u\n", key->sampler[i].prefilter); + if (LP_DEBUG & DEBUG_JIT) { + tgsi_dump(shader->base.tokens, 0); + if(key->depth.enabled) { + debug_printf("depth.format = %s\n", pf_name(key->zsbuf_format)); + debug_printf("depth.func = %s\n", debug_dump_func(key->depth.func, TRUE)); + debug_printf("depth.writemask = %u\n", key->depth.writemask); + } + if(key->alpha.enabled) { + debug_printf("alpha.func = %s\n", debug_dump_func(key->alpha.func, TRUE)); + debug_printf("alpha.ref_value = %f\n", key->alpha.ref_value); + } + if(key->blend.logicop_enable) { + debug_printf("blend.logicop_func = %u\n", key->blend.logicop_func); + } + else if(key->blend.blend_enable) { + debug_printf("blend.rgb_func = %s\n", debug_dump_blend_func (key->blend.rgb_func, TRUE)); + debug_printf("rgb_src_factor = %s\n", debug_dump_blend_factor(key->blend.rgb_src_factor, TRUE)); + debug_printf("rgb_dst_factor = %s\n", debug_dump_blend_factor(key->blend.rgb_dst_factor, TRUE)); + debug_printf("alpha_func = %s\n", debug_dump_blend_func (key->blend.alpha_func, TRUE)); + debug_printf("alpha_src_factor = %s\n", debug_dump_blend_factor(key->blend.alpha_src_factor, TRUE)); + debug_printf("alpha_dst_factor = %s\n", debug_dump_blend_factor(key->blend.alpha_dst_factor, TRUE)); + } + debug_printf("blend.colormask = 0x%x\n", key->blend.colormask); + for(i = 0; i < PIPE_MAX_SAMPLERS; ++i) { + if(key->sampler[i].format) { + debug_printf("sampler[%u] = \n", i); + debug_printf(" .format = %s\n", + pf_name(key->sampler[i].format)); + debug_printf(" .target = %s\n", + debug_dump_tex_target(key->sampler[i].target, TRUE)); + debug_printf(" .pot = %u %u %u\n", + key->sampler[i].pot_width, + key->sampler[i].pot_height, + key->sampler[i].pot_depth); + debug_printf(" .wrap = %s %s %s\n", + debug_dump_tex_wrap(key->sampler[i].wrap_s, TRUE), + debug_dump_tex_wrap(key->sampler[i].wrap_t, TRUE), + debug_dump_tex_wrap(key->sampler[i].wrap_r, TRUE)); + debug_printf(" .min_img_filter = %s\n", + debug_dump_tex_filter(key->sampler[i].min_img_filter, TRUE)); + debug_printf(" .min_mip_filter = %s\n", + debug_dump_tex_mipfilter(key->sampler[i].min_mip_filter, TRUE)); + debug_printf(" .mag_img_filter = %s\n", + debug_dump_tex_filter(key->sampler[i].mag_img_filter, TRUE)); + if(key->sampler[i].compare_mode) + debug_printf(" .compare_mode = %s\n", debug_dump_func(key->sampler[i].compare_func, TRUE)); + debug_printf(" .normalized_coords = %u\n", key->sampler[i].normalized_coords); + debug_printf(" .prefilter = %u\n", key->sampler[i].prefilter); + } } } -#endif - variant = CALLOC_STRUCT(lp_fragment_shader_variant); if(!variant) return NULL; @@ -599,8 +599,8 @@ generate_fragment(struct llvmpipe_context *lp, } lp_build_conv_mask(builder, fs_type, blend_type, - fs_mask, num_fs, - &blend_mask, 1); + fs_mask, num_fs, + &blend_mask, 1); /* * Blending. @@ -631,16 +631,15 @@ generate_fragment(struct llvmpipe_context *lp, LLVMRunFunctionPassManager(screen->pass, variant->function); -#ifdef DEBUG - LLVMDumpValue(variant->function); - debug_printf("\n"); -#endif + if (LP_DEBUG & DEBUG_JIT) { + LLVMDumpValue(variant->function); + debug_printf("\n"); + } variant->jit_function = (lp_jit_frag_func)LLVMGetPointerToGlobal(screen->engine, variant->function); -#ifdef DEBUG - lp_disassemble(variant->jit_function); -#endif + if (LP_DEBUG & DEBUG_ASM) + lp_disassemble(variant->jit_function); variant->next = shader->variants; shader->variants = variant; diff --git a/src/gallium/drivers/llvmpipe/lp_state_sampler.c b/src/gallium/drivers/llvmpipe/lp_state_sampler.c index 8333805a3f..d382f9ca87 100644 --- a/src/gallium/drivers/llvmpipe/lp_state_sampler.c +++ b/src/gallium/drivers/llvmpipe/lp_state_sampler.c @@ -78,6 +78,34 @@ llvmpipe_bind_sampler_states(struct pipe_context *pipe, void +llvmpipe_bind_vertex_sampler_states(struct pipe_context *pipe, + unsigned num_samplers, + void **samplers) +{ + struct llvmpipe_context *llvmpipe = llvmpipe_context(pipe); + unsigned i; + + assert(num_samplers <= PIPE_MAX_VERTEX_SAMPLERS); + + /* Check for no-op */ + if (num_samplers == llvmpipe->num_vertex_samplers && + !memcmp(llvmpipe->vertex_samplers, samplers, num_samplers * sizeof(void *))) + return; + + draw_flush(llvmpipe->draw); + + for (i = 0; i < num_samplers; ++i) + llvmpipe->vertex_samplers[i] = samplers[i]; + for (i = num_samplers; i < PIPE_MAX_VERTEX_SAMPLERS; ++i) + llvmpipe->vertex_samplers[i] = NULL; + + llvmpipe->num_vertex_samplers = num_samplers; + + llvmpipe->dirty |= LP_NEW_SAMPLER; +} + + +void llvmpipe_set_sampler_textures(struct pipe_context *pipe, unsigned num, struct pipe_texture **texture) { @@ -117,6 +145,37 @@ llvmpipe_set_sampler_textures(struct pipe_context *pipe, void +llvmpipe_set_vertex_sampler_textures(struct pipe_context *pipe, + unsigned num_textures, + struct pipe_texture **textures) +{ + struct llvmpipe_context *llvmpipe = llvmpipe_context(pipe); + uint i; + + assert(num_textures <= PIPE_MAX_VERTEX_SAMPLERS); + + /* Check for no-op */ + if (num_textures == llvmpipe->num_vertex_textures && + !memcmp(llvmpipe->vertex_textures, textures, num_textures * sizeof(struct pipe_texture *))) { + return; + } + + draw_flush(llvmpipe->draw); + + for (i = 0; i < PIPE_MAX_VERTEX_SAMPLERS; i++) { + struct pipe_texture *tex = i < num_textures ? textures[i] : NULL; + + pipe_texture_reference(&llvmpipe->vertex_textures[i], tex); + lp_tex_tile_cache_set_texture(llvmpipe->vertex_tex_cache[i], tex); + } + + llvmpipe->num_vertex_textures = num_textures; + + llvmpipe->dirty |= LP_NEW_TEXTURE; +} + + +void llvmpipe_delete_sampler_state(struct pipe_context *pipe, void *sampler) { diff --git a/src/gallium/drivers/llvmpipe/lp_tex_cache.c b/src/gallium/drivers/llvmpipe/lp_tex_cache.c index c7c4143bc6..a6d9a2c1ac 100644 --- a/src/gallium/drivers/llvmpipe/lp_tex_cache.c +++ b/src/gallium/drivers/llvmpipe/lp_tex_cache.c @@ -155,7 +155,6 @@ lp_tex_tile_cache_validate_texture(struct llvmpipe_tex_tile_cache *tc) if (lpt->timestamp != tc->timestamp) { /* texture was modified, invalidate all cached tiles */ uint i; - debug_printf("INV %d %d\n", tc->timestamp, lpt->timestamp); for (i = 0; i < NUM_ENTRIES; i++) { tc->entries[i].addr.bits.invalid = 1; } @@ -291,7 +290,7 @@ lp_find_cached_tex_tile(struct llvmpipe_tex_tile_cache *tc, assert(0); } - util_format_read_4ub(tc->tex_trans->format, + util_format_read_4ub(tc->tex_trans->texture->format, (uint8_t *)tile->color, sizeof tile->color[0], tc->tex_trans_map, tc->tex_trans->stride, x, y, w, h); diff --git a/src/gallium/drivers/llvmpipe/lp_texture.c b/src/gallium/drivers/llvmpipe/lp_texture.c index 9b19cac972..9e41bc4074 100644 --- a/src/gallium/drivers/llvmpipe/lp_texture.c +++ b/src/gallium/drivers/llvmpipe/lp_texture.c @@ -50,7 +50,6 @@ /* Simple, maximally packed layout. */ - /* Conventional allocation path for non-display textures: */ static boolean @@ -65,20 +64,15 @@ llvmpipe_texture_layout(struct llvmpipe_screen *screen, unsigned buffer_size = 0; - util_format_get_block(lpt->base.format, &lpt->base.block); - for (level = 0; level <= pt->last_level; level++) { unsigned nblocksx, nblocksy; - pt->nblocksx[level] = pf_get_nblocksx(&pt->block, width); - pt->nblocksy[level] = pf_get_nblocksy(&pt->block, height); - /* Allocate storage for whole quads. This is particularly important * for depth surfaces, which are currently stored in a swizzled format. */ - nblocksx = pf_get_nblocksx(&pt->block, align(width, 2)); - nblocksy = pf_get_nblocksy(&pt->block, align(height, 2)); + nblocksx = pf_get_nblocksx(pt->format, align(width, 2)); + nblocksy = pf_get_nblocksy(pt->format, align(height, 2)); - lpt->stride[level] = align(nblocksx*pt->block.size, 16); + lpt->stride[level] = align(nblocksx * pf_get_blocksize(pt->format), 16); lpt->level_offset[level] = buffer_size; @@ -102,10 +96,6 @@ llvmpipe_displaytarget_layout(struct llvmpipe_screen *screen, { struct llvmpipe_winsys *winsys = screen->winsys; - util_format_get_block(lpt->base.format, &lpt->base.block); - lpt->base.nblocksx[0] = pf_get_nblocksx(&lpt->base.block, lpt->base.width0); - lpt->base.nblocksy[0] = pf_get_nblocksy(&lpt->base.block, lpt->base.height0); - lpt->dt = winsys->displaytarget_create(winsys, lpt->base.format, lpt->base.width0, @@ -182,8 +172,6 @@ llvmpipe_texture_blanket(struct pipe_screen * screen, lpt->base = *base; pipe_reference_init(&lpt->base.reference, 1); lpt->base.screen = screen; - lpt->base.nblocksx[0] = pf_get_nblocksx(&lpt->base.block, lpt->base.width0); - lpt->base.nblocksy[0] = pf_get_nblocksy(&lpt->base.block, lpt->base.height0); lpt->stride[0] = stride[0]; pipe_buffer_reference(&lpt->buffer, buffer); @@ -257,11 +245,17 @@ llvmpipe_get_tex_surface(struct pipe_screen *screen, ps->level = level; ps->zslice = zslice; + /* XXX shouldn't that rather be + tex_height = align(ps->height, 2); + to account for alignment done in llvmpipe_texture_layout ? + */ if (pt->target == PIPE_TEXTURE_CUBE) { - ps->offset += face * pt->nblocksy[level] * lpt->stride[level]; + unsigned tex_height = ps->height; + ps->offset += face * pf_get_nblocksy(pt->format, tex_height) * lpt->stride[level]; } else if (pt->target == PIPE_TEXTURE_3D) { - ps->offset += zslice * pt->nblocksy[level] * lpt->stride[level]; + unsigned tex_height = ps->height; + ps->offset += zslice * pf_get_nblocksy(pt->format, tex_height) * lpt->stride[level]; } else { assert(face == 0); @@ -302,14 +296,10 @@ llvmpipe_get_tex_transfer(struct pipe_screen *screen, if (lpt) { struct pipe_transfer *pt = &lpt->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 = lptex->stride[level]; pt->usage = usage; pt->face = face; @@ -318,11 +308,17 @@ llvmpipe_get_tex_transfer(struct pipe_screen *screen, lpt->offset = lptex->level_offset[level]; + /* XXX shouldn't that rather be + tex_height = align(u_minify(texture->height0, level), 2) + to account for alignment done in llvmpipe_texture_layout ? + */ if (texture->target == PIPE_TEXTURE_CUBE) { - lpt->offset += face * pt->nblocksy * pt->stride; + unsigned tex_height = u_minify(texture->height0, level); + lpt->offset += face * pf_get_nblocksy(texture->format, tex_height) * pt->stride; } else if (texture->target == PIPE_TEXTURE_3D) { - lpt->offset += zslice * pt->nblocksy * pt->stride; + unsigned tex_height = u_minify(texture->height0, level); + lpt->offset += zslice * pf_get_nblocksy(texture->format, tex_height) * pt->stride; } else { assert(face == 0); @@ -354,9 +350,11 @@ llvmpipe_transfer_map( struct pipe_screen *_screen, struct llvmpipe_screen *screen = llvmpipe_screen(_screen); ubyte *map, *xfer_map; struct llvmpipe_texture *lpt; + enum pipe_format format; assert(transfer->texture); lpt = llvmpipe_texture(transfer->texture); + format = lpt->base.format; if(lpt->dt) { struct llvmpipe_winsys *winsys = screen->winsys; @@ -381,8 +379,8 @@ llvmpipe_transfer_map( struct pipe_screen *_screen, } xfer_map = map + llvmpipe_transfer(transfer)->offset + - transfer->y / transfer->block.height * transfer->stride + - transfer->x / transfer->block.width * transfer->block.size; + transfer->y / pf_get_blockheight(format) * transfer->stride + + transfer->x / pf_get_blockwidth(format) * pf_get_blocksize(format); /*printf("map = %p xfer map = %p\n", map, xfer_map);*/ return xfer_map; } diff --git a/src/gallium/drivers/llvmpipe/lp_tile_cache.c b/src/gallium/drivers/llvmpipe/lp_tile_cache.c index ec3e002d62..7a1ecf5107 100644 --- a/src/gallium/drivers/llvmpipe/lp_tile_cache.c +++ b/src/gallium/drivers/llvmpipe/lp_tile_cache.c @@ -252,13 +252,13 @@ lp_flush_tile_cache(struct llvmpipe_tile_cache *tc) case LP_TILE_STATUS_CLEAR: /* Actually clear the tiles which were flagged as being in a * clear state. */ - util_fill_rect(tc->transfer_map, &pt->block, pt->stride, + util_fill_rect(tc->transfer_map, pt->texture->format, pt->stride, x, y, w, h, tc->clear_val); break; case LP_TILE_STATUS_DEFINED: - lp_tile_write_4ub(pt->format, + lp_tile_write_4ub(pt->texture->format, tile->color, tc->transfer_map, pt->stride, x, y, w, h); @@ -291,6 +291,11 @@ lp_get_cached_tile(struct llvmpipe_tile_cache *tc, assert(tc->surface); assert(tc->transfer); + if(!tc->transfer_map) + lp_tile_cache_map_transfers(tc); + + assert(tc->transfer_map); + switch(tile->status) { case LP_TILE_STATUS_CLEAR: /* don't get tile from framebuffer, just clear it */ @@ -306,7 +311,7 @@ lp_get_cached_tile(struct llvmpipe_tile_cache *tc, y &= ~(TILE_SIZE - 1); if (!pipe_clip_tile(x, y, &w, &h, tc->transfer)) - lp_tile_read_4ub(pt->format, + lp_tile_read_4ub(pt->texture->format, tile->color, tc->transfer_map, tc->transfer->stride, x, y, w, h); diff --git a/src/gallium/drivers/nouveau/nouveau_stateobj.h b/src/gallium/drivers/nouveau/nouveau_stateobj.h index b595405357..9aee9e4956 100644 --- a/src/gallium/drivers/nouveau/nouveau_stateobj.h +++ b/src/gallium/drivers/nouveau/nouveau_stateobj.h @@ -48,13 +48,14 @@ so_ref(struct nouveau_stateobj *ref, struct nouveau_stateobj **pso) struct nouveau_stateobj *so = *pso; int i; - if (pipe_reference((struct pipe_reference**)pso, &ref->reference)) { + if (pipe_reference(&(*pso)->reference, &ref->reference)) { free(so->push); for (i = 0; i < so->cur_reloc; i++) nouveau_bo_ref(NULL, &so->reloc[i].bo); free(so->reloc); free(so); } + *pso = ref; } static INLINE void @@ -111,20 +112,30 @@ so_emit(struct nouveau_channel *chan, struct nouveau_stateobj *so) { struct nouveau_pushbuf *pb = chan->pushbuf; unsigned nr, i; + int ret = 0; nr = so->cur - so->push; - if (pb->remaining < nr) - nouveau_pushbuf_flush(chan, nr); + /* This will flush if we need space. + * We don't actually need the marker. + */ + if ((ret = nouveau_pushbuf_marker_emit(chan, nr, so->cur_reloc))) { + debug_printf("so_emit failed marker emit with error %d\n", ret); + return; + } pb->remaining -= nr; memcpy(pb->cur, so->push, nr * 4); for (i = 0; i < so->cur_reloc; i++) { struct nouveau_stateobj_reloc *r = &so->reloc[i]; - nouveau_pushbuf_emit_reloc(chan, pb->cur + r->offset, + if ((ret = nouveau_pushbuf_emit_reloc(chan, pb->cur + r->offset, r->bo, r->data, 0, r->flags, - r->vor, r->tor); + r->vor, r->tor))) { + debug_printf("so_emit failed reloc with error %d\n", ret); + goto out; + } } +out: pb->cur += nr; } @@ -133,26 +144,45 @@ so_emit_reloc_markers(struct nouveau_channel *chan, struct nouveau_stateobj *so) { struct nouveau_pushbuf *pb = chan->pushbuf; unsigned i; + int ret = 0; if (!so) return; i = so->cur_reloc << 1; - if (pb->remaining < i) - nouveau_pushbuf_flush(chan, i); + /* This will flush if we need space. + * We don't actually need the marker. + */ + if ((ret = nouveau_pushbuf_marker_emit(chan, i, i))) { + debug_printf("so_emit_reloc_markers failed marker emit with" \ + "error %d\n", ret); + return; + } pb->remaining -= i; for (i = 0; i < so->cur_reloc; i++) { struct nouveau_stateobj_reloc *r = &so->reloc[i]; - nouveau_pushbuf_emit_reloc(chan, pb->cur++, r->bo, r->packet, 0, + if ((ret = nouveau_pushbuf_emit_reloc(chan, pb->cur++, r->bo, + r->packet, 0, (r->flags & (NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RDWR)) | - NOUVEAU_BO_DUMMY, 0, 0); - nouveau_pushbuf_emit_reloc(chan, pb->cur++, r->bo, r->data, 0, + NOUVEAU_BO_DUMMY, 0, 0))) { + debug_printf("so_emit_reloc_markers failed reloc" \ + "with error %d\n", ret); + pb->remaining += ((so->cur_reloc - i) << 1); + return; + } + if ((ret = nouveau_pushbuf_emit_reloc(chan, pb->cur++, r->bo, + r->data, 0, r->flags | NOUVEAU_BO_DUMMY, - r->vor, r->tor); + r->vor, r->tor))) { + debug_printf("so_emit_reloc_markers failed reloc" \ + "with error %d\n", ret); + pb->remaining += ((so->cur_reloc - i) << 1) - 1; + return; + } } } diff --git a/src/gallium/drivers/nv04/nv04_miptree.c b/src/gallium/drivers/nv04/nv04_miptree.c index 4fd72c82e6..e0a6948aeb 100644 --- a/src/gallium/drivers/nv04/nv04_miptree.c +++ b/src/gallium/drivers/nv04/nv04_miptree.c @@ -10,28 +10,21 @@ static void nv04_miptree_layout(struct nv04_miptree *nv04mt) { struct pipe_texture *pt = &nv04mt->base; - uint width = pt->width0, height = pt->height0; uint offset = 0; int nr_faces, l; nr_faces = 1; for (l = 0; l <= pt->last_level; l++) { - - pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width); - pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); - nv04mt->level[l].pitch = pt->width0; nv04mt->level[l].pitch = (nv04mt->level[l].pitch + 63) & ~63; - - width = u_minify(width, 1); - height = u_minify(height, 1); } for (l = 0; l <= pt->last_level; l++) { - nv04mt->level[l].image_offset = CALLOC(nr_faces, sizeof(unsigned)); + /* XXX guess was obviously missing */ + nv04mt->level[l].image_offset[0] = offset; offset += nv04mt->level[l].pitch * u_minify(pt->height0, l); } @@ -62,7 +55,7 @@ nv04_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) FREE(mt); return NULL; } - + mt->bo = nouveau_bo(mt->buffer); return &mt->base; } @@ -88,6 +81,7 @@ nv04_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt, mt->level[0].image_offset = CALLOC(1, sizeof(unsigned)); pipe_buffer_reference(&mt->buffer, pb); + mt->bo = nouveau_bo(mt->buffer); return &mt->base; } @@ -128,7 +122,7 @@ nv04_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, ns->base.zslice = zslice; ns->pitch = nv04mt->level[level].pitch; - ns->base.offset = nv04mt->level[level].image_offset; + ns->base.offset = nv04mt->level[level].image_offset[0]; return &ns->base; } diff --git a/src/gallium/drivers/nv04/nv04_state.h b/src/gallium/drivers/nv04/nv04_state.h index 399f750dbe..81d1d2ebaa 100644 --- a/src/gallium/drivers/nv04/nv04_state.h +++ b/src/gallium/drivers/nv04/nv04_state.h @@ -31,6 +31,7 @@ struct nv04_rasterizer_state { struct nv04_miptree { struct pipe_texture base; + struct nouveau_bo *bo; struct pipe_buffer *buffer; uint total_size; diff --git a/src/gallium/drivers/nv04/nv04_surface_2d.c b/src/gallium/drivers/nv04/nv04_surface_2d.c index 8be134b83d..3020806c5d 100644 --- a/src/gallium/drivers/nv04/nv04_surface_2d.c +++ b/src/gallium/drivers/nv04/nv04_surface_2d.c @@ -133,6 +133,9 @@ nv04_surface_copy_swizzle(struct nv04_surface_2d *ctx, assert(sub_w == w || util_is_pot(sub_w)); assert(sub_h == h || util_is_pot(sub_h)); + MARK_RING (chan, 8 + ((w+sub_w)/sub_w)*((h+sub_h)/sub_h)*17, 2 + + ((w+sub_w)/sub_w)*((h+sub_h)/sub_h)*2); + BEGIN_RING(chan, swzsurf, NV04_SWIZZLED_SURFACE_DMA_IMAGE, 1); OUT_RELOCo(chan, dst_bo, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); @@ -155,10 +158,10 @@ nv04_surface_copy_swizzle(struct nv04_surface_2d *ctx, sub_w = MIN2(sub_w, w - x); /* Must be 64-byte aligned */ - assert(!((dst->offset + nv04_swizzle_bits(dx+x, dy+y) * dst->texture->block.size) & 63)); + assert(!((dst->offset + nv04_swizzle_bits(dx+x, dy+y) * pf_get_blocksize(dst->texture->format)) & 63)); BEGIN_RING(chan, swzsurf, NV04_SWIZZLED_SURFACE_OFFSET, 1); - OUT_RELOCl(chan, dst_bo, dst->offset + nv04_swizzle_bits(dx+x, dy+y) * dst->texture->block.size, + OUT_RELOCl(chan, dst_bo, dst->offset + nv04_swizzle_bits(dx+x, dy+y) * pf_get_blocksize(dst->texture->format), NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); BEGIN_RING(chan, sifm, NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION, 9); @@ -177,7 +180,7 @@ nv04_surface_copy_swizzle(struct nv04_surface_2d *ctx, OUT_RING (chan, src_pitch | NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CENTER | NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_POINT_SAMPLE); - OUT_RELOCl(chan, src_bo, src->offset + (sy+y) * src_pitch + (sx+x) * src->texture->block.size, + OUT_RELOCl(chan, src_bo, src->offset + (sy+y) * src_pitch + (sx+x) * pf_get_blocksize(src->texture->format), NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); OUT_RING (chan, 0); } @@ -198,11 +201,11 @@ nv04_surface_copy_m2mf(struct nv04_surface_2d *ctx, unsigned src_pitch = ((struct nv04_surface *)src)->pitch; unsigned dst_pitch = ((struct nv04_surface *)dst)->pitch; unsigned dst_offset = dst->offset + dy * dst_pitch + - dx * dst->texture->block.size; + dx * pf_get_blocksize(dst->texture->format); unsigned src_offset = src->offset + sy * src_pitch + - sx * src->texture->block.size; + sx * pf_get_blocksize(src->texture->format); - WAIT_RING (chan, 3 + ((h / 2047) + 1) * 9); + MARK_RING (chan, 3 + ((h / 2047) + 1) * 9, 2 + ((h / 2047) + 1) * 2); BEGIN_RING(chan, m2mf, NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2); OUT_RELOCo(chan, src_bo, NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); @@ -219,7 +222,7 @@ nv04_surface_copy_m2mf(struct nv04_surface_2d *ctx, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_WR); OUT_RING (chan, src_pitch); OUT_RING (chan, dst_pitch); - OUT_RING (chan, w * src->texture->block.size); + OUT_RING (chan, w * pf_get_blocksize(src->texture->format)); OUT_RING (chan, count); OUT_RING (chan, 0x0101); OUT_RING (chan, 0); @@ -250,7 +253,7 @@ nv04_surface_copy_blit(struct nv04_surface_2d *ctx, struct pipe_surface *dst, if (format < 0) return 1; - WAIT_RING (chan, 12); + MARK_RING (chan, 12, 4); BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); OUT_RELOCo(chan, src_bo, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD); OUT_RELOCo(chan, dst_bo, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); @@ -315,7 +318,7 @@ nv04_surface_fill(struct nv04_surface_2d *ctx, struct pipe_surface *dst, gdirect_format = nv04_rect_format(dst->format); assert(gdirect_format >= 0); - WAIT_RING (chan, 16); + MARK_RING (chan, 16, 4); BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2); OUT_RELOCo(chan, dst_bo, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); OUT_RELOCo(chan, dst_bo, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); diff --git a/src/gallium/drivers/nv04/nv04_transfer.c b/src/gallium/drivers/nv04/nv04_transfer.c index e6456429f4..d66d6c6346 100644 --- a/src/gallium/drivers/nv04/nv04_transfer.c +++ b/src/gallium/drivers/nv04/nv04_transfer.c @@ -11,7 +11,7 @@ struct nv04_transfer { struct pipe_transfer base; struct pipe_surface *surface; - bool direct; + boolean direct; }; static void @@ -24,9 +24,6 @@ nv04_compatible_transfer_tex(struct pipe_texture *pt, unsigned level, template->width0 = u_minify(pt->width0, level); template->height0 = u_minify(pt->height0, level); template->depth0 = 1; - template->block = pt->block; - template->nblocksx[0] = pt->nblocksx[level]; - template->nblocksy[0] = pt->nblocksx[level]; template->last_level = 0; template->nr_samples = pt->nr_samples; @@ -49,14 +46,10 @@ nv04_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; pipe_texture_reference(&tx->base.texture, pt); - tx->base.format = pt->format; tx->base.x = x; tx->base.y = y; tx->base.width = w; tx->base.height = h; - tx->base.block = pt->block; - tx->base.nblocksx = pt->nblocksx[level]; - tx->base.nblocksy = pt->nblocksy[level]; tx->base.stride = mt->level[level].pitch; tx->base.usage = usage; tx->base.face = face; @@ -158,7 +151,7 @@ nv04_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) pipe_transfer_buffer_flags(ptx)); return map + ns->base.offset + - ptx->y * ns->pitch + ptx->x * ptx->block.size; + ptx->y * ns->pitch + ptx->x * pf_get_blocksize(ptx->texture->format); } static void diff --git a/src/gallium/drivers/nv10/nv10_miptree.c b/src/gallium/drivers/nv10/nv10_miptree.c index b2a6c59b74..6a52b6af36 100644 --- a/src/gallium/drivers/nv10/nv10_miptree.c +++ b/src/gallium/drivers/nv10/nv10_miptree.c @@ -11,7 +11,7 @@ nv10_miptree_layout(struct nv10_miptree *nv10mt) { struct pipe_texture *pt = &nv10mt->base; boolean swizzled = FALSE; - uint width = pt->width0, height = pt->height0; + uint width = pt->width0; uint offset = 0; int nr_faces, l, f; @@ -22,21 +22,16 @@ nv10_miptree_layout(struct nv10_miptree *nv10mt) } for (l = 0; l <= pt->last_level; l++) { - - pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width); - pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); - if (swizzled) - nv10mt->level[l].pitch = pt->nblocksx[l] * pt->block.size; + nv10mt->level[l].pitch = pf_get_stride(pt->format, width); else - nv10mt->level[l].pitch = pt->nblocksx[0] * pt->block.size; + nv10mt->level[l].pitch = pf_get_stride(pt->format, pt->width0); nv10mt->level[l].pitch = (nv10mt->level[l].pitch + 63) & ~63; nv10mt->level[l].image_offset = CALLOC(nr_faces, sizeof(unsigned)); width = u_minify(width, 1); - height = u_minify(height, 1); } @@ -72,6 +67,7 @@ nv10_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt, mt->level[0].image_offset = CALLOC(1, sizeof(unsigned)); pipe_buffer_reference(&mt->buffer, pb); + mt->bo = nouveau_bo(mt->buffer); return &mt->base; } @@ -95,6 +91,7 @@ nv10_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) FREE(mt); return NULL; } + mt->bo = nouveau_bo(mt->buffer); return &mt->base; } diff --git a/src/gallium/drivers/nv10/nv10_state.h b/src/gallium/drivers/nv10/nv10_state.h index 3a3fd0d4f4..2524ac02e2 100644 --- a/src/gallium/drivers/nv10/nv10_state.h +++ b/src/gallium/drivers/nv10/nv10_state.h @@ -126,6 +126,7 @@ struct nv10_depth_stencil_alpha_state { struct nv10_miptree { struct pipe_texture base; + struct nouveau_bo *bo; struct pipe_buffer *buffer; uint total_size; diff --git a/src/gallium/drivers/nv10/nv10_transfer.c b/src/gallium/drivers/nv10/nv10_transfer.c index ec54297ab0..06bb513417 100644 --- a/src/gallium/drivers/nv10/nv10_transfer.c +++ b/src/gallium/drivers/nv10/nv10_transfer.c @@ -11,7 +11,7 @@ struct nv10_transfer { struct pipe_transfer base; struct pipe_surface *surface; - bool direct; + boolean direct; }; static void @@ -24,9 +24,6 @@ nv10_compatible_transfer_tex(struct pipe_texture *pt, unsigned level, template->width0 = u_minify(pt->width0, level); template->height0 = u_minify(pt->height0, level); template->depth0 = 1; - template->block = pt->block; - template->nblocksx[0] = pt->nblocksx[level]; - template->nblocksy[0] = pt->nblocksx[level]; template->last_level = 0; template->nr_samples = pt->nr_samples; @@ -49,14 +46,10 @@ nv10_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; pipe_texture_reference(&tx->base.texture, pt); - tx->base.format = pt->format; tx->base.x = x; tx->base.y = y; tx->base.width = w; tx->base.height = h; - tx->base.block = pt->block; - tx->base.nblocksx = pt->nblocksx[level]; - tx->base.nblocksy = pt->nblocksy[level]; tx->base.stride = mt->level[level].pitch; tx->base.usage = usage; tx->base.face = face; @@ -158,7 +151,7 @@ nv10_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) pipe_transfer_buffer_flags(ptx)); return map + ns->base.offset + - ptx->y * ns->pitch + ptx->x * ptx->block.size; + ptx->y * ns->pitch + ptx->x * pf_get_blocksize(ptx->texture->format); } static void diff --git a/src/gallium/drivers/nv20/nv20_miptree.c b/src/gallium/drivers/nv20/nv20_miptree.c index 554e28e47d..e2e01bd849 100644 --- a/src/gallium/drivers/nv20/nv20_miptree.c +++ b/src/gallium/drivers/nv20/nv20_miptree.c @@ -10,7 +10,7 @@ static void nv20_miptree_layout(struct nv20_miptree *nv20mt) { struct pipe_texture *pt = &nv20mt->base; - uint width = pt->width0, height = pt->height0; + uint width = pt->width0; uint offset = 0; int nr_faces, l, f; uint wide_pitch = pt->tex_usage & (PIPE_TEXTURE_USAGE_SAMPLER | @@ -26,19 +26,15 @@ nv20_miptree_layout(struct nv20_miptree *nv20mt) } for (l = 0; l <= pt->last_level; l++) { - pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width); - pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); - if (wide_pitch && (pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) - nv20mt->level[l].pitch = align(pt->width0 * pt->block.size, 64); + nv20mt->level[l].pitch = align(pf_get_stride(pt->format, pt->width0), 64); else - nv20mt->level[l].pitch = u_minify(pt->width0, l) * pt->block.size; + nv20mt->level[l].pitch = pf_get_stride(pt->format, width); nv20mt->level[l].image_offset = CALLOC(nr_faces, sizeof(unsigned)); width = u_minify(width, 1); - height = u_minify(height, 1); } for (f = 0; f < nr_faces; f++) { @@ -81,6 +77,7 @@ nv20_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt, mt->level[0].image_offset = CALLOC(1, sizeof(unsigned)); pipe_buffer_reference(&mt->buffer, pb); + mt->bo = nouveau_bo(mt->buffer); return &mt->base; } @@ -136,6 +133,7 @@ nv20_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt) FREE(mt); return NULL; } + mt->bo = nouveau_bo(mt->buffer); return &mt->base; } diff --git a/src/gallium/drivers/nv20/nv20_state.h b/src/gallium/drivers/nv20/nv20_state.h index 34f402fdcb..dde4106568 100644 --- a/src/gallium/drivers/nv20/nv20_state.h +++ b/src/gallium/drivers/nv20/nv20_state.h @@ -126,6 +126,7 @@ struct nv20_depth_stencil_alpha_state { struct nv20_miptree { struct pipe_texture base; + struct nouveau_bo *bo; struct pipe_buffer *buffer; uint total_size; diff --git a/src/gallium/drivers/nv20/nv20_transfer.c b/src/gallium/drivers/nv20/nv20_transfer.c index 87b5c14a3c..26a73c5143 100644 --- a/src/gallium/drivers/nv20/nv20_transfer.c +++ b/src/gallium/drivers/nv20/nv20_transfer.c @@ -11,7 +11,7 @@ struct nv20_transfer { struct pipe_transfer base; struct pipe_surface *surface; - bool direct; + boolean direct; }; static void @@ -24,9 +24,6 @@ nv20_compatible_transfer_tex(struct pipe_texture *pt, unsigned level, template->width0 = u_minify(pt->width0, level); template->height0 = u_minify(pt->height0, level); template->depth0 = 1; - template->block = pt->block; - template->nblocksx[0] = pt->nblocksx[level]; - template->nblocksy[0] = pt->nblocksx[level]; template->last_level = 0; template->nr_samples = pt->nr_samples; @@ -49,14 +46,10 @@ nv20_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; pipe_texture_reference(&tx->base.texture, pt); - tx->base.format = pt->format; tx->base.x = x; tx->base.y = y; tx->base.width = w; tx->base.height = h; - tx->base.block = pt->block; - tx->base.nblocksx = pt->nblocksx[level]; - tx->base.nblocksy = pt->nblocksy[level]; tx->base.stride = mt->level[level].pitch; tx->base.usage = usage; tx->base.face = face; @@ -158,7 +151,7 @@ nv20_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) pipe_transfer_buffer_flags(ptx)); return map + ns->base.offset + - ptx->y * ns->pitch + ptx->x * ptx->block.size; + ptx->y * ns->pitch + ptx->x * pf_get_blocksize(ptx->texture->format); } static void diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index d8300fd69f..46a821a48b 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -58,6 +58,9 @@ nv30_create(struct pipe_screen *pscreen, unsigned pctx_id) nv30->pipe.is_texture_referenced = nouveau_is_texture_referenced; nv30->pipe.is_buffer_referenced = nouveau_is_buffer_referenced; + screen->base.channel->user_private = nv30; + screen->base.channel->flush_notify = nv30_state_flush_notify; + nv30_init_query_functions(nv30); nv30_init_surface_functions(nv30); nv30_init_state_functions(nv30); diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 8d49366dfc..6f44b1c7fe 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -184,6 +184,7 @@ extern void nv30_fragtex_bind(struct nv30_context *); /* nv30_state.c and friends */ extern boolean nv30_state_validate(struct nv30_context *nv30); extern void nv30_state_emit(struct nv30_context *nv30); +extern void nv30_state_flush_notify(struct nouveau_channel *chan); extern struct nv30_state_entry nv30_state_rasterizer; extern struct nv30_state_entry nv30_state_scissor; extern struct nv30_state_entry nv30_state_stipple; diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c index b4c306d127..920fe64c32 100644 --- a/src/gallium/drivers/nv30/nv30_miptree.c +++ b/src/gallium/drivers/nv30/nv30_miptree.c @@ -9,7 +9,7 @@ static void nv30_miptree_layout(struct nv30_miptree *nv30mt) { struct pipe_texture *pt = &nv30mt->base; - uint width = pt->width0, height = pt->height0, depth = pt->depth0; + uint width = pt->width0; uint offset = 0; int nr_faces, l, f; uint wide_pitch = pt->tex_usage & (PIPE_TEXTURE_USAGE_SAMPLER | @@ -28,20 +28,15 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt) } for (l = 0; l <= pt->last_level; l++) { - pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width); - pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); - if (wide_pitch && (pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) - nv30mt->level[l].pitch = align(pt->width0 * pt->block.size, 64); + nv30mt->level[l].pitch = align(pf_get_stride(pt->format, pt->width0), 64); else - nv30mt->level[l].pitch = u_minify(pt->width0, l) * pt->block.size; + nv30mt->level[l].pitch = pf_get_stride(pt->format, width); nv30mt->level[l].image_offset = CALLOC(nr_faces, sizeof(unsigned)); width = u_minify(width, 1); - height = u_minify(height, 1); - depth = u_minify(depth, 1); } for (f = 0; f < nr_faces; f++) { @@ -120,6 +115,7 @@ nv30_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) FREE(mt); return NULL; } + mt->bo = nouveau_bo(mt->buffer); return &mt->base; } @@ -149,6 +145,7 @@ nv30_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt, mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR; pipe_buffer_reference(&mt->buffer, pb); + mt->bo = nouveau_bo(mt->buffer); return &mt->base; } diff --git a/src/gallium/drivers/nv30/nv30_state.h b/src/gallium/drivers/nv30/nv30_state.h index e6f23bf166..e42e872de7 100644 --- a/src/gallium/drivers/nv30/nv30_state.h +++ b/src/gallium/drivers/nv30/nv30_state.h @@ -72,6 +72,7 @@ struct nv30_fragment_program { struct nv30_miptree { struct pipe_texture base; + struct nouveau_bo *bo; struct pipe_buffer *buffer; uint total_size; diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c index 621b8846c8..ac52d946f0 100644 --- a/src/gallium/drivers/nv30/nv30_state_emit.c +++ b/src/gallium/drivers/nv30/nv30_state_emit.c @@ -41,7 +41,7 @@ nv30_state_emit(struct nv30_context *nv30) struct nouveau_channel *chan = nv30->screen->base.channel; struct nv30_state *state = &nv30->state; struct nv30_screen *screen = nv30->screen; - unsigned