Browse Source

iris: rewrite program cache to use u_upload_mgr

tags/19.1-branchpoint
Kenneth Graunke 7 years ago
parent
commit
9437e135ed
2 changed files with 41 additions and 111 deletions
  1. 13
    11
      src/gallium/drivers/iris/iris_context.h
  2. 28
    100
      src/gallium/drivers/iris/iris_program_cache.c

+ 13
- 11
src/gallium/drivers/iris/iris_context.h View File

@@ -34,6 +34,8 @@
struct iris_bo;
struct iris_batch;

#define IRIS_RESOURCE_FLAG_INSTRUCTION_CACHE (PIPE_RESOURCE_FLAG_DRV_PRIV << 0)

#define IRIS_MAX_TEXTURE_SAMPLERS 32
#define IRIS_MAX_VIEWPORTS 16

@@ -85,16 +87,15 @@ enum iris_program_cache_id {
IRIS_CACHE_BLORP_BLIT,
};

struct iris_program_cache {
struct hash_table *table;
struct iris_bo *bo;
void *map;
uint32_t next_offset;
};

struct iris_compiled_shader {
/** Offset to the start of the assembly in the program cache BO */
uint32_t prog_offset;
/** Buffer containing the uploaded assembly. */
struct pipe_resource *buffer;

/** Offset where the assembly lives in the BO. */
unsigned prog_offset;

/** Pointer to the assembly in the BO's map. */
void *map;

/** The program data (owned by the program cache hash table) */
struct brw_stage_prog_data *prog_data;
@@ -116,7 +117,8 @@ struct iris_context {
struct iris_compiled_shader *prog[MESA_SHADER_STAGES];
struct brw_vue_map *last_vue_map;

struct iris_program_cache cache;
struct u_upload_mgr *uploader;
struct hash_table *cache;
} shaders;

/** The main batch for rendering */
@@ -207,7 +209,7 @@ void iris_upload_and_bind_shader(struct iris_context *ice,
const void *key,
const void *assembly,
struct brw_stage_prog_data *prog_data);
const void *iris_find_previous_compile(struct iris_program_cache *cache,
const void *iris_find_previous_compile(const struct iris_context *ice,
enum iris_program_cache_id cache_id,
unsigned program_string_id);
#endif

+ 28
- 100
src/gallium/drivers/iris/iris_program_cache.c View File

@@ -22,12 +22,12 @@
*/
#include <stdio.h>
#include <errno.h>
#include <i915_drm.h>
#include "pipe/p_defines.h"
#include "pipe/p_state.h"
#include "pipe/p_context.h"
#include "pipe/p_screen.h"
#include "util/u_atomic.h"
#include "util/u_upload_mgr.h"
#include "compiler/nir/nir.h"
#include "compiler/nir/nir_builder.h"
#include "intel/compiler/brw_compiler.h"
@@ -42,7 +42,7 @@ struct keybox {
};

static struct keybox *
make_keybox(struct iris_program_cache *cache,
make_keybox(void *mem_ctx,
enum iris_program_cache_id cache_id,
const void *key)
{
@@ -57,7 +57,7 @@ make_keybox(struct iris_program_cache *cache,
};

struct keybox *keybox =
ralloc_size(cache->table, sizeof(struct keybox) + key_sizes[cache_id]);
ralloc_size(mem_ctx, sizeof(struct keybox) + key_sizes[cache_id]);

keybox->cache_id = cache_id;
keybox->size = key_sizes[cache_id];
@@ -121,11 +121,9 @@ iris_bind_cached_shader(struct iris_context *ice,
enum iris_program_cache_id cache_id,
const void *key)
{
struct iris_program_cache *cache = &ice->shaders.cache;

struct keybox *keybox = make_keybox(cache, cache_id, key);

struct hash_entry *entry = _mesa_hash_table_search(cache->table, keybox);
struct keybox *keybox = make_keybox(ice->shaders.cache, cache_id, key);
struct hash_entry *entry =
_mesa_hash_table_search(ice->shaders.cache, keybox);

if (!entry)
return false;
@@ -141,45 +139,12 @@ iris_bind_cached_shader(struct iris_context *ice,
return true;
}

static void
recreate_cache_bo(struct iris_context *ice, uint32_t size)
{
struct iris_program_cache *cache = &ice->shaders.cache;
struct iris_screen *screen = (struct iris_screen *)ice->ctx.screen;
struct iris_bo *old_bo = cache->bo;
void *old_map = cache->map;

cache->bo = iris_bo_alloc(screen->bufmgr, "program cache", size, 64);
cache->bo->kflags = EXEC_OBJECT_CAPTURE | EXEC_OBJECT_PINNED;
cache->map = iris_bo_map(&ice->dbg, cache->bo,
MAP_READ | MAP_WRITE | MAP_ASYNC | MAP_PERSISTENT);

/* Copy any existing data that needs to be saved. */
if (old_bo) {
perf_debug(&ice->dbg,
"Copying to larger program cache: %u kB -> %u kB\n",
(unsigned) old_bo->size / 1024,
(unsigned) cache->bo->size / 1024);

