[Mesa-dev] [PATCH 08/23] i965/vec4: Print disassembly after compaction.
Juha-Pekka Heikkilä
juhapekka.heikkila at gmail.com
Wed May 21 10:05:21 PDT 2014
On Mon, May 19, 2014 at 9:55 PM, Matt Turner <mattst88 at gmail.com> wrote:
> ---
> src/mesa/drivers/dri/i965/brw_vec4.h | 4 +-
> src/mesa/drivers/dri/i965/brw_vec4_generator.cpp | 109 +++++++++++++----------
> 2 files changed, 66 insertions(+), 47 deletions(-)
>
> diff --git a/src/mesa/drivers/dri/i965/brw_vec4.h b/src/mesa/drivers/dri/i965/brw_vec4.h
> index a86972a..3a1eb12 100644
> --- a/src/mesa/drivers/dri/i965/brw_vec4.h
> +++ b/src/mesa/drivers/dri/i965/brw_vec4.h
> @@ -36,6 +36,7 @@ extern "C" {
>
> #include "brw_context.h"
> #include "brw_eu.h"
> +#include "intel_asm_printer.h"
>
> #ifdef __cplusplus
> }; /* extern "C" */
> @@ -650,7 +651,8 @@ public:
> const unsigned *generate_assembly(exec_list *insts, unsigned *asm_size);
>
> private:
> - void generate_code(exec_list *instructions);
> + void generate_code(exec_list *instructions, int *num_annotations,
> + struct annotation **annotation);
> void generate_vec4_instruction(vec4_instruction *inst,
> struct brw_reg dst,
> struct brw_reg *src);
> diff --git a/src/mesa/drivers/dri/i965/brw_vec4_generator.cpp b/src/mesa/drivers/dri/i965/brw_vec4_generator.cpp
> index a91bfe7..2176de4 100644
> --- a/src/mesa/drivers/dri/i965/brw_vec4_generator.cpp
> +++ b/src/mesa/drivers/dri/i965/brw_vec4_generator.cpp
> @@ -21,6 +21,7 @@
> */
>
> #include "brw_vec4.h"
> +#include "brw_cfg.h"
>
> extern "C" {
> #include "brw_eu.h"
> @@ -1260,12 +1261,9 @@ vec4_generator::generate_vec4_instruction(vec4_instruction *instruction,
> }
>
> void
> -vec4_generator::generate_code(exec_list *instructions)
> +vec4_generator::generate_code(exec_list *instructions, int *num_annotations,
> + struct annotation **annotation)
> {
> - int last_native_insn_offset = 0;
> - const char *last_annotation_string = NULL;
> - const void *last_annotation_ir = NULL;
> -
> if (unlikely(debug_flag)) {
> if (shader_prog) {
> fprintf(stderr, "Native code for %s vertex shader %d:\n",
> @@ -1276,32 +1274,52 @@ vec4_generator::generate_code(exec_list *instructions)
> }
> }
>
> + int block_num = 0;
> + int ann_num = 0;
> + int ann_size = 1024;
> + cfg_t *cfg = NULL;
> + struct annotation *ann = NULL;
> +
> + if (unlikely(debug_flag)) {
> + cfg = new(mem_ctx) cfg_t(instructions);
> + ann = rzalloc_array(NULL, struct annotation, ann_size);
Same as with previous patch, Klocwork will tell me this is 'critical'
issue as cfg and ann are not checked for nullness.
