/*
- Copyright (C) Intel Corp. 2006. All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
-
- 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 (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 NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) 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.
-
- **********************************************************************/
- /*
- * Authors:
- * Keith Whitwell <keith@tungstengraphics.com>
- */
-
+ * Copyright (C) Intel Corp. 2006. All Rights Reserved.
+ * Intel funded Tungsten Graphics to
+ * develop this 3D driver.
+ *
+ * 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 (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 NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) 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.
+ */
#include "brw_context.h"
#include "brw_wm.h"
#include "brw_state.h"
+#include "main/enums.h"
#include "main/formats.h"
#include "main/fbobject.h"
#include "main/samplerobj.h"
+#include "main/framebuffer.h"
#include "program/prog_parameter.h"
+#include "program/program.h"
+#include "intel_mipmap_tree.h"
-#include "glsl/ralloc.h"
+#include "util/ralloc.h"
/**
* Return a bitfield where bit n is set if barycentric interpolation mode n
static unsigned
brw_compute_barycentric_interp_modes(struct brw_context *brw,
bool shade_model_flat,
+ bool persample_shading,
const struct gl_fragment_program *fprog)
{
unsigned barycentric_interp_modes = 0;
* modes are in use, and set the appropriate bits in
* barycentric_interp_modes.
*/
- for (attr = 0; attr < FRAG_ATTRIB_MAX; ++attr) {
+ for (attr = 0; attr < VARYING_SLOT_MAX; ++attr) {
enum glsl_interp_qualifier interp_qualifier =
fprog->InterpQualifier[attr];
- bool is_centroid = fprog->IsCentroid & BITFIELD64_BIT(attr);
- bool is_gl_Color = attr == FRAG_ATTRIB_COL0 || attr == FRAG_ATTRIB_COL1;
+ bool is_centroid = (fprog->IsCentroid & BITFIELD64_BIT(attr)) &&
+ !persample_shading;
+ bool is_sample = (fprog->IsSample & BITFIELD64_BIT(attr)) ||
+ persample_shading;
+ bool is_gl_Color = attr == VARYING_SLOT_COL0 || attr == VARYING_SLOT_COL1;
/* Ignore unused inputs. */
if (!(fprog->Base.InputsRead & BITFIELD64_BIT(attr)))
continue;
/* Ignore WPOS and FACE, because they don't require interpolation. */
- if (attr == FRAG_ATTRIB_WPOS || attr == FRAG_ATTRIB_FACE)
+ if (attr == VARYING_SLOT_POS || attr == VARYING_SLOT_FACE)
continue;
/* Determine the set (or sets) of barycentric coordinates needed to
if (is_centroid) {
barycentric_interp_modes |=
1 << BRW_WM_NONPERSPECTIVE_CENTROID_BARYCENTRIC;
+ } else if (is_sample) {
+ barycentric_interp_modes |=
+ 1 << BRW_WM_NONPERSPECTIVE_SAMPLE_BARYCENTRIC;
}
- if (!is_centroid || brw->needs_unlit_centroid_workaround) {
+ if ((!is_centroid && !is_sample) ||
+ brw->needs_unlit_centroid_workaround) {
barycentric_interp_modes |=
1 << BRW_WM_NONPERSPECTIVE_PIXEL_BARYCENTRIC;
}
if (is_centroid) {
barycentric_interp_modes |=
1 << BRW_WM_PERSPECTIVE_CENTROID_BARYCENTRIC;
+ } else if (is_sample) {
+ barycentric_interp_modes |=
+ 1 << BRW_WM_PERSPECTIVE_SAMPLE_BARYCENTRIC;
}
- if (!is_centroid || brw->needs_unlit_centroid_workaround) {
+ if ((!is_centroid && !is_sample) ||
+ brw->needs_unlit_centroid_workaround) {
barycentric_interp_modes |=
1 << BRW_WM_PERSPECTIVE_PIXEL_BARYCENTRIC;
}
return barycentric_interp_modes;
}
-
-void
-brw_wm_payload_setup(struct brw_context *brw,
- struct brw_wm_compile *c)
+static uint8_t
+computed_depth_mode(struct gl_fragment_program *fp)
{
- struct intel_context *intel = &brw->intel;
- bool uses_depth = (c->fp->program.Base.InputsRead &
- (1 << FRAG_ATTRIB_WPOS)) != 0;
- unsigned barycentric_interp_modes = c->prog_data.barycentric_interp_modes;
- int i;
-
- if (intel->gen >= 6) {
- /* R0-1: masks, pixel X/Y coordinates. */
- c->nr_payload_regs = 2;
- /* R2: only for 32-pixel dispatch.*/
-
- /* R3-26: barycentric interpolation coordinates. These appear in the
- * same order that they appear in the brw_wm_barycentric_interp_mode
- * enum. Each set of coordinates occupies 2 registers if dispatch width
- * == 8 and 4 registers if dispatch width == 16. Coordinates only
- * appear if they were enabled using the "Barycentric Interpolation
- * Mode" bits in WM_STATE.