/* Put the new BO just past the old one */
cache->bo->gtt_offset = ALIGN(old_bo->gtt_offset + old_bo->size, 4096);

memcpy(cache->map, old_map, cache->next_offset);

iris_bo_unreference(old_bo);
iris_bo_unmap(old_bo);
} else {
/* Put the initial cache BO...somewhere. */
cache->bo->gtt_offset = 4096 * 10;
}
}

const void *
iris_find_previous_compile(struct iris_program_cache *cache,
iris_find_previous_compile(const struct iris_context *ice,
enum iris_program_cache_id cache_id,
unsigned program_string_id)
{
hash_table_foreach(cache->table, entry) {
hash_table_foreach(ice->shaders.cache, entry) {
const struct keybox *keybox = entry->key;
if (keybox->cache_id == cache_id &&
get_program_string_id(cache_id, keybox->data) == program_string_id) {
@@ -198,48 +163,19 @@ iris_find_previous_compile(struct iris_program_cache *cache,
* in our backend. This saves space in the program cache buffer.
*/
static const struct iris_compiled_shader *
find_existing_assembly(const struct iris_program_cache *cache,
find_existing_assembly(struct hash_table *cache,
const void *assembly,
unsigned assembly_size)
{
hash_table_foreach(cache->table, entry) {
hash_table_foreach(cache, entry) {
const struct iris_compiled_shader *existing = entry->data;
if (existing->prog_data->program_size == assembly_size &&
memcmp(cache->map + existing->prog_offset,
assembly, assembly_size) == 0)
memcmp(existing->map, assembly, assembly_size) == 0)
return existing;
}
return NULL;
}

static uint32_t
upload_new_assembly(struct iris_context *ice,
const void *assembly,
unsigned size)
{
struct iris_program_cache *cache = &ice->shaders.cache;

/* Allocate space in the cache BO for our new program. */
if (cache->next_offset + size > cache->bo->size) {
uint32_t new_size = cache->bo->size * 2;

while (cache->next_offset + size > new_size)
new_size *= 2;

recreate_cache_bo(ice, new_size);
}

uint32_t offset = cache->next_offset;

/* Programs are always 64-byte aligned, so set up the next one now */
cache->next_offset = ALIGN(offset + size, 64);

/* Copy data to the buffer */
memcpy(cache->map + offset, assembly, size);

return offset;
}

/**
* Upload a new shader to the program cache, and bind it for use.
*
@@ -254,9 +190,9 @@ iris_upload_and_bind_shader(struct iris_context *ice,
{
struct iris_screen *screen = (void *) ice->ctx.screen;
struct gen_device_info *devinfo = &screen->devinfo;
struct iris_program_cache *cache = &ice->shaders.cache;
struct hash_table *cache = ice->shaders.cache;
struct iris_compiled_shader *shader =
ralloc_size(cache->table, sizeof(struct iris_compiled_shader) +
ralloc_size(cache, sizeof(struct iris_compiled_shader) +
ice->state.derived_program_state_size(cache_id));
const struct iris_compiled_shader *existing =
find_existing_assembly(cache, assembly, prog_data->program_size);
@@ -270,8 +206,10 @@ iris_upload_and_bind_shader(struct iris_context *ice,
if (existing) {
shader->prog_offset = existing->prog_offset;
} else {
shader->prog_offset =
upload_new_assembly(ice, assembly, prog_data->program_size);
shader->buffer = NULL;
u_upload_alloc(ice->shaders.uploader, 0, prog_data->program_size,
64, &shader->prog_offset, &shader->buffer, &shader->map);
memcpy(shader->map, assembly, prog_data->program_size);
}

shader->prog_data = prog_data;
@@ -284,7 +222,7 @@ iris_upload_and_bind_shader(struct iris_context *ice,
ice->state.set_derived_program_state(devinfo, cache_id, shader);

struct keybox *keybox = make_keybox(cache, cache_id, key);
_mesa_hash_table_insert(cache->table, keybox, shader);
_mesa_hash_table_insert(ice->shaders.cache, keybox, shader);

if (cache_id <= MESA_SHADER_STAGES) {
ice->shaders.prog[cache_id] = shader;
@@ -295,33 +233,24 @@ iris_upload_and_bind_shader(struct iris_context *ice,
void
iris_init_program_cache(struct iris_context *ice)
{
struct iris_program_cache *cache = &ice->shaders.cache;
ice->shaders.cache =
_mesa_hash_table_create(ice, keybox_hash, keybox_equals);

cache->table = _mesa_hash_table_create(ice, keybox_hash, keybox_equals);
recreate_cache_bo(ice, 16384);
ice->shaders.uploader =
u_upload_create(&ice->ctx, 16384, PIPE_BIND_CUSTOM, PIPE_USAGE_IMMUTABLE,
IRIS_RESOURCE_FLAG_INSTRUCTION_CACHE);
}

void
iris_destroy_program_cache(struct iris_context *ice)
{
struct iris_program_cache *cache = &ice->shaders.cache;

/* This can be NULL if context creation failed early on. */
if (cache->bo) {
iris_bo_unmap(cache->bo);
iris_bo_unreference(cache->bo);
cache->bo = NULL;
cache->map = NULL;
}

cache->next_offset = 0;

for (int i = 0; i < MESA_SHADER_STAGES; i++) {
ice->shaders.prog[i] = NULL;
}

ralloc_free(cache->table);
u_upload_destroy(ice->shaders.uploader);

ralloc_free(ice->shaders.cache);
}

static const char *
@@ -336,15 +265,14 @@ cache_name(enum iris_program_cache_id cache_id)
void
iris_print_program_cache(struct iris_context *ice)
{
struct iris_program_cache *cache = &ice->shaders.cache;
struct iris_screen *screen = (struct iris_screen *)ice->ctx.screen;
const struct gen_device_info *devinfo = &screen->devinfo;

hash_table_foreach(cache->table, entry) {
hash_table_foreach(ice->shaders.cache, entry) {
const struct keybox *keybox = entry->key;
struct iris_compiled_shader *shader = entry->data;
fprintf(stderr, "%s:\n", cache_name(keybox->cache_id));
brw_disassemble(devinfo, cache->map, shader->prog_offset,
brw_disassemble(devinfo, shader->map, 0,
shader->prog_data->program_size, stderr);
}
}

Loading…
Cancel
Save