ref: d05cf10fe718ebb09394d9c183ed046d05a8e6a2
dir: /vp9/decoder/vp9_dthread.c/
/*
* Copyright (c) 2014 The WebM project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#include "./vpx_config.h"
#include "vpx_mem/vpx_mem.h"
#include "vp9/common/vp9_reconinter.h"
#include "vp9/decoder/vp9_dthread.h"
#include "vp9/decoder/vp9_decoder.h"
// #define DEBUG_THREAD
#if CONFIG_MULTITHREAD
static INLINE void mutex_lock(pthread_mutex_t *const mutex) {
const int kMaxTryLocks = 4000;
int locked = 0;
int i;
for (i = 0; i < kMaxTryLocks; ++i) {
if (!pthread_mutex_trylock(mutex)) {
locked = 1;
break;
}
}
if (!locked)
pthread_mutex_lock(mutex);
}
#endif // CONFIG_MULTITHREAD
static INLINE void sync_read(VP9LfSync *const lf_sync, int r, int c) {
#if CONFIG_MULTITHREAD
const int nsync = lf_sync->sync_range;
if (r && !(c & (nsync - 1))) {
pthread_mutex_t *const mutex = &lf_sync->mutex_[r - 1];
mutex_lock(mutex);
while (c > lf_sync->cur_sb_col[r - 1] - nsync) {
pthread_cond_wait(&lf_sync->cond_[r - 1], mutex);
}
pthread_mutex_unlock(mutex);
}
#else
(void)lf_sync;
(void)r;
(void)c;
#endif // CONFIG_MULTITHREAD
}
static INLINE void sync_write(VP9LfSync *const lf_sync, int r, int c,
const int sb_cols) {
#if CONFIG_MULTITHREAD
const int nsync = lf_sync->sync_range;
int cur;
// Only signal when there are enough filtered SB for next row to run.
int sig = 1;
if (c < sb_cols - 1) {
cur = c;
if (c % nsync)
sig = 0;
} else {
cur = sb_cols + nsync;
}
if (sig) {
mutex_lock(&lf_sync->mutex_[r]);
lf_sync->cur_sb_col[r] = cur;
pthread_cond_signal(&lf_sync->cond_[r]);
pthread_mutex_unlock(&lf_sync->mutex_[r]);
}
#else
(void)lf_sync;
(void)r;
(void)c;
(void)sb_cols;
#endif // CONFIG_MULTITHREAD
}
// Implement row loopfiltering for each thread.
static void loop_filter_rows_mt(const YV12_BUFFER_CONFIG *const frame_buffer,
VP9_COMMON *const cm,
struct macroblockd_plane planes[MAX_MB_PLANE],
int start, int stop, int y_only,
VP9LfSync *const lf_sync, int num_lf_workers) {
const int num_planes = y_only ? 1 : MAX_MB_PLANE;
int r, c; // SB row and col
const int sb_cols = mi_cols_aligned_to_sb(cm->mi_cols) >> MI_BLOCK_SIZE_LOG2;
for (r = start; r < stop; r += num_lf_workers) {
const int mi_row = r << MI_BLOCK_SIZE_LOG2;
MODE_INFO **const mi = cm->mi_grid_visible + mi_row * cm->mi_stride;
for (c = 0; c < sb_cols; ++c) {
const int mi_col = c << MI_BLOCK_SIZE_LOG2;
LOOP_FILTER_MASK lfm;
int plane;
sync_read(lf_sync, r, c);
vp9_setup_dst_planes(planes, frame_buffer, mi_row, mi_col);
vp9_setup_mask(cm, mi_row, mi_col, mi + mi_col, cm->mi_stride, &lfm);
for (plane = 0; plane < num_planes; ++plane) {
vp9_filter_block_plane(cm, &planes[plane], mi_row, &lfm);
}
sync_write(lf_sync, r, c, sb_cols);
}
}
}
// Row-based multi-threaded loopfilter hook
static int loop_filter_row_worker(void *arg1, void *arg2) {
TileWorkerData *const tile_data = (TileWorkerData*)arg1;
LFWorkerData *const lf_data = &tile_data->lfdata;
(void) arg2;
loop_filter_rows_mt(lf_data->frame_buffer, lf_data->cm, lf_data->planes,
lf_data->start, lf_data->stop, lf_data->y_only,
lf_data->lf_sync, lf_data->num_lf_workers);
return 1;
}
// VP9 decoder: Implement multi-threaded loopfilter that uses the tile
// threads.
void vp9_loop_filter_frame_mt(YV12_BUFFER_CONFIG *frame,
VP9Decoder *pbi, VP9_COMMON *cm,
int frame_filter_level,
int y_only) {
VP9LfSync *const lf_sync = &pbi->lf_row_sync;
const VP9WorkerInterface *const winterface = vp9_get_worker_interface();
// Number of superblock rows and cols
const int sb_rows = mi_cols_aligned_to_sb(cm->mi_rows) >> MI_BLOCK_SIZE_LOG2;
const int tile_cols = 1 << cm->log2_tile_cols;
const int num_workers = MIN(pbi->max_threads & ~1, tile_cols);
int i;
// Allocate memory used in thread synchronization.