- */
- for (i = 0; i < BRW_WM_BARYCENTRIC_INTERP_MODE_COUNT; ++i) {
- if (barycentric_interp_modes & (1 << i)) {
- c->barycentric_coord_reg[i] = c->nr_payload_regs;
- c->nr_payload_regs += 2;
- if (c->dispatch_width == 16) {
- c->nr_payload_regs += 2;
- }
- }
- }
-
- /* R27: interpolated depth if uses source depth */
- if (uses_depth) {
- c->source_depth_reg = c->nr_payload_regs;
- c->nr_payload_regs++;
- if (c->dispatch_width == 16) {
- /* R28: interpolated depth if not 8-wide. */
- c->nr_payload_regs++;
- }
- }
- /* R29: interpolated W set if GEN6_WM_USES_SOURCE_W.
- */
- if (uses_depth) {
- c->source_w_reg = c->nr_payload_regs;
- c->nr_payload_regs++;
- if (c->dispatch_width == 16) {
- /* R30: interpolated W if not 8-wide. */
- c->nr_payload_regs++;
- }
- }
- /* R31: MSAA position offsets. */
- /* R32-: bary for 32-pixel. */
- /* R58-59: interp W for 32-pixel. */
-
- if (c->fp->program.Base.OutputsWritten &
- BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
- c->source_depth_to_render_target = true;
- c->computes_depth = true;
+ if (fp->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
+ switch (fp->FragDepthLayout) {
+ case FRAG_DEPTH_LAYOUT_NONE:
+ case FRAG_DEPTH_LAYOUT_ANY:
+ return BRW_PSCDEPTH_ON;
+ case FRAG_DEPTH_LAYOUT_GREATER:
+ return BRW_PSCDEPTH_ON_GE;
+ case FRAG_DEPTH_LAYOUT_LESS:
+ return BRW_PSCDEPTH_ON_LE;
+ case FRAG_DEPTH_LAYOUT_UNCHANGED:
+ return BRW_PSCDEPTH_OFF;
}
- } else {
- brw_wm_lookup_iz(intel, c);
}
+ return BRW_PSCDEPTH_OFF;
}
bool
-brw_wm_prog_data_compare(const void *in_a, const void *in_b,
- int aux_size, const void *in_key)
+brw_wm_prog_data_compare(const void *in_a, const void *in_b)
{
const struct brw_wm_prog_data *a = in_a;
const struct brw_wm_prog_data *b = in_b;
- /* Compare all the struct up to the pointers. */
- if (memcmp(a, b, offsetof(struct brw_wm_prog_data, param)))
- return false;
-
- if (memcmp(a->param, b->param, a->nr_params * sizeof(void *)))
+ /* Compare the base structure. */
+ if (!brw_stage_prog_data_compare(&a->base, &b->base))
return false;
- if (memcmp(a->pull_param, b->pull_param, a->nr_pull_params * sizeof(void *)))
+ /* Compare the rest of the structure. */
+ const unsigned offset = sizeof(struct brw_stage_prog_data);
+ if (memcmp(((char *) a) + offset, ((char *) b) + offset,
+ sizeof(struct brw_wm_prog_data) - offset))
return false;
return true;
}
-void
-brw_wm_prog_data_free(const void *in_prog_data)
-{
- const struct brw_wm_prog_data *prog_data = in_prog_data;
-
- ralloc_free((void *)prog_data->param);
- ralloc_free((void *)prog_data->pull_param);
-}
-
/**
* All Mesa program -> GPU code generation goes through this function.