> + }
> +
> foreach_list(node, instructions) {
> vec4_instruction *inst = (vec4_instruction *)node;
> struct brw_reg src[3], dst;
>
> if (unlikely(debug_flag)) {
> - if (last_annotation_ir != inst->ir) {
> - last_annotation_ir = inst->ir;
> - if (last_annotation_ir) {
> - fprintf(stderr, " ");
> - if (shader_prog) {
> - ((ir_instruction *) last_annotation_ir)->fprint(stderr);
> - } else {
> - const prog_instruction *vpi;
> - vpi = (const prog_instruction *) inst->ir;
> - fprintf(stderr, "%d: ", (int)(vpi - prog->Instructions));
> - _mesa_fprint_instruction_opt(stderr, vpi, 0,
> - PROG_PRINT_DEBUG, NULL);
> - }
> - fprintf(stderr, "\n");
> - }
> - }
> - if (last_annotation_string != inst->annotation) {
> - last_annotation_string = inst->annotation;
> - if (last_annotation_string)
> - fprintf(stderr, " %s\n", last_annotation_string);
> - }
> + if (ann_num == ann_size) {
> + ann_size *= 2;
> + ann = reralloc(NULL, ann, struct annotation, ann_size);
reralloc can return null here.
> + }
> +
> + ann[ann_num].offset = p->next_insn_offset;
> + ann[ann_num].ir = inst->ir;
> + ann[ann_num].annotation = inst->annotation;
> +
> + if (cfg->blocks[block_num]->start == inst) {
> + ann[ann_num].block_start = cfg->blocks[block_num];
> + }
> +
> + /* There is no hardware DO instruction on Gen6+, so since DO always
> + * starts a basic block, we need to set the .block_start of the next
> + * instruction's annotation with a pointer to the bblock started by
> + * the DO.
> + *
> + * There's also only complication from emitting an annotation without
> + * a corresponding hardware instruction to disassemble.
> + */
> + if (brw->gen >= 6 && inst->opcode == BRW_OPCODE_DO) {
> + ann_num--;
> + }
> +
> + if (cfg->blocks[block_num]->end == inst) {
> + ann[ann_num].block_end = cfg->blocks[block_num];
> + block_num++;
> + }
> + ann_num++;
> }
>
> for (unsigned int i = 0; i < 3; i++) {
> @@ -1332,38 +1350,37 @@ vec4_generator::generate_code(exec_list *instructions)
> if (inst->no_dd_check)
> last->header.dependency_control |= BRW_DEPENDENCY_NOTCHECKED;
> }
> -
> - if (unlikely(debug_flag)) {
> - brw_disassemble(brw, p->store,
> - last_native_insn_offset, p->next_insn_offset, stderr);
> - }
> -
> - last_native_insn_offset = p->next_insn_offset;
> - }
> -
> - if (unlikely(debug_flag)) {
> - fprintf(stderr, "\n");
> }
>
> brw_set_uip_jip(p);
>
> - /* OK, while the INTEL_DEBUG=vs above is very nice for debugging VS
> - * emit issues, it doesn't get the jump distances into the output,
> - * which is often something we want to debug. So this is here in
> - * case you're doing that.
> - */
> - if (0 && unlikely(debug_flag)) {
> - brw_disassemble(brw, p->store, 0, p->next_insn_offset, stderr);
> + if (unlikely(debug_flag)) {
> + if (ann_num == ann_size) {
> + ann = reralloc(NULL, ann, struct annotation, ann_size + 1);
null check here for ann
> + }
> + ann[ann_num].offset = p->next_insn_offset;
> }
> + *num_annotations = ann_num;
> + *annotation = ann;
> }
>
> const unsigned *
> vec4_generator::generate_assembly(exec_list *instructions,
> unsigned *assembly_size)
> {
> + struct annotation *annotation;
> + int num_annotations;
> +
> brw_set_access_mode(p, BRW_ALIGN_16);
> - generate_code(instructions);
> - brw_compact_instructions(p, 0, 0, NULL);
> + generate_code(instructions, &num_annotations, &annotation);
> + brw_compact_instructions(p, 0, num_annotations, annotation);
> +
> + if (unlikely(debug_flag)) {
> + dump_assembly(p->store, num_annotations, annotation, brw, prog,
> + brw_disassemble);
> + ralloc_free(annotation);
> + }
> +
> return brw_get_program(p, assembly_size);
> }
>
> --
> 1.8.3.2
other than these null checks this look good to me.
More information about the mesa-dev
mailing list