/**
* Compute the VUE map for vertex shader program.
- *
- * Note that consumers of this map using cache keys must include
- * prog_data->userclip and prog_data->outputs_written in their key
- * (generated by CACHE_NEW_VS_PROG).
*/
void
brw_compute_vue_map(struct brw_context *brw, struct brw_vue_map *vue_map,
- GLbitfield64 slots_valid, bool userclip_active)
+ GLbitfield64 slots_valid)
{
- const struct intel_context *intel = &brw->intel;
-
vue_map->slots_valid = slots_valid;
int i;
+ /* gl_Layer doesn't get its own varying slot--it's stored in the virst VUE
+ * slot (VARYING_SLOT_PSIZ).
+ */
+ slots_valid &= ~VARYING_BIT_LAYER;
+
/* Make sure that the values we store in vue_map->varying_to_slot and
* vue_map->slot_to_varying won't overflow the signed chars that are used
* to store them. Note that since vue_map->slot_to_varying sometimes holds
/* VUE header: format depends on chip generation and whether clipping is
* enabled.
*/
- switch (intel->gen) {
+ switch (brw->gen) {
case 4:
case 5:
/* There are 8 dwords in VUE header pre-Ironlake:
*/
assign_vue_slot(vue_map, VARYING_SLOT_PSIZ);
assign_vue_slot(vue_map, VARYING_SLOT_POS);
- if (userclip_active) {
+ if (slots_valid & BITFIELD64_BIT(VARYING_SLOT_CLIP_DIST0))
assign_vue_slot(vue_map, VARYING_SLOT_CLIP_DIST0);
+ if (slots_valid & BITFIELD64_BIT(VARYING_SLOT_CLIP_DIST1))
assign_vue_slot(vue_map, VARYING_SLOT_CLIP_DIST1);
- }
+
/* front and back colors need to be consecutive so that we can use
* ATTRIBUTE_SWIZZLE_INPUTATTR_FACING to swizzle them when doing
* two-sided color.
bool
-brw_vec4_prog_data_compare(const struct brw_vec4_prog_data *a,
- const struct brw_vec4_prog_data *b)
-{
- /* Compare all the struct up to the pointers. */
- if (memcmp(a, b, offsetof(struct brw_vec4_prog_data, param)))
- return false;
-
- if (memcmp(a->param, b->param, a->nr_params * sizeof(void *)))
- return false;
-
- if (memcmp(a->pull_param, b->pull_param, a->nr_pull_params * sizeof(void *)))
- return false;
-
- return true;
-}
-
-
-bool
-brw_vs_prog_data_compare(const void *in_a, const void *in_b,
- int aux_size, const void *in_key)
+brw_vs_prog_data_compare(const void *in_a, const void *in_b)
{
const struct brw_vs_prog_data *a = in_a;
const struct brw_vs_prog_data *b = in_b;
/* Compare the rest of the struct. */
const unsigned offset = sizeof(struct brw_vec4_prog_data);
- if (memcmp(((char *) &a) + offset, ((char *) &b) + offset,
+ if (memcmp(((char *) a) + offset, ((char *) b) + offset,
sizeof(struct brw_vs_prog_data) - offset)) {
return false;
}
struct brw_vertex_program *vp,
struct brw_vs_prog_key *key)
{
- struct intel_context *intel = &brw->intel;
GLuint program_size;
const GLuint *program;
struct brw_vs_compile c;
} else {
param_count = vp->program.Base.Parameters->NumParameters * 4;
}
- /* We also upload clip plane data as uniforms */
- param_count += MAX_CLIP_PLANES * 4;
+ /* vec4_visitor::setup_uniform_clipplane_values() also uploads user clip
+ * planes as uniforms.
+ */
+ param_count += c.key.base.nr_userclip_plane_consts * 4;
prog_data.base.param = rzalloc_array(NULL, const float *, param_count);
prog_data.base.pull_param = rzalloc_array(NULL, const float *, param_count);
prog_data.inputs_read |= VERT_BIT_EDGEFLAG;
}
- if (intel->gen < 6) {
+ if (brw->gen < 6) {
/* Put dummy slots into the VUE for the SF to put the replaced
* point sprite coords in. We shouldn't need these dummy slots,
* which take up precious URB space, but it would mean that the SF
if (c.key.point_coord_replace & (1 << i))
outputs_written |= BITFIELD64_BIT(VARYING_SLOT_TEX0 + i);
}
+
+ /* if back colors are written, allocate slots for front colors too */
+ if (outputs_written & BITFIELD64_BIT(VARYING_SLOT_BFC0))
+ outputs_written |= BITFIELD64_BIT(VARYING_SLOT_COL0);
+ if (outputs_written & BITFIELD64_BIT(VARYING_SLOT_BFC1))
+ outputs_written |= BITFIELD64_BIT(VARYING_SLOT_COL1);
+ }
+
+ /* In order for legacy clipping to work, we need to populate the clip
+ * distance varying slots whenever clipping is enabled, even if the vertex
+ * shader doesn't write to gl_ClipDistance.