* Depending on the instructions used (i.e. flow control instructions)
* we'll use one of two code generators.
*/
-bool do_wm_prog(struct brw_context *brw,
- struct gl_shader_program *prog,
- struct brw_fragment_program *fp,
- struct brw_wm_prog_key *key)
+bool
+brw_codegen_wm_prog(struct brw_context *brw,
+ struct gl_shader_program *prog,
+ struct brw_fragment_program *fp,
+ struct brw_wm_prog_key *key)
{
- struct intel_context *intel = &brw->intel;
- struct brw_wm_compile *c;
+ struct gl_context *ctx = &brw->ctx;
+ void *mem_ctx = ralloc_context(NULL);
+ struct brw_wm_prog_data prog_data;
const GLuint *program;
struct gl_shader *fs = NULL;
GLuint program_size;
if (prog)
fs = prog->_LinkedShaders[MESA_SHADER_FRAGMENT];
- c = rzalloc(NULL, struct brw_wm_compile);
+ memset(&prog_data, 0, sizeof(prog_data));
+ /* key->alpha_test_func means simulating alpha testing via discards,
+ * so the shader definitely kills pixels.
+ */
+ prog_data.uses_kill = fp->program.UsesKill || key->alpha_test_func;
+ prog_data.uses_omask =
+ fp->program.Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_SAMPLE_MASK);
+ prog_data.computed_depth_mode = computed_depth_mode(&fp->program);
+
+ prog_data.early_fragment_tests = fs && fs->EarlyFragmentTests;
+
+ /* Use ALT floating point mode for ARB programs so that 0^0 == 1. */
+ if (!prog)
+ prog_data.base.use_alt_mode = true;
/* Allocate the references to the uniforms that will end up in the
* prog_data associated with the compiled program, and which will be freed
*/
int param_count;
if (fs) {
- param_count = fs->num_uniform_components;
+ param_count = fs->num_uniform_components +
+ fs->NumImages * BRW_IMAGE_PARAM_SIZE;
+ prog_data.base.nr_image_params = fs->NumImages;
} else {
param_count = fp->program.Base.Parameters->NumParameters * 4;
}
/* The backend also sometimes adds params for texture size. */
- param_count += 2 * BRW_MAX_TEX_UNIT;
- c->prog_data.param = rzalloc_array(NULL, const float *, param_count);
- c->prog_data.pull_param = rzalloc_array(NULL, const float *, param_count);
-
- memcpy(&c->key, key, sizeof(*key));
-
- c->fp = fp;
-
- brw_init_compile(brw, &c->func, c);
-
- c->prog_data.barycentric_interp_modes =
- brw_compute_barycentric_interp_modes(brw, c->key.flat_shade,
+ param_count += 2 * ctx->Const.Program[MESA_SHADER_FRAGMENT].MaxTextureImageUnits;
+ prog_data.base.param =
+ rzalloc_array(NULL, const gl_constant_value *, param_count);
+ prog_data.base.pull_param =
+ rzalloc_array(NULL, const gl_constant_value *, param_count);
+ prog_data.base.image_param =
+ rzalloc_array(NULL, struct brw_image_param,
+ prog_data.base.nr_image_params);
+ prog_data.base.nr_params = param_count;
+
+ prog_data.barycentric_interp_modes =
+ brw_compute_barycentric_interp_modes(brw, key->flat_shade,
+ key->persample_shading,
&fp->program);
- brw_wm_fs_emit(brw, c, prog);
-
- /* Scratch space is used for register spilling */
- if (c->last_scratch) {
- perf_debug("Fragment shader triggered register spilling. "
- "Try reducing the number of live scalar values to "
- "improve performance.\n");
-
- c->prog_data.total_scratch = brw_get_scratch_size(c->last_scratch);
+ program = brw_wm_fs_emit(brw, mem_ctx, key, &prog_data,