// This always needs to be done even if frame_filter_level is 0.
if (!cm->current_video_frame || cm->last_height != cm->height) {
if (cm->last_height != cm->height) {
const int aligned_last_height =
ALIGN_POWER_OF_TWO(cm->last_height, MI_SIZE_LOG2);
const int last_sb_rows =
mi_cols_aligned_to_sb(aligned_last_height >> MI_SIZE_LOG2) >>
MI_BLOCK_SIZE_LOG2;
vp9_loop_filter_dealloc(lf_sync, last_sb_rows);
}
vp9_loop_filter_alloc(cm, lf_sync, sb_rows, cm->width);
}
if (!frame_filter_level) return;
vp9_loop_filter_frame_init(cm, frame_filter_level);
// Initialize cur_sb_col to -1 for all SB rows.
vpx_memset(lf_sync->cur_sb_col, -1, sizeof(*lf_sync->cur_sb_col) * sb_rows);
// Set up loopfilter thread data.
// The decoder is using num_workers instead of pbi->num_tile_workers
// because it has been observed that using more threads on the
// loopfilter, than there are tile columns in the frame will hurt
// performance on Android. This is because the system will only
// schedule the tile decode workers on cores equal to the number
// of tile columns. Then if the decoder tries to use more threads for the
// loopfilter, it will hurt performance because of contention. If the
// multithreading code changes in the future then the number of workers
// used by the loopfilter should be revisited.
for (i = 0; i < num_workers; ++i) {
VP9Worker *const worker = &pbi->tile_workers[i];
TileWorkerData *const tile_data = (TileWorkerData*)worker->data1;
LFWorkerData *const lf_data = &tile_data->lfdata;
worker->hook = (VP9WorkerHook)loop_filter_row_worker;
// Loopfilter data
lf_data->frame_buffer = frame;
lf_data->cm = cm;
vp9_copy(lf_data->planes, pbi->mb.plane);
lf_data->start = i;
lf_data->stop = sb_rows;
lf_data->y_only = y_only; // always do all planes in decoder
lf_data->lf_sync = lf_sync;
lf_data->num_lf_workers = num_workers;
// Start loopfiltering
if (i == num_workers - 1) {
winterface->execute(worker);
} else {
winterface->launch(worker);
}
}
// Wait till all rows are finished
for (i = 0; i < num_workers; ++i) {
winterface->sync(&pbi->tile_workers[i]);
}
}
// Set up nsync by width.
static int get_sync_range(int width) {
// nsync numbers are picked by testing. For example, for 4k
// video, using 4 gives best performance.
if (width < 640)
return 1;
else if (width <= 1280)
return 2;
else if (width <= 4096)
return 4;
else
return 8;
}
// Allocate memory for lf row synchronization
void vp9_loop_filter_alloc(VP9_COMMON *cm, VP9LfSync *lf_sync, int rows,
int width) {
#if CONFIG_MULTITHREAD
int i;
CHECK_MEM_ERROR(cm, lf_sync->mutex_,
vpx_malloc(sizeof(*lf_sync->mutex_) * rows));
for (i = 0; i < rows; ++i) {
pthread_mutex_init(&lf_sync->mutex_[i], NULL);
}
CHECK_MEM_ERROR(cm, lf_sync->cond_,
vpx_malloc(sizeof(*lf_sync->cond_) * rows));
for (i = 0; i < rows; ++i) {
pthread_cond_init(&lf_sync->cond_[i], NULL);
}
#endif // CONFIG_MULTITHREAD
CHECK_MEM_ERROR(cm, lf_sync->cur_sb_col,
vpx_malloc(sizeof(*lf_sync->cur_sb_col) * rows));
// Set up nsync.
lf_sync->sync_range = get_sync_range(width);
}
// Deallocate lf synchronization related mutex and data
void vp9_loop_filter_dealloc(VP9LfSync *lf_sync, int rows) {
#if !CONFIG_MULTITHREAD
(void)rows;
#endif // !CONFIG_MULTITHREAD
if (lf_sync != NULL) {
#if CONFIG_MULTITHREAD
int i;
if (lf_sync->mutex_ != NULL) {
for (i = 0; i < rows; ++i) {
pthread_mutex_destroy(&lf_sync->mutex_[i]);
}
vpx_free(lf_sync->mutex_);
}
if (lf_sync->cond_ != NULL) {
for (i = 0; i < rows; ++i) {
pthread_cond_destroy(&lf_sync->cond_[i]);
}
vpx_free(lf_sync->cond_);
}
#endif // CONFIG_MULTITHREAD
vpx_free(lf_sync->cur_sb_col);
// clear the structure as the source of this call may be a resize in which
// case this call will be followed by an _alloc() which may fail.