i, samplers; + unsigned i; uint64_t states; if (nv30->pctx_id != screen->cur_pctx) { @@ -63,6 +63,14 @@ nv30_state_emit(struct nv30_context *nv30) } state->dirty = 0; +} + +void +nv30_state_flush_notify(struct nouveau_channel *chan) +{ + struct nv30_context *nv30 = chan->user_private; + struct nv30_state *state = &nv30->state; + unsigned i, samplers; so_emit_reloc_markers(chan, state->hw[NV30_STATE_FB]); for (i = 0, samplers = state->fp_samplers; i < 16 && samplers; i++) { diff --git a/src/gallium/drivers/nv30/nv30_transfer.c b/src/gallium/drivers/nv30/nv30_transfer.c index 5e429b4d85..e29bfbd3ef 100644 --- a/src/gallium/drivers/nv30/nv30_transfer.c +++ b/src/gallium/drivers/nv30/nv30_transfer.c @@ -11,7 +11,7 @@ struct nv30_transfer { struct pipe_transfer base; struct pipe_surface *surface; - bool direct; + boolean direct; }; static void @@ -24,9 +24,6 @@ nv30_compatible_transfer_tex(struct pipe_texture *pt, unsigned level, template->width0 = u_minify(pt->width0, level); template->height0 = u_minify(pt->height0, level); template->depth0 = 1; - template->block = pt->block; - template->nblocksx[0] = pt->nblocksx[level]; - template->nblocksy[0] = pt->nblocksx[level]; template->last_level = 0; template->nr_samples = pt->nr_samples; @@ -49,14 +46,10 @@ nv30_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; pipe_texture_reference(&tx->base.texture, pt); - tx->base.format = pt->format; tx->base.x = x; tx->base.y = y; tx->base.width = w; tx->base.height = h; - tx->base.block = pt->block; - tx->base.nblocksx = pt->nblocksx[level]; - tx->base.nblocksy = pt->nblocksy[level]; tx->base.stride = mt->level[level].pitch; tx->base.usage = usage; tx->base.face = face; @@ -158,7 +151,7 @@ nv30_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) pipe_transfer_buffer_flags(ptx)); return map + ns->base.offset + - ptx->y * ns->pitch + ptx->x * ptx->block.size; + ptx->y * ns->pitch + ptx->x * pf_get_blocksize(ptx->texture->format); } static void diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c index 7f008274a4..eb9cce4c78 100644 --- a/src/gallium/drivers/nv40/nv40_context.c +++ b/src/gallium/drivers/nv40/nv40_context.c @@ -58,6 +58,9 @@ nv40_create(struct pipe_screen *pscreen, unsigned pctx_id) nv40->pipe.is_texture_referenced = nouveau_is_texture_referenced; nv40->pipe.is_buffer_referenced = nouveau_is_buffer_referenced; + screen->base.channel->user_private = nv40; + screen->base.channel->flush_notify = nv40_state_flush_notify; + nv40_init_query_functions(nv40); nv40_init_surface_functions(nv40); nv40_init_state_functions(nv40); diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index a3d594167a..cf33b64a86 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -204,6 +204,7 @@ extern void nv40_fragtex_bind(struct nv40_context *); extern boolean nv40_state_validate(struct nv40_context *nv40); extern boolean nv40_state_validate_swtnl(struct nv40_context *nv40); extern void nv40_state_emit(struct nv40_context *nv40); +extern void nv40_state_flush_notify(struct nouveau_channel *chan); extern struct nv40_state_entry nv40_state_rasterizer; extern struct nv40_state_entry nv40_state_scissor; extern struct nv40_state_entry nv40_state_stipple; diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c index f73bedff6d..89ddf373e9 100644 --- a/src/gallium/drivers/nv40/nv40_miptree.c +++ b/src/gallium/drivers/nv40/nv40_miptree.c @@ -5,11 +5,13 @@ #include "nv40_context.h" + + static void nv40_miptree_layout(struct nv40_miptree *mt) { struct pipe_texture *pt = &mt->base; - uint width = pt->width0, height = pt->height0, depth = pt->depth0; + uint width = pt->width0; uint offset = 0; int nr_faces, l, f; uint wide_pitch = pt->tex_usage & (PIPE_TEXTURE_USAGE_SAMPLER | @@ -28,20 +30,15 @@ nv40_miptree_layout(struct nv40_miptree *mt) } for (l = 0; l <= pt->last_level; l++) { - pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width); - pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); - if (wide_pitch && (pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) - mt->level[l].pitch = align(pt->width0 * pt->block.size, 64); + mt->level[l].pitch = align(pf_get_stride(pt->format, pt->width0), 64); else - mt->level[l].pitch = u_minify(pt->width0, l) * pt->block.size; + mt->level[l].pitch = pf_get_stride(pt->format, width); mt->level[l].image_offset = CALLOC(nr_faces, sizeof(unsigned)); width = u_minify(width, 1); - height = u_minify(height, 1); - depth = u_minify(depth, 1); } for (f = 0; f < nr_faces; f++) { @@ -114,7 +111,7 @@ nv40_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt) FREE(mt); return NULL; } - + mt->bo = nouveau_bo(mt->buffer); return &mt->base; } @@ -143,6 +140,7 @@ nv40_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt, mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR; pipe_buffer_reference(&mt->buffer, pb); + mt->bo = nouveau_bo(mt->buffer); return &mt->base; } diff --git a/src/gallium/drivers/nv40/nv40_state.h b/src/gallium/drivers/nv40/nv40_state.h index 8a9d8c8fdf..192074e747 100644 --- a/src/gallium/drivers/nv40/nv40_state.h +++ b/src/gallium/drivers/nv40/nv40_state.h @@ -75,6 +75,7 @@ struct nv40_fragment_program { struct nv40_miptree { struct pipe_texture base; + struct nouveau_bo *bo; struct pipe_buffer *buffer; uint total_size; diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c index 198692965d..ba0fbcb26a 100644 --- a/src/gallium/drivers/nv40/nv40_state_emit.c +++ b/src/gallium/drivers/nv40/nv40_state_emit.c @@ -57,7 +57,7 @@ nv40_state_emit(struct nv40_context *nv40) struct nouveau_channel *chan = nv40->screen->base.channel; struct nv40_state *state = &nv40->state; struct nv40_screen *screen = nv40->screen; - unsigned i, samplers; + unsigned i; uint64_t states; if (nv40->pctx_id != screen->cur_pctx) { @@ -87,6 +87,14 @@ nv40_state_emit(struct nv40_context *nv40) } state->dirty = 0; +} + +void +nv40_state_flush_notify(struct nouveau_channel *chan) +{ + struct nv40_context *nv40 = chan->user_private; + struct nv40_state *state = &nv40->state; + unsigned i, samplers; so_emit_reloc_markers(chan, state->hw[NV40_STATE_FB]); for (i = 0, samplers = state->fp_samplers; i < 16 && samplers; i++) { diff --git a/src/gallium/drivers/nv40/nv40_transfer.c b/src/gallium/drivers/nv40/nv40_transfer.c index 36e253c96f..ed5be1cf87 100644 --- a/src/gallium/drivers/nv40/nv40_transfer.c +++ b/src/gallium/drivers/nv40/nv40_transfer.c @@ -11,7 +11,7 @@ struct nv40_transfer { struct pipe_transfer base; struct pipe_surface *surface; - bool direct; + boolean direct; }; static void @@ -24,9 +24,6 @@ nv40_compatible_transfer_tex(struct pipe_texture *pt, unsigned level, template->width0 = u_minify(pt->width0, level); template->height0 = u_minify(pt->height0, level); template->depth0 = 1; - template->block = pt->block; - template->nblocksx[0] = pt->nblocksx[level]; - template->nblocksy[0] = pt->nblocksx[level]; template->last_level = 0; template->nr_samples = pt->nr_samples; @@ -49,14 +46,10 @@ nv40_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; pipe_texture_reference(&tx->base.texture, pt); - tx->base.format = pt->format; tx->base.x = x; tx->base.y = y; tx->base.width = w; tx->base.height = h; - tx->base.block = pt->block; - tx->base.nblocksx = pt->nblocksx[level]; - tx->base.nblocksy = pt->nblocksy[level]; tx->base.stride = mt->level[level].pitch; tx->base.usage = usage; tx->base.face = face; @@ -158,7 +151,7 @@ nv40_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) pipe_transfer_buffer_flags(ptx)); return map + ns->base.offset + - ptx->y * ns->pitch + ptx->x * ptx->block.size; + ptx->y * ns->pitch + ptx->x * pf_get_blocksize(ptx->texture->format); } static void diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index 4b0f062295..5578a5838f 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -65,7 +65,7 @@ struct nv50_rasterizer_stateobj { }; struct nv50_sampler_stateobj { - bool normalized; + boolean normalized; unsigned tsc[8]; }; @@ -126,7 +126,7 @@ struct nv50_state { unsigned viewport_bypass; struct nouveau_stateobj *tsc_upload; struct nouveau_stateobj *tic_upload; - unsigned miptree_nr; + unsigned miptree_nr[PIPE_SHADER_TYPES]; struct nouveau_stateobj *vertprog; struct nouveau_stateobj *fragprog; struct nouveau_stateobj *programs; @@ -162,10 +162,10 @@ struct nv50_context { unsigned vtxbuf_nr; struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS]; unsigned vtxelt_nr; - struct nv50_sampler_stateobj *sampler[PIPE_MAX_SAMPLERS]; - unsigned sampler_nr; - struct nv50_miptree *miptree[PIPE_MAX_SAMPLERS]; - unsigned miptree_nr; + struct nv50_sampler_stateobj *sampler[PIPE_SHADER_TYPES][PIPE_MAX_SAMPLERS]; + unsigned sampler_nr[PIPE_SHADER_TYPES]; + struct nv50_miptree *miptree[PIPE_SHADER_TYPES][PIPE_MAX_SAMPLERS]; + unsigned miptree_nr[PIPE_SHADER_TYPES]; uint16_t vbo_fifo; }; @@ -218,7 +218,7 @@ extern void nv50_state_flush_notify(struct nouveau_channel *chan); extern void nv50_so_init_sifc(struct nv50_context *nv50, struct nouveau_stateobj *so, struct nouveau_bo *bo, unsigned reloc, - unsigned size); + unsigned offset, unsigned size); /* nv50_tex.c */ extern void nv50_tex_validate(struct nv50_context *); diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c index 3d58746793..9e083b662d 100644 --- a/src/gallium/drivers/nv50/nv50_miptree.c +++ b/src/gallium/drivers/nv50/nv50_miptree.c @@ -55,6 +55,20 @@ get_tile_mode(unsigned ny, unsigned d) return tile_mode | 0x10; } +static INLINE unsigned +get_zslice_offset(unsigned tile_mode, unsigned z, unsigned pitch, unsigned nb_h) +{ + unsigned tile_h = get_tile_height(tile_mode); + unsigned tile_d = get_tile_depth(tile_mode); + + /* pitch_2d == to next slice within this volume-tile */ + /* pitch_3d == size (in bytes) of a volume-tile */ + unsigned pitch_2d = tile_h * 64; + unsigned pitch_3d = tile_d * align(nb_h, tile_h) * pitch; + + return (z % tile_d) * pitch_2d + (z / tile_d) * pitch_3d; +} + static struct pipe_texture * nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *tmp) { @@ -91,13 +105,11 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *tmp) for (l = 0; l <= pt->last_level; l++) { struct nv50_miptree_level *lvl = &mt->level[l]; - - pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width); - pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height); + unsigned nblocksy = pf_get_nblocksy(pt->format, height); lvl->image_offset = CALLOC(mt->image_nr, sizeof(int)); - lvl->pitch = align(pt->nblocksx[l] * pt->block.size, 64); - lvl->tile_mode = get_tile_mode(pt->nblocksy[l], depth); + lvl->pitch = align(pf_get_stride(pt->format, width), 64); + lvl->tile_mode = get_tile_mode(nblocksy, depth); width = u_minify(width, 1); height = u_minify(height, 1); @@ -118,7 +130,7 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *tmp) unsigned tile_d = get_tile_depth(lvl->tile_mode); size = lvl->pitch; - size *= align(pt->nblocksy[l], tile_h); + size *= align(pf_get_nblocksy(pt->format, u_minify(pt->height0, l)), tile_h); size *= align(u_minify(pt->depth0, l), tile_d); lvl->image_offset[i] = mt->total_size; @@ -132,6 +144,8 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *tmp) mt->level[0].tile_mode, tile_flags, &mt->base.bo); if (ret) { + for (l = 0; l < pt->last_level; ++l) + FREE(mt->level[l].image_offset); FREE(mt); return NULL; } @@ -171,6 +185,10 @@ static void nv50_miptree_destroy(struct pipe_texture *pt) { struct nv50_miptree *mt = nv50_miptree(pt); + unsigned l; + + for (l = 0; l < pt->last_level; ++l) + FREE(mt->level[l].image_offset); nouveau_bo_ref(NULL, &mt->base.bo); FREE(mt); @@ -184,15 +202,10 @@ nv50_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, struct nv50_miptree *mt = nv50_miptree(pt); struct nv50_miptree_level *lvl = &mt->level[level]; struct pipe_surface *ps; - int img; + unsigned img = 0; if (pt->target == PIPE_TEXTURE_CUBE) img = face; - else - if (pt->target == PIPE_TEXTURE_3D) - img = zslice; - else - img = 0; ps = CALLOC_STRUCT(pipe_surface); if (!ps) @@ -208,6 +221,12 @@ nv50_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt, ps->zslice = zslice; ps->offset = lvl->image_offset[img]; + if (pt->target == PIPE_TEXTURE_3D) { + unsigned nb_h = pf_get_nblocksy(pt->format, ps->height); + ps->offset += get_zslice_offset(lvl->tile_mode, zslice, + lvl->pitch, nb_h); + } + return ps; } diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index f0fe7e6168..e496cf4cad 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -119,7 +119,7 @@ struct nv50_pc { struct nv50_reg *param; int param_nr; struct nv50_reg *immd; - float *immd_buf; + uint32_t *immd_buf; int immd_nr; struct nv50_reg **addr; int addr_nr; @@ -131,6 +131,9 @@ struct nv50_pc { struct nv50_reg *r_brdc; struct nv50_reg *r_dst[4]; + struct nv50_reg reg_instances[16]; + unsigned reg_instance_nr; + unsigned interp_mode[32]; /* perspective interpolation registers */ struct nv50_reg *iv_p; @@ -150,6 +153,20 @@ struct nv50_pc { boolean allow32; }; +static INLINE struct nv50_reg * +reg_instance(struct nv50_pc *pc, struct nv50_reg *reg) +{ + struct nv50_reg *ri; + + assert(pc->reg_instance_nr < 16); + ri = &pc->reg_instances[pc->reg_instance_nr++]; + if (reg) { + *ri = *reg; + reg->mod = 0; + } + return ri; +} + static INLINE void ctor_reg(struct nv50_reg *reg, unsigned type, int index, int hw) { @@ -342,25 +359,34 @@ static void kill_temp_temp(struct nv50_pc *pc) { int i; - + for (i = 0; i < pc->temp_temp_nr; i++) free_temp(pc, pc->temp_temp[i]); pc->temp_temp_nr = 0; } static int -ctor_immd(struct nv50_pc *pc, float x, float y, float z, float w) +ctor_immd_4u32(struct nv50_pc *pc, + uint32_t x, uint32_t y, uint32_t z, uint32_t w) { - pc->immd_buf = REALLOC(pc->immd_buf, (pc->immd_nr * 4 * sizeof(float)), - (pc->immd_nr + 1) * 4 * sizeof(float)); + unsigned size = pc->immd_nr * 4 * sizeof(uint32_t); + + pc->immd_buf = REALLOC(pc->immd_buf, size, size + 4 * sizeof(uint32_t)); + pc->immd_buf[(pc->immd_nr * 4) + 0] = x; pc->immd_buf[(pc->immd_nr * 4) + 1] = y; pc->immd_buf[(pc->immd_nr * 4) + 2] = z; pc->immd_buf[(pc->immd_nr * 4) + 3] = w; - + return pc->immd_nr++; } +static INLINE int +ctor_immd_4f32(struct nv50_pc *pc, float x, float y, float z, float w) +{ + return ctor_immd_4u32(pc, fui(x), fui(y), fui(z), fui(w)); +} + static struct nv50_reg * alloc_immd(struct nv50_pc *pc, float f) { @@ -368,11 +394,11 @@ alloc_immd(struct nv50_pc *pc, float f) unsigned hw; for (hw = 0; hw < pc->immd_nr * 4; hw++) - if (pc->immd_buf[hw] == f) + if (pc->immd_buf[hw] == fui(f)) break; if (hw == pc->immd_nr * 4) - hw = ctor_immd(pc, f, -f, 0.5 * f, 0) * 4; + hw = ctor_immd_4f32(pc, f, -f, 0.5 * f, 0) * 4; ctor_reg(r, P_IMMD, -1, hw); return r; @@ -464,22 +490,24 @@ set_dst(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_program_exec *e) static INLINE void set_immd(struct nv50_pc *pc, struct nv50_reg *imm, struct nv50_program_exec *e) { - unsigned val; - float f = pc->immd_buf[imm->hw]; + union { + float f; + uint32_t ui; + } u; + u.ui = pc->immd_buf[imm->hw]; - if (imm->mod & NV50_MOD_ABS) - f = fabsf(f); - val = fui((imm->mod & NV50_MOD_NEG) ? -f : f); + u.f = (imm->mod & NV50_MOD_ABS) ? fabsf(u.f) : u.f; + u.f = (imm->mod & NV50_MOD_NEG) ? -u.f : u.f; set_long(pc, e); - /*XXX: can't be predicated - bits overlap.. catch cases where both - * are required and avoid them. */ + /* XXX: can't be predicated - bits overlap; cases where both + * are required should be avoided by using pc->allow32 */ set_pred(pc, 0, 0, e); set_pred_wr(pc, 0, 0, e); e->inst[1] |= 0x00000002 | 0x00000001; - e->inst[0] |= (val & 0x3f) << 16; - e->inst[1] |= (val >> 6) << 2; + e->inst[0] |= (u.ui & 0x3f) << 16; + e->inst[1] |= (u.ui >> 6) << 2; } static INLINE void @@ -644,7 +672,7 @@ emit_mov(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) if (src->type == P_IMMD || src->type == P_CONST) { set_long(pc, e); set_data(pc, src, 0x7f, 9, e); - e->inst[1] |= 0x20000000; /* src0 const? */ + e->inst[1] |= 0x20000000; /* mov from c[] */ } else { if (src->type == P_ATTR) { set_long(pc, e); @@ -659,9 +687,9 @@ emit_mov(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) if (is_long(e) && !is_immd(e)) { e->inst[1] |= 0x04000000; /* 32-bit */ - e->inst[1] |= 0x0000c000; /* "subsubop" 0x3 */ + e->inst[1] |= 0x0000c000; /* 32-bit c[] load / lane mask 0:1 */ if (!(e->inst[1] & 0x20000000)) - e->inst[1] |= 0x00030000; /* "subsubop" 0xf */ + e->inst[1] |= 0x00030000; /* lane mask 2:3 */ } else e->inst[0] |= 0x00008000; @@ -676,6 +704,17 @@ emit_mov_immdval(struct nv50_pc *pc, struct nv50_reg *dst, float f) FREE(imm); } +static void +emit_nop(struct nv50_pc *pc) +{ + struct nv50_program_exec *e = exec(pc); + + e->inst[0] = 0xf0000000; + set_long(pc, e); + e->inst[1] = 0xe0000000; + emit(pc, e); +} + static boolean check_swap_src_0_1(struct nv50_pc *pc, struct nv50_reg **s0, struct nv50_reg **s1) @@ -795,6 +834,33 @@ set_src_2(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e) } static void +emit_mov_from_pred(struct nv50_pc *pc, struct nv50_reg *dst, int pred) +{ + struct nv50_program_exec *e = exec(pc); + + assert(dst->type == P_TEMP); + e->inst[1] = 0x20000000 | (pred << 12); + set_long(pc, e); + set_dst(pc, dst, e); + + emit(pc, e); +} + +static void +emit_mov_to_pred(struct nv50_pc *pc, int pred, struct nv50_reg *src) +{ + struct nv50_program_exec *e = exec(pc); + + e->inst[0] = 0x000001fc; + e->inst[1] = 0xa0000008; + set_long(pc, e); + set_pred_wr(pc, 1, pred, e); + set_src_0_restricted(pc, src, e); + + emit(pc, e); +} + +static void emit_mul(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, struct nv50_reg *src1) { @@ -898,7 +964,6 @@ static INLINE void emit_sub(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, struct nv50_reg *src1) { - assert(src0 != src1); src1->mod ^= NV50_MOD_NEG; emit_add(pc, dst, src0, src1); src1->mod ^= NV50_MOD_NEG; @@ -967,7 +1032,6 @@ static INLINE void emit_msb(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0, struct nv50_reg *src1, struct nv50_reg *src2) { - assert(src2 != src0 && src2 != src1); src2->mod ^= NV50_MOD_NEG; emit_mad(pc, dst, src0, src1, src2); src2->mod ^= NV50_MOD_NEG; @@ -1257,9 +1321,68 @@ emit_kil(struct nv50_pc *pc, struct nv50_reg *src) emit(pc, e); } +static struct nv50_program_exec * +emit_branch(struct nv50_pc *pc, int pred, unsigned cc, + struct nv50_program_exec **join) +{ + struct nv50_program_exec *e = exec(pc); + + if (join) { + set_long(pc, e); + e->inst[0] |= 0xa0000002; + emit(pc, e); + *join = e; + e = exec(pc); + } + + set_long(pc, e); + e->inst[0] |= 0x10000002; + if (pred >= 0) + set_pred(pc, cc, pred, e); + emit(pc, e); + return pc->p->exec_tail; +} + +#define QOP_ADD 0 +#define QOP_SUBR 1 +#define QOP_SUB 2 +#define QOP_MOV_SRC1 3 + +/* For a quad of threads / top left, top right, bottom left, bottom right + * pixels, do a different operation, and take src0 from a specific thread. + */ +static void +emit_quadop(struct nv50_pc *pc, struct nv50_reg *dst, int wp, int lane_src0, + struct nv50_reg *src0, struct nv50_reg *src1, ubyte qop) +{ + struct nv50_program_exec *e = exec(pc); + + e->inst[0] = 0xc0000000; + e->inst[1] = 0x80000000; + set_long(pc, e); + e->inst[0] |= lane_src0 << 16; + set_src_0(pc, src0, e); + set_src_2(pc, src1, e); + + if (wp >= 0) + set_pred_wr(pc, 1, wp, e); + + if (dst) + set_dst(pc, dst, e); + else { + e->inst[0] |= 0x000001fc; + e->inst[1] |= 0x00000008; + } + + e->inst[0] |= (qop & 3) << 20; + e->inst[1] |= (qop >> 2) << 22; + + emit(pc, e); +} + static void load_cube_tex_coords(struct nv50_pc *pc, struct nv50_reg *t[4], - struct nv50_reg **src, boolean proj) + struct nv50_reg **src, unsigned arg, boolean proj) { int mod[3] = { src[0]->mod, src[1]->mod, src[2]->mod }; @@ -1276,6 +1399,10 @@ load_cube_tex_coords(struct nv50_pc *pc, struct nv50_reg *t[4], if (proj && 0 /* looks more correct without this */) emit_mul(pc, t[2], t[2], src[3]); + else + if (arg == 4) /* there is no textureProj(samplerCubeShadow) */ + emit_mov(pc, t[3], src[3]); + emit_flop(pc, 0, t[2], t[2]); emit_mul(pc, t[0], src[0], t[2]); @@ -1284,89 +1411,214 @@ load_cube_tex_coords(struct nv50_pc *pc, struct nv50_reg *t[4], } static void -emit_tex(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask, - struct nv50_reg **src, unsigned unit, unsigned type, boolean proj) +load_proj_tex_coords(struct nv50_pc *pc, struct nv50_reg *t[4], + struct nv50_reg **src, unsigned dim, unsigned arg) { - struct nv50_reg *t[4]; - struct nv50_program_exec *e; + unsigned c, mode; + + if (src[0]->type == P_TEMP && src[0]->rhw != -1) { + mode = pc->interp_mode[src[0]->index] | INTERP_PERSPECTIVE; - unsigned c, mode, dim; + t[3]->rhw = src[3]->rhw; + emit_interp(pc, t[3], NULL, (mode & INTERP_CENTROID)); + emit_flop(pc, 0, t[3], t[3]); + for (c = 0; c < dim; ++c) { + t[c]->rhw = src[c]->rhw; + emit_interp(pc, t[c], t[3], mode); + } + if (arg != dim) { /* depth reference value */ + t[dim]->rhw = src[2]->rhw; + emit_interp(pc, t[dim], t[3], mode); + } + } else { + /* XXX: for some reason the blob sometimes uses MAD + * (mad f32 $rX $rY $rZ neg $r63) + */ + emit_flop(pc, 0, t[3], src[3]); + for (c = 0; c < dim; ++c) + emit_mul(pc, t[c], src[c], t[3]); + if (arg != dim) /* depth reference value */ + emit_mul(pc, t[dim], src[2], t[3]); + } +} + +static INLINE void +get_tex_dim(unsigned type, unsigned *dim, unsigned *arg) +{ switch (type) { case TGSI_TEXTURE_1D: - dim = 1; + *arg = *dim = 1; + break; + case TGSI_TEXTURE_SHADOW1D: + *dim = 1; + *arg = 2; break; case TGSI_TEXTURE_UNKNOWN: case TGSI_TEXTURE_2D: - case TGSI_TEXTURE_SHADOW1D: /* XXX: x, z */ case TGSI_TEXTURE_RECT: - dim = 2; + *arg = *dim = 2; + break; + case TGSI_TEXTURE_SHADOW2D: + case TGSI_TEXTURE_SHADOWRECT: + *dim = 2; + *arg = 3; break; case TGSI_TEXTURE_3D: case TGSI_TEXTURE_CUBE: - case TGSI_TEXTURE_SHADOW2D: - case TGSI_TEXTURE_SHADOWRECT: /* XXX */ - dim = 3; + *dim = *arg = 3; break; default: assert(0); break; } +} - /* some cards need t[0]'s hw index to be a multiple of 4 */ - alloc_temp4(pc, t, 0); +/* We shouldn't execute TEXLOD if any of the pixels in a quad have + * different LOD values, so branch off groups of equal LOD. + */ +static void +emit_texlod_sequence(struct nv50_pc *pc, struct nv50_reg *tlod, + struct nv50_reg *src, struct nv50_program_exec *tex) +{ + struct nv50_program_exec *join_at; + unsigned i, target = pc->p->exec_size + 7 * 2; - if (type == TGSI_TEXTURE_CUBE) { - load_cube_tex_coords(pc, t, src, proj); - } else - if (proj) { - if (src[0]->type == P_TEMP && src[0]->rhw != -1) { - mode = pc->interp_mode[src[0]->index]; - - t[3]->rhw = src[3]->rhw; - emit_interp(pc, t[3], NULL, (mode & INTERP_CENTROID)); - emit_flop(pc, 0, t[3], t[3]); - - for (c = 0; c < dim; c++) { - t[c]->rhw = src[c]->rhw; - emit_interp(pc, t[c], t[3], - (mode | INTERP_PERSPECTIVE)); - } - } else { - emit_flop(pc, 0, t[3], src[3]); - for (c = 0; c < dim; c++) - emit_mul(pc, t[c], src[c], t[3]); + /* Subtract lod of each pixel from lod of top left pixel, jump + * texlod insn if result is 0, then repeat for 2 other pixels. + */ + emit_quadop(pc, NULL, 0, 0, tlod, tlod, 0x55); + emit_branch(pc, 0, 2, &join_at)->param.index = target; - /* XXX: for some reason the blob sometimes uses MAD: - * emit_mad(pc, t[c], src[0][c], t[3], t[3]) - * pc->p->exec_tail->inst[1] |= 0x080fc000; - */ + for (i = 1; i < 4; ++i) { + emit_quadop(pc, NULL, 0, i, tlod, tlod, 0x55); + emit_branch(pc, 0, 2, NULL)->param.index = target; + } + + emit_mov(pc, tlod, src); /* target */ + emit(pc, tex); /* texlod */ + + join_at->param.index = target + 2 * 2; + emit_nop(pc); + pc->p->exec_tail->inst[1] |= 2; /* join _after_ tex */ +} + +static void +emit_texbias_sequence(struct nv50_pc *pc, struct nv50_reg *t[4], unsigned arg, + struct nv50_program_exec *tex) +{ + struct nv50_program_exec *e; + struct nv50_reg imm_1248, *t123[4][4], *r_bits = alloc_temp(pc, NULL); + int r_pred = 0; + unsigned n, c, i, cc[4] = { 0x0a, 0x13, 0x11, 0x10 }; + + pc->allow32 = FALSE; + ctor_reg(&imm_1248, P_IMMD, -1, ctor_immd_4u32(pc, 1, 2, 4, 8) * 4); + + /* Subtract bias value of thread i from bias values of each thread, + * store result in r_pred, and set bit i in r_bits if result was 0. + */ + assert(arg < 4); + for (i = 0; i < 4; ++i, ++imm_1248.hw) { + emit_quadop(pc, NULL, r_pred, i, t[arg], t[arg], 0x55); + emit_mov(pc, r_bits, &imm_1248); + set_pred(pc, 2, r_pred, pc->p->exec_tail); + } + emit_mov_to_pred(pc, r_pred, r_bits); + + /* The lanes of a quad are now grouped by the bit in r_pred they have + * set. Put the input values for TEX into a new register set for each + * group and execute TEX only for a specific group. + * We cannot use the same register set for each group because we need + * the derivatives, which are implicitly calculated, to be correct. + */ + for (i = 1; i < 4; ++i) { + alloc_temp4(pc, t123[i], 0); + + for (c = 0; c <= arg; ++c) + emit_mov(pc, t123[i][c], t[c]); + + *(e = exec(pc)) = *(tex); + e->inst[0] &= ~0x01fc; + set_dst(pc, t123[i][0], e); + set_pred(pc, cc[i], r_pred, e); + emit(pc, e); + } + /* finally TEX on the original regs (where we kept the input) */ + set_pred(pc, cc[0], r_pred, tex); + emit(pc, tex); + + /* put the 3 * n other results into regs for lane 0 */ + n = popcnt4(((e->inst[0] >> 25) & 0x3) | ((e->inst[1] >> 12) & 0xc)); + for (i = 1; i < 4; ++i) { + for (c = 0; c < n; ++c) { + emit_mov(pc, t[c], t123[i][c]); + set_pred(pc, cc[i], r_pred, pc->p->exec_tail); } - } else { - for (c = 0; c < dim; c++) - emit_mov(pc, t[c], src[c]); + free_temp4(pc, t123[i]); } + emit_nop(pc); + free_temp(pc, r_bits); +} + +static void +emit_tex(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask, + struct nv50_reg **src, unsigned unit, unsigned type, + boolean proj, int bias_lod) +{ + struct nv50_reg *t[4]; + struct nv50_program_exec *e; + unsigned c, dim, arg; + + /* t[i] must be within a single 128 bit super-reg */ + alloc_temp4(pc, t, 0); + e = exec(pc); + e->inst[0] = 0xf0000000; set_long(pc, e); - e->inst[0] |= 0xf0000000; - e->inst[1] |= 0x00000004; set_dst(pc, t[0], e); - e->inst[0] |= (unit << 9); - if (dim == 2) - e->inst[0] |= 0x00400000; - else - if (dim == 3) { - e->inst[0] |= 0x00800000; - if (type == TGSI_TEXTURE_CUBE) - e->inst[0] |= 0x08000000; + /* TIC and TSC binding indices (TSC is ignored as TSC_LINKED = TRUE): */ + e->inst[0] |= (unit << 9) /* | (unit << 17) */; + + /* live flag (don't set if TEX results affect input to another TEX): */ + /* e->inst[0] |= 0x00000004; */ + + get_tex_dim(type, &dim, &arg); + + if (type == TGSI_TEXTURE_CUBE) { + e->inst[0] |= 0x08000000; + load_cube_tex_coords(pc, t, src, arg, proj); + } else + if (proj) + load_proj_tex_coords(pc, t, src, dim, arg); + else { + for (c = 0; c < dim; c++) + emit_mov(pc, t[c], src[c]); + if (arg != dim) /* depth reference value (always src.z here) */ + emit_mov(pc, t[dim], src[2]); } e->inst[0] |= (mask & 0x3) << 25; e->inst[1] |= (mask & 0xc) << 12; - emit(pc, e); + if (!bias_lod) { + e->inst[0] |= (arg - 1) << 22; + emit(pc, e); + } else + if (bias_lod < 0) { + e->inst[0] |= arg << 22; + e->inst[1] |= 0x20000000; /* texbias */ + emit_mov(pc, t[arg], src[3]); + emit_texbias_sequence(pc, t, arg, e); + } else { + e->inst[0] |= arg << 22; + e->inst[1] |= 0x40000000; /* texlod */ + emit_mov(pc, t[arg], src[3]); + emit_texlod_sequence(pc, t[arg], src[3], e); + } + #if 1 c = 0; if (mask & 1) emit_mov(pc, dst[0], t[c++]); @@ -1389,38 +1641,6 @@ emit_tex(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask, } static void -emit_branch(struct nv50_pc *pc, int pred, unsigned cc, - struct nv50_program_exec **join) -{ - struct nv50_program_exec *e = exec(pc); - - if (join) { - set_long(pc, e); - e->inst[0] |= 0xa0000002; - emit(pc, e); - *join = e; - e = exec(pc); - } - - set_long(pc, e); - e->inst[0] |= 0x10000002; - if (pred >= 0) - set_pred(pc, cc, pred, e); - emit(pc, e); -} - -static void -emit_nop(struct nv50_pc *pc) -{ - struct nv50_program_exec *e = exec(pc); - - e->inst[0] = 0xf0000000; - set_long(pc, e); - e->inst[1] = 0xe0000000; - emit(pc, e); -} - -static void emit_ddx(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) { struct nv50_program_exec *e = exec(pc); @@ -1515,8 +1735,6 @@ convert_to_long(struct nv50_pc *pc, struct nv50_program_exec *e) static boolean negate_supported(const struct tgsi_full_instruction *insn, int i) { - int s; - switch (insn->Instruction.Opcode) { case TGSI_OPCODE_DDY: case TGSI_OPCODE_DP3: @@ -1526,29 +1744,14 @@ negate_supported(const struct tgsi_full_instruction *insn, int i) case TGSI_OPCODE_ADD: case TGSI_OPCODE_SUB: case TGSI_OPCODE_MAD: - break; + return TRUE; case TGSI_OPCODE_POW: if (i == 1) - break; + return TRUE; return FALSE; default: return FALSE; } - - /* Watch out for possible multiple uses of an nv50_reg, we - * can't use nv50_reg::neg in these cases. - */ - for (s = 0; s < insn->Instruction.NumSrcRegs; ++s) { - if (s == i) - continue; - if ((insn->Src[s].Register.Index == - insn->Src[i].Register.Index) && - (insn->Src[s].Register.File == - insn->Src[i].Register.File)) - return FALSE; - } - - return TRUE; } /* Return a read mask for source registers deduced from opcode & write mask. */ @@ -1576,9 +1779,13 @@ nv50_tgsi_src_mask(const struct tgsi_full_instruction *insn, int c) case TGSI_OPCODE_RSQ: case TGSI_OPCODE_SCS: return 0x1; + case TGSI_OPCODE_IF: + return 0x1; case TGSI_OPCODE_LIT: return 0xb; case TGSI_OPCODE_TEX: + case TGSI_OPCODE_TXB: + case TGSI_OPCODE_TXL: case TGSI_OPCODE_TXP: { const struct tgsi_instruction_texture *tex; @@ -1587,13 +1794,17 @@ nv50_tgsi_src_mask(const struct tgsi_full_instruction *insn, int c) tex = &insn->Texture; mask = 0x7; - if (insn->Instruction.Opcode == TGSI_OPCODE_TXP) - mask |= 0x8; + if (insn->Instruction.Opcode != TGSI_OPCODE_TEX && + insn->Instruction.Opcode != TGSI_OPCODE_TXD) + mask |= 0x8; /* bias, lod or proj */ switch (tex->Texture) { case TGSI_TEXTURE_1D: mask &= 0x9; break; + case TGSI_TEXTURE_SHADOW1D: + mask &= 0x5; + break; case TGSI_TEXTURE_2D: mask &= 0xb; break; @@ -1676,7 +1887,7 @@ tgsi_src(struct nv50_pc *pc, int chan, const struct tgsi_full_src_register *src, /* Indicate indirection by setting r->acc < 0 and * use the index field to select the address reg. */ - r = MALLOC_STRUCT(nv50_reg); + r = reg_instance(pc, NULL); swz = tgsi_util_get_src_register_swizzle( &src->Indirect, 0); ctor_reg(r, P_CONST, @@ -1730,6 +1941,8 @@ tgsi_src(struct nv50_pc *pc, int chan, const struct tgsi_full_src_register *src, break; } + if (r && r->acc >= 0 && r != temp) + return reg_instance(pc, r); return r; } @@ -1785,6 +1998,8 @@ nv50_tgsi_dst_revdep(unsigned op, int s, int c) case TGSI_OPCODE_LIT: case TGSI_OPCODE_SCS: case TGSI_OPCODE_TEX: + case TGSI_OPCODE_TXB: + case TGSI_OPCODE_TXL: case TGSI_OPCODE_TXP: /* these take care of dangerous swizzles themselves */ return 0x0; @@ -2187,11 +2402,19 @@ nv50_program_tx_insn(struct nv50_pc *pc, break; case TGSI_OPCODE_TEX: emit_tex(pc, dst, mask, src[0], unit, - inst->Texture.Texture, FALSE); + inst->Texture.Texture, FALSE, 0); + break; + case TGSI_OPCODE_TXB: + emit_tex(pc, dst, mask, src[0], unit, + inst->Texture.Texture, FALSE, -1); + break; + case TGSI_OPCODE_TXL: + emit_tex(pc, dst, mask, src[0], unit, + inst->Texture.Texture, FALSE, 1); break; case TGSI_OPCODE_TXP: emit_tex(pc, dst, mask, src[0], unit, - inst->Texture.Texture, TRUE); + inst->Texture.Texture, TRUE, 0); break; case TGSI_OPCODE_TRUNC: for (c = 0; c < 4; c++) { @@ -2245,20 +2468,9 @@ nv50_program_tx_insn(struct nv50_pc *pc, } } - for (i = 0; i < inst->Instruction.NumSrcRegs; i++) { - for (c = 0; c < 4; c++) { - if (!src[i][c]) - continue; - src[i][c]->mod = 0; - if (src[i][c]->index == -1 && src[i][c]->type == P_IMMD) - FREE(src[i][c]); - else - if (src[i][c]->acc < 0 && src[i][c]->type == P_CONST) - FREE(src[i][c]); /* indirect constant */ - } - } - kill_temp_temp(pc); + pc->reg_instance_nr = 0; + return TRUE; } @@ -2541,10 +2753,10 @@ nv50_program_tx_prep(struct nv50_pc *pc) const struct tgsi_full_immediate *imm = &tp.FullToken.FullImmediate; - ctor_immd(pc, imm->u[0].Float, - imm->u[1].Float, - imm->u[2].Float, - imm->u[3].Float); + ctor_immd_4f32(pc, imm->u[0].Float, + imm->u[1].Float, + imm->u[2].Float, + imm->u[3].Float); } break; case TGSI_TOKEN_TYPE_DECLARATION: @@ -3024,7 +3236,7 @@ nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p) } static void -nv50_program_upload_data(struct nv50_context *nv50, float *map, +nv50_program_upload_data(struct nv50_context *nv50, uint32_t *map, unsigned start, unsigned count, unsigned cbuf) { struct nouveau_channel *chan = nv50->screen->base.channel; @@ -3072,8 +3284,8 @@ nv50_program_validate_data(struct nv50_context *nv50, struct nv50_program *p) if (p->param_nr) { unsigned cb; - float *map = pipe_buffer_map(pscreen, nv50->constbuf[p->type], - PIPE_BUFFER_USAGE_CPU_READ); + uint32_t *map = pipe_buffer_map(pscreen, nv50->constbuf[p->type], + PIPE_BUFFER_USAGE_CPU_READ); if (p->type == PIPE_SHADER_VERTEX) cb = NV50_CB_PVP; diff --git a/src/gallium/drivers/nv50/nv50_program.h b/src/gallium/drivers/nv50/nv50_program.h index 255c7c737e..4a90c372ce 100644 --- a/src/gallium/drivers/nv50/nv50_program.h +++ b/src/gallium/drivers/nv50/nv50_program.h @@ -37,7 +37,7 @@ struct nv50_program { struct nouveau_bo *bo; - float *immd; + uint32_t *immd; unsigned immd_nr; unsigned param_nr; diff --git a/src/gallium/drivers/nv50/nv50_query.c b/src/gallium/drivers/nv50/nv50_query.c index 5305c93d59..268c9823f7 100644 --- a/src/gallium/drivers/nv50/nv50_query.c +++ b/src/gallium/drivers/nv50/nv50_query.c @@ -93,7 +93,7 @@ nv50_query_end(struct pipe_context *pipe, struct pipe_query *pq) struct nouveau_grobj *tesla = nv50->screen->tesla; struct nv50_query *q = nv50_query(pq); - WAIT_RING (chan, 5); + MARK_RING (chan, 5, 2); /* flush on lack of space or relocs */ BEGIN_RING(chan, tesla, NV50TCL_QUERY_ADDRESS_HIGH, 4); OUT_RELOCh(chan, q->bo, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); OUT_RELOCl(chan, q->bo, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index e1b2f11239..d443ca3ad0 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -76,6 +76,7 @@ nv50_screen_is_format_supported(struct pipe_screen *pscreen, case PIPE_FORMAT_DXT3_RGBA: case PIPE_FORMAT_DXT5_RGBA: case PIPE_FORMAT_Z24S8_UNORM: + case PIPE_FORMAT_S8Z24_UNORM: case PIPE_FORMAT_Z32_FLOAT: case PIPE_FORMAT_R16G16B16A16_SNORM: case PIPE_FORMAT_R16G16B16A16_UNORM: @@ -97,6 +98,10 @@ nv50_screen_get_param(struct pipe_screen *pscreen, int param) switch (param) { case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS: return 32; + case PIPE_CAP_MAX_VERTEX_TEXTURE_UNITS: + return 32; + case PIPE_CAP_MAX_COMBINED_SAMPLERS: + return 64; case PIPE_CAP_NPOT_TEXTURES: return 1; case PIPE_CAP_TWO_SIDED_STENCIL: @@ -122,8 +127,6 @@ nv50_screen_get_param(struct pipe_screen *pscreen, int param) case PIPE_CAP_TEXTURE_MIRROR_CLAMP: case PIPE_CAP_TEXTURE_MIRROR_REPEAT: return 1; - case PIPE_CAP_MAX_VERTEX_TEXTURE_UNITS: - return 0; case PIPE_CAP_TGSI_CONT_SUPPORTED: return 0; case PIPE_CAP_BLEND_EQUATION_SEPARATE: @@ -315,6 +318,9 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev) so_method(so, screen->tesla, 0x1400, 1); so_data (so, 0xf); + /* max TIC (bits 4:8) & TSC (ignored) bindings, per program type */ + so_method(so, screen->tesla, 0x13b4, 1); + so_data (so, 0x54); so_method(so, screen->tesla, 0x13bc, 1); so_data (so, 0x54); /* origin is top left (set to 1 for bottom left) */ @@ -387,7 +393,8 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev) so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1); so_data (so, 0x00000131 | (NV50_CB_PFP << 12)); - ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, 64*8*4, &screen->tic); + ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, PIPE_SHADER_TYPES*32*32, + &screen->tic); if (ret) { nv50_screen_destroy(pscreen); return NULL; @@ -398,9 +405,10 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev) NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); so_reloc (so, screen->tic, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); - so_data (so, 0x000007ff); + so_data (so, PIPE_SHADER_TYPES * 32 - 1); - ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, 64*8*4, &screen->tsc); + ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, PIPE_SHADER_TYPES*32*32, + &screen->tsc); if (ret) { nv50_screen_destroy(pscreen); return NULL; @@ -411,7 +419,7 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev) NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0); so_reloc (so, screen->tsc, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0); - so_data (so, 0x00000000); + so_data (so, 0x00000000); /* ignored if TSC_LINKED (0x1234) = 1 */ /* Vertex array limits - max them out */ @@ -425,6 +433,7 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev) so_data (so, fui(0.0)); so_data (so, fui(1.0)); + /* no dynamic combination of TIC & TSC entries => only BIND_TIC used */ so_method(so, screen->tesla, 0x1234, 1); so_data (so, 1); diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c index 07318f2394..88aef52d08 100644 --- a/src/gallium/drivers/nv50/nv50_state.c +++ b/src/gallium/drivers/nv50/nv50_state.c @@ -196,8 +196,9 @@ nv50_sampler_state_create(struct pipe_context *pipe, } if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { - tsc[0] |= (1 << 8); - tsc[0] |= (nvgl_comparison_op(cso->compare_func) & 0x7); + /* XXX: must be deactivated for non-shadow textures */ + tsc[0] |= (1 << 9); + tsc[0] |= (nvgl_comparison_op(cso->compare_func) & 0x7) << 10; } limit = CLAMP(cso->lod_bias, -16.0, 15.0); @@ -215,41 +216,66 @@ nv50_sampler_state_create(struct pipe_context *pipe, return (void *)sso; } -static void -nv50_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler) +static INLINE void +nv50_sampler_state_bind(struct pipe_context *pipe, unsigned type, + unsigned nr, void **sampler) { struct nv50_context *nv50 = nv50_context(pipe); - int i; - nv50->sampler_nr = nr; - for (i = 0; i < nv50->sampler_nr; i++) - nv50->sampler[i] = sampler[i]; + memcpy(nv50->sampler[type], sampler, nr * sizeof(void *)); + nv50->sampler_nr[type] = nr; nv50->dirty |= NV50_NEW_SAMPLER; } static void +nv50_vp_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **s) +{ + nv50_sampler_state_bind(pipe, PIPE_SHADER_VERTEX, nr, s); +} + +static void +nv50_fp_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **s) +{ + nv50_sampler_state_bind(pipe, PIPE_SHADER_FRAGMENT, nr, s); +} + +static void nv50_sampler_state_delete(struct pipe_context *pipe, void *hwcso) { FREE(hwcso); } -static void -nv50_set_sampler_texture(struct pipe_context *pipe, unsigned nr, - struct pipe_texture **pt) +static INLINE void +nv50_set_sampler_texture(struct pipe_context *pipe, unsigned type, + unsigned nr, struct pipe_texture **pt) { struct nv50_context *nv50 = nv50_context(pipe); - int i; + unsigned i; for (i = 0; i < nr; i++) - pipe_texture_reference((void *)&nv50->miptree[i], pt[i]); - for (i = nr; i < nv50->miptree_nr; i++) - pipe_texture_reference((void *)&nv50->miptree[i], NULL); + pipe_texture_reference((void *)&nv50->miptree[type][i], pt[i]); + for (i = nr; i < nv50->miptree_nr[type]; i++) + pipe_texture_reference((void *)&nv50->miptree[type][i], NULL); - nv50->miptree_nr = nr; + nv50->miptree_nr[type] = nr; nv50->dirty |= NV50_NEW_TEXTURE; } +static void +nv50_set_vp_sampler_textures(struct pipe_context *pipe, + unsigned nr, struct pipe_texture **pt) +{ + nv50_set_sampler_texture(pipe, PIPE_SHADER_VERTEX, nr, pt); +} + +static void +nv50_set_fp_sampler_textures(struct pipe_context *pipe, + unsigned nr, struct pipe_texture **pt) +{ + nv50_set_sampler_texture(pipe, PIPE_SHADER_FRAGMENT, nr, pt); +} + static void * nv50_rasterizer_state_create(struct pipe_context *pipe, const struct pipe_rasterizer_state *cso) @@ -648,9 +674,11 @@ nv50_init_state_functions(struct nv50_context *nv50) nv50->pipe.delete_blend_state = nv50_blend_state_delete; nv50->pipe.create_sampler_state = nv50_sampler_state_create; - nv50->pipe.bind_fragment_sampler_states = nv50_sampler_state_bind; nv50->pipe.delete_sampler_state = nv50_sampler_state_delete; - nv50->pipe.set_fragment_sampler_textures = nv50_set_sampler_texture; + nv50->pipe.bind_fragment_sampler_states = nv50_fp_sampler_state_bind; + nv50->pipe.bind_vertex_sampler_states = nv50_vp_sampler_state_bind; + nv50->pipe.set_fragment_sampler_textures = nv50_set_fp_sampler_textures; + nv50->pipe.set_vertex_sampler_textures = nv50_set_vp_sampler_textures; nv50->pipe.create_rasterizer_state = nv50_rasterizer_state_create; nv50->pipe.bind_rasterizer_state = nv50_rasterizer_state_bind; diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index c871acaab8..871e8097b6 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -156,6 +156,30 @@ nv50_state_validate_fb(struct nv50_context *nv50) } static void +nv50_validate_samplers(struct nv50_context *nv50, struct nouveau_stateobj *so, + unsigned p) +{ + struct nouveau_grobj *eng2d = nv50->screen->eng2d; + unsigned i, j, dw = nv50->sampler_nr[p] * 8; + + if (!dw) + return; + nv50_so_init_sifc(nv50, so, nv50->screen->tsc, NOUVEAU_BO_VRAM, + p * (32 * 8 * 4), dw * 4); + + so_method(so, eng2d, NV50_2D_SIFC_DATA | (2 << 29), dw); + + for (i = 0; i < nv50->sampler_nr[p]; ++i) { + if (nv50->sampler[p][i]) + so_datap(so, nv50->sampler[p][i]->tsc, 8); + else { + for (j = 0; j < 8; ++j) /* you get punished */ + so_data(so, 0); /* ... for leaving holes */ + } + } +} + +static void nv50_state_emit(struct nv50_context *nv50) { struct nv50_screen *screen = nv50->screen; @@ -246,7 +270,6 @@ boolean nv50_state_validate(struct nv50_context *nv50) { struct nouveau_grobj *tesla = nv50->screen->tesla; - struct nouveau_grobj *eng2d = nv50->screen->eng2d; struct nouveau_stateobj *so; unsigned i; @@ -369,22 +392,16 @@ scissor_uptodate: viewport_uptodate: if (nv50->dirty & NV50_NEW_SAMPLER) { - unsigned i; + unsigned nr = 0; - so = so_new(nv50->sampler_nr * 9 + 23 + 4, 2); + for (i = 0; i < PIPE_SHADER_TYPES; ++i) + nr += nv50->sampler_nr[i]; - nv50_so_init_sifc(nv50, so, nv50->screen->tsc, NOUVEAU_BO_VRAM, - nv50->sampler_nr * 8 * 4); + so = so_new(nr * 8 + 24 * PIPE_SHADER_TYPES + 2, 4); - for (i = 0; i < nv50->sampler_nr; i++) { - if (!nv50->sampler[i]) - continue; - so_method(so, eng2d, NV50_2D_SIFC_DATA | (2 << 29), 8); - so_datap (so, nv50->sampler[i]->tsc, 8); - } + nv50_validate_samplers(nv50, so, PIPE_SHADER_VERTEX); + nv50_validate_samplers(nv50, so, PIPE_SHADER_FRAGMENT); - so_method(so, tesla, 0x1440, 1); /* sync SIFC */ - so_data (so, 0); so_method(so, tesla, 0x1334, 1); /* flush TSC */ so_data (so, 0); @@ -407,10 +424,13 @@ viewport_uptodate: void nv50_so_init_sifc(struct nv50_context *nv50, struct nouveau_stateobj *so, - struct nouveau_bo *bo, unsigned reloc, unsigned size) + struct nouveau_bo *bo, unsigned reloc, + unsigned offset, unsigned size) { struct nouveau_grobj *eng2d = nv50->screen->eng2d; + reloc |= NOUVEAU_BO_WR; + so_method(so, eng2d, NV50_2D_DST_FORMAT, 2); so_data (so, NV50_2D_DST_FORMAT_R8_UNORM); so_data (so, 1); @@ -418,8 +438,8 @@ void nv50_so_init_sifc(struct nv50_context *nv50, so_data (so, 262144); so_data (so, 65536); so_data (so, 1); - so_reloc (so, bo, 0, reloc | NOUVEAU_BO_WR | NOUVEAU_BO_HIGH, 0, 0); - so_reloc (so, bo, 0, reloc | NOUVEAU_BO_WR | NOUVEAU_BO_LOW, 0, 0); + so_reloc (so, bo, offset, reloc | NOUVEAU_BO_HIGH, 0, 0); + so_reloc (so, bo, offset, reloc | NOUVEAU_BO_LOW, 0, 0); so_method(so, eng2d, NV50_2D_SIFC_UNK0800, 2); so_data (so, 0); so_data (so, NV50_2D_SIFC_FORMAT_R8_UNORM); diff --git a/src/gallium/drivers/nv50/nv50_surface.c b/src/gallium/drivers/nv50/nv50_surface.c index 6bf6f773b0..79655fc08d 100644 --- a/src/gallium/drivers/nv50/nv50_surface.c +++ b/src/gallium/drivers/nv50/nv50_surface.c @@ -62,6 +62,7 @@ nv50_surface_set(struct nv50_screen *screen, struct pipe_surface *ps, int dst) return 1; if (!bo->tile_flags) { + MARK_RING (chan, 9, 2); /* flush on lack of space or relocs */ BEGIN_RING(chan, eng2d, mthd, 2); OUT_RING (chan, format); OUT_RING (chan, 1); @@ -72,6 +73,7 @@ nv50_surface_set(struct nv50_screen *screen, struct pipe_surface *ps, int dst) OUT_RELOCh(chan, bo, ps->offset, flags); OUT_RELOCl(chan, bo, ps->offset, flags); } else { + MARK_RING (chan, 11, 2); /* flush on lack of space or relocs */ BEGIN_RING(chan, eng2d, mthd, 5); OUT_RING (chan, format); OUT_RING (chan, 0); diff --git a/src/gallium/drivers/nv50/nv50_tex.c b/src/gallium/drivers/nv50/nv50_tex.c index 55a601deb8..c4ca096d6a 100644 --- a/src/gallium/drivers/nv50/nv50_tex.c +++ b/src/gallium/drivers/nv50/nv50_tex.c @@ -70,6 +70,7 @@ static const struct nv50_texture_format nv50_tex_format_list[] = _(DXT5_RGBA, UNORM, C0, C1, C2, C3, DXT5), _MIXED(Z24S8_UNORM, UINT, UNORM, UINT, UINT, C1, C1, C1, ONE, 24_8), + _MIXED(S8Z24_UNORM, UNORM, UINT, UINT, UINT, C0, C0, C0, ONE, 8_24), _(R16G16B16A16_SNORM, UNORM, C0, C1, C2, C3, 16_16_16_16), _(R16G16B16A16_UNORM, SNORM, C0, C1, C2, C3, 16_16_16_16), @@ -87,7 +88,7 @@ static const struct nv50_texture_format nv50_tex_format_list[] = static int nv50_tex_construct(struct nv50_context *nv50, struct nouveau_stateobj *so, - struct nv50_miptree *mt, int unit) + struct nv50_miptree *mt, int unit, unsigned p) { unsigned i; uint32_t mode; @@ -99,7 +100,7 @@ nv50_tex_construct(struct nv50_context *nv50, struct nouveau_stateobj *so, if (i == NV50_TEX_FORMAT_LIST_SIZE) return 1; - if (nv50->sampler[unit]->normalized) + if (nv50->sampler[p][unit]->normalized) mode = 0x50001000 | (1 << 31); else { mode = 0x50001000 | (7 << 14); @@ -146,48 +147,78 @@ nv50_tex_construct(struct nv50_context *nv50, struct nouveau_stateobj *so, return 0; } -void -nv50_tex_validate(struct nv50_context *nv50) +#ifndef NV50TCL_BIND_TIC +#define NV50TCL_BIND_TIC(n) (0x1448 + 8 * n) +#endif + +static boolean +nv50_validate_textures(struct nv50_context *nv50, struct nouveau_stateobj *so, + unsigned p) { + static const unsigned p_remap[PIPE_SHADER_TYPES] = { 0, 2 }; + struct nouveau_grobj *eng2d = nv50->screen->eng2d; struct nouveau_grobj *tesla = nv50->screen->tesla; - struct nouveau_stateobj *so; - unsigned i, unit, push; - - push = MAX2(nv50->miptree_nr, nv50->state.miptree_nr) * 2 + 23 + 6; - so = so_new(nv50->miptree_nr * 9 + push, nv50->miptree_nr * 2 + 2); + unsigned unit, j, p_hw = p_remap[p]; nv50_so_init_sifc(nv50, so, nv50->screen->tic, NOUVEAU_BO_VRAM, - nv50->miptree_nr * 8 * 4); + p * (32 * 8 * 4), nv50->miptree_nr[p] * 8 * 4); - for (i = 0, unit = 0; unit < nv50->miptree_nr; ++unit) { - struct nv50_miptree *mt = nv50->miptree[unit]; - - if (!mt) - continue; + for (unit = 0; unit < nv50->miptree_nr[p]; ++unit) { + struct nv50_miptree *mt = nv50->miptree[p][unit]; so_method(so, eng2d, NV50_2D_SIFC_DATA | (2 << 29), 8); - if (nv50_tex_construct(nv50, so, mt, unit)) { - NOUVEAU_ERR("failed tex validate\n"); - so_ref(NULL, &so); - return; + if (mt) { + if (nv50_tex_construct(nv50, so, mt, unit, p)) + return FALSE; + /* Set TEX insn $t src binding $unit in program type p + * to TIC, TSC entry (32 * p + unit), mark valid (1). + */ + so_method(so, tesla, NV50TCL_BIND_TIC(p_hw), 1); + so_data (so, ((32 * p + unit) << 9) | (unit << 1) | 1); + } else { + for (j = 0; j < 8; ++j) + so_data(so, 0); + so_method(so, tesla, NV50TCL_BIND_TIC(p_hw), 1); + so_data (so, (unit << 1) | 0); } + } + + for (; unit < nv50->state.miptree_nr[p]; unit++) { + /* Make other bindings invalid. */ + so_method(so, tesla, NV50TCL_BIND_TIC(p_hw), 1); + so_data (so, (unit << 1) | 0); + } + + nv50->state.miptree_nr[p] = nv50->miptree_nr[p]; + return TRUE; +} - so_method(so, tesla, NV50TCL_SET_SAMPLER_TEX, 1); - so_data (so, (i++ << NV50TCL_SET_SAMPLER_TEX_TIC_SHIFT) | - (unit << NV50TCL_SET_SAMPLER_TEX_SAMPLER_SHIFT) | - NV50TCL_SET_SAMPLER_TEX_VALID); +void +nv50_tex_validate(struct nv50_context *nv50) +{ + struct nouveau_stateobj *so; + struct nouveau_grobj *tesla = nv50->screen->tesla; + unsigned p, push, nrlc; + + for (nrlc = 0, push = 0, p = 0; p < PIPE_SHADER_TYPES; ++p) { + push += MAX2(nv50->miptree_nr[p], nv50->state.miptree_nr[p]); + nrlc += nv50->miptree_nr[p]; } + push = push * 11 + 23 * PIPE_SHADER_TYPES + 4; + nrlc = nrlc * 2 + 2 * PIPE_SHADER_TYPES; + + so = so_new(push, nrlc); + + if (nv50_validate_textures(nv50, so, PIPE_SHADER_VERTEX) == FALSE || + nv50_validate_textures(nv50, so, PIPE_SHADER_FRAGMENT) == FALSE) { + so_ref(NULL, &so); - for (; unit < nv50->state.miptree_nr; unit++) { - so_method(so, tesla, NV50TCL_SET_SAMPLER_TEX, 1); - so_data (so, - (unit << NV50TCL_SET_SAMPLER_TEX_SAMPLER_SHIFT) | 0); + NOUVEAU_ERR("failed tex validate\n"); + return; } /* not sure if the following really do what I think: */ - so_method(so, tesla, 0x1440, 1); /* sync SIFC */ - so_data (so, 0); so_method(so, tesla, 0x1330, 1); /* flush TIC */ so_data (so, 0); so_method(so, tesla, 0x1338, 1); /* flush texture caches */ @@ -195,6 +226,4 @@ nv50_tex_validate(struct nv50_context *nv50) so_ref(so, &nv50->state.tic_upload); so_ref(NULL, &so); - nv50->state.miptree_nr = nv50->miptree_nr; } - diff --git a/src/gallium/drivers/nv50/nv50_texture.h b/src/gallium/drivers/nv50/nv50_texture.h index d531e61132..b870302019 100644 --- a/src/gallium/drivers/nv50/nv50_texture.h +++ b/src/gallium/drivers/nv50/nv50_texture.h @@ -82,6 +82,7 @@ #define NV50TIC_0_0_FMT_RGTC1 0x00000027 #define NV50TIC_0_0_FMT_RGTC2 0x00000028 #define NV50TIC_0_0_FMT_24_8 0x00000029 +#define NV50TIC_0_0_FMT_8_24 0x0000002a #define NV50TIC_0_0_FMT_32_DEPTH 0x0000002f #define NV50TIC_0_0_FMT_32_8 0x00000030 diff --git a/src/gallium/drivers/nv50/nv50_transfer.c b/src/gallium/drivers/nv50/nv50_transfer.c index 39d65279fc..6240a0c757 100644 --- a/src/gallium/drivers/nv50/nv50_transfer.c +++ b/src/gallium/drivers/nv50/nv50_transfer.c @@ -16,16 +16,19 @@ struct nv50_transfer { int level_depth; int level_x; int level_y; + int level_z; + unsigned nblocksx; + unsigned nblocksy; }; static void nv50_transfer_rect_m2mf(struct pipe_screen *pscreen, struct nouveau_bo *src_bo, unsigned src_offset, int src_pitch, unsigned src_tile_mode, - int sx, int sy, int sw, int sh, int sd, + int sx, int sy, int sz, int sw, int sh, int sd, struct nouveau_bo *dst_bo, unsigned dst_offset, int dst_pitch, unsigned dst_tile_mode, - int dx, int dy, int dw, int dh, int dd, + int dx, int dy, int dz, int dw, int dh, int dd, int cpp, int width, int height, unsigned src_reloc, unsigned dst_reloc) { @@ -54,7 +57,7 @@ nv50_transfer_rect_m2mf(struct pipe_screen *pscreen, OUT_RING (chan, sw * cpp); OUT_RING (chan, sh); OUT_RING (chan, sd); - OUT_RING (chan, 0); + OUT_RING (chan, sz); /* copying only 1 zslice per call */ } if (!dst_bo->tile_flags) { @@ -73,13 +76,13 @@ nv50_transfer_rect_m2mf(struct pipe_screen *pscreen, OUT_RING (chan, dw * cpp); OUT_RING (chan, dh); OUT_RING (chan, dd); - OUT_RING (chan, 0); + OUT_RING (chan, dz); /* copying only 1 zslice per call */ } while (height) { int line_count = height > 2047 ? 2047 : height; - WAIT_RING (chan, 15); + MARK_RING (chan, 15, 4); /* flush on lack of space or relocs */ BEGIN_RING(chan, m2mf, NV50_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN_HIGH, 2); OUT_RELOCh(chan, src_bo, src_offset, src_reloc); @@ -116,20 +119,6 @@ nv50_transfer_rect_m2mf(struct pipe_screen *pscreen, } } -static INLINE unsigned -get_zslice_offset(unsigned tile_mode, unsigned z, unsigned pitch, unsigned ny) -{ - unsigned tile_h = get_tile_height(tile_mode); - unsigned tile_d = get_tile_depth(tile_mode); - - /* pitch_2d == to next slice within this volume-tile */ - /* pitch_3d == to next slice in next 2D array of blocks */ - unsigned pitch_2d = tile_h * 64; - unsigned pitch_3d = tile_d * align(ny, tile_h) * pitch; - - return (z % tile_d) * pitch_2d + (z / tile_d) * pitch_3d; -} - static struct pipe_transfer * nv50_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice, @@ -151,20 +140,11 @@ nv50_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; pipe_texture_reference(&tx->base.texture, pt); - tx->base.format = pt->format; + tx->nblocksx = pf_get_nblocksx(pt->format, u_minify(pt->width0, level)); + tx->nblocksy = pf_get_nblocksy(pt->format, u_minify(pt->height0, level)); tx->base.width = w; tx->base.height = h; - tx->base.block = pt->block; - if (!pt->nblocksx[level]) { - tx->base.nblocksx = pf_get_nblocksx(&pt->block, - u_minify(pt->width0, level)); - tx->base.nblocksy = pf_get_nblocksy(&pt->block, - u_minify(pt->height0, level)); - } else { - tx->base.nblocksx = pt->nblocksx[level]; - tx->base.nblocksy = pt->nblocksy[level]; - } - tx->base.stride = tx->base.nblocksx * pt->block.size; + tx->base.stride = tx->nblocksx * pf_get_blocksize(pt->format); tx->base.usage = usage; tx->level_pitch = lvl->pitch; @@ -173,34 +153,30 @@ nv50_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, tx->level_depth = u_minify(mt->base.base.depth0, level); tx->level_offset = lvl->image_offset[image]; tx->level_tiling = lvl->tile_mode; - tx->level_x = pf_get_nblocksx(&tx->base.block, x); - tx->level_y = pf_get_nblocksy(&tx->base.block, y); + tx->level_z = zslice; + tx->level_x = pf_get_nblocksx(pt->format, x); + tx->level_y = pf_get_nblocksy(pt->format, y); ret = nouveau_bo_new(dev, NOUVEAU_BO_GART | NOUVEAU_BO_MAP, 0, - tx->base.nblocksy * tx->base.stride, &tx->bo); + tx->nblocksy * tx->base.stride, &tx->bo); if (ret) { FREE(tx); return NULL; } - if (pt->target == PIPE_TEXTURE_3D) - tx->level_offset += get_zslice_offset(lvl->tile_mode, zslice, - lvl->pitch, - tx->base.nblocksy); - if (usage & PIPE_TRANSFER_READ) { - nx = pf_get_nblocksx(&tx->base.block, tx->base.width); - ny = pf_get_nblocksy(&tx->base.block, tx->base.height); + nx = pf_get_nblocksx(pt->format, tx->base.width); + ny = pf_get_nblocksy(pt->format, tx->base.height); nv50_transfer_rect_m2mf(pscreen, mt->base.bo, tx->level_offset, tx->level_pitch, tx->level_tiling, - x, y, - tx->base.nblocksx, tx->base.nblocksy, + x, y, zslice, + tx->nblocksx, tx->nblocksy, tx->level_depth, tx->bo, 0, tx->base.stride, tx->bo->tile_mode, - 0, 0, - tx->base.nblocksx, tx->base.nblocksy, 1, - tx->base.block.size, nx, ny, + 0, 0, 0, + tx->nblocksx, tx->nblocksy, 1, + pf_get_blocksize(pt->format), nx, ny, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART, NOUVEAU_BO_GART); } @@ -213,23 +189,24 @@ nv50_transfer_del(struct pipe_transfer *ptx) { struct nv50_transfer *tx = (struct nv50_transfer *)ptx; struct nv50_miptree *mt = nv50_miptree(ptx->texture); + struct pipe_texture *pt = ptx->texture; - unsigned nx = pf_get_nblocksx(&tx->base.block, tx->base.width); - unsigned ny = pf_get_nblocksy(&tx->base.block, tx->base.height); + unsigned nx = pf_get_nblocksx(pt->format, tx->base.width); + unsigned ny = pf_get_nblocksy(pt->format, tx->base.height); if (ptx->usage & PIPE_TRANSFER_WRITE) { - struct pipe_screen *pscreen = ptx->texture->screen; + struct pipe_screen *pscreen = pt->screen; nv50_transfer_rect_m2mf(pscreen, tx->bo, 0, tx->base.stride, tx->bo->tile_mode, - 0, 0, - tx->base.nblocksx, tx->base.nblocksy, 1, + 0, 0, 0, + tx->nblocksx, tx->nblocksy, 1, mt->base.bo, tx->level_offset, tx->level_pitch, tx->level_tiling, - tx->level_x, tx->level_y, - tx->base.nblocksx, tx->base.nblocksy, + tx->level_x, tx->level_y, tx->level_z, + tx->nblocksx, tx->nblocksy, tx->level_depth, - tx->base.block.size, nx, ny, + pf_get_blocksize(pt->format), nx, ny, NOUVEAU_BO_GART, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART); } @@ -288,7 +265,7 @@ nv50_upload_sifc(struct nv50_context *nv50, reloc |= NOUVEAU_BO_WR; - WAIT_RING (chan, 32); + MARK_RING (chan, 32, 2); /* flush on lack of space or relocs */ if (bo->tile_flags) { BEGIN_RING(chan, eng2d, NV50_2D_DST_FORMAT, 5); diff --git a/src/gallium/drivers/r300/Makefile b/src/gallium/drivers/r300/Makefile index 63ae5c2766..8cfd4147c2 100644 --- a/src/gallium/drivers/r300/Makefile +++ b/src/gallium/drivers/r300/Makefile @@ -4,8 +4,8 @@ include $(TOP)/configs/current LIBNAME = r300 C_SOURCES = \ + r300_blit.c \ r300_chipset.c \ - r300_clear.c \ r300_context.c \ r300_debug.c \ r300_emit.c \ @@ -36,4 +36,4 @@ include ../../Makefile.template .PHONY : $(COMPILER_ARCHIVE) $(COMPILER_ARCHIVE): - cd $(TOP)/src/mesa/drivers/dri/r300/compiler; make + $(MAKE) -C $(TOP)/src/mesa/drivers/dri/r300/compiler diff --git a/src/gallium/drivers/r300/SConscript b/src/gallium/drivers/r300/SConscript index 97989040d2..0d2de17be9 100644 --- a/src/gallium/drivers/r300/SConscript +++ b/src/gallium/drivers/r300/SConscript @@ -9,8 +9,8 @@ env.Append(CPPPATH = ['#/src/mesa/drivers/dri/r300/compiler', '#/include', '#/sr r300 = env.ConvenienceLibrary( target = 'r300', source = [ + 'r300_blit.c', 'r300_chipset.c', - 'r300_clear.c', 'r300_context.c', 'r300_debug.c', 'r300_emit.c', diff --git a/src/gallium/drivers/r300/r300_blit.c b/src/gallium/drivers/r300/r300_blit.c new file mode 100644 index 0000000000..ffe066d536 --- /dev/null +++ b/src/gallium/drivers/r300/r300_blit.c @@ -0,0 +1,130 @@ +/* + * Copyright 2009 Marek Olšák <maraeo@gmail.com> + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * on the rights to use, copy, modify, merge, publish, distribute, sub + * license, and/or sell copies of the Software, and to permit persons to whom + * the Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. */ + +#include "r300_blit.h" +#include "r300_context.h" + +#include "util/u_rect.h" + +static void r300_blitter_save_states(struct r300_context* r300) +{ + util_blitter_save_blend(r300->blitter, r300->blend_state); + util_blitter_save_depth_stencil_alpha(r300->blitter, r300->dsa_state); + util_blitter_save_rasterizer(r300->blitter, r300->rs_state); + util_blitter_save_fragment_shader(r300->blitter, r300->fs); + util_blitter_save_vertex_shader(r300->blitter, r300->vs); +} + +/* Clear currently bound buffers. */ +void r300_clear(struct pipe_context* pipe, + unsigned buffers, + const float* rgba, + double depth, + unsigned stencil) +{ + /* XXX Implement fastfill. + * + * If fastfill is enabled, a few facts should be considered: + * + * 1) Zbuffer must be micro-tiled and whole microtiles must be + * written. + * + * 2) ZB_DEPTHCLEARVALUE is used to clear a zbuffer and Z Mask must be + * equal to 0. + * + * 3) RB3D_COLOR_CLEAR_VALUE is used to clear a colorbuffer and + * RB3D_COLOR_CHANNEL_MASK must be equal to 0. + * + * 4) ZB_CB_CLEAR can be used to make the ZB units help in clearing + * the colorbuffer. The color clear value is supplied through both + * RB3D_COLOR_CLEAR_VALUE and ZB_DEPTHCLEARVALUE, and the colorbuffer + * must be set in ZB_DEPTHOFFSET and ZB_DEPTHPITCH in addition to + * RB3D_COLOROFFSET and RB3D_COLORPITCH. It's obvious that the zbuffer + * will not be cleared and multiple render targets cannot be cleared + * this way either. + * + * 5) For 16-bit integer buffering, compression causes a hung with one or + * two samples and should not be used. + * + * 6) Fastfill must not be used if reading of compressed Z data is disabled + * and writing of compressed Z data is enabled (RD/WR_COMP_ENABLE), + * i.e. it cannot be used to compress the zbuffer. + * (what the hell does that mean and how does it fit in clearing + * the buffers?) + * + * - Marek + */ + + struct r300_context* r300 = r300_context(pipe); + + r300_blitter_save_states(r300); + + util_blitter_clear(r300->blitter, + r300->framebuffer_state.width, + r300->framebuffer_state.height, + r300->framebuffer_state.nr_cbufs, + buffers, rgba, depth, stencil); +} + +/* Copy a block of pixels from one surface to another. */ +void r300_surface_copy(struct pipe_context* pipe, + struct pipe_surface* dst, + unsigned dstx, unsigned dsty, + struct pipe_surface* src, + unsigned srcx, unsigned srcy, + unsigned width, unsigned height) +{ + struct r300_context* r300 = r300_context(pipe); + + /* Yeah we have to save all those states to ensure this blitter operation + * is really transparent. The states will be restored by the blitter once + * copying is done. */ + r300_blitter_save_states(r300); + util_blitter_save_framebuffer(r300->blitter, &r300->framebuffer_state); + + util_blitter_save_fragment_sampler_states( + r300->blitter, r300->sampler_count, (void**)r300->sampler_states); + + util_blitter_save_fragment_sampler_textures( + r300->blitter, r300->texture_count, + (struct pipe_texture**)r300->textures); + + /* Do a copy */ + util_blitter_copy(r300->blitter, + dst, dstx, dsty, src, srcx, srcy, width, height, TRUE); +} + +/* Fill a region of a surface with a constant value. */ +void r300_surface_fill(struct pipe_context* pipe, + struct pipe_surface* dst, + unsigned dstx, unsigned dsty, + unsigned width, unsigned height, + unsigned value) +{ + struct r300_context* r300 = r300_context(pipe); + + r300_blitter_save_states(r300); + util_blitter_save_framebuffer(r300->blitter, &r300->framebuffer_state); + + util_blitter_fill(r300->blitter, + dst, dstx, dsty, width, height, value); +} diff --git a/src/gallium/drivers/r300/r300_clear.h b/src/gallium/drivers/r300/r300_blit.h index b8fcdf273c..029e4f98e7 100644 --- a/src/gallium/drivers/r300/r300_clear.h +++ b/src/gallium/drivers/r300/r300_blit.h @@ -1,5 +1,5 @@ /* - * Copyright 2008 Corbin Simpson <MostAwesomeDude@gmail.com> + * Copyright 2008 Marek Olšák <maraeo@gmail.com> * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), @@ -20,10 +20,11 @@ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE * USE OR OTHER DEALINGS IN THE SOFTWARE. */ -#ifndef R300_CLEAR_H -#define R300_CLEAR_H +#ifndef R300_BLIT_H +#define R300_BLIT_H struct pipe_context; +struct pipe_surface; void r300_clear(struct pipe_context* pipe, unsigned buffers, @@ -31,4 +32,17 @@ void r300_clear(struct pipe_context* pipe, double depth, unsigned stencil); -#endif /* R300_CLEAR_H */ +void r300_surface_copy(struct pipe_context* pipe, + struct pipe_surface* dst, + unsigned dstx, unsigned dsty, + struct pipe_surface* src, + unsigned srcx, unsigned srcy, + unsigned width, unsigned height); + +void r300_surface_fill(struct pipe_context* pipe, + struct pipe_surface* dst, + unsigned dstx, unsigned dsty, + unsigned width, unsigned height, + unsigned value); + +#endif /* R300_BLIT_H */ diff --git a/src/gallium/drivers/r300/r300_clear.c b/src/gallium/drivers/r300/r300_clear.c deleted file mode 100644 index 02d6d504fc..0000000000 --- a/src/gallium/drivers/r300/r300_clear.c +++ /dev/null @@ -1,38 +0,0 @@ -/* - * Copyright 2008 Corbin Simpson <MostAwesomeDude@gmail.com> - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * on the rights to use, copy, modify, merge, publish, distribute, sub - * license, and/or sell copies of the Software, and to permit persons to whom - * the Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice (including the next - * paragraph) shall be included in all copies or substantial portions of the - * Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL - * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, - * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR - * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE - * USE OR OTHER DEALINGS IN THE SOFTWARE. */ - -#include "r300_clear.h" -#include "r300_context.h" - -#include "util/u_clear.h" - -/* Clears currently bound buffers. */ -void r300_clear(struct pipe_context* pipe, - unsigned buffers, - const float* rgba, - double depth, - unsigned stencil) -{ - /* XXX we can and should do one clear if both color and zs are set */ - util_clear(pipe, &r300_context(pipe)->framebuffer_state, - buffers, rgba, depth, stencil); -} diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c index 68a17dcb63..d5c2d63d39 100644 --- a/src/gallium/drivers/r300/r300_context.c +++ b/src/gallium/drivers/r300/r300_context.c @@ -28,7 +28,7 @@ #include "util/u_memory.h" #include "util/u_simple_list.h" -#include "r300_clear.h" +#include "r300_blit.h" #include "r300_context.h" #include "r300_flush.h" #include "r300_query.h" @@ -36,8 +36,8 @@ #include "r300_screen.h" #include "r300_state_derived.h" #include "r300_state_invariant.h" - -#include "radeon_winsys.h" +#include "r300_texture.h" +#include "r300_winsys.h" static enum pipe_error r300_clear_hash_table(void* key, void* value, void* data) @@ -52,6 +52,8 @@ static void r300_destroy_context(struct pipe_context* context) struct r300_context* r300 = r300_context(context); struct r300_query* query, * temp; + util_blitter_destroy(r300->blitter); + util_hash_table_foreach(r300->shader_hash_table, r300_clear_hash_table, NULL); util_hash_table_destroy(r300->shader_hash_table); @@ -124,6 +126,8 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen, r300->context.destroy = r300_destroy_context; r300->context.clear = r300_clear; + r300->context.surface_copy = r300_surface_copy; + r300->context.surface_fill = r300_surface_fill; if (r300screen->caps->has_tcl) { r300->context.draw_arrays = r300_draw_arrays; @@ -175,5 +179,8 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen, r300->winsys->set_flush_cb(r300->winsys, r300_flush_cb, r300); r300->dirty_state = R300_NEW_KITCHEN_SINK; r300->dirty_hw++; + + r300->blitter = util_blitter_create(&r300->context); + return &r300->context; } diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h index dd3f6ac143..6bd2766730 100644 --- a/src/gallium/drivers/r300/r300_context.h +++ b/src/gallium/drivers/r300/r300_context.h @@ -25,6 +25,8 @@ #include "draw/draw_vertex.h" +#include "util/u_blitter.h" + #include "pipe/p_context.h" #include "pipe/p_inlines.h" @@ -98,9 +100,17 @@ struct r300_sampler_state { unsigned min_lod, max_lod; }; +struct r300_scissor_regs { + uint32_t top_left; /* R300_SC_SCISSORS_TL: 0x43e0 */ + uint32_t bottom_right; /* R300_SC_SCISSORS_BR: 0x43e4 */ + + /* Whether everything is culled by scissoring. */ + boolean empty_area; +}; + struct r300_scissor_state { - uint32_t scissor_top_left; /* R300_SC_SCISSORS_TL: 0x43e0 */ - uint32_t scissor_bottom_right; /* R300_SC_SCISSORS_BR: 0x43e4 */ + struct r300_scissor_regs framebuffer; + struct r300_scissor_regs scissor; }; struct r300_texture_state { @@ -240,6 +250,8 @@ struct r300_context { struct radeon_winsys* winsys; /* Draw module. Used mostly for SW TCL. */ struct draw_context* draw; + /* Accelerated blit support. */ + struct blitter_context* blitter; /* Vertex buffer for rendering. */ struct pipe_buffer* vbo; diff --git a/src/gallium/drivers/r300/r300_cs.h b/src/gallium/drivers/r300/r300_cs.h index 8b100375fd..d142fee050 100644 --- a/src/gallium/drivers/r300/r300_cs.h +++ b/src/gallium/drivers/r300/r300_cs.h @@ -55,7 +55,7 @@ int cs_count = 0; #define CHECK_CS(size) \ - cs_winsys->check_cs(cs_winsys, (size)) + assert(cs_winsys->check_cs(cs_winsys, (size))) #define BEGIN_CS(size) do { \ CHECK_CS(size); \ @@ -115,6 +115,15 @@ cs_count -= 3; \ } while (0) +#define OUT_CS_RELOC_NO_OFFSET(bo, rd, wd, flags) do { \ + DBG(cs_context_copy, DBG_CS, "r300: writing relocation for buffer %p, " \ + "domains (%d, %d, %d)\n", \ + bo, rd, wd, flags); \ + assert(bo); \ + cs_winsys->write_cs_reloc(cs_winsys, bo, rd, wd, flags); \ + cs_count -= 2; \ +} while (0) + #define END_CS do { \ if (VERY_VERBOSE_CS) { \ DBG(cs_context_copy, DBG_CS, "r300: END_CS in %s (%s:%d)\n", __FUNCTION__, \ diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c index 171859b8e4..52a8388ead 100644 --- a/src/gallium/drivers/r300/r300_emit.c +++ b/src/gallium/drivers/r300/r300_emit.c @@ -1,5 +1,6 @@ /* * Copyright 2008 Corbin Simpson <MostAwesomeDude@gmail.com> + * Copyright 2009 Marek Olšák <maraeo@gmail.com> * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), @@ -41,9 +42,16 @@ void r300_emit_blend_state(struct r300_context* r300, CS_LOCALS(r300); BEGIN_CS(8); OUT_CS_REG_SEQ(R300_RB3D_CBLEND, 3); - OUT_CS(blend->blend_control); - OUT_CS(blend->alpha_blend_control); - OUT_CS(blend->color_channel_mask); + if (r300->framebuffer_state.nr_cbufs) { + OUT_CS(blend->blend_control); + OUT_CS(blend->alpha_blend_control); + OUT_CS(blend->color_channel_mask); + } else { + OUT_CS(0); + OUT_CS(0); + OUT_CS(0); + /* XXX also disable fastfill here once it's supported */ + } OUT_CS_REG(R300_RB3D_ROPCNTL, blend->rop); OUT_CS_REG(R300_RB3D_DITHER_CTL, blend->dither); END_CS; @@ -277,7 +285,7 @@ void r500_emit_fragment_program_code(struct r300_context* r300, BEGIN_CS(13 + ((code->inst_end + 1) * 6)); - OUT_CS_REG(R500_US_CONFIG, 0); + OUT_CS_REG(R500_US_CONFIG, R500_ZERO_TIMES_ANYTHING_EQUALS_ZERO); OUT_CS_REG(R500_US_PIXSIZE, code->max_temp_idx); OUT_CS_REG(R500_US_CODE_RANGE, R500_US_CODE_RANGE_ADDR(0) | R500_US_CODE_RANGE_SIZE(code->inst_end)); @@ -331,7 +339,13 @@ void r300_emit_fb_state(struct r300_context* r300, int i; CS_LOCALS(r300); - BEGIN_CS((10 * fb->nr_cbufs) + (fb->zsbuf ? 