+ &fp->program, prog, &program_size);
+ if (program == NULL) {
+ ralloc_free(mem_ctx);
+ return false;
+ }
- brw_get_scratch_bo(intel, &brw->wm.scratch_bo,
- c->prog_data.total_scratch * brw->max_wm_threads);
+ if (prog_data.base.total_scratch) {
+ brw_get_scratch_bo(brw, &brw->wm.base.scratch_bo,
+ prog_data.base.total_scratch * brw->max_wm_threads);
}
if (unlikely(INTEL_DEBUG & DEBUG_WM))
fprintf(stderr, "\n");
- /* get the program
- */
- program = brw_get_program(&c->func, &program_size);
-
- brw_upload_cache(&brw->cache, BRW_WM_PROG,
- &c->key, sizeof(c->key),
+ brw_upload_cache(&brw->cache, BRW_CACHE_FS_PROG,
+ key, sizeof(struct brw_wm_prog_key),
program, program_size,
- &c->prog_data, sizeof(c->prog_data),
- &brw->wm.prog_offset, &brw->wm.prog_data);
+ &prog_data, sizeof(prog_data),
+ &brw->wm.base.prog_offset, &brw->wm.prog_data);
- ralloc_free(c);
+ ralloc_free(mem_ctx);
return true;
}
static bool
-key_debug(const char *name, int a, int b)
+key_debug(struct brw_context *brw, const char *name, int a, int b)
{
if (a != b) {
perf_debug(" %s %d->%d\n", name, a, b);
}
bool
-brw_debug_recompile_sampler_key(const struct brw_sampler_prog_key_data *old_key,
+brw_debug_recompile_sampler_key(struct brw_context *brw,
+ const struct brw_sampler_prog_key_data *old_key,
const struct brw_sampler_prog_key_data *key)
{
bool found = false;
for (unsigned int i = 0; i < MAX_SAMPLERS; i++) {
- found |= key_debug("EXT_texture_swizzle or DEPTH_TEXTURE_MODE",
+ found |= key_debug(brw, "EXT_texture_swizzle or DEPTH_TEXTURE_MODE",
old_key->swizzles[i], key->swizzles[i]);
}
- found |= key_debug("GL_CLAMP enabled on any texture unit's 1st coordinate",
+ found |= key_debug(brw, "GL_CLAMP enabled on any texture unit's 1st coordinate",
old_key->gl_clamp_mask[0], key->gl_clamp_mask[0]);
- found |= key_debug("GL_CLAMP enabled on any texture unit's 2nd coordinate",
+ found |= key_debug(brw, "GL_CLAMP enabled on any texture unit's 2nd coordinate",
old_key->gl_clamp_mask[1], key->gl_clamp_mask[1]);
- found |= key_debug("GL_CLAMP enabled on any texture unit's 3rd coordinate",
+ found |= key_debug(brw, "GL_CLAMP enabled on any texture unit's 3rd coordinate",
old_key->gl_clamp_mask[2], key->gl_clamp_mask[2]);
- found |= key_debug("GL_MESA_ycbcr texturing\n",
- old_key->yuvtex_mask, key->yuvtex_mask);
- found |= key_debug("GL_MESA_ycbcr UV swapping\n",
- old_key->yuvtex_swap_mask, key->yuvtex_swap_mask);
+ found |= key_debug(brw, "gather channel quirk on any texture unit",
+ old_key->gather_channel_quirk_mask, key->gather_channel_quirk_mask);
+ found |= key_debug(brw, "compressed multisample layout",
+ old_key->compressed_multisample_layout_mask,
+ key->compressed_multisample_layout_mask);
+
+ for (unsigned int i = 0; i < MAX_SAMPLERS; i++) {
+ found |= key_debug(brw, "textureGather workarounds",
+ old_key->gen6_gather_wa[i], key->gen6_gather_wa[i]);
+ }
return found;
}
for (unsigned int i = 0; i < brw->cache.size; i++) {
for (c = brw->cache.items[i]; c; c = c->next) {
- if (c->cache_id == BRW_WM_PROG) {
+ if (c->cache_id == BRW_CACHE_FS_PROG) {
old_key = c->key;
if (old_key->program_string_id == key->program_string_id)
}
if (!c) {