+ */
+ if (c.key.base.userclip_active) {
+ outputs_written |= BITFIELD64_BIT(VARYING_SLOT_CLIP_DIST0);
+ outputs_written |= BITFIELD64_BIT(VARYING_SLOT_CLIP_DIST1);
}
- brw_compute_vue_map(brw, &prog_data.base.vue_map, outputs_written,
- c.key.base.userclip_active);
+ brw_compute_vue_map(brw, &prog_data.base.vue_map, outputs_written);
if (0) {
_mesa_fprint_program_opt(stdout, &c.vp->program.Base, PROG_PRINT_DEBUG,
return false;
}
- if (prog_data.base.nr_pull_params)
- prog_data.base.num_surfaces = 1;
- if (c.vp->program.Base.SamplersUsed)
- prog_data.base.num_surfaces = SURF_INDEX_VS_TEXTURE(BRW_MAX_TEX_UNIT);
- if (prog &&
- prog->_LinkedShaders[MESA_SHADER_VERTEX]->NumUniformBlocks) {
- prog_data.base.num_surfaces =
- SURF_INDEX_VS_UBO(prog->_LinkedShaders[MESA_SHADER_VERTEX]->NumUniformBlocks);
- }
-
/* Scratch space is used for register spilling */
if (c.base.last_scratch) {
perf_debug("Vertex shader triggered register spilling. "
prog_data.base.total_scratch
= brw_get_scratch_size(c.base.last_scratch*REG_SIZE);
- brw_get_scratch_bo(brw, &brw->vs.scratch_bo,
+ brw_get_scratch_bo(brw, &brw->vs.base.scratch_bo,
prog_data.base.total_scratch * brw->max_vs_threads);
}
&c.key, sizeof(c.key),
program, program_size,
&prog_data, sizeof(prog_data),
- &brw->vs.prog_offset, &brw->vs.prog_data);
+ &brw->vs.base.prog_offset, &brw->vs.prog_data);
ralloc_free(mem_ctx);
return true;
static bool
key_debug(struct brw_context *brw, const char *name, int a, int b)
{
- struct intel_context *intel = &brw->intel;
if (a != b) {
perf_debug(" %s %d->%d\n", name, a, b);
return true;
struct gl_shader_program *prog,
const struct brw_vs_prog_key *key)
{
- struct intel_context *intel = &brw->intel;
struct brw_cache_item *c = NULL;
const struct brw_vs_prog_key *old_key = NULL;
bool found = false;
found |= key_debug(brw, "clip distance enable",
old_key->base.uses_clip_distance, key->base.uses_clip_distance);
- found |= key_debug(brw, "clip plane enable bitfield",
- old_key->base.userclip_planes_enabled_gen_4_5,
- key->base.userclip_planes_enabled_gen_4_5);
found |= key_debug(brw, "copy edgeflag",
old_key->copy_edgeflag, key->copy_edgeflag);
found |= key_debug(brw, "PointCoord replace",
}
}
+
+void
+brw_setup_vec4_key_clip_info(struct brw_context *brw,
+ struct brw_vec4_prog_key *key,
+ bool program_uses_clip_distance)
+{
+ struct gl_context *ctx = &brw->ctx;
+
+ key->userclip_active = (ctx->Transform.ClipPlanesEnabled != 0);
+ key->uses_clip_distance = program_uses_clip_distance;
+ if (key->userclip_active && !key->uses_clip_distance) {
+ key->nr_userclip_plane_consts
+ = _mesa_logbase2(ctx->Transform.ClipPlanesEnabled) + 1;
+ }
+}
+
+
static void brw_upload_vs_prog(struct brw_context *brw)
{
- struct intel_context *intel = &brw->intel;
- struct gl_context *ctx = &intel->ctx;
+ struct gl_context *ctx = &brw->ctx;
struct brw_vs_prog_key key;
/* BRW_NEW_VERTEX_PROGRAM */
struct brw_vertex_program *vp =
* the inputs it asks for, whether they are varying or not.
