radeonsi/compute: Clamp COMPUTE_TMPRING_SIZE.WAVES to: num_cu * 32

[mesa.git] / src / gallium / drivers / radeonsi / si_compute.c

diff --git a/src/gallium/drivers/radeonsi/si_compute.c b/src/gallium/drivers/radeonsi/si_compute.c
index be6441817963b1b6f99391194fee59a911f0cd2c..bf935dcfee7aa82543582c60705158afba7a0a4c 100644 (file)
--- a/src/gallium/drivers/radeonsi/si_compute.c
+++ b/src/gallium/drivers/radeonsi/si_compute.c
@@ -23,14 +23,15 @@
  */
 
 #include "util/u_memory.h"
+#include "radeon/r600_pipe_common.h"
+#include "radeon/radeon_elf_util.h"
+#include "radeon/radeon_llvm_util.h"
 
 #include "radeon/r600_cs.h"
 #include "si_pipe.h"
 #include "si_shader.h"
 #include "sid.h"
 
-#include "radeon/radeon_llvm_util.h"
-
 #define MAX_GLOBAL_BUFFERS 20
 #if HAVE_LLVM < 0x0305
 #define NUM_USER_SGPRS 2
@@ -44,14 +45,18 @@ struct si_compute {
        unsigned local_size;
        unsigned private_size;
        unsigned input_size;
-       unsigned num_kernels;
-       struct si_shader *kernels;
+       struct radeon_shader_binary binary;
+       struct si_shader program;
        unsigned num_user_sgprs;
 
        struct r600_resource *input_buffer;
        struct pipe_resource *global_buffers[MAX_GLOBAL_BUFFERS];
 
+#if HAVE_LLVM < 0x0306
+       unsigned num_kernels;
+       struct si_shader *kernels;
        LLVMContextRef llvm_ctx;
+#endif
 };
 
 static void *si_create_compute_state(
@@ -61,10 +66,7 @@ static void *si_create_compute_state(
        struct si_context *sctx = (struct si_context *)ctx;
        struct si_compute *program = CALLOC_STRUCT(si_compute);
        const struct pipe_llvm_program_header *header;
-       const unsigned char *code;
-       unsigned i;
-
-       program->llvm_ctx = LLVMContextCreate();
+       const char *code;
 
        header = cso->prog;
        code = cso->prog + sizeof(struct pipe_llvm_program_header);
@@ -74,17 +76,27 @@ static void *si_create_compute_state(
        program->private_size = cso->req_private_mem;
        program->input_size = cso->req_input_mem;
 
-       program->num_kernels = radeon_llvm_get_num_kernels(program->llvm_ctx, code,
-                                                       header->num_bytes);
-       program->kernels = CALLOC(sizeof(struct si_shader),
-                                                       program->num_kernels);
-       for (i = 0; i < program->num_kernels; i++) {
-               LLVMModuleRef mod = radeon_llvm_get_kernel_module(program->llvm_ctx, i,
-                                                       code, header->num_bytes);
-               si_compile_llvm(sctx->screen, &program->kernels[i], mod);
-               LLVMDisposeModule(mod);
+#if HAVE_LLVM < 0x0306
+       {
+               unsigned i;
+               program->llvm_ctx = LLVMContextCreate();
+               program->num_kernels = radeon_llvm_get_num_kernels(program->llvm_ctx,
+                                       code, header->num_bytes);
+               program->kernels = CALLOC(sizeof(struct si_shader),
+                                                        program->num_kernels);
+               for (i = 0; i < program->num_kernels; i++) {
+                       LLVMModuleRef mod = radeon_llvm_get_kernel_module(program->llvm_ctx, i,
+                                                        code, header->num_bytes);
+                       si_compile_llvm(sctx->screen, &program->kernels[i], mod);
+                       LLVMDisposeModule(mod);
+               }
        }
+#else
 
+       radeon_elf_read(code, header->num_bytes, &program->binary, true);
+       si_shader_binary_read(sctx->screen, &program->program, &program->binary);
+
+#endif
        program->input_buffer = si_resource_create_custom(sctx->b.b.screen,
                PIPE_USAGE_IMMUTABLE, program->input_size);
 
@@ -181,10 +193,15 @@ static void si_launch_grid(
        uint64_t shader_va;
        unsigned arg_user_sgpr_count = NUM_USER_SGPRS;
        unsigned i;
-       struct si_shader *shader = &program->kernels[pc];
+       struct si_shader *shader = &program->program;
        unsigned lds_blocks;
        unsigned num_waves_for_scratch;
 
+#if HAVE_LLVM < 0x0306
+       shader = &program->kernels[pc];
+#endif
+
+
        radeon_emit(cs, PKT3(PKT3_CONTEXT_CONTROL, 1, 0) | PKT3_SHADER_TYPE_S(1));
        radeon_emit(cs, 0x80000000);
        radeon_emit(cs, 0x80000000);
@@ -198,6 +215,11 @@ static void si_launch_grid(
 
        pm4->compute_pkt = true;
 
+#if HAVE_LLVM >= 0x0306
+       /* Read the config information */
+       si_shader_binary_read_config(&program->binary, &program->program, pc);
+#endif
+
        /* Upload the kernel arguments */
 
        /* The extra num_work_size_bytes are for work group / work item size information */
@@ -290,6 +312,10 @@ static void si_launch_grid(
        }
 
        shader_va = shader->bo->gpu_address;
+
+#if HAVE_LLVM >= 0x0306
+       shader_va += pc;
+#endif
        si_pm4_add_bo(pm4, shader->bo, RADEON_USAGE_READ, RADEON_PRIO_SHADER_DATA);
        si_pm4_set_reg(pm4, R_00B830_COMPUTE_PGM_LO, (shader_va >> 8) & 0xffffffff);
        si_pm4_set_reg(pm4, R_00B834_COMPUTE_PGM_HI, shader_va >> 40);
@@ -345,6 +371,9 @@ static void si_launch_grid(
                | S_00B85C_SH1_CU_EN(0xffff /* Default value */))
                ;
 
+       num_waves_for_scratch =
+               MIN2(num_waves_for_scratch,
+                    32 * sctx->screen->b.info.max_compute_units);
        si_pm4_set_reg(pm4, R_00B860_COMPUTE_TMPRING_SIZE,
                /* The maximum value for WAVES is 32 * num CU.
                 * If you program this value incorrectly, the GPU will hang if
@@ -388,6 +417,7 @@ static void si_delete_compute_state(struct pipe_context *ctx, void* state){
                return;
        }
 
+#if HAVE_LLVM < 0x0306
        if (program->kernels) {
                for (int i = 0; i < program->num_kernels; i++){
                        if (program->kernels[i].bo){
@@ -400,10 +430,16 @@ static void si_delete_compute_state(struct pipe_context *ctx, void* state){
        if (program->llvm_ctx){
                LLVMContextDispose(program->llvm_ctx);
        }
+#else
+       si_shader_destroy(ctx, &program->program);
+#endif
+
        pipe_resource_reference(
                (struct pipe_resource **)&program->input_buffer, NULL);
 
-       //And then free the program itself.
+       FREE(program->binary.code);
+       FREE(program->binary.config);
+       FREE(program->binary.rodata);
        FREE(program);
 }