- perf_debug(" Didn't find previous compile in the shader cache for "
- "debug\n");
+ perf_debug(" Didn't find previous compile in the shader cache for debug\n");
return;
}
- found |= key_debug("alphatest, computed depth, depth test, or depth write",
+ found |= key_debug(brw, "alphatest, computed depth, depth test, or "
+ "depth write",
old_key->iz_lookup, key->iz_lookup);
- found |= key_debug("depth statistics", old_key->stats_wm, key->stats_wm);
- found |= key_debug("flat shading", old_key->flat_shade, key->flat_shade);
- found |= key_debug("number of color buffers", old_key->nr_color_regions, key->nr_color_regions);
- found |= key_debug("sample alpha to coverage", old_key->sample_alpha_to_coverage, key->sample_alpha_to_coverage);
- found |= key_debug("rendering to FBO", old_key->render_to_fbo, key->render_to_fbo);
- found |= key_debug("fragment color clamping", old_key->clamp_fragment_color, key->clamp_fragment_color);
- found |= key_debug("line smoothing", old_key->line_aa, key->line_aa);
- found |= key_debug("proj_attrib_mask", old_key->proj_attrib_mask, key->proj_attrib_mask);
- found |= key_debug("renderbuffer height", old_key->drawable_height, key->drawable_height);
- found |= key_debug("vertex shader outputs", old_key->vp_outputs_written, key->vp_outputs_written);
-
- found |= brw_debug_recompile_sampler_key(&old_key->tex, &key->tex);
+ found |= key_debug(brw, "depth statistics",
+ old_key->stats_wm, key->stats_wm);
+ found |= key_debug(brw, "flat shading",
+ old_key->flat_shade, key->flat_shade);
+ found |= key_debug(brw, "per-sample shading",
+ old_key->persample_shading, key->persample_shading);
+ found |= key_debug(brw, "per-sample shading and 2x MSAA",
+ old_key->persample_2x, key->persample_2x);
+ found |= key_debug(brw, "number of color buffers",
+ old_key->nr_color_regions, key->nr_color_regions);
+ found |= key_debug(brw, "MRT alpha test or alpha-to-coverage",
+ old_key->replicate_alpha, key->replicate_alpha);
+ found |= key_debug(brw, "rendering to FBO",
+ old_key->render_to_fbo, key->render_to_fbo);
+ found |= key_debug(brw, "fragment color clamping",
+ old_key->clamp_fragment_color, key->clamp_fragment_color);
+ found |= key_debug(brw, "line smoothing",
+ old_key->line_aa, key->line_aa);
+ found |= key_debug(brw, "renderbuffer height",
+ old_key->drawable_height, key->drawable_height);
+ found |= key_debug(brw, "input slots valid",
+ old_key->input_slots_valid, key->input_slots_valid);
+ found |= key_debug(brw, "mrt alpha test function",
+ old_key->alpha_test_func, key->alpha_test_func);
+ found |= key_debug(brw, "mrt alpha test reference value",
+ old_key->alpha_test_ref, key->alpha_test_ref);
+
+ found |= brw_debug_recompile_sampler_key(brw, &old_key->tex, &key->tex);
if (!found) {
perf_debug(" Something else\n");
}
}
+static uint8_t
+gen6_gather_workaround(GLenum internalformat)
+{
+ switch (internalformat) {
+ case GL_R8I: return WA_SIGN | WA_8BIT;
+ case GL_R8UI: return WA_8BIT;
+ case GL_R16I: return WA_SIGN | WA_16BIT;
+ case GL_R16UI: return WA_16BIT;
+ default:
+ /* Note that even though GL_R32I and GL_R32UI have format overrides in
+ * the surface state, there is no shader w/a required.