10 : 0) + 4); + /* Shouldn't fail unless there is a bug in the state tracker. */ + assert(fb->nr_cbufs <= 4); + + BEGIN_CS((10 * fb->nr_cbufs) + (2 * (4 - fb->nr_cbufs)) + + (fb->zsbuf ? 10 : 0) + 6); + + /* Flush and free renderbuffer caches. */ OUT_CS_REG(R300_RB3D_DSTCACHE_CTLSTAT, R300_RB3D_DSTCACHE_CTLSTAT_DC_FREE_FREE_3D_TAGS | R300_RB3D_DSTCACHE_CTLSTAT_DC_FLUSH_FLUSH_DIRTY_3D); @@ -339,6 +353,10 @@ void r300_emit_fb_state(struct r300_context* r300, R300_ZB_ZCACHE_CTLSTAT_ZC_FLUSH_FLUSH_AND_FREE | R300_ZB_ZCACHE_CTLSTAT_ZC_FREE_FREE); + /* Set the number of colorbuffers. */ + OUT_CS_REG(R300_RB3D_CCTL, R300_RB3D_CCTL_NUM_MULTIWRITES(fb->nr_cbufs)); + + /* Set up colorbuffers. */ for (i = 0; i < fb->nr_cbufs; i++) { surf = fb->cbufs[i]; tex = (struct r300_texture*)surf->texture; @@ -356,6 +374,12 @@ void r300_emit_fb_state(struct r300_context* r300, r300_translate_out_fmt(surf->format)); } + /* Disable unused colorbuffers. */ + for (; i < 4; i++) { + OUT_CS_REG(R300_US_OUT_FMT_0 + (4 * i), R300_US_OUT_FMT_UNUSED); + } + + /* Set up a zbuffer. */ if (fb->zsbuf) { surf = fb->zsbuf; tex = (struct r300_texture*)surf->texture; @@ -383,8 +407,6 @@ static void r300_emit_query_start(struct r300_context *r300) if (!query) return; - /* XXX This will almost certainly not return good results - * for overlapping queries. */ BEGIN_CS(4); if (caps->family == CHIP_FAMILY_RV530) { OUT_CS_REG(RV530_FG_ZBREG_DEST, RV530_FG_ZBREG_DEST_PIPE_SELECT_ALL); @@ -566,18 +588,28 @@ void r300_emit_rs_block_state(struct r300_context* r300, END_CS; } -void r300_emit_scissor_state(struct r300_context* r300, - struct r300_scissor_state* scissor) +static void r300_emit_scissor_regs(struct r300_context* r300, + struct r300_scissor_regs* scissor) { CS_LOCALS(r300); BEGIN_CS(3); OUT_CS_REG_SEQ(R300_SC_SCISSORS_TL, 2); - OUT_CS(scissor->scissor_top_left); - OUT_CS(scissor->scissor_bottom_right); + OUT_CS(scissor->top_left); + OUT_CS(scissor->bottom_right); END_CS; } +void r300_emit_scissor_state(struct r300_context* r300, + struct r300_scissor_state* scissor) +{ + if (r300->rs_state->rs.scissor) { + r300_emit_scissor_regs(r300, &scissor->scissor); + } else { + r300_emit_scissor_regs(r300, &scissor->framebuffer); + } +} + void r300_emit_texture(struct r300_context* r300, struct r300_sampler_state* sampler, struct r300_texture* tex, @@ -616,50 +648,68 @@ void r300_emit_texture(struct r300_context* r300, END_CS; } -/* XXX I can't read this and that's not good */ -void r300_emit_aos(struct r300_context* r300, unsigned offset) +static boolean r300_validate_aos(struct r300_context *r300) { struct pipe_vertex_buffer *vbuf = r300->vertex_buffer; struct pipe_vertex_element *velem = r300->vertex_element; - CS_LOCALS(r300); int i; - unsigned aos_count = r300->vertex_element_count; + /* Check if formats and strides are aligned to the size of DWORD. */ + for (i = 0; i < r300->vertex_element_count; i++) { + if (vbuf[velem[i].vertex_buffer_index].stride % 4 != 0 || + pf_get_blocksize(velem[i].src_format) % 4 != 0) { + return FALSE; + } + } + return TRUE; +} + +void r300_emit_aos(struct r300_context* r300, unsigned offset) +{ + struct pipe_vertex_buffer *vb1, *vb2, *vbuf = r300->vertex_buffer; + struct pipe_vertex_element *velem = r300->vertex_element; + int i; + unsigned size1, size2, aos_count = r300->vertex_element_count; unsigned packet_size = (aos_count * 3 + 1) / 2; + CS_LOCALS(r300); + + /* XXX Move this checking to a more approriate place. */ + if (!r300_validate_aos(r300)) { + /* XXX We should fallback using Draw. */ + assert(0); + } + BEGIN_CS(2 + packet_size + aos_count * 2); OUT_CS_PKT3(R300_PACKET3_3D_LOAD_VBPNTR, packet_size); OUT_CS(aos_count); + for (i = 0; i < aos_count - 1; i += 2) { - int buf_num1 = velem[i].vertex_buffer_index; - int buf_num2 = velem[i+1].vertex_buffer_index; - assert(vbuf[buf_num1].stride % 4 == 0 && util_format_get_size(velem[i].src_format) % 4 == 0); - assert(vbuf[buf_num2].stride % 4 == 0 && util_format_get_size(velem[i+1].src_format) % 4 == 0); - OUT_CS((util_format_get_size(velem[i].src_format) >> 2) | (vbuf[buf_num1].stride << 6) | - (util_format_get_size(velem[i+1].src_format) << 14) | (vbuf[buf_num2].stride << 22)); - OUT_CS(vbuf[buf_num1].buffer_offset + velem[i].src_offset + - offset * vbuf[buf_num1].stride); - OUT_CS(vbuf[buf_num2].buffer_offset + velem[i+1].src_offset + - offset * vbuf[buf_num2].stride); + vb1 = &vbuf[velem[i].vertex_buffer_index]; + vb2 = &vbuf[velem[i+1].vertex_buffer_index]; + size1 = util_format_get_size(velem[i].src_format); + size2 = util_format_get_size(velem[i+1].src_format); + + OUT_CS(R300_VBPNTR_SIZE0(size1) | R300_VBPNTR_STRIDE0(vb1->stride) | + R300_VBPNTR_SIZE1(size2) | R300_VBPNTR_STRIDE1(vb2->stride)); + OUT_CS(vb1->buffer_offset + velem[i].src_offset + offset * vb1->stride); + OUT_CS(vb2->buffer_offset + velem[i+1].src_offset + offset * vb2->stride); } + if (aos_count & 1) { - int buf_num = velem[i].vertex_buffer_index; - assert(vbuf[buf_num].stride % 4 == 0 && util_format_get_size(velem[i].src_format) % 4 == 0); - OUT_CS((util_format_get_size(velem[i].src_format) >> 2) | (vbuf[buf_num].stride << 6)); - OUT_CS(vbuf[buf_num].buffer_offset + velem[i].src_offset + - offset * vbuf[buf_num].stride); + vb1 = &vbuf[velem[i].vertex_buffer_index]; + size1 = util_format_get_size(velem[i].src_format); + + OUT_CS(R300_VBPNTR_SIZE0(size1) | R300_VBPNTR_STRIDE0(vb1->stride)); + OUT_CS(vb1->buffer_offset + velem[i].src_offset + offset * vb1->stride); } - /* XXX bare CS reloc */ for (i = 0; i < aos_count; i++) { - cs_winsys->write_cs_reloc(cs_winsys, - vbuf[velem[i].vertex_buffer_index].buffer, - RADEON_GEM_DOMAIN_GTT, - 0, - 0); - cs_count -= 2; + OUT_CS_RELOC_NO_OFFSET(vbuf[velem[i].vertex_buffer_index].buffer, + RADEON_GEM_DOMAIN_GTT, 0, 0); } END_CS; } + #if 0 void r300_emit_draw_packet(struct r300_context* r300) { @@ -834,10 +884,21 @@ void r300_emit_viewport_state(struct r300_context* r300, void r300_emit_texture_count(struct r300_context* r300) { + uint32_t tx_enable = 0; + int i; CS_LOCALS(r300); + /* Notice that texture_count and sampler_count are just sizes + * of the respective arrays. We still have to check for the individual + * elements. */ + for (i = 0; i < MIN2(r300->sampler_count, r300->texture_count); i++) { + if (r300->textures[i]) { + tx_enable |= 1 << i; + } + } + BEGIN_CS(2); - OUT_CS_REG(R300_TX_ENABLE, (1 << r300->texture_count) - 1); + OUT_CS_REG(R300_TX_ENABLE, tx_enable); END_CS; } @@ -872,10 +933,17 @@ void r300_emit_dirty_state(struct r300_context* r300) return; } + /* Check size of CS. */ + /* Make sure we have at least 8*1024 spare dwords. */ + /* XXX It would be nice to know the number of dwords we really need to + * XXX emit. */ + if (!r300->winsys->check_cs(r300->winsys, 8*1024)) { + r300->context.flush(&r300->context, 0, NULL); + } + /* Clean out BOs. */ r300->winsys->reset_bos(r300->winsys); - /* XXX check size */ validate: /* Color buffers... */ for (i = 0; i < r300->framebuffer_state.nr_cbufs; i++) { diff --git a/src/gallium/drivers/r300/r300_reg.h b/src/gallium/drivers/r300/r300_reg.h index 85b1ea568a..d8d08fbe26 100644 --- a/src/gallium/drivers/r300/r300_reg.h +++ b/src/gallium/drivers/r300/r300_reg.h @@ -2145,6 +2145,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE. /* Unpipelined. */ #define R300_RB3D_CCTL 0x4e00 +# define R300_RB3D_CCTL_NUM_MULTIWRITES(x) (MAX2(((x)-1), 0) << 5) # define R300_RB3D_CCTL_NUM_MULTIWRITES_1_BUFFER (0 << 5) # define R300_RB3D_CCTL_NUM_MULTIWRITES_2_BUFFERS (1 << 5) # define R300_RB3D_CCTL_NUM_MULTIWRITES_3_BUFFERS (2 << 5) @@ -3293,6 +3294,11 @@ enum { */ #define R300_PACKET3_3D_LOAD_VBPNTR 0x00002F00 +# define R300_VBPNTR_SIZE0(x) ((x) >> 2) +# define R300_VBPNTR_STRIDE0(x) (((x) >> 2) << 8) +# define R300_VBPNTR_SIZE1(x) (((x) >> 2) << 16) +# define R300_VBPNTR_STRIDE1(x) (((x) >> 2) << 24) + #define R300_PACKET3_INDX_BUFFER 0x00003300 # define R300_INDX_BUFFER_DST_SHIFT 0 # define R300_INDX_BUFFER_SKIP_SHIFT 16 diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c index 4c5fb405c6..4b210f72db 100644 --- a/src/gallium/drivers/r300/r300_render.c +++ b/src/gallium/drivers/r300/r300_render.c @@ -70,14 +70,21 @@ uint32_t r300_translate_primitive(unsigned prim) } } +static boolean r300_nothing_to_draw(struct r300_context *r300) +{ + return r300->rs_state->rs.scissor && + r300->scissor_state->scissor.empty_area; +} + static void r300_emit_draw_arrays(struct r300_context *r300, unsigned mode, unsigned count) { CS_LOCALS(r300); - BEGIN_CS(4); - OUT_CS_REG(R300_VAP_VF_MAX_VTX_INDX, count); + BEGIN_CS(6); + OUT_CS_REG(R300_VAP_VF_MIN_VTX_INDX, 0); + OUT_CS_REG(R300_VAP_VF_MAX_VTX_INDX, count - 1); OUT_CS_PKT3(R300_PACKET3_3D_DRAW_VBUF_2, 0); OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_LIST | (count << 16) | r300_translate_primitive(mode)); @@ -102,7 +109,8 @@ static void r300_emit_draw_elements(struct r300_context *r300, assert((start * indexSize) % 4 == 0); assert(offset_dwords == 0); - BEGIN_CS(10); + BEGIN_CS(12); + OUT_CS_REG(R300_VAP_VF_MIN_VTX_INDX, minIndex); OUT_CS_REG(R300_VAP_VF_MAX_VTX_INDX, maxIndex); OUT_CS_PKT3(R300_PACKET3_3D_DRAW_INDX_2, 0); if (indexSize == 4) { @@ -173,10 +181,15 @@ boolean r300_draw_range_elements(struct pipe_context* pipe, return FALSE; } + if (count > 65535) { return FALSE; } + if (r300_nothing_to_draw(r300)) { + return TRUE; + } + r300_update_derived_state(r300); if (!r300_setup_vertex_buffers(r300)) { @@ -218,6 +231,10 @@ boolean r300_draw_arrays(struct pipe_context* pipe, unsigned mode, return FALSE; } + if (r300_nothing_to_draw(r300)) { + return TRUE; + } + r300_update_derived_state(r300); if (!r300_setup_vertex_buffers(r300)) { @@ -251,6 +268,10 @@ boolean r300_swtcl_draw_arrays(struct pipe_context* pipe, return FALSE; } + if (r300_nothing_to_draw(r300)) { + return TRUE; + } + for (i = 0; i < r300->vertex_buffer_count; i++) { void* buf = pipe_buffer_map(pipe->screen, r300->vertex_buffer[i].buffer, @@ -292,6 +313,10 @@ boolean r300_swtcl_draw_range_elements(struct pipe_context* pipe, return FALSE; } + if (r300_nothing_to_draw(r300)) { + return TRUE; + } + for (i = 0; i < r300->vertex_buffer_count; i++) { void* buf = pipe_buffer_map(pipe->screen, r300->vertex_buffer[i].buffer, diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c index 2e7b1423e6..feb571a23d 100644 --- a/src/gallium/drivers/r300/r300_screen.c +++ b/src/gallium/drivers/r300/r300_screen.c @@ -220,12 +220,18 @@ static boolean check_tex_format(enum pipe_format format, uint32_t usage, /* Z buffer or texture */ case PIPE_FORMAT_Z16_UNORM: + retval = usage & + (PIPE_TEXTURE_USAGE_DEPTH_STENCIL | + PIPE_TEXTURE_USAGE_SAMPLER); + break; + + /* 24bit Z buffer can only be used as a texture on R500. */ case PIPE_FORMAT_Z24X8_UNORM: /* Z buffer with stencil or texture */ case PIPE_FORMAT_Z24S8_UNORM: retval = usage & (PIPE_TEXTURE_USAGE_DEPTH_STENCIL | - PIPE_TEXTURE_USAGE_SAMPLER); + (is_r500 ? PIPE_TEXTURE_USAGE_SAMPLER : 0)); break; /* Definitely unsupported formats. */ @@ -312,14 +318,10 @@ r300_get_tex_transfer(struct pipe_screen *screen, trans = CALLOC_STRUCT(r300_transfer); if (trans) { pipe_texture_reference(&trans->transfer.texture, texture); - trans->transfer.format = texture->format; trans->transfer.x = x; trans->transfer.y = y; trans->transfer.width = w; trans->transfer.height = h; - trans->transfer.block = texture->block; - trans->transfer.nblocksx = texture->nblocksx[level]; - trans->transfer.nblocksy = texture->nblocksy[level]; trans->transfer.stride = r300_texture_get_stride(tex, level); trans->transfer.usage = usage; @@ -345,6 +347,7 @@ static void* r300_transfer_map(struct pipe_screen* screen, { struct r300_texture* tex = (struct r300_texture*)transfer->texture; char* map; + enum pipe_format format = tex->tex.format; map = pipe_buffer_map(screen, tex->buffer, pipe_transfer_buffer_flags(transfer)); @@ -354,8 +357,8 @@ static void* r300_transfer_map(struct pipe_screen* screen, } return map + r300_transfer(transfer)->offset + - transfer->y / transfer->block.height * transfer->stride + - transfer->x / transfer->block.width * transfer->block.size; + transfer->y / pf_get_blockheight(format) * transfer->stride + + transfer->x / pf_get_blockwidth(format) * pf_get_blocksize(format); } static void r300_transfer_unmap(struct pipe_screen* screen, diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c index 442af70e14..91cf972ede 100644 --- a/src/gallium/drivers/r300/r300_state.c +++ b/src/gallium/drivers/r300/r300_state.c @@ -151,9 +151,10 @@ static void r300_set_blend_color(struct pipe_context* pipe, const struct pipe_blend_color* color) { struct r300_context* r300 = r300_context(pipe); + union util_color uc; - util_pack_color(color->color, PIPE_FORMAT_A8R8G8B8_UNORM, - &r300->blend_color_state->blend_color); + util_pack_color(color->color, PIPE_FORMAT_A8R8G8B8_UNORM, &uc); + r300->blend_color_state->blend_color = uc.ui; /* XXX if FP16 blending is enabled, we should use the FP16 format */ r300->blend_color_state->blend_color_red_alpha = @@ -289,11 +290,37 @@ static void r300_set_edgeflags(struct pipe_context* pipe, /* XXX and even worse, I have no idea WTF the bitfield is */ } +static void r300_set_scissor_regs(const struct pipe_scissor_state* state, + struct r300_scissor_regs *scissor, + boolean is_r500) +{ + if (is_r500) { + scissor->top_left = + (state->minx << R300_SCISSORS_X_SHIFT) | + (state->miny << R300_SCISSORS_Y_SHIFT); + scissor->bottom_right = + ((state->maxx - 1) << R300_SCISSORS_X_SHIFT) | + ((state->maxy - 1) << R300_SCISSORS_Y_SHIFT); + } else { + /* Offset of 1440 in non-R500 chipsets. */ + scissor->top_left = + ((state->minx + 1440) << R300_SCISSORS_X_SHIFT) | + ((state->miny + 1440) << R300_SCISSORS_Y_SHIFT); + scissor->bottom_right = + (((state->maxx - 1) + 1440) << R300_SCISSORS_X_SHIFT) | + (((state->maxy - 1) + 1440) << R300_SCISSORS_Y_SHIFT); + } + + scissor->empty_area = state->minx >= state->maxx || + state->miny >= state->maxy; +} + static void r300_set_framebuffer_state(struct pipe_context* pipe, const struct pipe_framebuffer_state* state) { struct r300_context* r300 = r300_context(pipe); + struct pipe_scissor_state scissor; if (r300->draw) { draw_flush(r300->draw); @@ -301,7 +328,18 @@ static void r300->framebuffer_state = *state; + scissor.minx = scissor.miny = 0; + scissor.maxx = state->width; + scissor.maxy = state->height; + r300_set_scissor_regs(&scissor, &r300->scissor_state->framebuffer, + r300_screen(r300->context.screen)->caps->is_r500); + + /* Don't rely on the order of states being set for the first time. */ + if (!r300->rs_state || !r300->rs_state->rs.scissor) { + r300->dirty_state |= R300_NEW_SCISSOR; + } r300->dirty_state |= R300_NEW_FRAMEBUFFERS; + r300->dirty_state |= R300_NEW_BLEND; } /* Create fragment shader state. */ @@ -382,8 +420,6 @@ static void* r300_create_rs_state(struct pipe_context* pipe, if (state->bypass_vs_clip_and_viewport || !r300_screen(pipe->screen)->caps->has_tcl) { rs->vap_control_status |= R300_VAP_TCL_BYPASS; - } else { - rs->rs.bypass_vs_clip_and_viewport = TRUE; } rs->point_size = pack_float_16_6x(state->point_size) | @@ -513,6 +549,7 @@ static void* struct r300_context* r300 = r300_context(pipe); struct r300_sampler_state* sampler = CALLOC_STRUCT(r300_sampler_state); int lod_bias; + union util_color uc; sampler->filter0 |= (r300_translate_wrap(state->wrap_s) << R300_TX_WRAP_S_SHIFT) | @@ -534,8 +571,8 @@ static void* sampler->filter1 |= r300_anisotropy(state->max_anisotropy); - util_pack_color(state->border_color, PIPE_FORMAT_A8R8G8B8_UNORM, - &sampler->border_color); + util_pack_color(state->border_color, PIPE_FORMAT_A8R8G8B8_UNORM, &uc); + sampler->border_color = uc.ui; /* R500-specific fixups and optimizations */ if (r300_screen(r300->context.screen)->caps->is_r500) { @@ -590,8 +627,6 @@ static void r300_set_sampler_textures(struct pipe_context* pipe, return; } - r300->context.flush(&r300->context, 0, NULL); - for (i = 0; i < count; i++) { if (r300->textures[i] != (struct r300_texture*)texture[i]) { pipe_texture_reference((struct pipe_texture**)&r300->textures[i], @@ -623,24 +658,13 @@ static void r300_set_scissor_state(struct pipe_context* pipe, { struct r300_context* r300 = r300_context(pipe); - if (r300_screen(r300->context.screen)->caps->is_r500) { - r300->scissor_state->scissor_top_left = - (state->minx << R300_SCISSORS_X_SHIFT) | - (state->miny << R300_SCISSORS_Y_SHIFT); - r300->scissor_state->scissor_bottom_right = - ((state->maxx - 1) << R300_SCISSORS_X_SHIFT) | - ((state->maxy - 1) << R300_SCISSORS_Y_SHIFT); - } else { - /* Offset of 1440 in non-R500 chipsets. */ - r300->scissor_state->scissor_top_left = - ((state->minx + 1440) << R300_SCISSORS_X_SHIFT) | - ((state->miny + 1440) << R300_SCISSORS_Y_SHIFT); - r300->scissor_state->scissor_bottom_right = - (((state->maxx - 1) + 1440) << R300_SCISSORS_X_SHIFT) | - (((state->maxy - 1) + 1440) << R300_SCISSORS_Y_SHIFT); - } + r300_set_scissor_regs(state, &r300->scissor_state->scissor, + r300_screen(r300->context.screen)->caps->is_r500); - r300->dirty_state |= R300_NEW_SCISSOR; + /* Don't rely on the order of states being set for the first time. */ + if (!r300->rs_state || r300->rs_state->rs.scissor) { + r300->dirty_state |= R300_NEW_SCISSOR; + } } static void r300_set_viewport_state(struct pipe_context* pipe, diff --git a/src/gallium/drivers/r300/r300_state_derived.c b/src/gallium/drivers/r300/r300_state_derived.c index cd969d633b..29bc701a86 100644 --- a/src/gallium/drivers/r300/r300_state_derived.c +++ b/src/gallium/drivers/r300/r300_state_derived.c @@ -134,6 +134,16 @@ static void r300_vertex_psc(struct r300_context* r300) uint16_t type, swizzle; enum pipe_format format; unsigned i; + int identity[16] = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15}; + int* stream_tab; + + /* If TCL is bypassed, map vertex streams to equivalent VS output + * locations. */ + if (r300->rs_state->enable_vte) { + stream_tab = identity; + } else { + stream_tab = r300->vs->stream_loc_notcl; + } /* Vertex shaders have no semantics on their inputs, * so PSC should just route stuff based on the vertex elements, @@ -147,10 +157,10 @@ static void r300_vertex_psc(struct r300_context* r300) format = r300->vertex_element[i].src_format; type = r300_translate_vertex_data_type(format) | - (i << R300_DST_VEC_LOC_SHIFT); + (stream_tab[i] << R300_DST_VEC_LOC_SHIFT); swizzle = r300_translate_vertex_data_swizzle(format); - if (i % 2) { + if (i & 1) { vformat->vap_prog_stream_cntl[i >> 1] |= type << 16; vformat->vap_prog_stream_cntl_ext[i >> 1] |= swizzle << 16; } else { @@ -159,7 +169,6 @@ static void r300_vertex_psc(struct r300_context* r300) } } - assert(i <= 15); /* Set the last vector in the PSC. */ @@ -178,7 +187,7 @@ static void r300_swtcl_vertex_psc(struct r300_context* r300) uint16_t type, swizzle; enum pipe_format format; unsigned i, attrib_count; - int* vs_output_tab = r300->vs->output_stream_loc_swtcl; + int* vs_output_tab = r300->vs->stream_loc_notcl; /* For each Draw attribute, route it to the fragment shader according * to the vs_output_tab. */ @@ -462,6 +471,29 @@ static void r300_update_derived_shader_state(struct r300_context* r300) r300->dirty_state |= R300_NEW_RS_BLOCK; } +static boolean r300_dsa_writes_depth_stencil(struct r300_dsa_state* dsa) +{ + /* We are interested only in the cases when a new depth or stencil value + * can be written and changed. */ + + /* We might optionally check for [Z func: never] and inspect the stencil + * state in a similar fashion, but it's not terribly important. */ + return (dsa->z_buffer_control & R300_Z_WRITE_ENABLE) || + (dsa->stencil_ref_mask & R300_STENCILWRITEMASK_MASK) || + ((dsa->z_buffer_control & R500_STENCIL_REFMASK_FRONT_BACK) && + (dsa->stencil_ref_bf & R300_STENCILWRITEMASK_MASK)); +} + +static boolean r300_dsa_alpha_test_enabled(struct r300_dsa_state* dsa) +{ + /* We are interested only in the cases when alpha testing can kill + * a fragment. */ + uint32_t af = dsa->alpha_function; + + return (af & R300_FG_ALPHA_FUNC_ENABLE) && + (af & R300_FG_ALPHA_FUNC_ALWAYS) != R300_FG_ALPHA_FUNC_ALWAYS; +} + static void r300_update_ztop(struct r300_context* r300) { r300->ztop_state.z_buffer_top = R300_ZTOP_ENABLE; @@ -478,19 +510,25 @@ static void r300_update_ztop(struct r300_context* r300) * The docs claim that for the first three cases, if no ZS writes happen, * then ZTOP can be used. * + * (3) will never apply since we do not support chroma-keyed operations. + * (4) will need to be re-examined (and this comment updated) if/when + * Hyper-Z becomes supported. + * * Additionally, the following conditions require disabled ZTOP: - * ~) Depth writes in fragment shader - * ~) Outstanding occlusion queries + * 5) Depth writes in fragment shader + * 6) Outstanding occlusion queries * * ~C. */ - if (r300->dsa_state->alpha_function) { - r300->ztop_state.z_buffer_top = R300_ZTOP_DISABLE; - } else if (r300->fs->info.uses_kill) { + + /* ZS writes */ + if (r300_dsa_writes_depth_stencil(r300->dsa_state) && + (r300_dsa_alpha_test_enabled(r300->dsa_state) || /* (1) */ + r300->fs->info.uses_kill)) { /* (2) */ r300->ztop_state.z_buffer_top = R300_ZTOP_DISABLE; - } else if (r300_fragment_shader_writes_depth(r300->fs)) { + } else if (r300_fragment_shader_writes_depth(r300->fs)) { /* (5) */ r300->ztop_state.z_buffer_top = R300_ZTOP_DISABLE; - } else if (r300->query_current) { + } else if (r300->query_current) { /* (6) */ r300->ztop_state.z_buffer_top = R300_ZTOP_DISABLE; } } diff --git a/src/gallium/drivers/r300/r300_state_invariant.c b/src/gallium/drivers/r300/r300_state_invariant.c index 46d1cb39b5..bcd4c030f9 100644 --- a/src/gallium/drivers/r300/r300_state_invariant.c +++ b/src/gallium/drivers/r300/r300_state_invariant.c @@ -43,7 +43,7 @@ void r300_emit_invariant_state(struct r300_context* r300) struct r300_capabilities* caps = r300_screen(r300->context.screen)->caps; CS_LOCALS(r300); - BEGIN_CS(24 + (caps->has_tcl ? 2: 0)); + BEGIN_CS(20 + (caps->has_tcl ? 2: 0)); /*** Graphics Backend (GB) ***/ /* Various GB enables */ @@ -70,9 +70,6 @@ void r300_emit_invariant_state(struct r300_context* r300) OUT_CS_REG(R300_US_W_FMT, 0x0); /*** VAP ***/ - /* Max and min vertex index clamp. */ - OUT_CS_REG(R300_VAP_VF_MIN_VTX_INDX, 0x0); - OUT_CS_REG(R300_VAP_VF_MAX_VTX_INDX, 0xffffff); /* Sign/normalize control */ OUT_CS_REG(R300_VAP_PSC_SGN_NORM_CNTL, R300_SGN_NORM_NO_ZERO); /* TCL-only stuff */ @@ -84,15 +81,11 @@ void r300_emit_invariant_state(struct r300_context* r300) END_CS; /* XXX unsorted stuff from surface_fill */ - BEGIN_CS(56 + (caps->has_tcl ? 5 : 0) + (caps->is_r500 ? 4 : 0)); - /* Flush PVS. */ - OUT_CS_REG(R300_VAP_PVS_STATE_FLUSH_REG, 0x0); + BEGIN_CS(44 + (caps->has_tcl ? 7 : 0) + (caps->is_r500 ? 4 : 0)); - OUT_CS_REG(R300_SE_VTE_CNTL, R300_VPORT_X_SCALE_ENA | - R300_VPORT_X_OFFSET_ENA | R300_VPORT_Y_SCALE_ENA | - R300_VPORT_Y_OFFSET_ENA | R300_VPORT_Z_SCALE_ENA | - R300_VPORT_Z_OFFSET_ENA | R300_VTX_W0_FMT); if (caps->has_tcl) { + /*Flushing PVS is required before the VAP_GB registers can be changed*/ + OUT_CS_REG(R300_VAP_PVS_STATE_FLUSH_REG, 0); OUT_CS_REG_SEQ(R300_VAP_GB_VERT_CLIP_ADJ, 4); OUT_CS_32F(1.0); OUT_CS_32F(1.0); @@ -123,19 +116,15 @@ void r300_emit_invariant_state(struct r300_context* r300) OUT_CS_REG(R300_SU_DEPTH_OFFSET, 0x00000000); OUT_CS_REG(R300_SC_HYPERZ, 0x0000001C); OUT_CS_REG(R300_SC_EDGERULE, 0x2DA49525); - OUT_CS_REG(R300_RB3D_CCTL, 0x00000000); OUT_CS_REG(R300_RB3D_AARESOLVE_CTL, 0x00000000); if (caps->is_r500) { OUT_CS_REG(R500_RB3D_DISCARD_SRC_PIXEL_LTE_THRESHOLD, 0x00000000); OUT_CS_REG(R500_RB3D_DISCARD_SRC_PIXEL_GTE_THRESHOLD, 0xFFFFFFFF); } - OUT_CS_REG(R300_ZB_FORMAT, 0x00000002); - OUT_CS_REG(R300_ZB_ZCACHE_CTLSTAT, 0x00000003); OUT_CS_REG(R300_ZB_BW_CNTL, 0x00000000); OUT_CS_REG(R300_ZB_DEPTHCLEARVALUE, 0x00000000); OUT_CS_REG(R300_ZB_HIZ_OFFSET, 0x00000000); OUT_CS_REG(R300_ZB_HIZ_PITCH, 0x00000000); - OUT_CS_REG(R300_SE_VTE_CNTL, 0x0000043F); /* XXX */ OUT_CS_REG(R300_SC_CLIP_RULE, 0xaaaa); diff --git a/src/gallium/drivers/r300/r300_texture.c b/src/gallium/drivers/r300/r300_texture.c index 5538ec3918..170483b9bb 100644 --- a/src/gallium/drivers/r300/r300_texture.c +++ b/src/gallium/drivers/r300/r300_texture.c @@ -106,7 +106,7 @@ unsigned r300_texture_get_stride(struct r300_texture* tex, unsigned level) return 0; } - return align(pf_get_stride(&tex->tex.block, u_minify(tex->tex.width0, level)), 32); + return align(pf_get_stride(tex->tex.format, u_minify(tex->tex.width0, level)), 32); } static void r300_setup_miptree(struct r300_texture* tex) @@ -116,11 +116,10 @@ static void r300_setup_miptree(struct r300_texture* tex) int i; for (i = 0; i <= base->last_level; i++) { - base->nblocksx[i] = pf_get_nblocksx(&base->block, u_minify(base->width0, i)); - base->nblocksy[i] = pf_get_nblocksy(&base->block, u_minify(base->height0, i)); + unsigned nblocksy = pf_get_nblocksy(base->format, u_minify(base->height0, i)); stride = r300_texture_get_stride(tex, i); - layer_size = stride * base->nblocksy[i]; + layer_size = stride * nblocksy; if (base->target == PIPE_TEXTURE_CUBE) size = layer_size * 6; @@ -130,7 +129,7 @@ static void r300_setup_miptree(struct r300_texture* tex) tex->offset[i] = align(tex->size, 32); tex->size = tex->offset[i] + size; tex->layer_size[i] = layer_size; - tex->pitch[i] = stride / base->block.size; + tex->pitch[i] = stride / pf_get_blocksize(base->format); debug_printf("r300: Texture miptree: Level %d " "(%dx%dx%d px, pitch %d bytes)\n", @@ -246,7 +245,7 @@ static struct pipe_texture* tex->tex.screen = screen; tex->stride_override = *stride; - tex->pitch[0] = *stride / base->block.size; + tex->pitch[0] = *stride / pf_get_blocksize(base->format); r300_setup_flags(tex); r300_setup_texture_state(tex, r300_screen(screen)->caps->is_r500); @@ -284,7 +283,6 @@ r300_video_surface_create(struct pipe_screen *screen, template.width0 = util_next_power_of_two(width); template.height0 = util_next_power_of_two(height); template.depth0 = 1; - util_format_get_block(template.format, &template.block); template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_RENDER_TARGET; diff --git a/src/gallium/drivers/r300/r300_vs.c b/src/gallium/drivers/r300/r300_vs.c index 31248346bc..fa207c939c 100644 --- a/src/gallium/drivers/r300/r300_vs.c +++ b/src/gallium/drivers/r300/r300_vs.c @@ -143,35 +143,33 @@ static void r300_shader_vap_output_fmt( assert(gen_count <= 8); } -/* Set VS output stream locations for SWTCL. */ -static void r300_stream_locations_swtcl( +/* Sets up stream mapping to equivalent VS outputs if TCL is bypassed + * or isn't present. */ +static void r300_stream_locations_notcl( struct r300_shader_semantics* vs_outputs, - int* output_stream_loc) + int* stream_loc) { int i, tabi = 0, gen_count; - /* XXX Check whether the numbers (0, 1, 2+i, etc.) are correct. - * These should go to VAP_PROG_STREAM_CNTL/DST_VEC_LOC. */ - /* Position. */ - output_stream_loc[tabi++] = 0; + stream_loc[tabi++] = 0; /* Point size. */ if (vs_outputs->psize != ATTR_UNUSED) { - output_stream_loc[tabi++] = 1; + stream_loc[tabi++] = 1; } /* Colors. */ for (i = 0; i < ATTR_COLOR_COUNT; i++) { if (vs_outputs->color[i] != ATTR_UNUSED) { - output_stream_loc[tabi++] = 2 + i; + stream_loc[tabi++] = 2 + i; } } /* Back-face colors. */ for (i = 0; i < ATTR_COLOR_COUNT; i++) { if (vs_outputs->bcolor[i] != ATTR_UNUSED) { - output_stream_loc[tabi++] = 4 + i; + stream_loc[tabi++] = 4 + i; } } @@ -180,7 +178,7 @@ static void r300_stream_locations_swtcl( for (i = 0; i < ATTR_GENERIC_COUNT; i++) { if (vs_outputs->bcolor[i] != ATTR_UNUSED) { assert(tabi < 16); - output_stream_loc[tabi++] = 6 + gen_count; + stream_loc[tabi++] = 6 + gen_count; gen_count++; } } @@ -188,7 +186,7 @@ static void r300_stream_locations_swtcl( /* Fog coordinates. */ if (vs_outputs->fog != ATTR_UNUSED) { assert(tabi < 16); - output_stream_loc[tabi++] = 6 + gen_count; + stream_loc[tabi++] = 6 + gen_count; gen_count++; } @@ -196,7 +194,7 @@ static void r300_stream_locations_swtcl( assert(gen_count <= 8); for (; tabi < 16;) { - output_stream_loc[tabi++] = -1; + stream_loc[tabi++] = -1; } } @@ -254,10 +252,7 @@ void r300_translate_vertex_shader(struct r300_context* r300, /* Initialize. */ r300_shader_read_vs_outputs(&vs->info, &vs->outputs); r300_shader_vap_output_fmt(&vs->outputs, vs->hwfmt); - - if (!r300_screen(r300->context.screen)->caps->has_tcl) { - r300_stream_locations_swtcl(&vs->outputs, vs->output_stream_loc_swtcl); - } + r300_stream_locations_notcl(&vs->outputs, vs->stream_loc_notcl); /* Setup the compiler */ rc_init(&compiler.Base); @@ -283,7 +278,7 @@ void r300_translate_vertex_shader(struct r300_context* r300, /* Invoke the compiler */ r3xx_compile_vertex_program(&compiler); if (compiler.Base.Error) { - /* XXX Fail gracefully */ + /* XXX We should fallback using Draw. */ fprintf(stderr, "r300 VP: Compiler error\n"); abort(); } diff --git a/src/gallium/drivers/r300/r300_vs.h b/src/gallium/drivers/r300/r300_vs.h index 283dd5a9e8..67e9db5366 100644 --- a/src/gallium/drivers/r300/r300_vs.h +++ b/src/gallium/drivers/r300/r300_vs.h @@ -38,9 +38,11 @@ struct r300_vertex_shader { struct tgsi_shader_info info; struct r300_shader_semantics outputs; - int output_stream_loc_swtcl[16]; uint hwfmt[4]; + /* Stream locations for SWTCL or if TCL is bypassed. */ + int stream_loc_notcl[16]; + /* Has this shader been translated yet? */ boolean translated; diff --git a/src/gallium/drivers/r300/r300_winsys.h b/src/gallium/drivers/r300/r300_winsys.h index f86985841f..1ae6de70fe 100644 --- a/src/gallium/drivers/r300/r300_winsys.h +++ b/src/gallium/drivers/r300/r300_winsys.h @@ -35,6 +35,8 @@ extern "C" { #include "pipe/p_state.h" #include "pipe/internal/p_winsys_screen.h" +#include "radeon_winsys.h" + struct pipe_context* r300_create_context(struct pipe_screen* screen, struct radeon_winsys* radeon_winsys); diff --git a/src/gallium/drivers/softpipe/sp_clear.c b/src/gallium/drivers/softpipe/sp_clear.c index 8fac8e6e05..f98087deb8 100644 --- a/src/gallium/drivers/softpipe/sp_clear.c +++ b/src/gallium/drivers/softpipe/sp_clear.c @@ -48,6 +48,7 @@ softpipe_clear(struct pipe_context *pipe, unsigned buffers, const float *rgba, double depth, unsigned stencil) { struct softpipe_context *softpipe = softpipe_context(pipe); + union util_color uc; unsigned cv; uint i; @@ -62,12 +63,12 @@ softpipe_clear(struct pipe_context *pipe, unsigned buffers, const float *rgba, for (i = 0; i < softpipe->framebuffer.nr_cbufs; i++) { struct pipe_surface *ps = softpipe->framebuffer.cbufs[i]; - util_pack_color(rgba, ps->format, &cv); - sp_tile_cache_clear(softpipe->cbuf_cache[i], rgba, cv); + util_pack_color(rgba, ps->format, &uc); + sp_tile_cache_clear(softpipe->cbuf_cache[i], rgba, uc.ui); #if !TILE_CLEAR_OPTIMIZATION /* non-cached surface */ - pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, cv); + pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, uc.ui); #endif } } diff --git a/src/gallium/drivers/softpipe/sp_screen.c b/src/gallium/drivers/softpipe/sp_screen.c index 6bf3df8e6a..bd3532de4f 100644 --- a/src/gallium/drivers/softpipe/sp_screen.c +++ b/src/gallium/drivers/softpipe/sp_screen.c @@ -149,6 +149,7 @@ softpipe_is_format_supported( struct pipe_screen *screen, case PIPE_FORMAT_B6UG5SR5S_NORM: case PIPE_FORMAT_X8UB8UG8SR8S_NORM: case PIPE_FORMAT_A8B8G8R8_SNORM: + case PIPE_FORMAT_NONE: return FALSE; default: return TRUE; diff --git a/src/gallium/drivers/softpipe/sp_texture.c b/src/gallium/drivers/softpipe/sp_texture.c index 0f3323ff2f..9e83410fcd 100644 --- a/src/gallium/drivers/softpipe/sp_texture.c +++ b/src/gallium/drivers/softpipe/sp_texture.c @@ -65,13 +65,11 @@ softpipe_texture_layout(struct pipe_screen *screen, pt->depth0 = depth; for (level = 0; level <= pt->last_level; level++) { - pt->nblocksx[level] = pf_get_nblocksx(&pt->block, width); - pt->nblocksy[level] = pf_get_nblocksy(&pt->block, height); - spt->stride[level] = pt->nblocksx[level]*pt->block.size; + spt->stride[level] = pf_get_stride(pt->format, width); spt->level_offset[level] = buffer_size; - buffer_size += (pt->nblocksy[level] * + buffer_size += (pf_get_nblocksy(pt->format, height) * ((pt->target == PIPE_TEXTURE_CUBE) ? 