*/
key.base.program_string_id = vp->id;
- key.base.userclip_active = (ctx->Transform.ClipPlanesEnabled != 0);
- key.base.uses_clip_distance = vp->program.UsesClipDistance;
- if (key.base.userclip_active && !key.base.uses_clip_distance) {
- if (intel->gen < 6) {
- key.base.nr_userclip_plane_consts
- = _mesa_bitcount_64(ctx->Transform.ClipPlanesEnabled);
- key.base.userclip_planes_enabled_gen_4_5
- = ctx->Transform.ClipPlanesEnabled;
- } else {
- key.base.nr_userclip_plane_consts
- = _mesa_logbase2(ctx->Transform.ClipPlanesEnabled) + 1;
- }
- }
+ brw_setup_vec4_key_clip_info(brw, &key.base, vp->program.UsesClipDistance);
/* _NEW_POLYGON */
- if (intel->gen < 6) {
+ if (brw->gen < 6) {
key.copy_edgeflag = (ctx->Polygon.FrontMode != GL_FILL ||
ctx->Polygon.BackMode != GL_FILL);
}
key.base.clamp_vertex_color = ctx->Light._ClampVertexColor;
/* _NEW_POINT */
- if (intel->gen < 6 && ctx->Point.PointSprite) {
+ if (brw->gen < 6 && ctx->Point.PointSprite) {
for (i = 0; i < 8; i++) {
if (ctx->Point.CoordReplace[i])
key.point_coord_replace |= (1 << i);
}
/* _NEW_TEXTURE */
- brw_populate_sampler_prog_key_data(ctx, prog, &key.base.tex);
+ brw_populate_sampler_prog_key_data(ctx, prog, brw->vs.base.sampler_count,
+ &key.base.tex);
/* BRW_NEW_VERTICES */
- if (intel->gen < 8 && !intel->is_haswell) {
+ if (brw->gen < 8 && !brw->is_haswell) {
/* Prior to Haswell, the hardware can't natively support GL_FIXED or
* 2_10_10_10_REV vertex formats. Set appropriate workaround flags.
*/
if (!brw_search_cache(&brw->cache, BRW_VS_PROG,
&key, sizeof(key),
- &brw->vs.prog_offset, &brw->vs.prog_data)) {
+ &brw->vs.base.prog_offset, &brw->vs.prog_data)) {
bool success = do_vs_prog(brw, ctx->Shader.CurrentVertexProgram,
vp, &key);
-
+ (void) success;
assert(success);
}
+ brw->vs.base.prog_data = &brw->vs.prog_data->base.base;
+
if (memcmp(&brw->vs.prog_data->base.vue_map, &brw->vue_map_geom_out,
sizeof(brw->vue_map_geom_out)) != 0) {
- brw->vue_map_geom_out = brw->vs.prog_data->base.vue_map;
- brw->state.dirty.brw |= BRW_NEW_VUE_MAP_GEOM_OUT;
+ brw->vue_map_vs = brw->vs.prog_data->base.vue_map;
+ brw->state.dirty.brw |= BRW_NEW_VUE_MAP_VS;
+ if (brw->gen < 7) {
+ /* No geometry shader support, so the VS VUE map is the VUE map for
+ * the output of the "geometry" portion of the pipeline.
+ */
+ brw->vue_map_geom_out = brw->vue_map_vs;
+ brw->state.dirty.brw |= BRW_NEW_VUE_MAP_GEOM_OUT;
+ }
}
}
{
struct brw_context *brw = brw_context(ctx);
struct brw_vs_prog_key key;
- uint32_t old_prog_offset = brw->vs.prog_offset;
+ uint32_t old_prog_offset = brw->vs.base.prog_offset;
struct brw_vs_prog_data *old_prog_data = brw->vs.prog_data;
bool success;
key.base.program_string_id = bvp->id;
key.base.clamp_vertex_color = ctx->API == API_OPENGL_COMPAT;
- for (int i = 0; i < MAX_SAMPLERS; i++) {
+ unsigned sampler_count = _mesa_fls(vp->Base.SamplersUsed);
+ for (unsigned i = 0; i < sampler_count; i++) {
if (vp->Base.ShadowSamplers & (1 << i)) {
/* Assume DEPTH_TEXTURE_MODE is the default: X, X, X, 1 */
key.base.tex.swizzles[i] =
success = do_vs_prog(brw, prog, bvp, &key);
- brw->vs.prog_offset = old_prog_offset;
+ brw->vs.base.prog_offset = old_prog_offset;
brw->vs.prog_data = old_prog_data;
return success;
}
-void
-brw_vec4_prog_data_free(const struct brw_vec4_prog_data *prog_data)
-{
- ralloc_free((void *)prog_data->param);
- ralloc_free((void *)prog_data->pull_param);
-}
-
-
void
brw_vs_prog_data_free(const void *in_prog_data)
{
projects / mesa.git / blobdiff