+ */
+ return 0;
+ }
+}
+
void
brw_populate_sampler_prog_key_data(struct gl_context *ctx,
const struct gl_program *prog,
+ unsigned sampler_count,
struct brw_sampler_prog_key_data *key)
{
- struct intel_context *intel = intel_context(ctx);
+ struct brw_context *brw = brw_context(ctx);
- for (int s = 0; s < MAX_SAMPLERS; s++) {
+ for (int s = 0; s < sampler_count; s++) {
key->swizzles[s] = SWIZZLE_NOOP;
if (!(prog->SamplersUsed & (1 << s)))
int unit_id = prog->SamplerUnits[s];
const struct gl_texture_unit *unit = &ctx->Texture.Unit[unit_id];
- if (unit->_ReallyEnabled && unit->_Current->Target != GL_TEXTURE_BUFFER) {
+ if (unit->_Current && unit->_Current->Target != GL_TEXTURE_BUFFER) {
const struct gl_texture_object *t = unit->_Current;
const struct gl_texture_image *img = t->Image[0][t->BaseLevel];
struct gl_sampler_object *sampler = _mesa_get_samplerobj(ctx, unit_id);
/* Haswell handles texture swizzling as surface format overrides
* (except for GL_ALPHA); all other platforms need MOVs in the shader.
*/
- if (!intel->is_haswell || alpha_depth)
- key->swizzles[s] = brw_get_texture_swizzle(t);
-
- if (img->InternalFormat == GL_YCBCR_MESA) {
- key->yuvtex_mask |= 1 << s;
- if (img->TexFormat == MESA_FORMAT_YCBCR)
- key->yuvtex_swap_mask |= 1 << s;
- }
+ if (alpha_depth || (brw->gen < 8 && !brw->is_haswell))
+ key->swizzles[s] = brw_get_texture_swizzle(ctx, t);
- if (sampler->MinFilter != GL_NEAREST &&
+ if (brw->gen < 8 &&
+ sampler->MinFilter != GL_NEAREST &&
sampler->MagFilter != GL_NEAREST) {
if (sampler->WrapS == GL_CLAMP)
key->gl_clamp_mask[0] |= 1 << s;
if (sampler->WrapR == GL_CLAMP)
key->gl_clamp_mask[2] |= 1 << s;
}
+
+ /* gather4's channel select for green from RG32F is broken; requires
+ * a shader w/a on IVB; fixable with just SCS on HSW.
+ */
+ if (brw->gen == 7 && !brw->is_haswell && prog->UsesGather) {
+ if (img->InternalFormat == GL_RG32F)
+ key->gather_channel_quirk_mask |= 1 << s;
+ }
+
+ /* Gen6's gather4 is broken for UINT/SINT; we treat them as
+ * UNORM/FLOAT instead and fix it in the shader.
+ */
+ if (brw->gen == 6 && prog->UsesGather) {
+ key->gen6_gather_wa[s] = gen6_gather_workaround(img->InternalFormat);
+ }
+
+ /* If this is a multisample sampler, and uses the CMS MSAA layout,
+ * then we need to emit slightly different code to first sample the
+ * MCS surface.
+ */
+ struct intel_texture_object *intel_tex =
+ intel_texture_object((struct gl_texture_object *)t);
+
+ if (brw->gen >= 7 &&
+ intel_tex->mt->msaa_layout == INTEL_MSAA_LAYOUT_CMS) {
+ key->compressed_multisample_layout_mask |= 1 << s;
+ }
}
}
}
-static void brw_wm_populate_key( struct brw_context *brw,
- struct brw_wm_prog_key *key )
+static bool
+brw_wm_state_dirty (struct brw_context *brw)
{
- struct gl_context *ctx = &brw->intel.ctx;