6 : depth) * spt->stride[level]); @@ -99,9 +97,6 @@ softpipe_displaytarget_layout(struct pipe_screen *screen, PIPE_BUFFER_USAGE_GPU_READ_WRITE); unsigned tex_usage = spt->base.tex_usage; - spt->base.nblocksx[0] = pf_get_nblocksx(&spt->base.block, spt->base.width0); - spt->base.nblocksy[0] = pf_get_nblocksy(&spt->base.block, spt->base.height0); - spt->buffer = screen->surface_buffer_create( screen, spt->base.width0, spt->base.height0, @@ -177,8 +172,6 @@ softpipe_texture_blanket(struct pipe_screen * screen, spt->base = *base; pipe_reference_init(&spt->base.reference, 1); spt->base.screen = screen; - spt->base.nblocksx[0] = pf_get_nblocksx(&spt->base.block, spt->base.width0); - spt->base.nblocksy[0] = pf_get_nblocksy(&spt->base.block, spt->base.height0); spt->stride[0] = stride[0]; pipe_buffer_reference(&spt->buffer, buffer); @@ -246,10 +239,12 @@ softpipe_get_tex_surface(struct pipe_screen *screen, ps->zslice = zslice; if (pt->target == PIPE_TEXTURE_CUBE) { - ps->offset += face * pt->nblocksy[level] * spt->stride[level]; + ps->offset += face * pf_get_nblocksy(pt->format, u_minify(pt->height0, level)) * + spt->stride[level]; } else if (pt->target == PIPE_TEXTURE_3D) { - ps->offset += zslice * pt->nblocksy[level] * spt->stride[level]; + ps->offset += zslice * pf_get_nblocksy(pt->format, u_minify(pt->height0, level)) * + spt->stride[level]; } else { assert(face == 0); @@ -304,15 +299,12 @@ softpipe_get_tex_transfer(struct pipe_screen *screen, spt = CALLOC_STRUCT(softpipe_transfer); if (spt) { struct pipe_transfer *pt = &spt->base; + int nblocksy = pf_get_nblocksy(texture->format, u_minify(texture->height0, level)); 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 = sptex->stride[level]; pt->usage = usage; pt->face = face; @@ -322,10 +314,10 @@ softpipe_get_tex_transfer(struct pipe_screen *screen, spt->offset = sptex->level_offset[level]; if (texture->target == PIPE_TEXTURE_CUBE) { - spt->offset += face * pt->nblocksy * pt->stride; + spt->offset += face * nblocksy * pt->stride; } else if (texture->target == PIPE_TEXTURE_3D) { - spt->offset += zslice * pt->nblocksy * pt->stride; + spt->offset += zslice * nblocksy * pt->stride; } else { assert(face == 0); @@ -363,9 +355,11 @@ softpipe_transfer_map( struct pipe_screen *screen, { ubyte *map, *xfer_map; struct softpipe_texture *spt; + enum pipe_format format; assert(transfer->texture); spt = softpipe_texture(transfer->texture); + format = transfer->texture->format; map = pipe_buffer_map(screen, spt->buffer, pipe_transfer_buffer_flags(transfer)); if (map == NULL) @@ -382,8 +376,8 @@ softpipe_transfer_map( struct pipe_screen *screen, } xfer_map = map + softpipe_transfer(transfer)->offset + - transfer->y / transfer->block.height * transfer->stride + - transfer->x / transfer->block.width * transfer->block.size; + transfer->y / pf_get_blockheight(format) * transfer->stride + + transfer->x / pf_get_blockwidth(format) * pf_get_blocksize(format); /*printf("map = %p xfer map = %p\n", map, xfer_map);*/ return xfer_map; } @@ -440,7 +434,6 @@ softpipe_video_surface_create(struct pipe_screen *screen, template.width0 = util_next_power_of_two(width); template.height0 = util_next_power_of_two(height); template.depth0 = 1; - util_format_get_block(template.format, &template.block); template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_RENDER_TARGET; sp_vsfc->tex = screen->texture_create(screen, &template); diff --git a/src/gallium/drivers/softpipe/sp_tile_cache.c b/src/gallium/drivers/softpipe/sp_tile_cache.c index cde2260073..49e1ab0ca7 100644 --- a/src/gallium/drivers/softpipe/sp_tile_cache.c +++ b/src/gallium/drivers/softpipe/sp_tile_cache.c @@ -285,8 +285,9 @@ sp_tile_cache_flush_clear(struct softpipe_tile_cache *tc) uint x, y; uint numCleared = 0; + assert(pt->texture); /* clear the scratch tile to the clear value */ - clear_tile(&tc->tile, pt->format, tc->clear_val); + clear_tile(&tc->tile, pt->texture->format, tc->clear_val); /* push the tile to all positions marked as clear */ for (y = 0; y < h; y += TILE_SIZE) { @@ -373,6 +374,7 @@ sp_find_cached_tile(struct softpipe_tile_cache *tc, if (addr.value != tile->addr.value) { + assert(pt->texture); if (tile->addr.bits.invalid == 0) { /* put dirty tile back in framebuffer */ if (tc->depth_stencil) { @@ -396,10 +398,10 @@ sp_find_cached_tile(struct softpipe_tile_cache *tc, if (is_clear_flag_set(tc->clear_flags, addr)) { /* don't get tile from framebuffer, just clear it */ if (tc->depth_stencil) { - clear_tile(tile, pt->format, tc->clear_val); + clear_tile(tile, pt->texture->format, tc->clear_val); } else { - clear_tile_rgba(tile, pt->format, tc->clear_color); + clear_tile_rgba(tile, pt->texture->format, tc->clear_color); } clear_clear_flag(tc->clear_flags, addr); } diff --git a/src/gallium/drivers/softpipe/sp_winsys.h b/src/gallium/drivers/softpipe/sp_winsys.h index 9e571862b7..f203ded29e 100644 --- a/src/gallium/drivers/softpipe/sp_winsys.h +++ b/src/gallium/drivers/softpipe/sp_winsys.h @@ -34,15 +34,17 @@ #ifndef SP_WINSYS_H #define SP_WINSYS_H - #ifdef __cplusplus extern "C" { #endif +#include "pipe/p_defines.h" struct pipe_screen; struct pipe_winsys; struct pipe_context; +struct pipe_texture; +struct pipe_buffer; struct pipe_context *softpipe_create( struct pipe_screen * ); diff --git a/src/gallium/drivers/svga/include/svga_escape.h b/src/gallium/drivers/svga/include/svga_escape.h new file mode 100644 index 0000000000..7b85e9b8c8 --- /dev/null +++ b/src/gallium/drivers/svga/include/svga_escape.h @@ -0,0 +1,89 @@ +/********************************************************** + * Copyright 2007-2009 VMware, Inc. All rights reserved. + * + * Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, copy, + * modify, merge, publish, distribute, sublicense, and/or sell copies + * of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + * + **********************************************************/ + +/* + * svga_escape.h -- + * + * Definitions for our own (vendor-specific) SVGA Escape commands. + */ + +#ifndef _SVGA_ESCAPE_H_ +#define _SVGA_ESCAPE_H_ + + +/* + * Namespace IDs for the escape command + */ + +#define SVGA_ESCAPE_NSID_VMWARE 0x00000000 +#define SVGA_ESCAPE_NSID_DEVEL 0xFFFFFFFF + + +/* + * Within SVGA_ESCAPE_NSID_VMWARE, we multiplex commands according to + * the first DWORD of escape data (after the nsID and size). As a + * guideline we're using the high word and low word as a major and + * minor command number, respectively. + * + * Major command number allocation: + * + * 0000: Reserved + * 0001: SVGA_ESCAPE_VMWARE_LOG (svga_binary_logger.h) + * 0002: SVGA_ESCAPE_VMWARE_VIDEO (svga_overlay.h) + * 0003: SVGA_ESCAPE_VMWARE_HINT (svga_escape.h) + */ + +#define SVGA_ESCAPE_VMWARE_MAJOR_MASK 0xFFFF0000 + + +/* + * SVGA Hint commands. + * + * These escapes let the SVGA driver provide optional information to + * he host about the state of the guest or guest applications. The + * host can use these hints to make user interface or performance + * decisions. + * + * Notes: + * + * - SVGA_ESCAPE_VMWARE_HINT_FULLSCREEN is deprecated for guests + * that use the SVGA Screen Object extension. Instead of sending + * this escape, use the SVGA_SCREEN_FULLSCREEN_HINT flag on your + * Screen Object. + */ + +#define SVGA_ESCAPE_VMWARE_HINT 0x00030000 +#define SVGA_ESCAPE_VMWARE_HINT_FULLSCREEN 0x00030001 // Deprecated + +typedef +struct { + uint32 command; + uint32 fullscreen; + struct { + int32 x, y; + } monitorPosition; +} SVGAEscapeHintFullscreen; + +#endif /* _SVGA_ESCAPE_H_ */ diff --git a/src/gallium/drivers/svga/include/svga_overlay.h b/src/gallium/drivers/svga/include/svga_overlay.h new file mode 100644 index 0000000000..82c1d3ff3e --- /dev/null +++ b/src/gallium/drivers/svga/include/svga_overlay.h @@ -0,0 +1,201 @@ +/********************************************************** + * Copyright 2007-2009 VMware, Inc. All rights reserved. + * + * Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, copy, + * modify, merge, publish, distribute, sublicense, and/or sell copies + * of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + * + **********************************************************/ + +/* + * svga_overlay.h -- + * + * Definitions for video-overlay support. + */ + +#ifndef _SVGA_OVERLAY_H_ +#define _SVGA_OVERLAY_H_ + +#include "svga_reg.h" + +/* + * Video formats we support + */ + +#define VMWARE_FOURCC_YV12 0x32315659 // 'Y' 'V' '1' '2' +#define VMWARE_FOURCC_YUY2 0x32595559 // 'Y' 'U' 'Y' '2' +#define VMWARE_FOURCC_UYVY 0x59565955 // 'U' 'Y' 'V' 'Y' + +typedef enum { + SVGA_OVERLAY_FORMAT_INVALID = 0, + SVGA_OVERLAY_FORMAT_YV12 = VMWARE_FOURCC_YV12, + SVGA_OVERLAY_FORMAT_YUY2 = VMWARE_FOURCC_YUY2, + SVGA_OVERLAY_FORMAT_UYVY = VMWARE_FOURCC_UYVY, +} SVGAOverlayFormat; + +#define SVGA_VIDEO_COLORKEY_MASK 0x00ffffff + +#define SVGA_ESCAPE_VMWARE_VIDEO 0x00020000 + +#define SVGA_ESCAPE_VMWARE_VIDEO_SET_REGS 0x00020001 + /* FIFO escape layout: + * Type, Stream Id, (Register Id, Value) pairs */ + +#define SVGA_ESCAPE_VMWARE_VIDEO_FLUSH 0x00020002 + /* FIFO escape layout: + * Type, Stream Id */ + +typedef +struct SVGAEscapeVideoSetRegs { + struct { + uint32 cmdType; + uint32 streamId; + } header; + + // May include zero or more items. + struct { + uint32 registerId; + uint32 value; + } items[1]; +} SVGAEscapeVideoSetRegs; + +typedef +struct SVGAEscapeVideoFlush { + uint32 cmdType; + uint32 streamId; +} SVGAEscapeVideoFlush; + + +/* + * Struct definitions for the video overlay commands built on + * SVGAFifoCmdEscape. + */ +typedef +struct { + uint32 command; + uint32 overlay; +} SVGAFifoEscapeCmdVideoBase; + +typedef +struct { + SVGAFifoEscapeCmdVideoBase videoCmd; +} SVGAFifoEscapeCmdVideoFlush; + +typedef +struct { + SVGAFifoEscapeCmdVideoBase videoCmd; + struct { + uint32 regId; + uint32 value; + } items[1]; +} SVGAFifoEscapeCmdVideoSetRegs; + +typedef +struct { + SVGAFifoEscapeCmdVideoBase videoCmd; + struct { + uint32 regId; + uint32 value; + } items[SVGA_VIDEO_NUM_REGS]; +} SVGAFifoEscapeCmdVideoSetAllRegs; + + +/* + *---------------------------------------------------------------------- + * + * VMwareVideoGetAttributes -- + * + * Computes the size, pitches and offsets for YUV frames. + * + * Results: + * TRUE on success; otherwise FALSE on failure. + * + * Side effects: + * Pitches and offsets for the given YUV frame are put in 'pitches' + * and 'offsets' respectively. They are both optional though. + * + *---------------------------------------------------------------------- + */ + +static INLINE Bool +VMwareVideoGetAttributes(const SVGAOverlayFormat format, // IN + uint32 *width, // IN / OUT + uint32 *height, // IN / OUT + uint32 *size, // OUT + uint32 *pitches, // OUT (optional) + uint32 *offsets) // OUT (optional) +{ + int tmp; + + *width = (*width + 1) & ~1; + + if (offsets) { + offsets[0] = 0; + } + + switch (format) { + case VMWARE_FOURCC_YV12: + *height = (*height + 1) & ~1; + *size = (*width + 3) & ~3; + + if (pitches) { + pitches[0] = *size; + } + + *size *= *height; + + if (offsets) { + offsets[1] = *size; + } + + tmp = ((*width >> 1) + 3) & ~3; + + if (pitches) { + pitches[1] = pitches[2] = tmp; + } + + tmp *= (*height >> 1); + *size += tmp; + + if (offsets) { + offsets[2] = *size; + } + + *size += tmp; + break; + + case VMWARE_FOURCC_YUY2: + case VMWARE_FOURCC_UYVY: + *size = *width * 2; + + if (pitches) { + pitches[0] = *size; + } + + *size *= *height; + break; + + default: + return FALSE; + } + + return TRUE; +} + +#endif // _SVGA_OVERLAY_H_ diff --git a/src/gallium/drivers/svga/svga_pipe_clear.c b/src/gallium/drivers/svga/svga_pipe_clear.c index 6195c3897e..409b3b41cb 100644 --- a/src/gallium/drivers/svga/svga_pipe_clear.c +++ b/src/gallium/drivers/svga/svga_pipe_clear.c @@ -46,7 +46,7 @@ try_clear(struct svga_context *svga, boolean restore_viewport = FALSE; SVGA3dClearFlag flags = 0; struct pipe_framebuffer_state *fb = &svga->curr.framebuffer; - unsigned color = 0; + union util_color uc; ret = svga_update_state(svga, SVGA_STATE_HW_CLEAR); if (ret) @@ -54,7 +54,7 @@ try_clear(struct svga_context *svga, if ((buffers & PIPE_CLEAR_COLOR) && fb->cbufs[0]) { flags |= SVGA3D_CLEAR_COLOR; - util_pack_color(rgba, PIPE_FORMAT_A8R8G8B8_UNORM, &color); + util_pack_color(rgba, PIPE_FORMAT_A8R8G8B8_UNORM, &uc); rect.w = fb->cbufs[0]->width; rect.h = fb->cbufs[0]->height; @@ -77,7 +77,7 @@ try_clear(struct svga_context *svga, return ret; } - ret = SVGA3D_ClearRect(svga->swc, flags, color, depth, stencil, + ret = SVGA3D_ClearRect(svga->swc, flags, uc.ui, depth, stencil, rect.x, rect.y, rect.w, rect.h); if (ret != PIPE_OK) return ret; diff --git a/src/gallium/drivers/svga/svga_pipe_sampler.c b/src/gallium/drivers/svga/svga_pipe_sampler.c index b4e57c5d15..78053e755e 100644 --- a/src/gallium/drivers/svga/svga_pipe_sampler.c +++ b/src/gallium/drivers/svga/svga_pipe_sampler.c @@ -101,6 +101,7 @@ svga_create_sampler_state(struct pipe_context *pipe, { struct svga_context *svga = svga_context(pipe); struct svga_sampler_state *cso = CALLOC_STRUCT( svga_sampler_state ); + union util_color uc; cso->mipfilter = translate_mip_filter(sampler->min_mip_filter); cso->magfilter = translate_img_filter( sampler->mag_img_filter ); @@ -121,8 +122,8 @@ svga_create_sampler_state(struct pipe_context *pipe, ubyte a = float_to_ubyte(sampler->border_color[3]); util_pack_color_ub( r, g, b, a, - PIPE_FORMAT_B8G8R8A8_UNORM, - &cso->bordercolor ); + PIPE_FORMAT_B8G8R8A8_UNORM, &uc); + cso->bordercolor = uc.ui; } /* No SVGA3D support for: diff --git a/src/gallium/drivers/svga/svga_screen_buffer.c b/src/gallium/drivers/svga/svga_screen_buffer.c index 1f8a889672..58a1aba464 100644 --- a/src/gallium/drivers/svga/svga_screen_buffer.c +++ b/src/gallium/drivers/svga/svga_screen_buffer.c @@ -356,7 +356,8 @@ svga_buffer_upload_flush(struct svga_context *svga, sbuf->hw.boxes = NULL; /* Decrement reference count */ - pipe_buffer_reference((struct pipe_buffer **)&sbuf, NULL); + pipe_reference(&(sbuf->base.reference), NULL); + sbuf = NULL; } diff --git a/src/gallium/drivers/svga/svga_screen_cache.c b/src/gallium/drivers/svga/svga_screen_cache.c index 8a06383f61..eff36e0bcc 100644 --- a/src/gallium/drivers/svga/svga_screen_cache.c +++ b/src/gallium/drivers/svga/svga_screen_cache.c @@ -277,6 +277,15 @@ svga_screen_surface_create(struct svga_screen *svgascreen, while(size < key->size.width) size <<= 1; key->size.width = size; + /* Since we're reusing buffers we're effectively transforming all + * of them into dynamic buffers. + * + * It would be nice to not cache long lived static buffers. But there + * is no way to detect the long lived from short lived ones yet. A + * good heuristic would be buffer size. + */ + key->flags &= ~SVGA3D_SURFACE_HINT_STATIC; + key->flags |= SVGA3D_SURFACE_HINT_DYNAMIC; } handle = svga_screen_cache_lookup(svgascreen, key); diff --git a/src/gallium/drivers/svga/svga_screen_texture.c b/src/gallium/drivers/svga/svga_screen_texture.c index b899796101..ab93dab223 100644 --- a/src/gallium/drivers/svga/svga_screen_texture.c +++ b/src/gallium/drivers/svga/svga_screen_texture.c @@ -159,7 +159,8 @@ svga_transfer_dma_band(struct svga_transfer *st, st->base.x + st->base.width, y + h, st->base.zslice + 1, - texture->base.block.size*8/(texture->base.block.width*texture->base.block.height)); + pf_get_blocksize(texture->base.format)*8/ + (pf_get_blockwidth(texture->base.format)*pf_get_blockheight(texture->base.format))); box.x = st->base.x; box.y = y; @@ -209,7 +210,8 @@ svga_transfer_dma(struct svga_transfer *st, } else { unsigned y, h, srcy; - h = st->hw_nblocksy * st->base.block.height; + unsigned blockheight = pf_get_blockheight(st->base.texture->format); + h = st->hw_nblocksy * blockheight; srcy = 0; for(y = 0; y < st->base.height; y += h) { unsigned offset, length; @@ -219,11 +221,11 @@ svga_transfer_dma(struct svga_transfer *st, h = st->base.height - y; /* Transfer band must be aligned to pixel block boundaries */ - assert(y % st->base.block.height == 0); - assert(h % st->base.block.height == 0); + assert(y % blockheight == 0); + assert(h % blockheight == 0); - offset = y * st->base.stride / st->base.block.height; - length = h * st->base.stride / st->base.block.height; + offset = y * st->base.stride / blockheight; + length = h * st->base.stride / blockheight; sw = (uint8_t *)st->swbuf + offset; @@ -286,8 +288,6 @@ svga_texture_create(struct pipe_screen *screen, height = templat->height0; depth = templat->depth0; for(level = 0; level <= templat->last_level; ++level) { - tex->base.nblocksx[level] = pf_get_nblocksx(&tex->base.block, width); - tex->base.nblocksy[level] = pf_get_nblocksy(&tex->base.block, height); width = u_minify(width, 1); height = u_minify(height, 1); depth = u_minify(depth, 1); @@ -526,7 +526,7 @@ svga_texture_view_surface(struct pipe_context *pipe, { struct svga_screen *ss = svga_screen(tex->base.screen); struct svga_winsys_surface *handle; - int i, j; + uint32_t i, j; unsigned z_offset = 0; SVGA_DBG(DEBUG_PERF, @@ -658,13 +658,11 @@ svga_get_tex_surface(struct pipe_screen *screen, s->real_level = 0; s->real_zslice = 0; } else { - struct svga_winsys_screen *sws = svga_winsys_screen(screen); - SVGA_DBG(DEBUG_VIEWS, "svga: Surface view: no %p, level %u, face %u, z %u, %p\n", pt, level, face, zslice, s); memset(&s->key, 0, sizeof s->key); - sws->surface_reference(sws, &s->handle, tex->handle); + s->handle = tex->handle; s->real_face = face; s->real_level = level; s->real_zslice = zslice; @@ -678,11 +676,14 @@ static void svga_tex_surface_destroy(struct pipe_surface *surf) { struct svga_surface *s = svga_surface(surf); + struct svga_texture *t = svga_texture(surf->texture); struct svga_screen *ss = svga_screen(surf->texture->screen); - SVGA_DBG(DEBUG_DMA, "unref sid %p (tex surface)\n", s->handle); - assert(s->key.cachable == 0); - svga_screen_surface_destroy(ss, &s->key, &s->handle); + if(s->handle != t->handle) { + SVGA_DBG(DEBUG_DMA, "unref sid %p (tex surface)\n", s->handle); + svga_screen_surface_destroy(ss, &s->key, &s->handle); + } + pipe_texture_reference(&surf->texture, NULL); FREE(surf); } @@ -771,6 +772,8 @@ svga_get_tex_transfer(struct pipe_screen *screen, struct svga_screen *ss = svga_screen(screen); struct svga_winsys_screen *sws = ss->sws; struct svga_transfer *st; + unsigned nblocksx = pf_get_nblocksx(texture->format, w); + unsigned nblocksy = pf_get_nblocksy(texture->format, h); /* We can't map texture storage directly */ if (usage & PIPE_TRANSFER_MAP_DIRECTLY) @@ -780,21 +783,17 @@ svga_get_tex_transfer(struct pipe_screen *screen, if (!st) return NULL; - st->base.format = texture->format; - st->base.block = texture->block; st->base.x = x; st->base.y = y; st->base.width = w; st->base.height = h; - st->base.nblocksx = pf_get_nblocksx(&texture->block, w); - st->base.nblocksy = pf_get_nblocksy(&texture->block, h); - st->base.stride = st->base.nblocksx*st->base.block.size; + st->base.stride = nblocksx*pf_get_blocksize(texture->format); st->base.usage = usage; st->base.face = face; st->base.level = level; st->base.zslice = zslice; - st->hw_nblocksy = st->base.nblocksy; + st->hw_nblocksy = nblocksy; st->hwbuf = svga_winsys_buffer_create(ss, 1, @@ -810,15 +809,15 @@ svga_get_tex_transfer(struct pipe_screen *screen, if(!st->hwbuf) goto no_hwbuf; - if(st->hw_nblocksy < st->base.nblocksy) { + if(st->hw_nblocksy < nblocksy) { /* We couldn't allocate a hardware buffer big enough for the transfer, * so allocate regular malloc memory instead */ debug_printf("%s: failed to allocate %u KB of DMA, splitting into %u x %u KB DMA transfers\n", __FUNCTION__, - (st->base.nblocksy*st->base.stride + 1023)/1024, - (st->base.nblocksy + st->hw_nblocksy - 1)/st->hw_nblocksy, + (nblocksy*st->base.stride + 1023)/1024, + (nblocksy + st->hw_nblocksy - 1)/st->hw_nblocksy, (st->hw_nblocksy*st->base.stride + 1023)/1024); - st->swbuf = MALLOC(st->base.nblocksy*st->base.stride); + st->swbuf = MALLOC(nblocksy*st->base.stride); if(!st->swbuf) goto no_swbuf; } @@ -912,7 +911,6 @@ svga_get_tex_sampler_view(struct pipe_context *pipe, struct pipe_texture *pt, unsigned min_lod, unsigned max_lod) { struct svga_screen *ss = svga_screen(pt->screen); - struct svga_winsys_screen *sws = ss->sws; struct svga_texture *tex = svga_texture(pt); struct svga_sampler_view *sv = NULL; SVGA3dSurfaceFormat format = svga_translate_format(pt->format); @@ -963,7 +961,7 @@ svga_get_tex_sampler_view(struct pipe_context *pipe, struct pipe_texture *pt, sv = CALLOC_STRUCT(svga_sampler_view); pipe_reference_init(&sv->reference, 1); - sv->texture = tex; + pipe_texture_reference(&sv->texture, pt); sv->min_lod = min_lod; sv->max_lod = max_lod; @@ -978,7 +976,7 @@ svga_get_tex_sampler_view(struct pipe_context *pipe, struct pipe_texture *pt, pt->depth0, pt->last_level); sv->key.cachable = 0; - sws->surface_reference(sws, &sv->handle, tex->handle); + sv->handle = tex->handle; return sv; } @@ -1001,7 +999,7 @@ svga_get_tex_sampler_view(struct pipe_context *pipe, struct pipe_texture *pt, if (!sv->handle) { assert(0); sv->key.cachable = 0; - sws->surface_reference(sws, &sv->handle, tex->handle); + sv->handle = tex->handle; return sv; } @@ -1015,14 +1013,14 @@ svga_get_tex_sampler_view(struct pipe_context *pipe, struct pipe_texture *pt, void svga_validate_sampler_view(struct svga_context *svga, struct svga_sampler_view *v) { - struct svga_texture *tex = v->texture; + struct svga_texture *tex = svga_texture(v->texture); unsigned numFaces; unsigned age = 0; int i, k; assert(svga); - if (v->handle == v->texture->handle) + if (v->handle == tex->handle) return; age = tex->age; @@ -1050,11 +1048,14 @@ svga_validate_sampler_view(struct svga_context *svga, struct svga_sampler_view * void svga_destroy_sampler_view_priv(struct svga_sampler_view *v) { - struct svga_screen *ss = svga_screen(v->texture->base.screen); - - SVGA_DBG(DEBUG_DMA, "unref sid %p (sampler view)\n", v->handle); - svga_screen_surface_destroy(ss, &v->key, &v->handle); + struct svga_texture *tex = svga_texture(v->texture); + if(v->handle != tex->handle) { + struct svga_screen *ss = svga_screen(v->texture->screen); + SVGA_DBG(DEBUG_DMA, "unref sid %p (sampler view)\n", v->handle); + svga_screen_surface_destroy(ss, &v->key, &v->handle); + } + pipe_texture_reference(&v->texture, NULL); FREE(v); } @@ -1070,8 +1071,7 @@ svga_screen_buffer_from_texture(struct pipe_texture *texture, svga_translate_format(texture->format), stex->handle); - *stride = pf_get_nblocksx(&texture->block, texture->width0) * - texture->block.size; + *stride = pf_get_stride(texture->format, texture->width0); return *buffer != NULL; } diff --git a/src/gallium/drivers/svga/svga_screen_texture.h b/src/gallium/drivers/svga/svga_screen_texture.h index 1cc4063e65..89ae24219f 100644 --- a/src/gallium/drivers/svga/svga_screen_texture.h +++ b/src/gallium/drivers/svga/svga_screen_texture.h @@ -61,7 +61,7 @@ struct svga_sampler_view { struct pipe_reference reference; - struct svga_texture *texture; + struct pipe_texture *texture; int min_lod; int max_lod; @@ -94,6 +94,13 @@ struct svga_texture * operation. */ struct svga_host_surface_cache_key key; + + /** + * Handle for the host side surface. + * + * This handle is owned by this texture. Views should hold on to a reference + * to this texture and never destroy this handle directly. + */ struct svga_winsys_surface *handle; }; @@ -164,8 +171,9 @@ svga_sampler_view_reference(struct svga_sampler_view **ptr, struct svga_sampler_ { struct svga_sampler_view *old = *ptr; - if (pipe_reference((struct pipe_reference **)ptr, &v->reference)) + if (pipe_reference(&(*ptr)->reference, &v->reference)) svga_destroy_sampler_view_priv(old); + *ptr = v; } extern void diff --git a/src/gallium/drivers/svga/svga_state_constants.c b/src/gallium/drivers/svga/svga_state_constants.c index 209ed28245..6b0e511cec 100644 --- a/src/gallium/drivers/svga/svga_state_constants.c +++ b/src/gallium/drivers/svga/svga_state_constants.c @@ -231,7 +231,8 @@ static int emit_vs_consts( struct svga_context *svga, struct svga_tracked_state svga_hw_vs_parameters = { "hw vs params", - (SVGA_NEW_VS_CONST_BUFFER | + (SVGA_NEW_PRESCALE | + SVGA_NEW_VS_CONST_BUFFER | SVGA_NEW_ZERO_STRIDE | SVGA_NEW_VS_RESULT), emit_vs_consts diff --git a/src/gallium/drivers/svga/svgadump/svga_dump.c b/src/gallium/drivers/svga/svgadump/svga_dump.c index 910afa2528..e6d4a74e86 100644 --- a/src/gallium/drivers/svga/svgadump/svga_dump.c +++ b/src/gallium/drivers/svga/svgadump/svga_dump.c @@ -42,554 +42,554 @@ dump_SVGA3dVertexDecl(const SVGA3dVertexDecl *cmd) { switch((*cmd).identity.type) { case SVGA3D_DECLTYPE_FLOAT1: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT1\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT1\n"); break; case SVGA3D_DECLTYPE_FLOAT2: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT2\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT2\n"); break; case SVGA3D_DECLTYPE_FLOAT3: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT3\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT3\n"); break; case SVGA3D_DECLTYPE_FLOAT4: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT4\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT4\n"); break; case SVGA3D_DECLTYPE_D3DCOLOR: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_D3DCOLOR\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_D3DCOLOR\n"); break; case SVGA3D_DECLTYPE_UBYTE4: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_UBYTE4\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_UBYTE4\n"); break; case SVGA3D_DECLTYPE_SHORT2: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_SHORT2\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_SHORT2\n"); break; case SVGA3D_DECLTYPE_SHORT4: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_SHORT4\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_SHORT4\n"); break; case SVGA3D_DECLTYPE_UBYTE4N: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_UBYTE4N\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_UBYTE4N\n"); break; case SVGA3D_DECLTYPE_SHORT2N: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_SHORT2N\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_SHORT2N\n"); break; case SVGA3D_DECLTYPE_SHORT4N: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_SHORT4N\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_SHORT4N\n"); break; case SVGA3D_DECLTYPE_USHORT2N: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_USHORT2N\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_USHORT2N\n"); break; case SVGA3D_DECLTYPE_USHORT4N: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_USHORT4N\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_USHORT4N\n"); break; case SVGA3D_DECLTYPE_UDEC3: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_UDEC3\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_UDEC3\n"); break; case SVGA3D_DECLTYPE_DEC3N: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_DEC3N\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_DEC3N\n"); break; case SVGA3D_DECLTYPE_FLOAT16_2: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT16_2\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT16_2\n"); break; case SVGA3D_DECLTYPE_FLOAT16_4: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT16_4\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT16_4\n"); break; case SVGA3D_DECLTYPE_MAX: - debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_MAX\n"); + _debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_MAX\n"); break; default: - debug_printf("\t\t.identity.type = %i\n", (*cmd).identity.type); + _debug_printf("\t\t.identity.type = %i\n", (*cmd).identity.type); break; } switch((*cmd).identity.method) { case SVGA3D_DECLMETHOD_DEFAULT: - debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_DEFAULT\n"); + _debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_DEFAULT\n"); break; case SVGA3D_DECLMETHOD_PARTIALU: - debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_PARTIALU\n"); + _debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_PARTIALU\n"); break; case SVGA3D_DECLMETHOD_PARTIALV: - debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_PARTIALV\n"); + _debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_PARTIALV\n"); break; case SVGA3D_DECLMETHOD_CROSSUV: - debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_CROSSUV\n"); + _debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_CROSSUV\n"); break; case SVGA3D_DECLMETHOD_UV: - debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_UV\n"); + _debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_UV\n"); break; case SVGA3D_DECLMETHOD_LOOKUP: - debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_LOOKUP\n"); + _debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_LOOKUP\n"); break; case SVGA3D_DECLMETHOD_LOOKUPPRESAMPLED: - debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_LOOKUPPRESAMPLED\n"); + _debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_LOOKUPPRESAMPLED\n"); break; default: - debug_printf("\t\t.identity.method = %i\n", (*cmd).identity.method); + _debug_printf("\t\t.identity.method = %i\n", (*cmd).identity.method); break; } switch((*cmd).identity.usage) { case SVGA3D_DECLUSAGE_POSITION: - debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_POSITION\n"); + _debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_POSITION\n"); break; case SVGA3D_DECLUSAGE_BLENDWEIGHT: - debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_BLENDWEIGHT\n"); + _debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_BLENDWEIGHT\n"); break; case SVGA3D_DECLUSAGE_BLENDINDICES: - debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_BLENDINDICES\n"); + _debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_BLENDINDICES\n"); break; case SVGA3D_DECLUSAGE_NORMAL: - debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_NORMAL\n"); + _debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_NORMAL\n"); break; case SVGA3D_DECLUSAGE_PSIZE: - debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_PSIZE\n"); + _debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_PSIZE\n"); break; case SVGA3D_DECLUSAGE_TEXCOORD: - debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_TEXCOORD\n"); + _debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_TEXCOORD\n"); break; case SVGA3D_DECLUSAGE_TANGENT: - debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_TANGENT\n"); + _debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_TANGENT\n"); break; case SVGA3D_DECLUSAGE_BINORMAL: - debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_BINORMAL\n"); + _debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_BINORMAL\n"); break; case SVGA3D_DECLUSAGE_TESSFACTOR: - debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_TESSFACTOR\n"); + _debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_TESSFACTOR\n"); break; case SVGA3D_DECLUSAGE_POSITIONT: - debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_POSITIONT\n"); + _debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_POSITIONT\n"); break; case SVGA3D_DECLUSAGE_COLOR: - debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_COLOR\n"); + _debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_COLOR\n"); break; case SVGA3D_DECLUSAGE_FOG: - debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_FOG\n"); + _debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_FOG\n"); break; case SVGA3D_DECLUSAGE_DEPTH: - debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_DEPTH\n"); + _debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_DEPTH\n"); break; case SVGA3D_DECLUSAGE_SAMPLE: - debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_SAMPLE\n"); + _debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_SAMPLE\n"); break; case SVGA3D_DECLUSAGE_MAX: - debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_MAX\n"); + _debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_MAX\n"); break; default: - debug_printf("\t\t.identity.usage = %i\n", (*cmd).identity.usage); + _debug_printf("\t\t.identity.usage = %i\n", (*cmd).identity.usage); break; } - debug_printf("\t\t.identity.usageIndex = %u\n", (*cmd).identity.usageIndex); - debug_printf("\t\t.array.surfaceId = %u\n", (*cmd).array.surfaceId); - debug_printf("\t\t.array.offset = %u\n", (*cmd).array.offset); - debug_printf("\t\t.array.stride = %u\n", (*cmd).array.stride); - debug_printf("\t\t.rangeHint.first = %u\n", (*cmd).rangeHint.first); - debug_printf("\t\t.rangeHint.last = %u\n", (*cmd).rangeHint.last); + _debug_printf("\t\t.identity.usageIndex = %u\n", (*cmd).identity.usageIndex); + _debug_printf("\t\t.array.surfaceId = %u\n", (*cmd).array.surfaceId); + _debug_printf("\t\t.array.offset = %u\n", (*cmd).array.offset); + _debug_printf("\t\t.array.stride = %u\n", (*cmd).array.stride); + _debug_printf("\t\t.rangeHint.first = %u\n", (*cmd).rangeHint.first); + _debug_printf("\t\t.rangeHint.last = %u\n", (*cmd).rangeHint.last); } static void dump_SVGA3dTextureState(const SVGA3dTextureState *cmd) { - debug_printf("\t\t.stage = %u\n", (*cmd).stage); + _debug_printf("\t\t.stage = %u\n", (*cmd).stage); switch((*cmd).name) { case SVGA3D_TS_INVALID: - debug_printf("\t\t.name = SVGA3D_TS_INVALID\n"); + _debug_printf("\t\t.name = SVGA3D_TS_INVALID\n"); break; case SVGA3D_TS_BIND_TEXTURE: - debug_printf("\t\t.name = SVGA3D_TS_BIND_TEXTURE\n"); + _debug_printf("\t\t.name = SVGA3D_TS_BIND_TEXTURE\n"); break; case SVGA3D_TS_COLOROP: - debug_printf("\t\t.name = SVGA3D_TS_COLOROP\n"); + _debug_printf("\t\t.name = SVGA3D_TS_COLOROP\n"); break; case SVGA3D_TS_COLORARG1: - debug_printf("\t\t.name = SVGA3D_TS_COLORARG1\n"); + _debug_printf("\t\t.name = SVGA3D_TS_COLORARG1\n"); break; case SVGA3D_TS_COLORARG2: - debug_printf("\t\t.name = SVGA3D_TS_COLORARG2\n"); + _debug_printf("\t\t.name = SVGA3D_TS_COLORARG2\n"); break; case SVGA3D_TS_ALPHAOP: - debug_printf("\t\t.name = SVGA3D_TS_ALPHAOP\n"); + _debug_printf("\t\t.name = SVGA3D_TS_ALPHAOP\n"); break; case SVGA3D_TS_ALPHAARG1: - debug_printf("\t\t.name = SVGA3D_TS_ALPHAARG1\n"); + _debug_printf("\t\t.name = SVGA3D_TS_ALPHAARG1\n"); break; case SVGA3D_TS_ALPHAARG2: - debug_printf("\t\t.name = SVGA3D_TS_ALPHAARG2\n"); + _debug_printf("\t\t.name = SVGA3D_TS_ALPHAARG2\n"); break; case SVGA3D_TS_ADDRESSU: - debug_printf("\t\t.name = SVGA3D_TS_ADDRESSU\n"); + _debug_printf("\t\t.name = SVGA3D_TS_ADDRESSU\n"); break; case SVGA3D_TS_ADDRESSV: - debug_printf("\t\t.name = SVGA3D_TS_ADDRESSV\n"); + _debug_printf("\t\t.name = SVGA3D_TS_ADDRESSV\n"); break; case SVGA3D_TS_MIPFILTER: - debug_printf("\t\t.name = SVGA3D_TS_MIPFILTER\n"); + _debug_printf("\t\t.name = SVGA3D_TS_MIPFILTER\n"); break; case SVGA3D_TS_MAGFILTER: - debug_printf("\t\t.name = SVGA3D_TS_MAGFILTER\n"); + _debug_printf("\t\t.name = SVGA3D_TS_MAGFILTER\n"); break; case SVGA3D_TS_MINFILTER: - debug_printf("\t\t.name = SVGA3D_TS_MINFILTER\n"); + _debug_printf("\t\t.name = SVGA3D_TS_MINFILTER\n"); break; case SVGA3D_TS_BORDERCOLOR: - debug_printf("\t\t.name = SVGA3D_TS_BORDERCOLOR\n"); + _debug_printf("\t\t.name = SVGA3D_TS_BORDERCOLOR\n"); break; case SVGA3D_TS_TEXCOORDINDEX: - debug_printf("\t\t.name = SVGA3D_TS_TEXCOORDINDEX\n"); + _debug_printf("\t\t.name = SVGA3D_TS_TEXCOORDINDEX\n"); break; case SVGA3D_TS_TEXTURETRANSFORMFLAGS: - debug_printf("\t\t.name = SVGA3D_TS_TEXTURETRANSFORMFLAGS\n"); + _debug_printf("\t\t.name = SVGA3D_TS_TEXTURETRANSFORMFLAGS\n"); break; case SVGA3D_TS_TEXCOORDGEN: - debug_printf("\t\t.name = SVGA3D_TS_TEXCOORDGEN\n"); + _debug_printf("\t\t.name = SVGA3D_TS_TEXCOORDGEN\n"); break; case SVGA3D_TS_BUMPENVMAT00: - debug_printf("\t\t.name = SVGA3D_TS_BUMPENVMAT00\n"); + _debug_printf("\t\t.name = SVGA3D_TS_BUMPENVMAT00\n"); break; case SVGA3D_TS_BUMPENVMAT01: - debug_printf("\t\t.name = SVGA3D_TS_BUMPENVMAT01\n"); + _debug_printf("\t\t.name = SVGA3D_TS_BUMPENVMAT01\n"); break; case SVGA3D_TS_BUMPENVMAT10: - debug_printf("\t\t.name = SVGA3D_TS_BUMPENVMAT10\n"); + _debug_printf("\t\t.name = SVGA3D_TS_BUMPENVMAT10\n"); break; case SVGA3D_TS_BUMPENVMAT11: - debug_printf("\t\t.name = SVGA3D_TS_BUMPENVMAT11\n"); + _debug_printf("\t\t.name = SVGA3D_TS_BUMPENVMAT11\n"); break; case SVGA3D_TS_TEXTURE_MIPMAP_LEVEL: - debug_printf("\t\t.name = SVGA3D_TS_TEXTURE_MIPMAP_LEVEL\n"); + _debug_printf("\t\t.name = SVGA3D_TS_TEXTURE_MIPMAP_LEVEL\n"); break; case SVGA3D_TS_TEXTURE_LOD_BIAS: - debug_printf("\t\t.name = SVGA3D_TS_TEXTURE_LOD_BIAS\n"); + _debug_printf("\t\t.name = SVGA3D_TS_TEXTURE_LOD_BIAS\n"); break; case SVGA3D_TS_TEXTURE_ANISOTROPIC_LEVEL: - debug_printf("\t\t.name = SVGA3D_TS_TEXTURE_ANISOTROPIC_LEVEL\n"); + _debug_printf("\t\t.name = SVGA3D_TS_TEXTURE_ANISOTROPIC_LEVEL\n"); break; case SVGA3D_TS_ADDRESSW: - debug_printf("\t\t.name = SVGA3D_TS_ADDRESSW\n"); + _debug_printf("\t\t.name = SVGA3D_TS_ADDRESSW\n"); break; case SVGA3D_TS_GAMMA: - debug_printf("\t\t.name = SVGA3D_TS_GAMMA\n"); + _debug_printf("\t\t.name = SVGA3D_TS_GAMMA\n"); break; case SVGA3D_TS_BUMPENVLSCALE: - debug_printf("\t\t.name = SVGA3D_TS_BUMPENVLSCALE\n"); + _debug_printf("\t\t.name = SVGA3D_TS_BUMPENVLSCALE\n"); break; case SVGA3D_TS_BUMPENVLOFFSET: - debug_printf("\t\t.name = SVGA3D_TS_BUMPENVLOFFSET\n"); + _debug_printf("\t\t.name = SVGA3D_TS_BUMPENVLOFFSET\n"); break; case SVGA3D_TS_COLORARG0: - debug_printf("\t\t.name = SVGA3D_TS_COLORARG0\n"); + _debug_printf("\t\t.name = SVGA3D_TS_COLORARG0\n"); break; case SVGA3D_TS_ALPHAARG0: - debug_printf("\t\t.name = SVGA3D_TS_ALPHAARG0\n"); + _debug_printf("\t\t.name = SVGA3D_TS_ALPHAARG0\n"); break; case SVGA3D_TS_MAX: - debug_printf("\t\t.name = SVGA3D_TS_MAX\n"); + _debug_printf("\t\t.name = SVGA3D_TS_MAX\n"); break; default: - debug_printf("\t\t.name = %i\n", (*cmd).name); + _debug_printf("\t\t.name = %i\n", (*cmd).name); break; } - debug_printf("\t\t.value = %u\n", (*cmd).value); - debug_printf("\t\t.floatValue = %f\n", (*cmd).floatValue); + _debug_printf("\t\t.value = %u\n", (*cmd).value); + _debug_printf("\t\t.floatValue = %f\n", (*cmd).floatValue); } static void dump_SVGA3dCopyBox(const SVGA3dCopyBox *cmd) { - debug_printf("\t\t.x = %u\n", (*cmd).x); - debug_printf("\t\t.y = %u\n", (*cmd).y); - debug_printf("\t\t.z = %u\n", (*cmd).z); - debug_printf("\t\t.w = %u\n", (*cmd).w); - debug_printf("\t\t.h = %u\n", (*cmd).h); - debug_printf("\t\t.d = %u\n", (*cmd).d); - debug_printf("\t\t.srcx = %u\n", (*cmd).srcx); - debug_printf("\t\t.srcy = %u\n", (*cmd).srcy); - debug_printf("\t\t.srcz = %u\n", (*cmd).srcz); + _debug_printf("\t\t.x = %u\n", (*cmd).x); + _debug_printf("\t\t.y = %u\n", (*cmd).y); + _debug_printf("\t\t.z = %u\n", (*cmd).z); + _debug_printf("\t\t.w = %u\n", (*cmd).w); + _debug_printf("\t\t.h = %u\n", (*cmd).h); + _debug_printf("\t\t.d = %u\n", (*cmd).d); + _debug_printf("\t\t.srcx = %u\n", (*cmd).srcx); + _debug_printf("\t\t.srcy = %u\n", (*cmd).srcy); + _debug_printf("\t\t.srcz = %u\n", (*cmd).srcz); } static void dump_SVGA3dCmdSetClipPlane(const SVGA3dCmdSetClipPlane *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); - debug_printf("\t\t.index = %u\n", (*cmd).index); - debug_printf("\t\t.plane[0] = %f\n", (*cmd).plane[0]); - debug_printf("\t\t.plane[1] = %f\n", (*cmd).plane[1]); - debug_printf("\t\t.plane[2] = %f\n", (*cmd).plane[2]); - debug_printf("\t\t.plane[3] = %f\n", (*cmd).plane[3]); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.index = %u\n", (*cmd).index); + _debug_printf("\t\t.plane[0] = %f\n", (*cmd).plane[0]); + _debug_printf("\t\t.plane[1] = %f\n", (*cmd).plane[1]); + _debug_printf("\t\t.plane[2] = %f\n", (*cmd).plane[2]); + _debug_printf("\t\t.plane[3] = %f\n", (*cmd).plane[3]); } static void dump_SVGA3dCmdWaitForQuery(const SVGA3dCmdWaitForQuery *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); switch((*cmd).type) { case SVGA3D_QUERYTYPE_OCCLUSION: - debug_printf("\t\t.type = SVGA3D_QUERYTYPE_OCCLUSION\n"); + _debug_printf("\t\t.type = SVGA3D_QUERYTYPE_OCCLUSION\n"); break; case SVGA3D_QUERYTYPE_MAX: - debug_printf("\t\t.type = SVGA3D_QUERYTYPE_MAX\n"); + _debug_printf("\t\t.type = SVGA3D_QUERYTYPE_MAX\n"); break; default: - debug_printf("\t\t.type = %i\n", (*cmd).type); + _debug_printf("\t\t.type = %i\n", (*cmd).type); break; } - debug_printf("\t\t.guestResult.gmrId = %u\n", (*cmd).guestResult.gmrId); - debug_printf("\t\t.guestResult.offset = %u\n", (*cmd).guestResult.offset); + _debug_printf("\t\t.guestResult.gmrId = %u\n", (*cmd).guestResult.gmrId); + _debug_printf("\t\t.guestResult.offset = %u\n", (*cmd).guestResult.offset); } static void dump_SVGA3dCmdSetRenderTarget(const SVGA3dCmdSetRenderTarget *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); switch((*cmd).type) { case SVGA3D_RT_DEPTH: - debug_printf("\t\t.type = SVGA3D_RT_DEPTH\n"); + _debug_printf("\t\t.type = SVGA3D_RT_DEPTH\n"); break; case SVGA3D_RT_STENCIL: - debug_printf("\t\t.type = SVGA3D_RT_STENCIL\n"); + _debug_printf("\t\t.type = SVGA3D_RT_STENCIL\n"); break; default: - debug_printf("\t\t.type = SVGA3D_RT_COLOR%u\n", (*cmd).type - SVGA3D_RT_COLOR0); + _debug_printf("\t\t.type = SVGA3D_RT_COLOR%u\n", (*cmd).type - SVGA3D_RT_COLOR0); break; } - debug_printf("\t\t.target.sid = %u\n", (*cmd).target.sid); - debug_printf("\t\t.target.face = %u\n", (*cmd).target.face); - debug_printf("\t\t.target.mipmap = %u\n", (*cmd).target.mipmap); + _debug_printf("\t\t.target.sid = %u\n", (*cmd).target.sid); + _debug_printf("\t\t.target.face = %u\n", (*cmd).target.face); + _debug_printf("\t\t.target.mipmap = %u\n", (*cmd).target.mipmap); } static void dump_SVGA3dCmdSetTextureState(const SVGA3dCmdSetTextureState *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); } static void dump_SVGA3dCmdSurfaceCopy(const SVGA3dCmdSurfaceCopy *cmd) { - debug_printf("\t\t.src.sid = %u\n", (*cmd).src.sid); - debug_printf("\t\t.src.face = %u\n", (*cmd).src.face); - debug_printf("\t\t.src.mipmap = %u\n", (*cmd).src.mipmap); - debug_printf("\t\t.dest.sid = %u\n", (*cmd).dest.sid); - debug_printf("\t\t.dest.face = %u\n", (*cmd).dest.face); - debug_printf("\t\t.dest.mipmap = %u\n", (*cmd).dest.mipmap); + _debug_printf("\t\t.src.sid = %u\n", (*cmd).src.sid); + _debug_printf("\t\t.src.face = %u\n", (*cmd).src.face); + _debug_printf("\t\t.src.mipmap = %u\n", (*cmd).src.mipmap); + _debug_printf("\t\t.dest.sid = %u\n", (*cmd).dest.sid); + _debug_printf("\t\t.dest.face = %u\n", (*cmd).dest.face); + _debug_printf("\t\t.dest.mipmap = %u\n", (*cmd).dest.mipmap); } static void dump_SVGA3dCmdSetMaterial(const SVGA3dCmdSetMaterial *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); switch((*cmd).face) { case SVGA3D_FACE_INVALID: - debug_printf("\t\t.face = SVGA3D_FACE_INVALID\n"); + _debug_printf("\t\t.face = SVGA3D_FACE_INVALID\n"); break; case SVGA3D_FACE_NONE: - debug_printf("\t\t.face = SVGA3D_FACE_NONE\n"); + _debug_printf("\t\t.face = SVGA3D_FACE_NONE\n"); break; case SVGA3D_FACE_FRONT: - debug_printf("\t\t.face = SVGA3D_FACE_FRONT\n"); + _debug_printf("\t\t.face = SVGA3D_FACE_FRONT\n"); break; case SVGA3D_FACE_BACK: - debug_printf("\t\t.face = SVGA3D_FACE_BACK\n"); + _debug_printf("\t\t.face = SVGA3D_FACE_BACK\n"); break; case SVGA3D_FACE_FRONT_BACK: - debug_printf("\t\t.face = SVGA3D_FACE_FRONT_BACK\n"); + _debug_printf("\t\t.face = SVGA3D_FACE_FRONT_BACK\n"); break; case SVGA3D_FACE_MAX: - debug_printf("\t\t.face = SVGA3D_FACE_MAX\n"); + _debug_printf("\t\t.face = SVGA3D_FACE_MAX\n"); break; default: - debug_printf("\t\t.face = %i\n", (*cmd).face); + _debug_printf("\t\t.face = %i\n", (*cmd).face); break; } - debug_printf("\t\t.material.diffuse[0] = %f\n", (*cmd).material.diffuse[0]); - debug_printf("\t\t.material.diffuse[1] = %f\n", (*cmd).material.diffuse[1]); - debug_printf("\t\t.material.diffuse[2] = %f\n", (*cmd).material.diffuse[2]); - debug_printf("\t\t.material.diffuse[3] = %f\n", (*cmd).material.diffuse[3]); - debug_printf("\t\t.material.ambient[0] = %f\n", (*cmd).material.ambient[0]); - debug_printf("\t\t.material.ambient[1] = %f\n", (*cmd).material.ambient[1]); - debug_printf("\t\t.material.ambient[2] = %f\n", (*cmd).material.ambient[2]); - debug_printf("\t\t.material.ambient[3] = %f\n", (*cmd).material.ambient[3]); - debug_printf("\t\t.material.specular[0] = %f\n", (*cmd).material.specular[0]); - debug_printf("\t\t.material.specular[1] = %f\n", (*cmd).material.specular[1]); - debug_printf("\t\t.material.specular[2] = %f\n", (*cmd).material.specular[2]); - debug_printf("\t\t.material.specular[3] = %f\n", (*cmd).material.specular[3]); - debug_printf("\t\t.material.emissive[0] = %f\n", (*cmd).material.emissive[0]); - debug_printf("\t\t.material.emissive[1] = %f\n", (*cmd).material.emissive[1]); - debug_printf("\t\t.material.emissive[2] = %f\n", (*cmd).material.emissive[2]); - debug_printf("\t\t.material.emissive[3] = %f\n", (*cmd).material.emissive[3]); - debug_printf("\t\t.material.shininess = %f\n", (*cmd).material.shininess); + _debug_printf("\t\t.material.diffuse[0] = %f\n", (*cmd).material.diffuse[0]); + _debug_printf("\t\t.material.diffuse[1] = %f\n", (*cmd).material.diffuse[1]); + _debug_printf("\t\t.material.diffuse[2] = %f\n", (*cmd).material.diffuse[2]); + _debug_printf("\t\t.material.diffuse[3] = %f\n", (*cmd).material.diffuse[3]); + _debug_printf("\t\t.material.ambient[0] = %f\n", (*cmd).material.ambient[0]); + _debug_printf("\t\t.material.ambient[1] = %f\n", (*cmd).material.ambient[1]); + _debug_printf("\t\t.material.ambient[2] = %f\n", (*cmd).material.ambient[2]); + _debug_printf("\t\t.material.ambient[3] = %f\n", (*cmd).material.ambient[3]); + _debug_printf("\t\t.material.specular[0] = %f\n", (*cmd).material.specular[0]); + _debug_printf("\t\t.material.specular[1] = %f\n", (*cmd).material.specular[1]); + _debug_printf("\t\t.material.specular[2] = %f\n", (*cmd).material.specular[2]); + _debug_printf("\t\t.material.specular[3] = %f\n", (*cmd).material.specular[3]); + _debug_printf("\t\t.material.emissive[0] = %f\n", (*cmd).material.emissive[0]); + _debug_printf("\t\t.material.emissive[1] = %f\n", (*cmd).material.emissive[1]); + _debug_printf("\t\t.material.emissive[2] = %f\n", (*cmd).material.emissive[2]); + _debug_printf("\t\t.material.emissive[3] = %f\n", (*cmd).material.emissive[3]); + _debug_printf("\t\t.material.shininess = %f\n", (*cmd).material.shininess); } static void dump_SVGA3dCmdSetLightData(const SVGA3dCmdSetLightData *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); - debug_printf("\t\t.index = %u\n", (*cmd).index); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.index = %u\n", (*cmd).index); switch((*cmd).data.type) { case SVGA3D_LIGHTTYPE_INVALID: - debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_INVALID\n"); + _debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_INVALID\n"); break; case SVGA3D_LIGHTTYPE_POINT: - debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_POINT\n"); + _debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_POINT\n"); break; case SVGA3D_LIGHTTYPE_SPOT1: - debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_SPOT1\n"); + _debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_SPOT1\n"); break; case SVGA3D_LIGHTTYPE_SPOT2: - debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_SPOT2\n"); + _debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_SPOT2\n"); break; case SVGA3D_LIGHTTYPE_DIRECTIONAL: - debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_DIRECTIONAL\n"); + _debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_DIRECTIONAL\n"); break; case SVGA3D_LIGHTTYPE_MAX: - debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_MAX\n"); + _debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_MAX\n"); break; default: - debug_printf("\t\t.data.type = %i\n", (*cmd).data.type); + _debug_printf("\t\t.data.type = %i\n", (*cmd).data.type); break; } - debug_printf("\t\t.data.inWorldSpace = %u\n", (*cmd).data.inWorldSpace); - debug_printf("\t\t.data.diffuse[0] = %f\n", (*cmd).data.diffuse[0]); - debug_printf("\t\t.data.diffuse[1] = %f\n", (*cmd).data.diffuse[1]); - debug_printf("\t\t.data.diffuse[2] = %f\n", (*cmd).data.diffuse[2]); - debug_printf("\t\t.data.diffuse[3] = %f\n", (*cmd).data.diffuse[3]); - debug_printf("\t\t.data.specular[0] = %f\n", (*cmd).data.specular[0]); - debug_printf("\t\t.data.specular[1] = %f\n", (*cmd).data.specular[1]); - debug_printf("\t\t.data.specular[2] = %f\n", (*cmd).data.specular[2]); - debug_printf("\t\t.data.specular[3] = %f\n", (*cmd).data.specular[3]); - debug_printf("\t\t.data.ambient[0] = %f\n", (*cmd).data.ambient[0]); - debug_printf("\t\t.data.ambient[1] = %f\n", (*cmd).data.ambient[1]); - debug_printf("\t\t.data.ambient[2] = %f\n", (*cmd).data.ambient[2]); - debug_printf("\t\t.data.ambient[3] = %f\n", (*cmd).data.ambient[3]); - debug_printf("\t\t.data.position[0] = %f\n", (*cmd).data.position[0]); - debug_printf("\t\t.data.position[1] = %f\n", (*cmd).data.position[1]); - debug_printf("\t\t.data.position[2] = %f\n", (*cmd).data.position[2]); - debug_printf("\t\t.data.position[3] = %f\n", (*cmd).data.position[3]); - debug_printf("\t\t.data.direction[0] = %f\n", (*cmd).data.direction[0]); - debug_printf("\t\t.data.direction[1] = %f\n", (*cmd).data.direction[1]); - debug_printf("\t\t.data.direction[2] = %f\n", (*cmd).data.direction[2]); - debug_printf("\t\t.data.direction[3] = %f\n", (*cmd).data.direction[3]); - debug_printf("\t\t.data.range = %f\n", (*cmd).data.range); - debug_printf("\t\t.data.falloff = %f\n", (*cmd).data.falloff); - debug_printf("\t\t.data.attenuation0 = %f\n", (*cmd).data.attenuation0); - debug_printf("\t\t.data.attenuation1 = %f\n", (*cmd).data.attenuation1); - debug_printf("\t\t.data.attenuation2 = %f\n", (*cmd).data.attenuation2); - debug_printf("\t\t.data.theta = %f\n", (*cmd).data.theta); - debug_printf("\t\t.data.phi = %f\n", (*cmd).data.phi); + _debug_printf("\t\t.data.inWorldSpace = %u\n", (*cmd).data.inWorldSpace); + _debug_printf("\t\t.data.diffuse[0] = %f\n", (*cmd).data.diffuse[0]); + _debug_printf("\t\t.data.diffuse[1] = %f\n", (*cmd).data.diffuse[1]); + _debug_printf("\t\t.data.diffuse[2] = %f\n", (*cmd).data.diffuse[2]); + _debug_printf("\t\t.data.diffuse[3] = %f\n", (*cmd).data.diffuse[3]); + _debug_printf("\t\t.data.specular[0] = %f\n", (*cmd).data.specular[0]); + _debug_printf("\t\t.data.specular[1] = %f\n", (*cmd).data.specular[1]); + _debug_printf("\t\t.data.specular[2] = %f\n", (*cmd).data.specular[2]); + _debug_printf("\t\t.data.specular[3] = %f\n", (*cmd).data.specular[3]); + _debug_printf("\t\t.data.ambient[0] = %f\n", (*cmd).data.ambient[0]); + _debug_printf("\t\t.data.ambient[1] = %f\n", (*cmd).data.ambient[1]); + _debug_printf("\t\t.data.ambient[2] = %f\n", (*cmd).data.ambient[2]); + _debug_printf("\t\t.data.ambient[3] = %f\n", (*cmd).data.ambient[3]); + _debug_printf("\t\t.data.position[0] = %f\n", (*cmd).data.position[0]); + _debug_printf("\t\t.data.position[1] = %f\n", (*cmd).data.position[1]); + _debug_printf("\t\t.data.position[2] = %f\n", (*cmd).data.position[2]); + _debug_printf("\t\t.data.position[3] = %f\n", (*cmd).data.position[3]); + _debug_printf("\t\t.data.direction[0] = %f\n", (*cmd).data.direction[0]); + _debug_printf("\t\t.data.direction[1] = %f\n", (*cmd).data.direction[1]); + _debug_printf("\t\t.data.direction[2] = %f\n", (*cmd).data.direction[2]); + _debug_printf("\t\t.data.direction[3] = %f\n", (*cmd).data.direction[3]); + _debug_printf("\t\t.data.range = %f\n", (*cmd).data.range); + _debug_printf("\t\t.data.falloff = %f\n", (*cmd).data.falloff); + _debug_printf("\t\t.data.attenuation0 = %f\n", (*cmd).data.attenuation0); + _debug_printf("\t\t.data.attenuation1 = %f\n", (*cmd).data.attenuation1); + _debug_printf("\t\t.data.attenuation2 = %f\n", (*cmd).data.attenuation2); + _debug_printf("\t\t.data.theta = %f\n", (*cmd).data.theta); + _debug_printf("\t\t.data.phi = %f\n", (*cmd).data.phi); } static void dump_SVGA3dCmdSetViewport(const SVGA3dCmdSetViewport *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); - debug_printf("\t\t.rect.x = %u\n", (*cmd).rect.x); - debug_printf("\t\t.rect.y = %u\n", (*cmd).rect.y); - debug_printf("\t\t.rect.w = %u\n", (*cmd).rect.w); - debug_printf("\t\t.rect.h = %u\n", (*cmd).rect.h); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.rect.x = %u\n", (*cmd).rect.x); + _debug_printf("\t\t.rect.y = %u\n", (*cmd).rect.y); + _debug_printf("\t\t.rect.w = %u\n", (*cmd).rect.w); + _debug_printf("\t\t.rect.h = %u\n", (*cmd).rect.h); } static void dump_SVGA3dCmdSetScissorRect(const SVGA3dCmdSetScissorRect *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); - debug_printf("\t\t.rect.x = %u\n", (*cmd).rect.x); - debug_printf("\t\t.rect.y = %u\n", (*cmd).rect.y); - debug_printf("\t\t.rect.w = %u\n", (*cmd).rect.w); - debug_printf("\t\t.rect.h = %u\n", (*cmd).rect.h); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.rect.x = %u\n", (*cmd).rect.x); + _debug_printf("\t\t.rect.y = %u\n", (*cmd).rect.y); + _debug_printf("\t\t.rect.w = %u\n", (*cmd).rect.w); + _debug_printf("\t\t.rect.h = %u\n", (*cmd).rect.h); } static void dump_SVGA3dCopyRect(const SVGA3dCopyRect *cmd) { - debug_printf("\t\t.x = %u\n", (*cmd).x); - debug_printf("\t\t.y = %u\n", (*cmd).y); - debug_printf("\t\t.w = %u\n", (*cmd).w); - debug_printf("\t\t.h = %u\n", (*cmd).h); - debug_printf("\t\t.srcx = %u\n", (*cmd).srcx); - debug_printf("\t\t.srcy = %u\n", (*cmd).srcy); + _debug_printf("\t\t.x = %u\n", (*cmd).x); + _debug_printf("\t\t.y = %u\n", (*cmd).y); + _debug_printf("\t\t.w = %u\n", (*cmd).w); + _debug_printf("\t\t.h = %u\n", (*cmd).h); + _debug_printf("\t\t.srcx = %u\n", (*cmd).srcx); + _debug_printf("\t\t.srcy = %u\n", (*cmd).srcy); } static void dump_SVGA3dCmdSetShader(const SVGA3dCmdSetShader *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); switch((*cmd).type) { case SVGA3D_SHADERTYPE_COMPILED_DX8: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_COMPILED_DX8\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_COMPILED_DX8\n"); break; case SVGA3D_SHADERTYPE_VS: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_VS\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_VS\n"); break; case SVGA3D_SHADERTYPE_PS: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_PS\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_PS\n"); break; case SVGA3D_SHADERTYPE_MAX: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_MAX\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_MAX\n"); break; default: - debug_printf("\t\t.type = %i\n", (*cmd).type); + _debug_printf("\t\t.type = %i\n", (*cmd).type); break; } - debug_printf("\t\t.shid = %u\n", (*cmd).shid); + _debug_printf("\t\t.shid = %u\n", (*cmd).shid); } static void dump_SVGA3dCmdEndQuery(const SVGA3dCmdEndQuery *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); switch((*cmd).type) { case SVGA3D_QUERYTYPE_OCCLUSION: - debug_printf("\t\t.type = SVGA3D_QUERYTYPE_OCCLUSION\n"); + _debug_printf("\t\t.type = SVGA3D_QUERYTYPE_OCCLUSION\n"); break; case SVGA3D_QUERYTYPE_MAX: - debug_printf("\t\t.type = SVGA3D_QUERYTYPE_MAX\n"); + _debug_printf("\t\t.type = SVGA3D_QUERYTYPE_MAX\n"); break; default: - debug_printf("\t\t.type = %i\n", (*cmd).type); + _debug_printf("\t\t.type = %i\n", (*cmd).type); break; } - debug_printf("\t\t.guestResult.gmrId = %u\n", (*cmd).guestResult.gmrId); - debug_printf("\t\t.guestResult.offset = %u\n", (*cmd).guestResult.offset); + _debug_printf("\t\t.guestResult.gmrId = %u\n", (*cmd).guestResult.gmrId); + _debug_printf("\t\t.guestResult.offset = %u\n", (*cmd).guestResult.offset); } static void dump_SVGA3dSize(const SVGA3dSize *cmd) { - debug_printf("\t\t.width = %u\n", (*cmd).width); - debug_printf("\t\t.height = %u\n", (*cmd).height); - debug_printf("\t\t.depth = %u\n", (*cmd).depth); + _debug_printf("\t\t.width = %u\n", (*cmd).width); + _debug_printf("\t\t.height = %u\n", (*cmd).height); + _debug_printf("\t\t.depth = %u\n", (*cmd).depth); } static void dump_SVGA3dCmdDestroySurface(const SVGA3dCmdDestroySurface *cmd) { - debug_printf("\t\t.sid = %u\n", (*cmd).sid); + _debug_printf("\t\t.sid = %u\n", (*cmd).sid); } static void dump_SVGA3dCmdDefineContext(const SVGA3dCmdDefineContext *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); } static void dump_SVGA3dRect(const SVGA3dRect *cmd) { - debug_printf("\t\t.x = %u\n", (*cmd).x); - debug_printf("\t\t.y = %u\n", (*cmd).y); - debug_printf("\t\t.w = %u\n", (*cmd).w); - debug_printf("\t\t.h = %u\n", (*cmd).h); + _debug_printf("\t\t.x = %u\n", (*cmd).x); + _debug_printf("\t\t.y = %u\n", (*cmd).y); + _debug_printf("\t\t.w = %u\n", (*cmd).w); + _debug_printf("\t\t.h = %u\n", (*cmd).h); } static void dump_SVGA3dCmdBeginQuery(const SVGA3dCmdBeginQuery *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); switch((*cmd).type) { case SVGA3D_QUERYTYPE_OCCLUSION: - debug_printf("\t\t.type = SVGA3D_QUERYTYPE_OCCLUSION\n"); + _debug_printf("\t\t.type = SVGA3D_QUERYTYPE_OCCLUSION\n"); break; case SVGA3D_QUERYTYPE_MAX: - debug_printf("\t\t.type = SVGA3D_QUERYTYPE_MAX\n"); + _debug_printf("\t\t.type = SVGA3D_QUERYTYPE_MAX\n"); break; default: - debug_printf("\t\t.type = %i\n", (*cmd).type); + _debug_printf("\t\t.type = %i\n", (*cmd).type); break; } } @@ -599,336 +599,336 @@ dump_SVGA3dRenderState(const SVGA3dRenderState *cmd) { switch((*cmd).state) { case SVGA3D_RS_INVALID: - debug_printf("\t\t.state = SVGA3D_RS_INVALID\n"); + _debug_printf("\t\t.state = SVGA3D_RS_INVALID\n"); break; case SVGA3D_RS_ZENABLE: - debug_printf("\t\t.state = SVGA3D_RS_ZENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_ZENABLE\n"); break; case SVGA3D_RS_ZWRITEENABLE: - debug_printf("\t\t.state = SVGA3D_RS_ZWRITEENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_ZWRITEENABLE\n"); break; case SVGA3D_RS_ALPHATESTENABLE: - debug_printf("\t\t.state = SVGA3D_RS_ALPHATESTENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_ALPHATESTENABLE\n"); break; case SVGA3D_RS_DITHERENABLE: - debug_printf("\t\t.state = SVGA3D_RS_DITHERENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_DITHERENABLE\n"); break; case SVGA3D_RS_BLENDENABLE: - debug_printf("\t\t.state = SVGA3D_RS_BLENDENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_BLENDENABLE\n"); break; case SVGA3D_RS_FOGENABLE: - debug_printf("\t\t.state = SVGA3D_RS_FOGENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_FOGENABLE\n"); break; case SVGA3D_RS_SPECULARENABLE: - debug_printf("\t\t.state = SVGA3D_RS_SPECULARENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_SPECULARENABLE\n"); break; case SVGA3D_RS_STENCILENABLE: - debug_printf("\t\t.state = SVGA3D_RS_STENCILENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_STENCILENABLE\n"); break; case SVGA3D_RS_LIGHTINGENABLE: - debug_printf("\t\t.state = SVGA3D_RS_LIGHTINGENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_LIGHTINGENABLE\n"); break; case SVGA3D_RS_NORMALIZENORMALS: - debug_printf("\t\t.state = SVGA3D_RS_NORMALIZENORMALS\n"); + _debug_printf("\t\t.state = SVGA3D_RS_NORMALIZENORMALS\n"); break; case SVGA3D_RS_POINTSPRITEENABLE: - debug_printf("\t\t.state = SVGA3D_RS_POINTSPRITEENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_POINTSPRITEENABLE\n"); break; case SVGA3D_RS_POINTSCALEENABLE: - debug_printf("\t\t.state = SVGA3D_RS_POINTSCALEENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_POINTSCALEENABLE\n"); break; case SVGA3D_RS_STENCILREF: - debug_printf("\t\t.state = SVGA3D_RS_STENCILREF\n"); + _debug_printf("\t\t.state = SVGA3D_RS_STENCILREF\n"); break; case SVGA3D_RS_STENCILMASK: - debug_printf("\t\t.state = SVGA3D_RS_STENCILMASK\n"); + _debug_printf("\t\t.state = SVGA3D_RS_STENCILMASK\n"); break; case SVGA3D_RS_STENCILWRITEMASK: - debug_printf("\t\t.state = SVGA3D_RS_STENCILWRITEMASK\n"); + _debug_printf("\t\t.state = SVGA3D_RS_STENCILWRITEMASK\n"); break; case SVGA3D_RS_FOGSTART: - debug_printf("\t\t.state = SVGA3D_RS_FOGSTART\n"); + _debug_printf("\t\t.state = SVGA3D_RS_FOGSTART\n"); break; case SVGA3D_RS_FOGEND: - debug_printf("\t\t.state = SVGA3D_RS_FOGEND\n"); + _debug_printf("\t\t.state = SVGA3D_RS_FOGEND\n"); break; case SVGA3D_RS_FOGDENSITY: - debug_printf("\t\t.state = SVGA3D_RS_FOGDENSITY\n"); + _debug_printf("\t\t.state = SVGA3D_RS_FOGDENSITY\n"); break; case SVGA3D_RS_POINTSIZE: - debug_printf("\t\t.state = SVGA3D_RS_POINTSIZE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_POINTSIZE\n"); break; case SVGA3D_RS_POINTSIZEMIN: - debug_printf("\t\t.state = SVGA3D_RS_POINTSIZEMIN\n"); + _debug_printf("\t\t.state = SVGA3D_RS_POINTSIZEMIN\n"); break; case SVGA3D_RS_POINTSIZEMAX: - debug_printf("\t\t.state = SVGA3D_RS_POINTSIZEMAX\n"); + _debug_printf("\t\t.state = SVGA3D_RS_POINTSIZEMAX\n"); break; case SVGA3D_RS_POINTSCALE_A: - debug_printf("\t\t.state = SVGA3D_RS_POINTSCALE_A\n"); + _debug_printf("\t\t.state = SVGA3D_RS_POINTSCALE_A\n"); break; case SVGA3D_RS_POINTSCALE_B: - debug_printf("\t\t.state = SVGA3D_RS_POINTSCALE_B\n"); + _debug_printf("\t\t.state = SVGA3D_RS_POINTSCALE_B\n"); break; case SVGA3D_RS_POINTSCALE_C: - debug_printf("\t\t.state = SVGA3D_RS_POINTSCALE_C\n"); + _debug_printf("\t\t.state = SVGA3D_RS_POINTSCALE_C\n"); break; case SVGA3D_RS_FOGCOLOR: - debug_printf("\t\t.state = SVGA3D_RS_FOGCOLOR\n"); + _debug_printf("\t\t.state = SVGA3D_RS_FOGCOLOR\n"); break; case SVGA3D_RS_AMBIENT: - debug_printf("\t\t.state = SVGA3D_RS_AMBIENT\n"); + _debug_printf("\t\t.state = SVGA3D_RS_AMBIENT\n"); break; case SVGA3D_RS_CLIPPLANEENABLE: - debug_printf("\t\t.state = SVGA3D_RS_CLIPPLANEENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_CLIPPLANEENABLE\n"); break; case SVGA3D_RS_FOGMODE: - debug_printf("\t\t.state = SVGA3D_RS_FOGMODE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_FOGMODE\n"); break; case SVGA3D_RS_FILLMODE: - debug_printf("\t\t.state = SVGA3D_RS_FILLMODE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_FILLMODE\n"); break; case SVGA3D_RS_SHADEMODE: - debug_printf("\t\t.state = SVGA3D_RS_SHADEMODE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_SHADEMODE\n"); break; case SVGA3D_RS_LINEPATTERN: - debug_printf("\t\t.state = SVGA3D_RS_LINEPATTERN\n"); + _debug_printf("\t\t.state = SVGA3D_RS_LINEPATTERN\n"); break; case SVGA3D_RS_SRCBLEND: - debug_printf("\t\t.state = SVGA3D_RS_SRCBLEND\n"); + _debug_printf("\t\t.state = SVGA3D_RS_SRCBLEND\n"); break; case SVGA3D_RS_DSTBLEND: - debug_printf("\t\t.state = SVGA3D_RS_DSTBLEND\n"); + _debug_printf("\t\t.state = SVGA3D_RS_DSTBLEND\n"); break; case SVGA3D_RS_BLENDEQUATION: - debug_printf("\t\t.state = SVGA3D_RS_BLENDEQUATION\n"); + _debug_printf("\t\t.state = SVGA3D_RS_BLENDEQUATION\n"); break; case SVGA3D_RS_CULLMODE: - debug_printf("\t\t.state = SVGA3D_RS_CULLMODE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_CULLMODE\n"); break; case SVGA3D_RS_ZFUNC: - debug_printf("\t\t.state = SVGA3D_RS_ZFUNC\n"); + _debug_printf("\t\t.state = SVGA3D_RS_ZFUNC\n"); break; case SVGA3D_RS_ALPHAFUNC: - debug_printf("\t\t.state = SVGA3D_RS_ALPHAFUNC\n"); + _debug_printf("\t\t.state = SVGA3D_RS_ALPHAFUNC\n"); break; case SVGA3D_RS_STENCILFUNC: - debug_printf("\t\t.state = SVGA3D_RS_STENCILFUNC\n"); + _debug_printf("\t\t.state = SVGA3D_RS_STENCILFUNC\n"); break; case SVGA3D_RS_STENCILFAIL: - debug_printf("\t\t.state = SVGA3D_RS_STENCILFAIL\n"); + _debug_printf("\t\t.state = SVGA3D_RS_STENCILFAIL\n"); break; case SVGA3D_RS_STENCILZFAIL: - debug_printf("\t\t.state = SVGA3D_RS_STENCILZFAIL\n"); + _debug_printf("\t\t.state = SVGA3D_RS_STENCILZFAIL\n"); break; case SVGA3D_RS_STENCILPASS: - debug_printf("\t\t.state = SVGA3D_RS_STENCILPASS\n"); + _debug_printf("\t\t.state = SVGA3D_RS_STENCILPASS\n"); break; case SVGA3D_RS_ALPHAREF: - debug_printf("\t\t.state = SVGA3D_RS_ALPHAREF\n"); + _debug_printf("\t\t.state = SVGA3D_RS_ALPHAREF\n"); break; case SVGA3D_RS_FRONTWINDING: - debug_printf("\t\t.state = SVGA3D_RS_FRONTWINDING\n"); + _debug_printf("\t\t.state = SVGA3D_RS_FRONTWINDING\n"); break; case SVGA3D_RS_COORDINATETYPE: - debug_printf("\t\t.state = SVGA3D_RS_COORDINATETYPE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_COORDINATETYPE\n"); break; case SVGA3D_RS_ZBIAS: - debug_printf("\t\t.state = SVGA3D_RS_ZBIAS\n"); + _debug_printf("\t\t.state = SVGA3D_RS_ZBIAS\n"); break; case SVGA3D_RS_RANGEFOGENABLE: - debug_printf("\t\t.state = SVGA3D_RS_RANGEFOGENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_RANGEFOGENABLE\n"); break; case SVGA3D_RS_COLORWRITEENABLE: - debug_printf("\t\t.state = SVGA3D_RS_COLORWRITEENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_COLORWRITEENABLE\n"); break; case SVGA3D_RS_VERTEXMATERIALENABLE: - debug_printf("\t\t.state = SVGA3D_RS_VERTEXMATERIALENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_VERTEXMATERIALENABLE\n"); break; case SVGA3D_RS_DIFFUSEMATERIALSOURCE: - debug_printf("\t\t.state = SVGA3D_RS_DIFFUSEMATERIALSOURCE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_DIFFUSEMATERIALSOURCE\n"); break; case SVGA3D_RS_SPECULARMATERIALSOURCE: - debug_printf("\t\t.state = SVGA3D_RS_SPECULARMATERIALSOURCE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_SPECULARMATERIALSOURCE\n"); break; case SVGA3D_RS_AMBIENTMATERIALSOURCE: - debug_printf("\t\t.state = SVGA3D_RS_AMBIENTMATERIALSOURCE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_AMBIENTMATERIALSOURCE\n"); break; case SVGA3D_RS_EMISSIVEMATERIALSOURCE: - debug_printf("\t\t.state = SVGA3D_RS_EMISSIVEMATERIALSOURCE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_EMISSIVEMATERIALSOURCE\n"); break; case SVGA3D_RS_TEXTUREFACTOR: - debug_printf("\t\t.state = SVGA3D_RS_TEXTUREFACTOR\n"); + _debug_printf("\t\t.state = SVGA3D_RS_TEXTUREFACTOR\n"); break; case SVGA3D_RS_LOCALVIEWER: - debug_printf("\t\t.state = SVGA3D_RS_LOCALVIEWER\n"); + _debug_printf("\t\t.state = SVGA3D_RS_LOCALVIEWER\n"); break; case SVGA3D_RS_SCISSORTESTENABLE: - debug_printf("\t\t.state = SVGA3D_RS_SCISSORTESTENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_SCISSORTESTENABLE\n"); break; case SVGA3D_RS_BLENDCOLOR: - debug_printf("\t\t.state = SVGA3D_RS_BLENDCOLOR\n"); + _debug_printf("\t\t.state = SVGA3D_RS_BLENDCOLOR\n"); break; case SVGA3D_RS_STENCILENABLE2SIDED: - debug_printf("\t\t.state = SVGA3D_RS_STENCILENABLE2SIDED\n"); + _debug_printf("\t\t.state = SVGA3D_RS_STENCILENABLE2SIDED\n"); break; case SVGA3D_RS_CCWSTENCILFUNC: - debug_printf("\t\t.state = SVGA3D_RS_CCWSTENCILFUNC\n"); + _debug_printf("\t\t.state = SVGA3D_RS_CCWSTENCILFUNC\n"); break; case SVGA3D_RS_CCWSTENCILFAIL: - debug_printf("\t\t.state = SVGA3D_RS_CCWSTENCILFAIL\n"); + _debug_printf("\t\t.state = SVGA3D_RS_CCWSTENCILFAIL\n"); break; case SVGA3D_RS_CCWSTENCILZFAIL: - debug_printf("\t\t.state = SVGA3D_RS_CCWSTENCILZFAIL\n"); + _debug_printf("\t\t.state = SVGA3D_RS_CCWSTENCILZFAIL\n"); break; case SVGA3D_RS_CCWSTENCILPASS: - debug_printf("\t\t.state = SVGA3D_RS_CCWSTENCILPASS\n"); + _debug_printf("\t\t.state = SVGA3D_RS_CCWSTENCILPASS\n"); break; case SVGA3D_RS_VERTEXBLEND: - debug_printf("\t\t.state = SVGA3D_RS_VERTEXBLEND\n"); + _debug_printf("\t\t.state = SVGA3D_RS_VERTEXBLEND\n"); break; case SVGA3D_RS_SLOPESCALEDEPTHBIAS: - debug_printf("\t\t.state = SVGA3D_RS_SLOPESCALEDEPTHBIAS\n"); + _debug_printf("\t\t.state = SVGA3D_RS_SLOPESCALEDEPTHBIAS\n"); break; case SVGA3D_RS_DEPTHBIAS: - debug_printf("\t\t.state = SVGA3D_RS_DEPTHBIAS\n"); + _debug_printf("\t\t.state = SVGA3D_RS_DEPTHBIAS\n"); break; case SVGA3D_RS_OUTPUTGAMMA: - debug_printf("\t\t.state = SVGA3D_RS_OUTPUTGAMMA\n"); + _debug_printf("\t\t.state = SVGA3D_RS_OUTPUTGAMMA\n"); break; case