i965: Calculate thread_count in brw_alloc_stage_scratch

[mesa.git] / src / mesa / drivers / dri / i965 / brw_vs.c

diff --git a/src/mesa/drivers/dri/i965/brw_vs.c b/src/mesa/drivers/dri/i965/brw_vs.c
index a2f79a66ef9992502c907ab85661671d7d340a89..de0ba79f3577589b3a5ea753dcfcba2852a4f0ef 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_vs.c
+++ b/src/mesa/drivers/dri/i965/brw_vs.c
@@ -159,7 +159,6 @@ brw_codegen_vs_prog(struct brw_context *brw,
 {
    const struct brw_compiler *compiler = brw->screen->compiler;
    const struct gen_device_info *devinfo = &brw->screen->devinfo;
-   GLuint program_size;
    const GLuint *program;
    struct brw_vs_prog_data prog_data;
    struct brw_stage_prog_data *stage_prog_data = &prog_data.base.base;
@@ -178,41 +177,19 @@ brw_codegen_vs_prog(struct brw_context *brw,
    brw_assign_common_binding_table_offsets(devinfo, &vp->program,
                                            &prog_data.base.base, 0);
 
-   /* Allocate the references to the uniforms that will end up in the
-    * prog_data associated with the compiled program, and which will be freed
-    * by the state cache.
-    */
-   int param_count = vp->program.nir->num_uniforms / 4;
-
-   prog_data.base.base.nr_image_params = vp->program.info.num_images;
-
-   /* vec4_visitor::setup_uniform_clipplane_values() also uploads user clip
-    * planes as uniforms.
-    */
-   param_count += key->nr_userclip_plane_consts * 4;
-
-   stage_prog_data->param =
-      rzalloc_array(NULL, const gl_constant_value *, param_count);
-   stage_prog_data->pull_param =
-      rzalloc_array(NULL, const gl_constant_value *, param_count);
-   stage_prog_data->image_param =
-      rzalloc_array(NULL, struct brw_image_param,
-                    stage_prog_data->nr_image_params);
-   stage_prog_data->nr_params = param_count;
-
    if (!vp->program.is_arb_asm) {
-      brw_nir_setup_glsl_uniforms(vp->program.nir, &vp->program,
+      brw_nir_setup_glsl_uniforms(mem_ctx, vp->program.nir, &vp->program,
                                   &prog_data.base.base,
                                   compiler->scalar_stage[MESA_SHADER_VERTEX]);
       brw_nir_analyze_ubo_ranges(compiler, vp->program.nir,
                                  prog_data.base.base.ubo_ranges);
    } else {
-      brw_nir_setup_arb_uniforms(vp->program.nir, &vp->program,
+      brw_nir_setup_arb_uniforms(mem_ctx, vp->program.nir, &vp->program,
                                  &prog_data.base.base);
    }
 
    uint64_t outputs_written =
-      brw_vs_outputs_written(brw, key, vp->program.info.outputs_written);
+      brw_vs_outputs_written(brw, key, vp->program.nir->info.outputs_written);
 
    brw_compute_vue_map(devinfo,
                        &prog_data.base.vue_map, outputs_written,
@@ -244,9 +221,8 @@ brw_codegen_vs_prog(struct brw_context *brw,
    char *error_str;
    program = brw_compile_vs(compiler, brw, mem_ctx, key, &prog_data,
                             vp->program.nir,
-                            brw_select_clip_planes(&brw->ctx),
                             !_mesa_is_gles3(&brw->ctx),
-                            st_index, &program_size, &error_str);
+                            st_index, &error_str);
    if (program == NULL) {
       if (!vp->program.is_arb_asm) {
          vp->program.sh.data->LinkStatus = linking_failure;
@@ -272,14 +248,16 @@ brw_codegen_vs_prog(struct brw_context *brw,
 
    /* Scratch space is used for register spilling */
    brw_alloc_stage_scratch(brw, &brw->vs.base,
-                           prog_data.base.base.total_scratch,
-                           devinfo->max_vs_threads);
+                           prog_data.base.base.total_scratch);
 
+   /* The param and pull_param arrays will be freed by the shader cache. */
+   ralloc_steal(NULL, prog_data.base.base.param);
+   ralloc_steal(NULL, prog_data.base.base.pull_param);
    brw_upload_cache(&brw->cache, BRW_CACHE_VS_PROG,
-                   key, sizeof(struct brw_vs_prog_key),
-                   program, program_size,
-                   &prog_data, sizeof(prog_data),
-                   &brw->vs.base.prog_offset, &brw->vs.base.prog_data);
+                    key, sizeof(struct brw_vs_prog_key),
+                    program, prog_data.base.base.program_size,
+                    &prog_data, sizeof(prog_data),
+                    &brw->vs.base.prog_offset, &brw->vs.base.prog_data);
    ralloc_free(mem_ctx);
 
    return true;
@@ -305,8 +283,8 @@ brw_vs_populate_key(struct brw_context *brw,
 {
    struct gl_context *ctx = &brw->ctx;
    /* BRW_NEW_VERTEX_PROGRAM */
-   struct brw_program *vp = (struct brw_program *)brw->vertex_program;
-   struct gl_program *prog = (struct gl_program *) brw->vertex_program;
+   struct gl_program *prog = brw->programs[MESA_SHADER_VERTEX];
+   struct brw_program *vp = (struct brw_program *) prog;
    const struct gen_device_info *devinfo = &brw->screen->devinfo;
 
    memset(key, 0, sizeof(*key));
@@ -345,7 +323,7 @@ brw_vs_populate_key(struct brw_context *brw,
    brw_populate_sampler_prog_key_data(ctx, prog, &key->tex);
 
    /* BRW_NEW_VS_ATTRIB_WORKAROUNDS */
-   if (devinfo->gen < 8 && !brw->is_haswell) {
+   if (devinfo->gen < 8 && !devinfo->is_haswell) {
       memcpy(key->gl_attrib_wa_flags, brw->vb.attrib_wa_flags,
              sizeof(brw->vb.attrib_wa_flags));
    }
@@ -356,7 +334,8 @@ brw_upload_vs_prog(struct brw_context *brw)
 {
    struct brw_vs_prog_key key;
    /* BRW_NEW_VERTEX_PROGRAM */
-   struct brw_program *vp = (struct brw_program *)brw->vertex_program;
+   struct brw_program *vp =
+      (struct brw_program *) brw->programs[MESA_SHADER_VERTEX];
 
    if (!brw_vs_state_dirty(brw))
       return;