- struct intel_context *intel = &brw->intel;
+ return brw_state_dirty(brw,
+ _NEW_BUFFERS |
+ _NEW_COLOR |
+ _NEW_DEPTH |
+ _NEW_FRAG_CLAMP |
+ _NEW_HINT |
+ _NEW_LIGHT |
+ _NEW_LINE |
+ _NEW_MULTISAMPLE |
+ _NEW_POLYGON |
+ _NEW_STENCIL |
+ _NEW_TEXTURE,
+ BRW_NEW_FRAGMENT_PROGRAM |
+ BRW_NEW_REDUCED_PRIMITIVE |
+ BRW_NEW_STATS_WM |
+ BRW_NEW_VUE_MAP_GEOM_OUT);
+}
+
+static void
+brw_wm_populate_key(struct brw_context *brw, struct brw_wm_prog_key *key)
+{
+ struct gl_context *ctx = &brw->ctx;
/* BRW_NEW_FRAGMENT_PROGRAM */
- const struct brw_fragment_program *fp =
- (struct brw_fragment_program *)brw->fragment_program;
+ const struct brw_fragment_program *fp =
+ (struct brw_fragment_program *) brw->fragment_program;
const struct gl_program *prog = (struct gl_program *) brw->fragment_program;
GLuint lookup = 0;
GLuint line_aa;
bool program_uses_dfdy = fp->program.UsesDFdy;
+ const bool multisample_fbo = _mesa_geometric_samples(ctx->DrawBuffer) > 1;
memset(key, 0, sizeof(*key));
/* Build the index for table lookup
*/
- if (intel->gen < 6) {
+ if (brw->gen < 6) {
/* _NEW_COLOR */
if (fp->program.UsesKill || ctx->Color.AlphaEnabled)
lookup |= IZ_PS_KILL_ALPHATEST_BIT;
if (ctx->Depth.Test && ctx->Depth.Mask) /* ?? */
lookup |= IZ_DEPTH_WRITE_ENABLE_BIT;
- /* _NEW_STENCIL */
+ /* _NEW_STENCIL | _NEW_BUFFERS */
if (ctx->Stencil._Enabled) {
lookup |= IZ_STENCIL_TEST_ENABLE_BIT;
/* _NEW_LINE, _NEW_POLYGON, BRW_NEW_REDUCED_PRIMITIVE */
if (ctx->Line.SmoothFlag) {
- if (brw->intel.reduced_primitive == GL_LINES) {
+ if (brw->reduced_primitive == GL_LINES) {
line_aa = AA_ALWAYS;
}
- else if (brw->intel.reduced_primitive == GL_TRIANGLES) {
+ else if (brw->reduced_primitive == GL_TRIANGLES) {
if (ctx->Polygon.FrontMode == GL_LINE) {
line_aa = AA_SOMETIMES;
key->line_aa = line_aa;
- if (intel->gen < 6)
- key->stats_wm = brw->intel.stats_wm;
+ /* _NEW_HINT */
+ key->high_quality_derivatives =
+ ctx->Hint.FragmentShaderDerivative == GL_NICEST;
- /* BRW_NEW_WM_INPUT_DIMENSIONS */
- /* Only set this for fixed function. The optimization it enables isn't
- * useful for programs using shaders.
- */
- if (ctx->Shader.CurrentFragmentProgram)
- key->proj_attrib_mask = 0xffffffff;
- else
- key->proj_attrib_mask = brw->wm.input_size_masks[4-1];
+ if (brw->gen < 6)
+ key->stats_wm = brw->stats_wm;
/* _NEW_LIGHT */
key->flat_shade = (ctx->Light.ShadeModel == GL_FLAT);
key->clamp_fragment_color = ctx->Color._ClampFragmentColor;
/* _NEW_TEXTURE */
- brw_populate_sampler_prog_key_data(ctx, prog, &key->tex);
+ brw_populate_sampler_prog_key_data(ctx, prog, brw->wm.base.sampler_count,
+ &key->tex);
/* _NEW_BUFFERS */
/*
* For DRI2 the origin_x/y will always be (0,0) but we still need the
* drawable height in order to invert the Y axis.