SVGA3D_RS_ZVISIBLE: - debug_printf("\t\t.state = SVGA3D_RS_ZVISIBLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_ZVISIBLE\n"); break; case SVGA3D_RS_LASTPIXEL: - debug_printf("\t\t.state = SVGA3D_RS_LASTPIXEL\n"); + _debug_printf("\t\t.state = SVGA3D_RS_LASTPIXEL\n"); break; case SVGA3D_RS_CLIPPING: - debug_printf("\t\t.state = SVGA3D_RS_CLIPPING\n"); + _debug_printf("\t\t.state = SVGA3D_RS_CLIPPING\n"); break; case SVGA3D_RS_WRAP0: - debug_printf("\t\t.state = SVGA3D_RS_WRAP0\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP0\n"); break; case SVGA3D_RS_WRAP1: - debug_printf("\t\t.state = SVGA3D_RS_WRAP1\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP1\n"); break; case SVGA3D_RS_WRAP2: - debug_printf("\t\t.state = SVGA3D_RS_WRAP2\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP2\n"); break; case SVGA3D_RS_WRAP3: - debug_printf("\t\t.state = SVGA3D_RS_WRAP3\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP3\n"); break; case SVGA3D_RS_WRAP4: - debug_printf("\t\t.state = SVGA3D_RS_WRAP4\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP4\n"); break; case SVGA3D_RS_WRAP5: - debug_printf("\t\t.state = SVGA3D_RS_WRAP5\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP5\n"); break; case SVGA3D_RS_WRAP6: - debug_printf("\t\t.state = SVGA3D_RS_WRAP6\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP6\n"); break; case SVGA3D_RS_WRAP7: - debug_printf("\t\t.state = SVGA3D_RS_WRAP7\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP7\n"); break; case SVGA3D_RS_WRAP8: - debug_printf("\t\t.state = SVGA3D_RS_WRAP8\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP8\n"); break; case SVGA3D_RS_WRAP9: - debug_printf("\t\t.state = SVGA3D_RS_WRAP9\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP9\n"); break; case SVGA3D_RS_WRAP10: - debug_printf("\t\t.state = SVGA3D_RS_WRAP10\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP10\n"); break; case SVGA3D_RS_WRAP11: - debug_printf("\t\t.state = SVGA3D_RS_WRAP11\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP11\n"); break; case SVGA3D_RS_WRAP12: - debug_printf("\t\t.state = SVGA3D_RS_WRAP12\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP12\n"); break; case SVGA3D_RS_WRAP13: - debug_printf("\t\t.state = SVGA3D_RS_WRAP13\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP13\n"); break; case SVGA3D_RS_WRAP14: - debug_printf("\t\t.state = SVGA3D_RS_WRAP14\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP14\n"); break; case SVGA3D_RS_WRAP15: - debug_printf("\t\t.state = SVGA3D_RS_WRAP15\n"); + _debug_printf("\t\t.state = SVGA3D_RS_WRAP15\n"); break; case SVGA3D_RS_MULTISAMPLEANTIALIAS: - debug_printf("\t\t.state = SVGA3D_RS_MULTISAMPLEANTIALIAS\n"); + _debug_printf("\t\t.state = SVGA3D_RS_MULTISAMPLEANTIALIAS\n"); break; case SVGA3D_RS_MULTISAMPLEMASK: - debug_printf("\t\t.state = SVGA3D_RS_MULTISAMPLEMASK\n"); + _debug_printf("\t\t.state = SVGA3D_RS_MULTISAMPLEMASK\n"); break; case SVGA3D_RS_INDEXEDVERTEXBLENDENABLE: - debug_printf("\t\t.state = SVGA3D_RS_INDEXEDVERTEXBLENDENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_INDEXEDVERTEXBLENDENABLE\n"); break; case SVGA3D_RS_TWEENFACTOR: - debug_printf("\t\t.state = SVGA3D_RS_TWEENFACTOR\n"); + _debug_printf("\t\t.state = SVGA3D_RS_TWEENFACTOR\n"); break; case SVGA3D_RS_ANTIALIASEDLINEENABLE: - debug_printf("\t\t.state = SVGA3D_RS_ANTIALIASEDLINEENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_ANTIALIASEDLINEENABLE\n"); break; case SVGA3D_RS_COLORWRITEENABLE1: - debug_printf("\t\t.state = SVGA3D_RS_COLORWRITEENABLE1\n"); + _debug_printf("\t\t.state = SVGA3D_RS_COLORWRITEENABLE1\n"); break; case SVGA3D_RS_COLORWRITEENABLE2: - debug_printf("\t\t.state = SVGA3D_RS_COLORWRITEENABLE2\n"); + _debug_printf("\t\t.state = SVGA3D_RS_COLORWRITEENABLE2\n"); break; case SVGA3D_RS_COLORWRITEENABLE3: - debug_printf("\t\t.state = SVGA3D_RS_COLORWRITEENABLE3\n"); + _debug_printf("\t\t.state = SVGA3D_RS_COLORWRITEENABLE3\n"); break; case SVGA3D_RS_SEPARATEALPHABLENDENABLE: - debug_printf("\t\t.state = SVGA3D_RS_SEPARATEALPHABLENDENABLE\n"); + _debug_printf("\t\t.state = SVGA3D_RS_SEPARATEALPHABLENDENABLE\n"); break; case SVGA3D_RS_SRCBLENDALPHA: - debug_printf("\t\t.state = SVGA3D_RS_SRCBLENDALPHA\n"); + _debug_printf("\t\t.state = SVGA3D_RS_SRCBLENDALPHA\n"); break; case SVGA3D_RS_DSTBLENDALPHA: - debug_printf("\t\t.state = SVGA3D_RS_DSTBLENDALPHA\n"); + _debug_printf("\t\t.state = SVGA3D_RS_DSTBLENDALPHA\n"); break; case SVGA3D_RS_BLENDEQUATIONALPHA: - debug_printf("\t\t.state = SVGA3D_RS_BLENDEQUATIONALPHA\n"); + _debug_printf("\t\t.state = SVGA3D_RS_BLENDEQUATIONALPHA\n"); break; case SVGA3D_RS_MAX: - debug_printf("\t\t.state = SVGA3D_RS_MAX\n"); + _debug_printf("\t\t.state = SVGA3D_RS_MAX\n"); break; default: - debug_printf("\t\t.state = %i\n", (*cmd).state); + _debug_printf("\t\t.state = %i\n", (*cmd).state); break; } - debug_printf("\t\t.uintValue = %u\n", (*cmd).uintValue); - debug_printf("\t\t.floatValue = %f\n", (*cmd).floatValue); + _debug_printf("\t\t.uintValue = %u\n", (*cmd).uintValue); + _debug_printf("\t\t.floatValue = %f\n", (*cmd).floatValue); } static void dump_SVGA3dVertexDivisor(const SVGA3dVertexDivisor *cmd) { - debug_printf("\t\t.value = %u\n", (*cmd).value); - debug_printf("\t\t.count = %u\n", (*cmd).count); - debug_printf("\t\t.indexedData = %u\n", (*cmd).indexedData); - debug_printf("\t\t.instanceData = %u\n", (*cmd).instanceData); + _debug_printf("\t\t.value = %u\n", (*cmd).value); + _debug_printf("\t\t.count = %u\n", (*cmd).count); + _debug_printf("\t\t.indexedData = %u\n", (*cmd).indexedData); + _debug_printf("\t\t.instanceData = %u\n", (*cmd).instanceData); } static void dump_SVGA3dCmdDefineShader(const SVGA3dCmdDefineShader *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); - debug_printf("\t\t.shid = %u\n", (*cmd).shid); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.shid = %u\n", (*cmd).shid); switch((*cmd).type) { case SVGA3D_SHADERTYPE_COMPILED_DX8: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_COMPILED_DX8\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_COMPILED_DX8\n"); break; case SVGA3D_SHADERTYPE_VS: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_VS\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_VS\n"); break; case SVGA3D_SHADERTYPE_PS: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_PS\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_PS\n"); break; case SVGA3D_SHADERTYPE_MAX: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_MAX\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_MAX\n"); break; default: - debug_printf("\t\t.type = %i\n", (*cmd).type); + _debug_printf("\t\t.type = %i\n", (*cmd).type); break; } } @@ -936,53 +936,53 @@ dump_SVGA3dCmdDefineShader(const SVGA3dCmdDefineShader *cmd) static void dump_SVGA3dCmdSetShaderConst(const SVGA3dCmdSetShaderConst *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); - debug_printf("\t\t.reg = %u\n", (*cmd).reg); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.reg = %u\n", (*cmd).reg); switch((*cmd).type) { case SVGA3D_SHADERTYPE_COMPILED_DX8: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_COMPILED_DX8\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_COMPILED_DX8\n"); break; case SVGA3D_SHADERTYPE_VS: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_VS\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_VS\n"); break; case SVGA3D_SHADERTYPE_PS: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_PS\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_PS\n"); break; case SVGA3D_SHADERTYPE_MAX: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_MAX\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_MAX\n"); break; default: - debug_printf("\t\t.type = %i\n", (*cmd).type); + _debug_printf("\t\t.type = %i\n", (*cmd).type); break; } switch((*cmd).ctype) { case SVGA3D_CONST_TYPE_FLOAT: - debug_printf("\t\t.ctype = SVGA3D_CONST_TYPE_FLOAT\n"); - debug_printf("\t\t.values[0] = %f\n", *(const float *)&(*cmd).values[0]); - debug_printf("\t\t.values[1] = %f\n", *(const float *)&(*cmd).values[1]); - debug_printf("\t\t.values[2] = %f\n", *(const float *)&(*cmd).values[2]); - debug_printf("\t\t.values[3] = %f\n", *(const float *)&(*cmd).values[3]); + _debug_printf("\t\t.ctype = SVGA3D_CONST_TYPE_FLOAT\n"); + _debug_printf("\t\t.values[0] = %f\n", *(const float *)&(*cmd).values[0]); + _debug_printf("\t\t.values[1] = %f\n", *(const float *)&(*cmd).values[1]); + _debug_printf("\t\t.values[2] = %f\n", *(const float *)&(*cmd).values[2]); + _debug_printf("\t\t.values[3] = %f\n", *(const float *)&(*cmd).values[3]); break; case SVGA3D_CONST_TYPE_INT: - debug_printf("\t\t.ctype = SVGA3D_CONST_TYPE_INT\n"); - debug_printf("\t\t.values[0] = %u\n", (*cmd).values[0]); - debug_printf("\t\t.values[1] = %u\n", (*cmd).values[1]); - debug_printf("\t\t.values[2] = %u\n", (*cmd).values[2]); - debug_printf("\t\t.values[3] = %u\n", (*cmd).values[3]); + _debug_printf("\t\t.ctype = SVGA3D_CONST_TYPE_INT\n"); + _debug_printf("\t\t.values[0] = %u\n", (*cmd).values[0]); + _debug_printf("\t\t.values[1] = %u\n", (*cmd).values[1]); + _debug_printf("\t\t.values[2] = %u\n", (*cmd).values[2]); + _debug_printf("\t\t.values[3] = %u\n", (*cmd).values[3]); break; case SVGA3D_CONST_TYPE_BOOL: - debug_printf("\t\t.ctype = SVGA3D_CONST_TYPE_BOOL\n"); - debug_printf("\t\t.values[0] = %u\n", (*cmd).values[0]); - debug_printf("\t\t.values[1] = %u\n", (*cmd).values[1]); - debug_printf("\t\t.values[2] = %u\n", (*cmd).values[2]); - debug_printf("\t\t.values[3] = %u\n", (*cmd).values[3]); + _debug_printf("\t\t.ctype = SVGA3D_CONST_TYPE_BOOL\n"); + _debug_printf("\t\t.values[0] = %u\n", (*cmd).values[0]); + _debug_printf("\t\t.values[1] = %u\n", (*cmd).values[1]); + _debug_printf("\t\t.values[2] = %u\n", (*cmd).values[2]); + _debug_printf("\t\t.values[3] = %u\n", (*cmd).values[3]); break; default: - debug_printf("\t\t.ctype = %i\n", (*cmd).ctype); - debug_printf("\t\t.values[0] = %u\n", (*cmd).values[0]); - debug_printf("\t\t.values[1] = %u\n", (*cmd).values[1]); - debug_printf("\t\t.values[2] = %u\n", (*cmd).values[2]); - debug_printf("\t\t.values[3] = %u\n", (*cmd).values[3]); + _debug_printf("\t\t.ctype = %i\n", (*cmd).ctype); + _debug_printf("\t\t.values[0] = %u\n", (*cmd).values[0]); + _debug_printf("\t\t.values[1] = %u\n", (*cmd).values[1]); + _debug_printf("\t\t.values[2] = %u\n", (*cmd).values[2]); + _debug_printf("\t\t.values[3] = %u\n", (*cmd).values[3]); break; } } @@ -990,25 +990,25 @@ dump_SVGA3dCmdSetShaderConst(const SVGA3dCmdSetShaderConst *cmd) static void dump_SVGA3dCmdSetZRange(const SVGA3dCmdSetZRange *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); - debug_printf("\t\t.zRange.min = %f\n", (*cmd).zRange.min); - debug_printf("\t\t.zRange.max = %f\n", (*cmd).zRange.max); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.zRange.min = %f\n", (*cmd).zRange.min); + _debug_printf("\t\t.zRange.max = %f\n", (*cmd).zRange.max); } static void dump_SVGA3dCmdDrawPrimitives(const SVGA3dCmdDrawPrimitives *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); - debug_printf("\t\t.numVertexDecls = %u\n", (*cmd).numVertexDecls); - debug_printf("\t\t.numRanges = %u\n", (*cmd).numRanges); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.numVertexDecls = %u\n", (*cmd).numVertexDecls); + _debug_printf("\t\t.numRanges = %u\n", (*cmd).numRanges); } static void dump_SVGA3dCmdSetLightEnabled(const SVGA3dCmdSetLightEnabled *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); - debug_printf("\t\t.index = %u\n", (*cmd).index); - debug_printf("\t\t.enabled = %u\n", (*cmd).enabled); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.index = %u\n", (*cmd).index); + _debug_printf("\t\t.enabled = %u\n", (*cmd).enabled); } static void @@ -1016,86 +1016,86 @@ dump_SVGA3dPrimitiveRange(const SVGA3dPrimitiveRange *cmd) { switch((*cmd).primType) { case SVGA3D_PRIMITIVE_INVALID: - debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_INVALID\n"); + _debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_INVALID\n"); break; case SVGA3D_PRIMITIVE_TRIANGLELIST: - debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_TRIANGLELIST\n"); + _debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_TRIANGLELIST\n"); break; case SVGA3D_PRIMITIVE_POINTLIST: - debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_POINTLIST\n"); + _debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_POINTLIST\n"); break; case SVGA3D_PRIMITIVE_LINELIST: - debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_LINELIST\n"); + _debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_LINELIST\n"); break; case SVGA3D_PRIMITIVE_LINESTRIP: - debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_LINESTRIP\n"); + _debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_LINESTRIP\n"); break; case SVGA3D_PRIMITIVE_TRIANGLESTRIP: - debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_TRIANGLESTRIP\n"); + _debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_TRIANGLESTRIP\n"); break; case SVGA3D_PRIMITIVE_TRIANGLEFAN: - debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_TRIANGLEFAN\n"); + _debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_TRIANGLEFAN\n"); break; case SVGA3D_PRIMITIVE_MAX: - debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_MAX\n"); + _debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_MAX\n"); break; default: - debug_printf("\t\t.primType = %i\n", (*cmd).primType); + _debug_printf("\t\t.primType = %i\n", (*cmd).primType); break; } - debug_printf("\t\t.primitiveCount = %u\n", (*cmd).primitiveCount); - debug_printf("\t\t.indexArray.surfaceId = %u\n", (*cmd).indexArray.surfaceId); - debug_printf("\t\t.indexArray.offset = %u\n", (*cmd).indexArray.offset); - debug_printf("\t\t.indexArray.stride = %u\n", (*cmd).indexArray.stride); - debug_printf("\t\t.indexWidth = %u\n", (*cmd).indexWidth); - debug_printf("\t\t.indexBias = %i\n", (*cmd).indexBias); + _debug_printf("\t\t.primitiveCount = %u\n", (*cmd).primitiveCount); + _debug_printf("\t\t.indexArray.surfaceId = %u\n", (*cmd).indexArray.surfaceId); + _debug_printf("\t\t.indexArray.offset = %u\n", (*cmd).indexArray.offset); + _debug_printf("\t\t.indexArray.stride = %u\n", (*cmd).indexArray.stride); + _debug_printf("\t\t.indexWidth = %u\n", (*cmd).indexWidth); + _debug_printf("\t\t.indexBias = %i\n", (*cmd).indexBias); } static void dump_SVGA3dCmdPresent(const SVGA3dCmdPresent *cmd) { - debug_printf("\t\t.sid = %u\n", (*cmd).sid); + _debug_printf("\t\t.sid = %u\n", (*cmd).sid); } static void dump_SVGA3dCmdSetRenderState(const SVGA3dCmdSetRenderState *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); } static void dump_SVGA3dCmdSurfaceStretchBlt(const SVGA3dCmdSurfaceStretchBlt *cmd) { - debug_printf("\t\t.src.sid = %u\n", (*cmd).src.sid); - debug_printf("\t\t.src.face = %u\n", (*cmd).src.face); - debug_printf("\t\t.src.mipmap = %u\n", (*cmd).src.mipmap); - debug_printf("\t\t.dest.sid = %u\n", (*cmd).dest.sid); - debug_printf("\t\t.dest.face = %u\n", (*cmd).dest.face); - debug_printf("\t\t.dest.mipmap = %u\n", (*cmd).dest.mipmap); - debug_printf("\t\t.boxSrc.x = %u\n", (*cmd).boxSrc.x); - debug_printf("\t\t.boxSrc.y = %u\n", (*cmd).boxSrc.y); - debug_printf("\t\t.boxSrc.z = %u\n", (*cmd).boxSrc.z); - debug_printf("\t\t.boxSrc.w = %u\n", (*cmd).boxSrc.w); - debug_printf("\t\t.boxSrc.h = %u\n", (*cmd).boxSrc.h); - debug_printf("\t\t.boxSrc.d = %u\n", (*cmd).boxSrc.d); - debug_printf("\t\t.boxDest.x = %u\n", (*cmd).boxDest.x); - debug_printf("\t\t.boxDest.y = %u\n", (*cmd).boxDest.y); - debug_printf("\t\t.boxDest.z = %u\n", (*cmd).boxDest.z); - debug_printf("\t\t.boxDest.w = %u\n", (*cmd).boxDest.w); - debug_printf("\t\t.boxDest.h = %u\n", (*cmd).boxDest.h); - debug_printf("\t\t.boxDest.d = %u\n", (*cmd).boxDest.d); + _debug_printf("\t\t.src.sid = %u\n", (*cmd).src.sid); + _debug_printf("\t\t.src.face = %u\n", (*cmd).src.face); + _debug_printf("\t\t.src.mipmap = %u\n", (*cmd).src.mipmap); + _debug_printf("\t\t.dest.sid = %u\n", (*cmd).dest.sid); + _debug_printf("\t\t.dest.face = %u\n", (*cmd).dest.face); + _debug_printf("\t\t.dest.mipmap = %u\n", (*cmd).dest.mipmap); + _debug_printf("\t\t.boxSrc.x = %u\n", (*cmd).boxSrc.x); + _debug_printf("\t\t.boxSrc.y = %u\n", (*cmd).boxSrc.y); + _debug_printf("\t\t.boxSrc.z = %u\n", (*cmd).boxSrc.z); + _debug_printf("\t\t.boxSrc.w = %u\n", (*cmd).boxSrc.w); + _debug_printf("\t\t.boxSrc.h = %u\n", (*cmd).boxSrc.h); + _debug_printf("\t\t.boxSrc.d = %u\n", (*cmd).boxSrc.d); + _debug_printf("\t\t.boxDest.x = %u\n", (*cmd).boxDest.x); + _debug_printf("\t\t.boxDest.y = %u\n", (*cmd).boxDest.y); + _debug_printf("\t\t.boxDest.z = %u\n", (*cmd).boxDest.z); + _debug_printf("\t\t.boxDest.w = %u\n", (*cmd).boxDest.w); + _debug_printf("\t\t.boxDest.h = %u\n", (*cmd).boxDest.h); + _debug_printf("\t\t.boxDest.d = %u\n", (*cmd).boxDest.d); switch((*cmd).mode) { case SVGA3D_STRETCH_BLT_POINT: - debug_printf("\t\t.mode = SVGA3D_STRETCH_BLT_POINT\n"); + _debug_printf("\t\t.mode = SVGA3D_STRETCH_BLT_POINT\n"); break; case SVGA3D_STRETCH_BLT_LINEAR: - debug_printf("\t\t.mode = SVGA3D_STRETCH_BLT_LINEAR\n"); + _debug_printf("\t\t.mode = SVGA3D_STRETCH_BLT_LINEAR\n"); break; case SVGA3D_STRETCH_BLT_MAX: - debug_printf("\t\t.mode = SVGA3D_STRETCH_BLT_MAX\n"); + _debug_printf("\t\t.mode = SVGA3D_STRETCH_BLT_MAX\n"); break; default: - debug_printf("\t\t.mode = %i\n", (*cmd).mode); + _debug_printf("\t\t.mode = %i\n", (*cmd).mode); break; } } @@ -1103,21 +1103,21 @@ dump_SVGA3dCmdSurfaceStretchBlt(const SVGA3dCmdSurfaceStretchBlt *cmd) static void dump_SVGA3dCmdSurfaceDMA(const SVGA3dCmdSurfaceDMA *cmd) { - debug_printf("\t\t.guest.ptr.gmrId = %u\n", (*cmd).guest.ptr.gmrId); - debug_printf("\t\t.guest.ptr.offset = %u\n", (*cmd).guest.ptr.offset); - debug_printf("\t\t.guest.pitch = %u\n", (*cmd).guest.pitch); - debug_printf("\t\t.host.sid = %u\n", (*cmd).host.sid); - debug_printf("\t\t.host.face = %u\n", (*cmd).host.face); - debug_printf("\t\t.host.mipmap = %u\n", (*cmd).host.mipmap); + _debug_printf("\t\t.guest.ptr.gmrId = %u\n", (*cmd).guest.ptr.gmrId); + _debug_printf("\t\t.guest.ptr.offset = %u\n", (*cmd).guest.ptr.offset); + _debug_printf("\t\t.guest.pitch = %u\n", (*cmd).guest.pitch); + _debug_printf("\t\t.host.sid = %u\n", (*cmd).host.sid); + _debug_printf("\t\t.host.face = %u\n", (*cmd).host.face); + _debug_printf("\t\t.host.mipmap = %u\n", (*cmd).host.mipmap); switch((*cmd).transfer) { case SVGA3D_WRITE_HOST_VRAM: - debug_printf("\t\t.transfer = SVGA3D_WRITE_HOST_VRAM\n"); + _debug_printf("\t\t.transfer = SVGA3D_WRITE_HOST_VRAM\n"); break; case SVGA3D_READ_HOST_VRAM: - debug_printf("\t\t.transfer = SVGA3D_READ_HOST_VRAM\n"); + _debug_printf("\t\t.transfer = SVGA3D_READ_HOST_VRAM\n"); break; default: - debug_printf("\t\t.transfer = %i\n", (*cmd).transfer); + _debug_printf("\t\t.transfer = %i\n", (*cmd).transfer); break; } } @@ -1125,107 +1125,107 @@ dump_SVGA3dCmdSurfaceDMA(const SVGA3dCmdSurfaceDMA *cmd) static void dump_SVGA3dCmdSurfaceDMASuffix(const SVGA3dCmdSurfaceDMASuffix *cmd) { - debug_printf("\t\t.suffixSize = %u\n", (*cmd).suffixSize); - debug_printf("\t\t.maximumOffset = %u\n", (*cmd).maximumOffset); - debug_printf("\t\t.flags.discard = %u\n", (*cmd).flags.discard); - debug_printf("\t\t.flags.unsynchronized = %u\n", (*cmd).flags.unsynchronized); + _debug_printf("\t\t.suffixSize = %u\n", (*cmd).suffixSize); + _debug_printf("\t\t.maximumOffset = %u\n", (*cmd).maximumOffset); + _debug_printf("\t\t.flags.discard = %u\n", (*cmd).flags.discard); + _debug_printf("\t\t.flags.unsynchronized = %u\n", (*cmd).flags.unsynchronized); } static void dump_SVGA3dCmdSetTransform(const SVGA3dCmdSetTransform *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); switch((*cmd).type) { case SVGA3D_TRANSFORM_INVALID: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_INVALID\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_INVALID\n"); break; case SVGA3D_TRANSFORM_WORLD: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_WORLD\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_WORLD\n"); break; case SVGA3D_TRANSFORM_VIEW: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_VIEW\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_VIEW\n"); break; case SVGA3D_TRANSFORM_PROJECTION: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_PROJECTION\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_PROJECTION\n"); break; case SVGA3D_TRANSFORM_TEXTURE0: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE0\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE0\n"); break; case SVGA3D_TRANSFORM_TEXTURE1: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE1\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE1\n"); break; case SVGA3D_TRANSFORM_TEXTURE2: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE2\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE2\n"); break; case SVGA3D_TRANSFORM_TEXTURE3: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE3\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE3\n"); break; case SVGA3D_TRANSFORM_TEXTURE4: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE4\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE4\n"); break; case SVGA3D_TRANSFORM_TEXTURE5: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE5\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE5\n"); break; case SVGA3D_TRANSFORM_TEXTURE6: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE6\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE6\n"); break; case SVGA3D_TRANSFORM_TEXTURE7: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE7\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE7\n"); break; case SVGA3D_TRANSFORM_WORLD1: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_WORLD1\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_WORLD1\n"); break; case SVGA3D_TRANSFORM_WORLD2: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_WORLD2\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_WORLD2\n"); break; case SVGA3D_TRANSFORM_WORLD3: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_WORLD3\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_WORLD3\n"); break; case SVGA3D_TRANSFORM_MAX: - debug_printf("\t\t.type = SVGA3D_TRANSFORM_MAX\n"); + _debug_printf("\t\t.type = SVGA3D_TRANSFORM_MAX\n"); break; default: - debug_printf("\t\t.type = %i\n", (*cmd).type); + _debug_printf("\t\t.type = %i\n", (*cmd).type); break; } - debug_printf("\t\t.matrix[0] = %f\n", (*cmd).matrix[0]); - debug_printf("\t\t.matrix[1] = %f\n", (*cmd).matrix[1]); - debug_printf("\t\t.matrix[2] = %f\n", (*cmd).matrix[2]); - debug_printf("\t\t.matrix[3] = %f\n", (*cmd).matrix[3]); - debug_printf("\t\t.matrix[4] = %f\n", (*cmd).matrix[4]); - debug_printf("\t\t.matrix[5] = %f\n", (*cmd).matrix[5]); - debug_printf("\t\t.matrix[6] = %f\n", (*cmd).matrix[6]); - debug_printf("\t\t.matrix[7] = %f\n", (*cmd).matrix[7]); - debug_printf("\t\t.matrix[8] = %f\n", (*cmd).matrix[8]); - debug_printf("\t\t.matrix[9] = %f\n", (*cmd).matrix[9]); - debug_printf("\t\t.matrix[10] = %f\n", (*cmd).matrix[10]); - debug_printf("\t\t.matrix[11] = %f\n", (*cmd).matrix[11]); - debug_printf("\t\t.matrix[12] = %f\n", (*cmd).matrix[12]); - debug_printf("\t\t.matrix[13] = %f\n", (*cmd).matrix[13]); - debug_printf("\t\t.matrix[14] = %f\n", (*cmd).matrix[14]); - debug_printf("\t\t.matrix[15] = %f\n", (*cmd).matrix[15]); + _debug_printf("\t\t.matrix[0] = %f\n", (*cmd).matrix[0]); + _debug_printf("\t\t.matrix[1] = %f\n", (*cmd).matrix[1]); + _debug_printf("\t\t.matrix[2] = %f\n", (*cmd).matrix[2]); + _debug_printf("\t\t.matrix[3] = %f\n", (*cmd).matrix[3]); + _debug_printf("\t\t.matrix[4] = %f\n", (*cmd).matrix[4]); + _debug_printf("\t\t.matrix[5] = %f\n", (*cmd).matrix[5]); + _debug_printf("\t\t.matrix[6] = %f\n", (*cmd).matrix[6]); + _debug_printf("\t\t.matrix[7] = %f\n", (*cmd).matrix[7]); + _debug_printf("\t\t.matrix[8] = %f\n", (*cmd).matrix[8]); + _debug_printf("\t\t.matrix[9] = %f\n", (*cmd).matrix[9]); + _debug_printf("\t\t.matrix[10] = %f\n", (*cmd).matrix[10]); + _debug_printf("\t\t.matrix[11] = %f\n", (*cmd).matrix[11]); + _debug_printf("\t\t.matrix[12] = %f\n", (*cmd).matrix[12]); + _debug_printf("\t\t.matrix[13] = %f\n", (*cmd).matrix[13]); + _debug_printf("\t\t.matrix[14] = %f\n", (*cmd).matrix[14]); + _debug_printf("\t\t.matrix[15] = %f\n", (*cmd).matrix[15]); } static void dump_SVGA3dCmdDestroyShader(const SVGA3dCmdDestroyShader *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); - debug_printf("\t\t.shid = %u\n", (*cmd).shid); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.shid = %u\n", (*cmd).shid); switch((*cmd).type) { case SVGA3D_SHADERTYPE_COMPILED_DX8: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_COMPILED_DX8\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_COMPILED_DX8\n"); break; case SVGA3D_SHADERTYPE_VS: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_VS\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_VS\n"); break; case SVGA3D_SHADERTYPE_PS: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_PS\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_PS\n"); break; case SVGA3D_SHADERTYPE_MAX: - debug_printf("\t\t.type = SVGA3D_SHADERTYPE_MAX\n"); + _debug_printf("\t\t.type = SVGA3D_SHADERTYPE_MAX\n"); break; default: - debug_printf("\t\t.type = %i\n", (*cmd).type); + _debug_printf("\t\t.type = %i\n", (*cmd).type); break; } } @@ -1233,187 +1233,213 @@ dump_SVGA3dCmdDestroyShader(const SVGA3dCmdDestroyShader *cmd) static void dump_SVGA3dCmdDestroyContext(const SVGA3dCmdDestroyContext *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); } static void dump_SVGA3dCmdClear(const SVGA3dCmdClear *cmd) { - debug_printf("\t\t.cid = %u\n", (*cmd).cid); + _debug_printf("\t\t.cid = %u\n", (*cmd).cid); switch((*cmd).clearFlag) { case SVGA3D_CLEAR_COLOR: - debug_printf("\t\t.clearFlag = SVGA3D_CLEAR_COLOR\n"); + _debug_printf("\t\t.clearFlag = SVGA3D_CLEAR_COLOR\n"); break; case SVGA3D_CLEAR_DEPTH: - debug_printf("\t\t.clearFlag = SVGA3D_CLEAR_DEPTH\n"); + _debug_printf("\t\t.clearFlag = SVGA3D_CLEAR_DEPTH\n"); break; case SVGA3D_CLEAR_STENCIL: - debug_printf("\t\t.clearFlag = SVGA3D_CLEAR_STENCIL\n"); + _debug_printf("\t\t.clearFlag = SVGA3D_CLEAR_STENCIL\n"); break; default: - debug_printf("\t\t.clearFlag = %i\n", (*cmd).clearFlag); + _debug_printf("\t\t.clearFlag = %i\n", (*cmd).clearFlag); break; } - debug_printf("\t\t.color = %u\n", (*cmd).color); - debug_printf("\t\t.depth = %f\n", (*cmd).depth); - debug_printf("\t\t.stencil = %u\n", (*cmd).stencil); + _debug_printf("\t\t.color = %u\n", (*cmd).color); + _debug_printf("\t\t.depth = %f\n", (*cmd).depth); + _debug_printf("\t\t.stencil = %u\n", (*cmd).stencil); } static void dump_SVGA3dCmdDefineSurface(const SVGA3dCmdDefineSurface *cmd) { - debug_printf("\t\t.sid = %u\n", (*cmd).sid); + _debug_printf("\t\t.sid = %u\n", (*cmd).sid); switch((*cmd).surfaceFlags) { case SVGA3D_SURFACE_CUBEMAP: - debug_printf("\t\t.surfaceFlags = SVGA3D_SURFACE_CUBEMAP\n"); + _debug_printf("\t\t.surfaceFlags = SVGA3D_SURFACE_CUBEMAP\n"); break; case SVGA3D_SURFACE_HINT_STATIC: - debug_printf("\t\t.surfaceFlags = SVGA3D_SURFACE_HINT_STATIC\n"); + _debug_printf("\t\t.surfaceFlags = SVGA3D_SURFACE_HINT_STATIC\n"); break; case SVGA3D_SURFACE_HINT_DYNAMIC: - debug_printf("\t\t.surfaceFlags = SVGA3D_SURFACE_HINT_DYNAMIC\n"); + _debug_printf("\t\t.surfaceFlags = SVGA3D_SURFACE_HINT_DYNAMIC\n"); break; case SVGA3D_SURFACE_HINT_INDEXBUFFER: - debug_printf("\t\t.surfaceFlags = SVGA3D_SURFACE_HINT_INDEXBUFFER\n"); + _debug_printf("\t\t.surfaceFlags = SVGA3D_SURFACE_HINT_INDEXBUFFER\n"); break; case SVGA3D_SURFACE_HINT_VERTEXBUFFER: - debug_printf("\t\t.surfaceFlags = SVGA3D_SURFACE_HINT_VERTEXBUFFER\n"); + _debug_printf("\t\t.surfaceFlags = SVGA3D_SURFACE_HINT_VERTEXBUFFER\n"); break; default: - debug_printf("\t\t.surfaceFlags = %i\n", (*cmd).surfaceFlags); + _debug_printf("\t\t.surfaceFlags = %i\n", (*cmd).surfaceFlags); break; } switch((*cmd).format) { case SVGA3D_FORMAT_INVALID: - debug_printf("\t\t.format = SVGA3D_FORMAT_INVALID\n"); + _debug_printf("\t\t.format = SVGA3D_FORMAT_INVALID\n"); break; case SVGA3D_X8R8G8B8: - debug_printf("\t\t.format = SVGA3D_X8R8G8B8\n"); + _debug_printf("\t\t.format = SVGA3D_X8R8G8B8\n"); break; case SVGA3D_A8R8G8B8: - debug_printf("\t\t.format = SVGA3D_A8R8G8B8\n"); + _debug_printf("\t\t.format = SVGA3D_A8R8G8B8\n"); break; case SVGA3D_R5G6B5: - debug_printf("\t\t.format = SVGA3D_R5G6B5\n"); + _debug_printf("\t\t.format = SVGA3D_R5G6B5\n"); break; case SVGA3D_X1R5G5B5: - debug_printf("\t\t.format = SVGA3D_X1R5G5B5\n"); + _debug_printf("\t\t.format = SVGA3D_X1R5G5B5\n"); break; case SVGA3D_A1R5G5B5: - debug_printf("\t\t.format = SVGA3D_A1R5G5B5\n"); + _debug_printf("\t\t.format = SVGA3D_A1R5G5B5\n"); break; case SVGA3D_A4R4G4B4: - debug_printf("\t\t.format = SVGA3D_A4R4G4B4\n"); + _debug_printf("\t\t.format = SVGA3D_A4R4G4B4\n"); break; case SVGA3D_Z_D32: - debug_printf("\t\t.format = SVGA3D_Z_D32\n"); + _debug_printf("\t\t.format = SVGA3D_Z_D32\n"); break; case SVGA3D_Z_D16: - debug_printf("\t\t.format = SVGA3D_Z_D16\n"); + _debug_printf("\t\t.format = SVGA3D_Z_D16\n"); break; case SVGA3D_Z_D24S8: - debug_printf("\t\t.format = SVGA3D_Z_D24S8\n"); + _debug_printf("\t\t.format = SVGA3D_Z_D24S8\n"); break; case SVGA3D_Z_D15S1: - debug_printf("\t\t.format = SVGA3D_Z_D15S1\n"); + _debug_printf("\t\t.format = SVGA3D_Z_D15S1\n"); break; case SVGA3D_LUMINANCE8: - debug_printf("\t\t.format = SVGA3D_LUMINANCE8\n"); + _debug_printf("\t\t.format = SVGA3D_LUMINANCE8\n"); break; case SVGA3D_LUMINANCE4_ALPHA4: - debug_printf("\t\t.format = SVGA3D_LUMINANCE4_ALPHA4\n"); + _debug_printf("\t\t.format = SVGA3D_LUMINANCE4_ALPHA4\n"); break; case SVGA3D_LUMINANCE16: - debug_printf("\t\t.format = SVGA3D_LUMINANCE16\n"); + _debug_printf("\t\t.format = SVGA3D_LUMINANCE16\n"); break; case SVGA3D_LUMINANCE8_ALPHA8: - debug_printf("\t\t.format = SVGA3D_LUMINANCE8_ALPHA8\n"); + _debug_printf("\t\t.format = SVGA3D_LUMINANCE8_ALPHA8\n"); break; case SVGA3D_DXT1: - debug_printf("\t\t.format = SVGA3D_DXT1\n"); + _debug_printf("\t\t.format = SVGA3D_DXT1\n"); break; case SVGA3D_DXT2: - debug_printf("\t\t.format = SVGA3D_DXT2\n"); + _debug_printf("\t\t.format = SVGA3D_DXT2\n"); break; case SVGA3D_DXT3: - debug_printf("\t\t.format = SVGA3D_DXT3\n"); + _debug_printf("\t\t.format = SVGA3D_DXT3\n"); break; case SVGA3D_DXT4: - debug_printf("\t\t.format = SVGA3D_DXT4\n"); + _debug_printf("\t\t.format = SVGA3D_DXT4\n"); break; case SVGA3D_DXT5: - debug_printf("\t\t.format = SVGA3D_DXT5\n"); + _debug_printf("\t\t.format = SVGA3D_DXT5\n"); break; case SVGA3D_BUMPU8V8: - debug_printf("\t\t.format = SVGA3D_BUMPU8V8\n"); + _debug_printf("\t\t.format = SVGA3D_BUMPU8V8\n"); break; case SVGA3D_BUMPL6V5U5: - debug_printf("\t\t.format = SVGA3D_BUMPL6V5U5\n"); + _debug_printf("\t\t.format = SVGA3D_BUMPL6V5U5\n"); break; case SVGA3D_BUMPX8L8V8U8: - debug_printf("\t\t.format = SVGA3D_BUMPX8L8V8U8\n"); + _debug_printf("\t\t.format = SVGA3D_BUMPX8L8V8U8\n"); break; case SVGA3D_BUMPL8V8U8: - debug_printf("\t\t.format = SVGA3D_BUMPL8V8U8\n"); + _debug_printf("\t\t.format = SVGA3D_BUMPL8V8U8\n"); break; case SVGA3D_ARGB_S10E5: - debug_printf("\t\t.format = SVGA3D_ARGB_S10E5\n"); + _debug_printf("\t\t.format = SVGA3D_ARGB_S10E5\n"); break; case SVGA3D_ARGB_S23E8: - debug_printf("\t\t.format = SVGA3D_ARGB_S23E8\n"); + _debug_printf("\t\t.format = SVGA3D_ARGB_S23E8\n"); break; case SVGA3D_A2R10G10B10: - debug_printf("\t\t.format = SVGA3D_A2R10G10B10\n"); + _debug_printf("\t\t.format = SVGA3D_A2R10G10B10\n"); break; case SVGA3D_V8U8: - debug_printf("\t\t.format = SVGA3D_V8U8\n"); + _debug_printf("\t\t.format = SVGA3D_V8U8\n"); break; case SVGA3D_Q8W8V8U8: - debug_printf("\t\t.format = SVGA3D_Q8W8V8U8\n"); + _debug_printf("\t\t.format = SVGA3D_Q8W8V8U8\n"); break; case SVGA3D_CxV8U8: - debug_printf("\t\t.format = SVGA3D_CxV8U8\n"); + _debug_printf("\t\t.format = SVGA3D_CxV8U8\n"); break; case SVGA3D_X8L8V8U8: - debug_printf("\t\t.format = SVGA3D_X8L8V8U8\n"); + _debug_printf("\t\t.format = SVGA3D_X8L8V8U8\n"); break; case SVGA3D_A2W10V10U10: - debug_printf("\t\t.format = SVGA3D_A2W10V10U10\n"); + _debug_printf("\t\t.format = SVGA3D_A2W10V10U10\n"); break; case SVGA3D_ALPHA8: - debug_printf("\t\t.format = SVGA3D_ALPHA8\n"); + _debug_printf("\t\t.format = SVGA3D_ALPHA8\n"); break; case SVGA3D_R_S10E5: - debug_printf("\t\t.format = SVGA3D_R_S10E5\n"); + _debug_printf("\t\t.format = SVGA3D_R_S10E5\n"); break; case SVGA3D_R_S23E8: - debug_printf("\t\t.format = SVGA3D_R_S23E8\n"); + _debug_printf("\t\t.format = SVGA3D_R_S23E8\n"); break; case SVGA3D_RG_S10E5: - debug_printf("\t\t.format = SVGA3D_RG_S10E5\n"); + _debug_printf("\t\t.format = SVGA3D_RG_S10E5\n"); break; case SVGA3D_RG_S23E8: - debug_printf("\t\t.format = SVGA3D_RG_S23E8\n"); + _debug_printf("\t\t.format = SVGA3D_RG_S23E8\n"); break; case SVGA3D_BUFFER: - debug_printf("\t\t.format = SVGA3D_BUFFER\n"); + _debug_printf("\t\t.format = SVGA3D_BUFFER\n"); break; case SVGA3D_Z_D24X8: - debug_printf("\t\t.format = SVGA3D_Z_D24X8\n"); + _debug_printf("\t\t.format = SVGA3D_Z_D24X8\n"); break; case SVGA3D_FORMAT_MAX: - debug_printf("\t\t.format = SVGA3D_FORMAT_MAX\n"); + _debug_printf("\t\t.format = SVGA3D_FORMAT_MAX\n"); break; default: - debug_printf("\t\t.format = %i\n", (*cmd).format); + _debug_printf("\t\t.format = %i\n", (*cmd).format); break; } - debug_printf("\t\t.face[0].numMipLevels = %u\n", (*cmd).face[0].numMipLevels); - debug_printf("\t\t.face[1].numMipLevels = %u\n", (*cmd).face[1].numMipLevels); - debug_printf("\t\t.face[2].numMipLevels = %u\n", (*cmd).face[2].numMipLevels); - debug_printf("\t\t.face[3].numMipLevels = %u\n", (*cmd).face[3].numMipLevels); - debug_printf("\t\t.face[4].numMipLevels = %u\n", (*cmd).face[4].numMipLevels); - debug_printf("\t\t.face[5].numMipLevels = %u\n", (*cmd).face[5].numMipLevels); + _debug_printf("\t\t.face[0].numMipLevels = %u\n", (*cmd).face[0].numMipLevels); + _debug_printf("\t\t.face[1].numMipLevels = %u\n", (*cmd).face[1].numMipLevels); + _debug_printf("\t\t.face[2].numMipLevels = %u\n", (*cmd).face[2].numMipLevels); + _debug_printf("\t\t.face[3].numMipLevels = %u\n", (*cmd).face[3].numMipLevels); + _debug_printf("\t\t.face[4].numMipLevels = %u\n", (*cmd).face[4].numMipLevels); + _debug_printf("\t\t.face[5].numMipLevels = %u\n", (*cmd).face[5].numMipLevels); +} + +static void +dump_SVGASignedRect(const SVGASignedRect *cmd) +{ + _debug_printf("\t\t.left = %i\n", (*cmd).left); + _debug_printf("\t\t.top = %i\n", (*cmd).top); + _debug_printf("\t\t.right = %i\n", (*cmd).right); + _debug_printf("\t\t.bottom = %i\n", (*cmd).bottom); +} + +static void +dump_SVGA3dCmdBlitSurfaceToScreen(const SVGA3dCmdBlitSurfaceToScreen *cmd) +{ + _debug_printf("\t\t.srcImage.sid = %u\n", (*cmd).srcImage.sid); + _debug_printf("\t\t.srcImage.face = %u\n", (*cmd).srcImage.face); + _debug_printf("\t\t.srcImage.mipmap = %u\n", (*cmd).srcImage.mipmap); + _debug_printf("\t\t.srcRect.left = %i\n", (*cmd).srcRect.left); + _debug_printf("\t\t.srcRect.top = %i\n", (*cmd).srcRect.top); + _debug_printf("\t\t.srcRect.right = %i\n", (*cmd).srcRect.right); + _debug_printf("\t\t.srcRect.bottom = %i\n", (*cmd).srcRect.bottom); + _debug_printf("\t\t.destScreenId = %u\n", (*cmd).destScreenId); + _debug_printf("\t\t.destRect.left = %i\n", (*cmd).destRect.left); + _debug_printf("\t\t.destRect.top = %i\n", (*cmd).destRect.top); + _debug_printf("\t\t.destRect.right = %i\n", (*cmd).destRect.right); + _debug_printf("\t\t.destRect.bottom = %i\n", (*cmd).destRect.bottom); } @@ -1438,7 +1464,7 @@ svga_dump_commands(const void *commands, uint32_t size) switch(cmd_id) { case SVGA_3D_CMD_SURFACE_DEFINE: - debug_printf("\tSVGA_3D_CMD_SURFACE_DEFINE\n"); + _debug_printf("\tSVGA_3D_CMD_SURFACE_DEFINE\n"); { const SVGA3dCmdDefineSurface *cmd = (const SVGA3dCmdDefineSurface *)body; dump_SVGA3dCmdDefineSurface(cmd); @@ -1450,7 +1476,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SURFACE_DESTROY: - debug_printf("\tSVGA_3D_CMD_SURFACE_DESTROY\n"); + _debug_printf("\tSVGA_3D_CMD_SURFACE_DESTROY\n"); { const SVGA3dCmdDestroySurface *cmd = (const SVGA3dCmdDestroySurface *)body; dump_SVGA3dCmdDestroySurface(cmd); @@ -1458,7 +1484,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SURFACE_COPY: - debug_printf("\tSVGA_3D_CMD_SURFACE_COPY\n"); + _debug_printf("\tSVGA_3D_CMD_SURFACE_COPY\n"); { const SVGA3dCmdSurfaceCopy *cmd = (const SVGA3dCmdSurfaceCopy *)body; dump_SVGA3dCmdSurfaceCopy(cmd); @@ -1470,7 +1496,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SURFACE_STRETCHBLT: - debug_printf("\tSVGA_3D_CMD_SURFACE_STRETCHBLT\n"); + _debug_printf("\tSVGA_3D_CMD_SURFACE_STRETCHBLT\n"); { const SVGA3dCmdSurfaceStretchBlt *cmd = (const SVGA3dCmdSurfaceStretchBlt *)body; dump_SVGA3dCmdSurfaceStretchBlt(cmd); @@ -1478,7 +1504,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SURFACE_DMA: - debug_printf("\tSVGA_3D_CMD_SURFACE_DMA\n"); + _debug_printf("\tSVGA_3D_CMD_SURFACE_DMA\n"); { const SVGA3dCmdSurfaceDMA *cmd = (const SVGA3dCmdSurfaceDMA *)body; dump_SVGA3dCmdSurfaceDMA(cmd); @@ -1494,7 +1520,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_CONTEXT_DEFINE: - debug_printf("\tSVGA_3D_CMD_CONTEXT_DEFINE\n"); + _debug_printf("\tSVGA_3D_CMD_CONTEXT_DEFINE\n"); { const SVGA3dCmdDefineContext *cmd = (const SVGA3dCmdDefineContext *)body; dump_SVGA3dCmdDefineContext(cmd); @@ -1502,7 +1528,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_CONTEXT_DESTROY: - debug_printf("\tSVGA_3D_CMD_CONTEXT_DESTROY\n"); + _debug_printf("\tSVGA_3D_CMD_CONTEXT_DESTROY\n"); { const SVGA3dCmdDestroyContext *cmd = (const SVGA3dCmdDestroyContext *)body; dump_SVGA3dCmdDestroyContext(cmd); @@ -1510,7 +1536,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SETTRANSFORM: - debug_printf("\tSVGA_3D_CMD_SETTRANSFORM\n"); + _debug_printf("\tSVGA_3D_CMD_SETTRANSFORM\n"); { const SVGA3dCmdSetTransform *cmd = (const SVGA3dCmdSetTransform *)body; dump_SVGA3dCmdSetTransform(cmd); @@ -1518,7 +1544,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SETZRANGE: - debug_printf("\tSVGA_3D_CMD_SETZRANGE\n"); + _debug_printf("\tSVGA_3D_CMD_SETZRANGE\n"); { const SVGA3dCmdSetZRange *cmd = (const SVGA3dCmdSetZRange *)body; dump_SVGA3dCmdSetZRange(cmd); @@ -1526,7 +1552,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SETRENDERSTATE: - debug_printf("\tSVGA_3D_CMD_SETRENDERSTATE\n"); + _debug_printf("\tSVGA_3D_CMD_SETRENDERSTATE\n"); { const SVGA3dCmdSetRenderState *cmd = (const SVGA3dCmdSetRenderState *)body; dump_SVGA3dCmdSetRenderState(cmd); @@ -1538,7 +1564,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SETRENDERTARGET: - debug_printf("\tSVGA_3D_CMD_SETRENDERTARGET\n"); + _debug_printf("\tSVGA_3D_CMD_SETRENDERTARGET\n"); { const SVGA3dCmdSetRenderTarget *cmd = (const SVGA3dCmdSetRenderTarget *)body; dump_SVGA3dCmdSetRenderTarget(cmd); @@ -1546,7 +1572,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SETTEXTURESTATE: - debug_printf("\tSVGA_3D_CMD_SETTEXTURESTATE\n"); + _debug_printf("\tSVGA_3D_CMD_SETTEXTURESTATE\n"); { const SVGA3dCmdSetTextureState *cmd = (const SVGA3dCmdSetTextureState *)body; dump_SVGA3dCmdSetTextureState(cmd); @@ -1558,7 +1584,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SETMATERIAL: - debug_printf("\tSVGA_3D_CMD_SETMATERIAL\n"); + _debug_printf("\tSVGA_3D_CMD_SETMATERIAL\n"); { const SVGA3dCmdSetMaterial *cmd = (const SVGA3dCmdSetMaterial *)body; dump_SVGA3dCmdSetMaterial(cmd); @@ -1566,7 +1592,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SETLIGHTDATA: - debug_printf("\tSVGA_3D_CMD_SETLIGHTDATA\n"); + _debug_printf("\tSVGA_3D_CMD_SETLIGHTDATA\n"); { const SVGA3dCmdSetLightData *cmd = (const SVGA3dCmdSetLightData *)body; dump_SVGA3dCmdSetLightData(cmd); @@ -1574,7 +1600,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SETLIGHTENABLED: - debug_printf("\tSVGA_3D_CMD_SETLIGHTENABLED\n"); + _debug_printf("\tSVGA_3D_CMD_SETLIGHTENABLED\n"); { const SVGA3dCmdSetLightEnabled *cmd = (const SVGA3dCmdSetLightEnabled *)body; dump_SVGA3dCmdSetLightEnabled(cmd); @@ -1582,7 +1608,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SETVIEWPORT: - debug_printf("\tSVGA_3D_CMD_SETVIEWPORT\n"); + _debug_printf("\tSVGA_3D_CMD_SETVIEWPORT\n"); { const SVGA3dCmdSetViewport *cmd = (const SVGA3dCmdSetViewport *)body; dump_SVGA3dCmdSetViewport(cmd); @@ -1590,7 +1616,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SETCLIPPLANE: - debug_printf("\tSVGA_3D_CMD_SETCLIPPLANE\n"); + _debug_printf("\tSVGA_3D_CMD_SETCLIPPLANE\n"); { const SVGA3dCmdSetClipPlane *cmd = (const SVGA3dCmdSetClipPlane *)body; dump_SVGA3dCmdSetClipPlane(cmd); @@ -1598,7 +1624,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_CLEAR: - debug_printf("\tSVGA_3D_CMD_CLEAR\n"); + _debug_printf("\tSVGA_3D_CMD_CLEAR\n"); { const SVGA3dCmdClear *cmd = (const SVGA3dCmdClear *)body; dump_SVGA3dCmdClear(cmd); @@ -1610,7 +1636,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_PRESENT: - debug_printf("\tSVGA_3D_CMD_PRESENT\n"); + _debug_printf("\tSVGA_3D_CMD_PRESENT\n"); { const SVGA3dCmdPresent *cmd = (const SVGA3dCmdPresent *)body; dump_SVGA3dCmdPresent(cmd); @@ -1622,7 +1648,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SHADER_DEFINE: - debug_printf("\tSVGA_3D_CMD_SHADER_DEFINE\n"); + _debug_printf("\tSVGA_3D_CMD_SHADER_DEFINE\n"); { const SVGA3dCmdDefineShader *cmd = (const SVGA3dCmdDefineShader *)body; dump_SVGA3dCmdDefineShader(cmd); @@ -1634,7 +1660,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SHADER_DESTROY: - debug_printf("\tSVGA_3D_CMD_SHADER_DESTROY\n"); + _debug_printf("\tSVGA_3D_CMD_SHADER_DESTROY\n"); { const SVGA3dCmdDestroyShader *cmd = (const SVGA3dCmdDestroyShader *)body; dump_SVGA3dCmdDestroyShader(cmd); @@ -1642,7 +1668,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SET_SHADER: - debug_printf("\tSVGA_3D_CMD_SET_SHADER\n"); + _debug_printf("\tSVGA_3D_CMD_SET_SHADER\n"); { const SVGA3dCmdSetShader *cmd = (const SVGA3dCmdSetShader *)body; dump_SVGA3dCmdSetShader(cmd); @@ -1650,7 +1676,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SET_SHADER_CONST: - debug_printf("\tSVGA_3D_CMD_SET_SHADER_CONST\n"); + _debug_printf("\tSVGA_3D_CMD_SET_SHADER_CONST\n"); { const SVGA3dCmdSetShaderConst *cmd = (const SVGA3dCmdSetShaderConst *)body; dump_SVGA3dCmdSetShaderConst(cmd); @@ -1658,7 +1684,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_DRAW_PRIMITIVES: - debug_printf("\tSVGA_3D_CMD_DRAW_PRIMITIVES\n"); + _debug_printf("\tSVGA_3D_CMD_DRAW_PRIMITIVES\n"); { const SVGA3dCmdDrawPrimitives *cmd = (const SVGA3dCmdDrawPrimitives *)body; unsigned i, j; @@ -1679,7 +1705,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_SETSCISSORRECT: - debug_printf("\tSVGA_3D_CMD_SETSCISSORRECT\n"); + _debug_printf("\tSVGA_3D_CMD_SETSCISSORRECT\n"); { const SVGA3dCmdSetScissorRect *cmd = (const SVGA3dCmdSetScissorRect *)body; dump_SVGA3dCmdSetScissorRect(cmd); @@ -1687,7 +1713,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_BEGIN_QUERY: - debug_printf("\tSVGA_3D_CMD_BEGIN_QUERY\n"); + _debug_printf("\tSVGA_3D_CMD_BEGIN_QUERY\n"); { const SVGA3dCmdBeginQuery *cmd = (const SVGA3dCmdBeginQuery *)body; dump_SVGA3dCmdBeginQuery(cmd); @@ -1695,7 +1721,7 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_END_QUERY: - debug_printf("\tSVGA_3D_CMD_END_QUERY\n"); + _debug_printf("\tSVGA_3D_CMD_END_QUERY\n"); { const SVGA3dCmdEndQuery *cmd = (const SVGA3dCmdEndQuery *)body; dump_SVGA3dCmdEndQuery(cmd); @@ -1703,32 +1729,44 @@ svga_dump_commands(const void *commands, uint32_t size) } break; case SVGA_3D_CMD_WAIT_FOR_QUERY: - debug_printf("\tSVGA_3D_CMD_WAIT_FOR_QUERY\n"); + _debug_printf("\tSVGA_3D_CMD_WAIT_FOR_QUERY\n"); { const SVGA3dCmdWaitForQuery *cmd = (const SVGA3dCmdWaitForQuery *)body; dump_SVGA3dCmdWaitForQuery(cmd); body = (const uint8_t *)&cmd[1]; } break; + case SVGA_3D_CMD_BLIT_SURFACE_TO_SCREEN: + _debug_printf("\tSVGA_3D_CMD_BLIT_SURFACE_TO_SCREEN\n"); + { + const SVGA3dCmdBlitSurfaceToScreen *cmd = (const SVGA3dCmdBlitSurfaceToScreen *)body; + dump_SVGA3dCmdBlitSurfaceToScreen(cmd); + body = (const uint8_t *)&cmd[1]; + while(body + sizeof(SVGASignedRect) <= next) { + dump_SVGASignedRect((const SVGASignedRect *)body); + body += sizeof(SVGASignedRect); + } + } + break; default: - debug_printf("\t0x%08x\n", cmd_id); + _debug_printf("\t0x%08x\n", cmd_id); break; } while(body + sizeof(uint32_t) <= next) { - debug_printf("\t\t0x%08x\n", *(const uint32_t *)body); + _debug_printf("\t\t0x%08x\n", *(const uint32_t *)body); body += sizeof(uint32_t); } while(body + sizeof(uint32_t) <= next) - debug_printf("\t\t0x%02x\n", *body++); + _debug_printf("\t\t0x%02x\n", *body++); } else if(cmd_id == SVGA_CMD_FENCE) { - debug_printf("\tSVGA_CMD_FENCE\n"); - debug_printf("\t\t0x%08x\n", ((const uint32_t *)next)[1]); + _debug_printf("\tSVGA_CMD_FENCE\n"); + _debug_printf("\t\t0x%08x\n", ((const uint32_t *)next)[1]); next += 2*sizeof(uint32_t); } else { - debug_printf("\t0x%08x\n", cmd_id); + _debug_printf("\t0x%08x\n", cmd_id); next += sizeof(uint32_t); } } diff --git a/src/gallium/drivers/svga/svgadump/svga_dump.py b/src/gallium/drivers/svga/svgadump/svga_dump.py index 288e753296..a1ada29ef8 100755 --- a/src/gallium/drivers/svga/svgadump/svga_dump.py +++ b/src/gallium/drivers/svga/svgadump/svga_dump.py @@ -71,14 +71,14 @@ class decl_dumper_t(decl_visitor.decl_visitor_t): print ' switch(%s) {' % ("(*cmd)" + self._instance,) for name, value in self.decl.values: print ' case %s:' % (name,) - print ' debug_printf("\\t\\t%s = %s\\n");' % (self._instance, name) + print ' _debug_printf("\\t\\t%s = %s\\n");' % (self._instance, name) print ' break;' print ' default:' - print ' debug_printf("\\t\\t%s = %%i\\n", %s);' % (self._instance, "(*cmd)" + self._instance) + print ' _debug_printf("\\t\\t%s = %%i\\n", %s);' % (self._instance, "(*cmd)" + self._instance) print ' break;' print ' }' else: - print ' debug_printf("\\t\\t%s = %%i\\n", %s);' % (self._instance, "(*cmd)" + self._instance) + print ' _debug_printf("\\t\\t%s = %%i\\n", %s);' % (self._instance, "(*cmd)" + self._instance) def dump_decl(instance, decl): @@ -154,7 +154,7 @@ class type_dumper_t(type_visitor.type_visitor_t): dump_decl(self.instance, decl) def print_instance(self, format): - print ' debug_printf("\\t\\t%s = %s\\n", %s);' % (self.instance, format, "(*cmd)" + self.instance) + print ' _debug_printf("\\t\\t%s = %s\\n", %s);' % (self.instance, format, "(*cmd)" + self.instance) def dump_type(instance, type_): @@ -202,6 +202,7 @@ cmds = [ ('SVGA_3D_CMD_END_QUERY', 'SVGA3dCmdEndQuery', (), None), ('SVGA_3D_CMD_WAIT_FOR_QUERY', 'SVGA3dCmdWaitForQuery', (), None), #('SVGA_3D_CMD_PRESENT_READBACK', None, (), None), + ('SVGA_3D_CMD_BLIT_SURFACE_TO_SCREEN', 'SVGA3dCmdBlitSurfaceToScreen', (), 'SVGASignedRect'), ] def dump_cmds(): @@ -230,7 +231,7 @@ svga_dump_commands(const void *commands, uint32_t size) indexes = 'ijklmn' for id, header, body, footer in cmds: print ' case %s:' % id - print ' debug_printf("\\t%s\\n");' % id + print ' _debug_printf("\\t%s\\n");' % id print ' {' print ' const %s *cmd = (const %s *)body;' % (header, header) if len(body): @@ -255,25 +256,25 @@ svga_dump_commands(const void *commands, uint32_t size) print ' }' print ' break;' print ' default:' - print ' debug_printf("\\t0x%08x\\n", cmd_id);' + print ' _debug_printf("\\t0x%08x\\n", cmd_id);' print ' break;' print ' }' print r''' while(body + sizeof(uint32_t) <= next) { - debug_printf("\t\t0x%08x\n", *(const uint32_t *)body); + _debug_printf("\t\t0x%08x\n", *(const uint32_t *)body); body += sizeof(uint32_t); } while(body + sizeof(uint32_t) <= next) - debug_printf("\t\t0x%02x\n", *body++); + _debug_printf("\t\t0x%02x\n", *body++); } else if(cmd_id == SVGA_CMD_FENCE) { - debug_printf("\tSVGA_CMD_FENCE\n"); - debug_printf("\t\t0x%08x\n", ((const uint32_t *)next)[1]); + _debug_printf("\tSVGA_CMD_FENCE\n"); + _debug_printf("\t\t0x%08x\n", ((const uint32_t *)next)[1]); next += 2*sizeof(uint32_t); } else { - debug_printf("\t0x%08x\n", cmd_id); + _debug_printf("\t0x%08x\n", cmd_id); next += sizeof(uint32_t); } } @@ -294,18 +295,18 @@ def main(): print '#include "svga_shader_dump.h"' print '#include "svga3d_reg.h"' print - print '#include "pipe/p_debug.h"' + print '#include "util/u_debug.h"' print '#include "svga_dump.h"' print config = parser.config_t( - include_paths = ['include'], + include_paths = ['../../../include', '../include'], compiler = 'gcc', ) headers = [ - 'include/svga_types.h', - 'include/svga3d_reg.h', + 'svga_types.h', + 'svga3d_reg.h', ] decls = parser.parse(headers, config, parser.COMPILATION_MODE.ALL_AT_ONCE) diff --git a/src/gallium/drivers/svga/svgadump/svga_shader_dump.c b/src/gallium/drivers/svga/svgadump/svga_shader_dump.c index b0e7fdf378..70e27d86d3 100644 --- a/src/gallium/drivers/svga/svgadump/svga_shader_dump.c +++ b/src/gallium/drivers/svga/svgadump/svga_shader_dump.c @@ -50,16 +50,16 @@ static void dump_op( struct sh_op op, const char *mnemonic ) assert( op.is_reg == 0 ); if (op.coissue) - debug_printf( "+" ); - debug_printf( "%s", mnemonic ); + _debug_printf( "+" ); + _debug_printf( "%s", mnemonic ); switch (op.control) { case 0: break; case SVGA3DOPCONT_PROJECT: - debug_printf( "p" ); + _debug_printf( "p" ); break; case SVGA3DOPCONT_BIAS: - debug_printf( "b" ); + _debug_printf( "b" ); break; default: assert( 0 ); @@ -72,28 +72,28 @@ static void dump_comp_op( struct sh_op op, const char *mnemonic ) assert( op.is_reg == 0 ); if (op.coissue) - debug_printf( "+" ); - debug_printf( "%s", mnemonic ); + _debug_printf( "+" ); + _debug_printf( "%s", mnemonic ); switch (op.control) { case SVGA3DOPCOMP_RESERVED0: break; case SVGA3DOPCOMP_GT: - debug_printf("_gt"); + _debug_printf("_gt"); break; case SVGA3DOPCOMP_EQ: - debug_printf("_eq"); + _debug_printf("_eq"); break; case SVGA3DOPCOMP_GE: - debug_printf("_ge"); + _debug_printf("_ge"); break; case SVGA3DOPCOMP_LT: - debug_printf("_lt"); + _debug_printf("_lt"); break; case SVGA3DOPCOMPC_NE: - debug_printf("_ne"); + _debug_printf("_ne"); break; case SVGA3DOPCOMP_LE: - debug_printf("_le"); + _debug_printf("_le"); break; case SVGA3DOPCOMP_RESERVED1: default: @@ -109,93 +109,93 @@ static void dump_reg( struct sh_reg reg, struct sh_srcreg *indreg, const struct switch (sh_reg_type( reg )) { case SVGA3DREG_TEMP: - debug_printf( "r%u", reg.number ); + _debug_printf( "r%u", reg.number ); break; case SVGA3DREG_INPUT: - debug_printf( "v%u", reg.number ); + _debug_printf( "v%u", reg.number ); break; case SVGA3DREG_CONST: if (reg.relative) { if (sh_srcreg_type( *indreg ) == SVGA3DREG_LOOP) - debug_printf( "c[aL+%u]", reg.number ); + _debug_printf( "c[aL+%u]", reg.number ); else - debug_printf( "c[a%u.x+%u]", indreg->number, reg.number ); + _debug_printf( "c[a%u.x+%u]", indreg->number, reg.number ); } else - debug_printf( "c%u", reg.number ); + _debug_printf( "c%u", reg.number ); break; case SVGA3DREG_ADDR: /* VS */ /* SVGA3DREG_TEXTURE */ /* PS */ if (di->is_ps) - debug_printf( "t%u", reg.number ); + _debug_printf( "t%u", reg.number ); else - debug_printf( "a%u", reg.number ); + _debug_printf( "a%u", reg.number ); break; case SVGA3DREG_RASTOUT: switch (reg.number) { case 0 /*POSITION*/: - debug_printf( "oPos" ); + _debug_printf( "oPos" ); break; case 1 /*FOG*/: - debug_printf( "oFog" ); + _debug_printf( "oFog" ); break; case 2 /*POINT_SIZE*/: - debug_printf( "oPts" ); + _debug_printf( "oPts" ); break; default: assert( 0 ); - debug_printf( "???" ); + _debug_printf( "???" ); } break; case SVGA3DREG_ATTROUT: assert( reg.number < 2 ); - debug_printf( "oD%u", reg.number ); + _debug_printf( "oD%u", reg.number ); break; case SVGA3DREG_TEXCRDOUT: /* SVGA3DREG_OUTPUT */ - debug_printf( "oT%u", reg.number ); + _debug_printf( "oT%u", reg.number ); break; case SVGA3DREG_COLOROUT: - debug_printf( "oC%u", reg.number ); + _debug_printf( "oC%u", reg.number ); break; case SVGA3DREG_DEPTHOUT: - debug_printf( "oD%u", reg.number ); + _debug_printf( "oD%u", reg.number ); break; case SVGA3DREG_SAMPLER: - debug_printf( "s%u", reg.number ); + _debug_printf( "s%u", reg.number ); break; case SVGA3DREG_CONSTBOOL: assert( !reg.relative ); - debug_printf( "b%u", reg.number ); + _debug_printf( "b%u", reg.number ); break; case SVGA3DREG_CONSTINT: assert( !reg.relative ); - debug_printf( "i%u", reg.number ); + _debug_printf( "i%u", reg.number ); break; case SVGA3DREG_LOOP: assert( reg.number == 0 ); - debug_printf( "aL" ); + _debug_printf( "aL" ); break; case SVGA3DREG_MISCTYPE: switch (reg.number) { case SVGA3DMISCREG_POSITION: - debug_printf( "vPos" ); + _debug_printf( "vPos" ); break; case SVGA3DMISCREG_FACE: - debug_printf( "vFace" ); + _debug_printf( "vFace" ); break; default: assert(0); @@ -204,46 +204,46 @@ static void dump_reg( struct sh_reg reg, struct sh_srcreg *indreg, const struct break; case SVGA3DREG_LABEL: - debug_printf( "l%u", reg.number ); + _debug_printf( "l%u", reg.number ); break; case SVGA3DREG_PREDICATE: - debug_printf( "p%u", reg.number ); + _debug_printf( "p%u", reg.number ); break; default: assert( 0 ); - debug_printf( "???" ); + _debug_printf( "???" ); } } static void dump_cdata( struct sh_cdata cdata ) { - debug_printf( "%f, %f, %f, %f", cdata.xyzw[0], cdata.xyzw[1], cdata.xyzw[2], cdata.xyzw[3] ); + _debug_printf( "%f, %f, %f, %f", cdata.xyzw[0], cdata.xyzw[1], cdata.xyzw[2], cdata.xyzw[3] ); } static void dump_idata( struct sh_idata idata ) { - debug_printf( "%d, %d, %d, %d", idata.xyzw[0], idata.xyzw[1], idata.xyzw[2], idata.xyzw[3] ); + _debug_printf( "%d, %d, %d, %d", idata.xyzw[0], idata.xyzw[1], idata.xyzw[2], idata.xyzw[3] ); } static void dump_bdata( boolean bdata ) { - debug_printf( bdata ? "TRUE" : "FALSE" ); + _debug_printf( bdata ? "TRUE" : "FALSE" ); } static void dump_sampleinfo( struct ps_sampleinfo sampleinfo ) { switch (sampleinfo.texture_type) { case SVGA3DSAMP_2D: - debug_printf( "_2d" ); + _debug_printf( "_2d" ); break; case SVGA3DSAMP_CUBE: - debug_printf( "_cube" ); + _debug_printf( "_cube" ); break; case SVGA3DSAMP_VOLUME: - debug_printf( "_volume" ); + _debug_printf( "_volume" ); break; default: assert( 0 ); @@ -255,46 +255,46 @@ static void dump_usageinfo( struct vs_semantic semantic ) { switch (semantic.usage) { case SVGA3D_DECLUSAGE_POSITION: - debug_printf("_position" ); + _debug_printf("_position" ); break; case SVGA3D_DECLUSAGE_BLENDWEIGHT: - debug_printf("_blendweight" ); + _debug_printf("_blendweight" ); break; case SVGA3D_DECLUSAGE_BLENDINDICES: - debug_printf("_blendindices" ); + _debug_printf("_blendindices" ); break; case SVGA3D_DECLUSAGE_NORMAL: - debug_printf("_normal" ); + _debug_printf("_normal" ); break; case SVGA3D_DECLUSAGE_PSIZE: - debug_printf("_psize" ); + _debug_printf("_psize" ); break; case SVGA3D_DECLUSAGE_TEXCOORD: - debug_printf("_texcoord"); + _debug_printf("_texcoord"); break; case SVGA3D_DECLUSAGE_TANGENT: - debug_printf("_tangent" ); + _debug_printf("_tangent" ); break; case SVGA3D_DECLUSAGE_BINORMAL: - debug_printf("_binormal" ); + _debug_printf("_binormal" ); break; case SVGA3D_DECLUSAGE_TESSFACTOR: - debug_printf("_tessfactor" ); + _debug_printf("_tessfactor" ); break; case SVGA3D_DECLUSAGE_POSITIONT: - debug_printf("_positiont" ); + _debug_printf("_positiont" ); break; case SVGA3D_DECLUSAGE_COLOR: - debug_printf("_color" ); + _debug_printf("_color" ); break; case SVGA3D_DECLUSAGE_FOG: - debug_printf("_fog" ); + _debug_printf("_fog" ); break; case SVGA3D_DECLUSAGE_DEPTH: - debug_printf("_depth" ); + _debug_printf("_depth" ); break; case SVGA3D_DECLUSAGE_SAMPLE: - debug_printf("_sample"); + _debug_printf("_sample"); break; default: assert( 0 ); @@ -302,7 +302,7 @@ static void dump_usageinfo( struct vs_semantic semantic ) } if (semantic.usage_index != 0) { - debug_printf("%d", semantic.usage_index ); + _debug_printf("%d", semantic.usage_index ); } } @@ -316,47 +316,47 @@ static void dump_dstreg( struct sh_dstreg dstreg, const struct dump_info *di ) assert( (dstreg.modifier & (SVGA3DDSTMOD_SATURATE | SVGA3DDSTMOD_PARTIALPRECISION)) == dstreg.modifier ); if (dstreg.modifier & SVGA3DDSTMOD_SATURATE) - debug_printf( "_sat" ); + _debug_printf( "_sat" ); if (dstreg.modifier & SVGA3DDSTMOD_PARTIALPRECISION) - debug_printf( "_pp" ); + _debug_printf( "_pp" ); switch (dstreg.shift_scale) { case 0: break; case 1: - debug_printf( "_x2" ); + _debug_printf( "_x2" ); break; case 2: - debug_printf( "_x4" ); + _debug_printf( "_x4" ); break; case 3: - debug_printf( "_x8" ); + _debug_printf( "_x8" ); break; case 13: - debug_printf( "_d8" ); + _debug_printf( "_d8" ); break; case 14: - debug_printf( "_d4" ); + _debug_printf( "_d4" ); break; case 15: - debug_printf( "_d2" ); + _debug_printf( "_d2" ); break; default: assert( 0 ); } - debug_printf( " " ); + _debug_printf( " " ); u.dstreg = dstreg; dump_reg( u.reg, NULL, di ); if (dstreg.write_mask != SVGA3DWRITEMASK_ALL) { - debug_printf( "." ); + _debug_printf( "." ); if (dstreg.write_mask & SVGA3DWRITEMASK_0) - debug_printf( "x" ); + _debug_printf( "x" ); if (dstreg.write_mask & SVGA3DWRITEMASK_1) - debug_printf( "y" ); + _debug_printf( "y" ); if (dstreg.write_mask & SVGA3DWRITEMASK_2) - debug_printf( "z" ); + _debug_printf( "z" ); if (dstreg.write_mask & SVGA3DWRITEMASK_3) - debug_printf( "w" ); + _debug_printf( "w" ); } } @@ -372,19 +372,19 @@ static void dump_srcreg( struct sh_srcreg srcreg, struct sh_srcreg *indreg, cons case SVGA3DSRCMOD_BIASNEG: case SVGA3DSRCMOD_SIGNNEG: case SVGA3DSRCMOD_X2NEG: - debug_printf( "-" ); + _debug_printf( "-" ); break; case SVGA3DSRCMOD_ABS: - debug_printf( "|" ); + _debug_printf( "|" ); break; case SVGA3DSRCMOD_ABSNEG: - debug_printf( "-|" ); + _debug_printf( "-|" ); break; case SVGA3DSRCMOD_COMP: - debug_printf( "1-" ); + _debug_printf( "1-" ); break; case SVGA3DSRCMOD_NOT: - debug_printf( "!" ); + _debug_printf( "!" ); } u.srcreg = srcreg; @@ -397,39 +397,39 @@ static void dump_srcreg( struct sh_srcreg srcreg, struct sh_srcreg *indreg, cons break; case SVGA3DSRCMOD_ABS: case SVGA3DSRCMOD_ABSNEG: - debug_printf( "|" ); + _debug_printf( "|" ); break; case SVGA3DSRCMOD_BIAS: case SVGA3DSRCMOD_BIASNEG: - debug_printf( "_bias" ); + _debug_printf( "_bias" ); break; case SVGA3DSRCMOD_SIGN: case SVGA3DSRCMOD_SIGNNEG: - debug_printf( "_bx2" ); + _debug_printf( "_bx2" ); break; case SVGA3DSRCMOD_X2: case SVGA3DSRCMOD_X2NEG: - debug_printf( "_x2" ); + _debug_printf( "_x2" ); break; case SVGA3DSRCMOD_DZ: - debug_printf( "_dz" ); + _debug_printf( "_dz" ); break; case SVGA3DSRCMOD_DW: - debug_printf( "_dw" ); + _debug_printf( "_dw" ); break; default: assert( 0 ); } if (srcreg.swizzle_x != 0 || srcreg.swizzle_y != 1 || srcreg.swizzle_z != 2 || srcreg.swizzle_w != 3) { - debug_printf( "." ); + _debug_printf( "." ); if (srcreg.swizzle_x == srcreg.swizzle_y && srcreg.swizzle_y == srcreg.swizzle_z && srcreg.swizzle_z == srcreg.swizzle_w) { - debug_printf( "%c", "xyzw"[srcreg.swizzle_x] ); + _debug_printf( "%c", "xyzw"[srcreg.swizzle_x] ); } else { - debug_printf( "%c", "xyzw"[srcreg.swizzle_x] ); - debug_printf( "%c", "xyzw"[srcreg.swizzle_y] ); - debug_printf( "%c", "xyzw"[srcreg.swizzle_z] ); - debug_printf( "%c", "xyzw"[srcreg.swizzle_w] ); + _debug_printf( "%c", "xyzw"[srcreg.swizzle_x] ); + _debug_printf( "%c", "xyzw"[srcreg.swizzle_y] ); + _debug_printf( "%c", "xyzw"[srcreg.swizzle_z] ); + _debug_printf( "%c", "xyzw"[srcreg.swizzle_w] ); } } } @@ -447,15 +447,15 @@ svga_shader_dump( if (do_binary) { for (i = 0; i < dwords; i++) - debug_printf(" 0x%08x,\n", assem[i]); + _debug_printf(" 0x%08x,\n", assem[i]); - debug_printf("\n\n"); + _debug_printf("\n\n"); } di.version.value = *assem++; di.is_ps = (di.version.type == SVGA3D_PS_TYPE); - debug_printf( + _debug_printf( "%s_%u_%u\n", di.is_ps ? "ps" : "vs", di.version.major, @@ -465,7 +465,7 @@ svga_shader_dump( struct sh_op op = *(struct sh_op *) assem; if (assem - start >= dwords) { - debug_printf("... ran off end of buffer\n"); + _debug_printf("... ran off end of buffer\n"); assert(0); return; } @@ -475,7 +475,7 @@ svga_shader_dump( { struct sh_dcl dcl = *(struct sh_dcl *) assem; - debug_printf( "dcl" ); + _debug_printf( "dcl" ); if (sh_dstreg_type( dcl.reg ) == SVGA3DREG_SAMPLER) dump_sampleinfo( dcl.u.ps.sampleinfo ); else if (di.is_ps) { @@ -486,7 +486,7 @@ svga_shader_dump( else dump_usageinfo( dcl.u.vs.semantic ); dump_dstreg( dcl.reg, &di ); - debug_printf( "\n" ); + _debug_printf( "\n" ); assem += sizeof( struct sh_dcl ) / sizeof( unsigned ); } break; @@ -495,11 +495,11 @@ svga_shader_dump( { struct sh_defb defb = *(struct sh_defb *) assem; - debug_printf( "defb " ); + _debug_printf( "defb " ); dump_reg( defb.reg, NULL, &di ); - debug_printf( ", " ); + _debug_printf( ", " ); dump_bdata( defb.data ); - debug_printf( "\n" ); + _debug_printf( "\n" ); assem += sizeof( struct sh_defb ) / sizeof( unsigned ); } break; @@ -508,11 +508,11 @@ svga_shader_dump( { struct sh_defi defi = *(struct sh_defi *) assem; - debug_printf( "defi " ); + _debug_printf( "defi " ); dump_reg( defi.reg, NULL, &di ); - debug_printf( ", " ); + _debug_printf( ", " ); dump_idata( defi.idata ); - debug_printf( "\n" ); + _debug_printf( "\n" ); assem += sizeof( struct sh_defi ) / sizeof( unsigned ); } break; @@ -528,11 +528,11 @@ svga_shader_dump( else { struct sh_unaryop unaryop = *(struct sh_unaryop *) assem; dump_dstreg( unaryop.dst, &di ); - debug_printf( ", " ); + _debug_printf( ", " ); dump_srcreg( unaryop.src, NULL, &di ); assem += sizeof( struct sh_unaryop ) / sizeof( unsigned ); } - debug_printf( "\n" ); + _debug_printf( "\n" ); break; case SVGA3DOP_TEX: @@ -549,7 +549,7 @@ svga_shader_dump( struct sh_unaryop unaryop = *(struct sh_unaryop *) assem; dump_dstreg( unaryop.dst, &di ); - debug_printf( ", " ); + _debug_printf( ", " ); dump_srcreg( unaryop.src, NULL, &di ); assem += sizeof( struct sh_unaryop ) / sizeof( unsigned ); } @@ -559,30 +559,30 @@ svga_shader_dump( dump_op( op, "texld" ); dump_dstreg( binaryop.dst, &di ); - debug_printf( ", " ); + _debug_printf( ", " ); dump_srcreg( binaryop.src0, NULL, &di ); - debug_printf( ", " ); + _debug_printf( ", " ); dump_srcreg( binaryop.src1, NULL, &di ); assem += sizeof( struct sh_binaryop ) / sizeof( unsigned ); } - debug_printf( "\n" ); + _debug_printf( "\n" ); break; case SVGA3DOP_DEF: { struct sh_def def = *(struct sh_def *) assem; - debug_printf( "def " ); + _debug_printf( "def " ); dump_reg( def.reg, NULL, &di ); - debug_printf( ", " ); + _debug_printf( ", " ); dump_cdata( def.cdata ); - debug_printf( "\n" ); + _debug_printf( "\n" ); assem += sizeof( struct sh_def ) / sizeof( unsigned ); } break; case SVGA3DOP_PHASE: - debug_printf( "phase\n" ); + _debug_printf( "phase\n" ); assem += sizeof( struct sh_op ) / sizeof( unsigned ); break; @@ -596,12 +596,12 @@ svga_shader_dump( break; case SVGA3DOP_RET: - debug_printf( "ret\n" ); + _debug_printf( "ret\n" ); assem += sizeof( struct sh_op ) / sizeof( unsigned ); break; case SVGA3DOP_END: - debug_printf( "end\n" ); + _debug_printf( "end\n" ); finished = TRUE; break; @@ -640,14 +640,14 @@ svga_shader_dump( } if (not_first_arg) - debug_printf( ", " ); + _debug_printf( ", " ); else - debug_printf( " " ); + _debug_printf( " " ); dump_srcreg( srcreg, &indreg, &di ); not_first_arg = TRUE; } - debug_printf( "\n" ); + _debug_printf( "\n" ); } } } diff --git a/src/gallium/drivers/trace/tr_dump_state.c b/src/gallium/drivers/trace/tr_dump_state.c index 6d58209294..0102cc1876 100644 --- a/src/gallium/drivers/trace/tr_dump_state.c +++ b/src/gallium/drivers/trace/tr_dump_state.c @@ -43,19 +43,6 @@ void trace_dump_format(enum pipe_format format) } -void trace_dump_block(const struct pipe_format_block *block) -{ - if (!trace_dumping_enabled_locked()) - return; - - trace_dump_struct_begin("pipe_format_block"); - trace_dump_member(uint, block, size); - trace_dump_member(uint, block, width); - trace_dump_member(uint, block, height); - trace_dump_struct_end(); -} - - static void trace_dump_reference(const struct pipe_reference *reference) { if (!trace_dumping_enabled_locked()) @@ -94,10 +81,6 @@ void trace_dump_template(const struct pipe_texture *templat) trace_dump_uint(templat->depth0); trace_dump_member_end(); - trace_dump_member_begin("block"); - trace_dump_block(&templat->block); - trace_dump_member_end(); - trace_dump_member(uint, templat, last_level); trace_dump_member(uint, templat, tex_usage); @@ -483,16 +466,9 @@ void trace_dump_transfer(const struct pipe_transfer *state) trace_dump_struct_begin("pipe_transfer"); - trace_dump_member(format, state, format); trace_dump_member(uint, state, width); trace_dump_member(uint, state, height); - trace_dump_member_begin("block"); - trace_dump_block(&state->block); - trace_dump_member_end(); - - trace_dump_member(uint, state, nblocksx); - trace_dump_member(uint, state, nblocksy); trace_dump_member(uint, state, stride); trace_dump_member(uint, state, usage); diff --git a/src/gallium/drivers/trace/tr_dump_state.h b/src/gallium/drivers/trace/tr_dump_state.h index 05b821adb6..07ad6fbb20 100644 --- a/src/gallium/drivers/trace/tr_dump_state.h +++ b/src/gallium/drivers/trace/tr_dump_state.h @@ -35,11 +35,8 @@ void trace_dump_format(enum pipe_format format); -void trace_dump_block(const struct pipe_format_block *block); - void trace_dump_template(const struct pipe_texture *templat); - void trace_dump_rasterizer_state(const struct pipe_rasterizer_state *state); void trace_dump_poly_stipple(const struct pipe_poly_stipple *state); diff --git a/src/gallium/drivers/trace/tr_rbug.c b/src/gallium/drivers/trace/tr_rbug.c index b59458c0e3..af1d7f3224 100644 --- a/src/gallium/drivers/trace/tr_rbug.c +++ b/src/gallium/drivers/trace/tr_rbug.c @@ -203,7 +203,9 @@ trace_rbug_texture_info(struct trace_rbug *tr_rbug, struct rbug_header *header, &t->width0, 1, &t->height0, 1, &t->depth0, 1, - t->block.width, t->block.height, t->block.size, + pf_get_blockwidth(t->format), + pf_get_blockheight(t->format), + pf_get_blocksize(t->format), t->last_level, t->nr_samples, t->tex_usage, @@ -251,9 +253,12 @@ trace_rbug_texture_read(struct trace_rbug *tr_rbug, struct rbug_header *header, map = screen->transfer_map(screen, t); rbug_send_texture_read_reply(tr_rbug->con, serial, - t->format, - t->block.width, t->block.height, t->block.size, - (uint8_t*)map, t->stride * t->nblocksy, + t->texture->format, + pf_get_blockwidth(t->texture->format), + pf_get_blockheight(t->texture->format), + pf_get_blocksize(t->texture->format), + (uint8_t*)map, + t->stride * pf_get_nblocksy(t->texture->format, t->height), t->stride, NULL); diff --git a/src/gallium/drivers/trace/tr_screen.c b/src/gallium/drivers/trace/tr_screen.c index 7da9bd3866..f69f7da000 100644 --- a/src/gallium/drivers/trace/tr_screen.c +++ b/src/gallium/drivers/trace/tr_screen.c @@ -35,6 +35,7 @@ #include "tr_screen.h" #include "pipe/p_inlines.h" +#include "pipe/p_format.h" static boolean trace = FALSE; @@ -424,7 +425,7 @@ trace_screen_transfer_unmap(struct pipe_screen *_screen, struct pipe_transfer *transfer = tr_trans->transfer; if(tr_trans->map) { - size_t size = transfer->nblocksy * transfer->stride; + size_t size = pf_get_nblocksy(transfer->texture->format, transfer->width) * transfer->stride; trace_dump_call_begin("pipe_screen", "transfer_write"); |