i965: Calculate thread_count in brw_alloc_stage_scratch

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

diff --git a/src/mesa/drivers/dri/i965/brw_cs.c b/src/mesa/drivers/dri/i965/brw_cs.c
index 0c505b3e505d1bdcfc3414529da978383da60940..1d34a8a79d18cd18b58f218decd0c25a05c93788 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_cs.c
+++ b/src/mesa/drivers/dri/i965/brw_cs.c
@@ -53,10 +53,8 @@ brw_codegen_cs_prog(struct brw_context *brw,
                     struct brw_cs_prog_key *key)
 {
    const struct gen_device_info *devinfo = &brw->screen->devinfo;
-   struct gl_context *ctx = &brw->ctx;
    const GLuint *program;
    void *mem_ctx = ralloc_context(NULL);
-   GLuint program_size;
    struct brw_cs_prog_data prog_data;
    bool start_busy = false;
    double start_time = 0;
@@ -78,23 +76,8 @@ brw_codegen_cs_prog(struct brw_context *brw,
 
    assign_cs_binding_table_offsets(devinfo, &cp->program, &prog_data);
 
-   /* 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 = cp->program.nir->num_uniforms / 4;
-
-   /* The backend also sometimes add a param for the thread local id. */
-   prog_data.thread_local_id_index = param_count++;
-
-   /* The backend also sometimes adds params for texture size. */
-   param_count += 2 * ctx->Const.Program[MESA_SHADER_COMPUTE].MaxTextureImageUnits;
-   prog_data.base.param = rzalloc_array(NULL, uint32_t, param_count);
-   prog_data.base.pull_param = rzalloc_array(NULL, uint32_t, param_count);
-   prog_data.base.nr_params = param_count;
-
-   brw_nir_setup_glsl_uniforms(cp->program.nir, &cp->program,&prog_data.base,
-                               true);
+   brw_nir_setup_glsl_uniforms(mem_ctx, cp->program.nir,
+                               &cp->program, &prog_data.base, true);
 
    if (unlikely(brw->perf_debug)) {
       start_busy = (brw->batch.last_bo &&
@@ -109,7 +92,7 @@ brw_codegen_cs_prog(struct brw_context *brw,
    char *error_str;
    program = brw_compile_cs(brw->screen->compiler, brw, mem_ctx, key,
                             &prog_data, cp->program.nir, st_index,
-                            &program_size, &error_str);
+                            &error_str);
    if (program == NULL) {
       cp->program.sh.data->LinkStatus = linking_failure;
       ralloc_strcat(&cp->program.sh.data->InfoLog, error_str);
@@ -131,33 +114,14 @@ brw_codegen_cs_prog(struct brw_context *brw,
       }
    }
 
-   const unsigned subslices = MAX2(brw->screen->subslice_total, 1);
-
-   /* WaCSScratchSize:hsw
-    *
-    * Haswell's scratch space address calculation appears to be sparse
-    * rather than tightly packed.  The Thread ID has bits indicating
-    * which subslice, EU within a subslice, and thread within an EU
-    * it is.  There's a maximum of two slices and two subslices, so these
-    * can be stored with a single bit.  Even though there are only 10 EUs
-    * per subslice, this is stored in 4 bits, so there's an effective
-    * maximum value of 16 EUs.  Similarly, although there are only 7
-    * threads per EU, this is stored in a 3 bit number, giving an effective
-    * maximum value of 8 threads per EU.
-    *
-    * This means that we need to use 16 * 8 instead of 10 * 7 for the
-    * number of threads per subslice.
-    */
-   const unsigned scratch_ids_per_subslice =
-      devinfo->is_haswell ? 16 * 8 : devinfo->max_cs_threads;
-
-   brw_alloc_stage_scratch(brw, &brw->cs.base,
-                           prog_data.base.total_scratch,
-                           scratch_ids_per_subslice * subslices);
+   brw_alloc_stage_scratch(brw, &brw->cs.base, prog_data.base.total_scratch);
 
+   /* The param and pull_param arrays will be freed by the shader cache. */
+   ralloc_steal(NULL, prog_data.base.param);
+   ralloc_steal(NULL, prog_data.base.pull_param);
    brw_upload_cache(&brw->cache, BRW_CACHE_CS_PROG,
                     key, sizeof(*key),
-                    program, program_size,
+                    program, prog_data.base.program_size,
                     &prog_data, sizeof(prog_data),
                     &brw->cs.base.prog_offset, &brw->cs.base.prog_data);
    ralloc_free(mem_ctx);