*/
- if (fp->program.Base.InputsRead & FRAG_BIT_WPOS) {
- key->drawable_height = ctx->DrawBuffer->Height;
+ if (fp->program.Base.InputsRead & VARYING_BIT_POS) {
+ key->drawable_height = _mesa_geometric_height(ctx->DrawBuffer);
}
- if ((fp->program.Base.InputsRead & FRAG_BIT_WPOS) || program_uses_dfdy) {
+ if ((fp->program.Base.InputsRead & VARYING_BIT_POS) || program_uses_dfdy) {
key->render_to_fbo = _mesa_is_user_fbo(ctx->DrawBuffer);
}
/* _NEW_BUFFERS */
key->nr_color_regions = ctx->DrawBuffer->_NumColorDrawBuffers;
- /* _NEW_MULTISAMPLE */
- key->sample_alpha_to_coverage = ctx->Multisample.SampleAlphaToCoverage;
- /* CACHE_NEW_VS_PROG */
- if (intel->gen < 6)
- key->vp_outputs_written = brw->vs.prog_data->outputs_written;
+ /* _NEW_MULTISAMPLE, _NEW_COLOR, _NEW_BUFFERS */
+ key->replicate_alpha = ctx->DrawBuffer->_NumColorDrawBuffers > 1 &&
+ (ctx->Multisample.SampleAlphaToCoverage || ctx->Color.AlphaEnabled);
+
+ /* _NEW_BUFFERS _NEW_MULTISAMPLE */
+ /* Ignore sample qualifier while computing this flag. */
+ key->persample_shading =
+ _mesa_get_min_invocations_per_fragment(ctx, &fp->program, true) > 1;
+ if (key->persample_shading)
+ key->persample_2x = _mesa_geometric_samples(ctx->DrawBuffer) == 2;
+
+ key->compute_pos_offset =
+ _mesa_get_min_invocations_per_fragment(ctx, &fp->program, false) > 1 &&
+ fp->program.Base.SystemValuesRead & SYSTEM_BIT_SAMPLE_POS;
+
+ key->compute_sample_id =
+ multisample_fbo &&
+ ctx->Multisample.Enabled &&
+ (fp->program.Base.SystemValuesRead & SYSTEM_BIT_SAMPLE_ID);
+
+ /* BRW_NEW_VUE_MAP_GEOM_OUT */
+ if (brw->gen < 6 || _mesa_bitcount_64(fp->program.Base.InputsRead &
+ BRW_FS_VARYING_INPUT_MASK) > 16)
+ key->input_slots_valid = brw->vue_map_geom_out.slots_valid;
+
+
+ /* _NEW_COLOR | _NEW_BUFFERS */
+ /* Pre-gen6, the hardware alpha test always used each render
+ * target's alpha to do alpha test, as opposed to render target 0's alpha
+ * like GL requires. Fix that by building the alpha test into the
+ * shader, and we'll skip enabling the fixed function alpha test.
+ */
+ if (brw->gen < 6 && ctx->DrawBuffer->_NumColorDrawBuffers > 1 &&
+ ctx->Color.AlphaEnabled) {
+ key->alpha_test_func = ctx->Color.AlphaFunc;
+ key->alpha_test_ref = ctx->Color.AlphaRef;
+ }
/* The unique fragment program ID */
key->program_string_id = fp->id;
}
-
-static void
+void
brw_upload_wm_prog(struct brw_context *brw)
{
- struct intel_context *intel = &brw->intel;
- struct gl_context *ctx = &intel->ctx;
+ struct gl_context *ctx = &brw->ctx;
+ struct gl_shader_program *current = ctx->_Shader->_CurrentFragmentProgram;
struct brw_wm_prog_key key;
struct brw_fragment_program *fp = (struct brw_fragment_program *)
brw->fragment_program;
+ if (!brw_wm_state_dirty(brw))
+ return;
+
brw_wm_populate_key(brw, &key);
- if (!brw_search_cache(&brw->cache, BRW_WM_PROG,
+ if (!brw_search_cache(&brw->cache, BRW_CACHE_FS_PROG,
&key, sizeof(key),
- &brw->wm.prog_offset, &brw->wm.prog_data)) {
- bool success = do_wm_prog(brw, ctx->Shader._CurrentFragmentProgram, fp,
- &key);
+ &brw->wm.base.prog_offset, &brw->wm.prog_data)) {
+ bool success = brw_codegen_wm_prog(brw, current, fp, &key);
(void) success;
assert(success);
}
+ brw->wm.base.prog_data = &brw->wm.prog_data->base;
}
-
-
-const struct brw_tracked_state brw_wm_prog = {
- .dirty = {
- .mesa = (_NEW_COLOR |
- _NEW_DEPTH |
- _NEW_STENCIL |
- _NEW_POLYGON |
- _NEW_LINE |
- _NEW_LIGHT |
- _NEW_FRAG_CLAMP |
- _NEW_BUFFERS |
- _NEW_TEXTURE |
- _NEW_MULTISAMPLE),
- .brw = (BRW_NEW_FRAGMENT_PROGRAM |
- BRW_NEW_WM_INPUT_DIMENSIONS |
- BRW_NEW_REDUCED_PRIMITIVE),
- .cache = CACHE_NEW_VS_PROG,
- },
- .emit = brw_upload_wm_prog
-};
-