vp9_zero(*lf_sync);
}
}
// TODO(hkuang): Clean up all the #ifdef in this file.
void vp9_frameworker_lock_stats(VP9Worker *const worker) {
#if CONFIG_MULTITHREAD
FrameWorkerData *const worker_data = worker->data1;
pthread_mutex_lock(&worker_data->stats_mutex);
#else
(void)worker;
#endif
}
void vp9_frameworker_unlock_stats(VP9Worker *const worker) {
#if CONFIG_MULTITHREAD
FrameWorkerData *const worker_data = worker->data1;
pthread_mutex_unlock(&worker_data->stats_mutex);
#else
(void)worker;
#endif
}
void vp9_frameworker_signal_stats(VP9Worker *const worker) {
#if CONFIG_MULTITHREAD
FrameWorkerData *const worker_data = worker->data1;
// TODO(hkuang): Investigate using broadcast or signal.
pthread_cond_signal(&worker_data->stats_cond);
#else
(void)worker;
#endif
}
// TODO(hkuang): Remove worker parameter as it is only used in debug code.
void vp9_frameworker_wait(VP9Worker *const worker, RefCntBuffer *const ref_buf,
int row) {
#if CONFIG_MULTITHREAD
if (!ref_buf)
return;
// Enabling the following line of code will get harmless tsan error but
// will get best performance.
// if (ref_buf->row >= row && ref_buf->buf.corrupted != 1) return;
{
// Find the worker thread that owns the reference frame. If the reference
// frame has been fully decoded, it may not have owner.
VP9Worker *const ref_worker = ref_buf->frame_worker_owner;
FrameWorkerData *const ref_worker_data =
(FrameWorkerData *)ref_worker->data1;
const VP9Decoder *const pbi = ref_worker_data->pbi;
#ifdef DEBUG_THREAD
{
FrameWorkerData *const worker_data = (FrameWorkerData *)worker->data1;
printf("%d %p worker is waiting for %d %p worker (%d) ref %d \r\n",
worker_data->worker_id, worker, ref_worker_data->worker_id,
ref_buf->frame_worker_owner, row, ref_buf->row);
}
#endif
vp9_frameworker_lock_stats(ref_worker);
while (ref_buf->row < row && pbi->cur_buf == ref_buf &&
ref_buf->buf.corrupted != 1) {
pthread_cond_wait(&ref_worker_data->stats_cond,
&ref_worker_data->stats_mutex);
}
if (ref_buf->buf.corrupted == 1) {
FrameWorkerData *const worker_data = (FrameWorkerData *)worker->data1;
vp9_frameworker_unlock_stats(ref_worker);
vpx_internal_error(&worker_data->pbi->common.error,
VPX_CODEC_CORRUPT_FRAME,
"Worker %p failed to decode frame", worker);
}
vp9_frameworker_unlock_stats(ref_worker);
}
#else
(void)ref_buf;
(void)row;
(void)ref_buf;
#endif // CONFIG_MULTITHREAD
}
void vp9_frameworker_broadcast(RefCntBuffer *const buf, int row) {
#if CONFIG_MULTITHREAD
VP9Worker *worker = buf->frame_worker_owner;
#ifdef DEBUG_THREAD
{
FrameWorkerData *const worker_data = (FrameWorkerData *)worker->data1;
printf("%d %p worker decode to (%d) \r\n", worker_data->worker_id,
buf->frame_worker_owner, row);
}
#endif
vp9_frameworker_lock_stats(worker);
buf->row = row;
vp9_frameworker_signal_stats(worker);
vp9_frameworker_unlock_stats(worker);
#else
(void)buf;
(void)row;
#endif // CONFIG_MULTITHREAD
}
void vp9_frameworker_copy_context(VP9Worker *const dst_worker,
VP9Worker *const src_worker) {
#if CONFIG_MULTITHREAD
FrameWorkerData *const src_worker_data = (FrameWorkerData *)src_worker->data1;
FrameWorkerData *const dst_worker_data = (FrameWorkerData *)dst_worker->data1;
VP9_COMMON *const src_cm = &src_worker_data->pbi->common;
VP9_COMMON *const dst_cm = &dst_worker_data->pbi->common;
int i;
// Wait until source frame's context is ready.
vp9_frameworker_lock_stats(src_worker);
while (!src_worker_data->frame_context_ready) {
pthread_cond_wait(&src_worker_data->stats_cond,
&src_worker_data->stats_mutex);
}
// src worker may have already finished decoding a frame and swapped the mi.
// TODO(hkuang): Remove following code after implenment no ModeInfo decoding.
if (src_worker_data->frame_decoded) {
dst_cm->prev_mip = src_cm->prev_mip;
dst_cm->prev_mi = src_cm->prev_mi;
dst_cm->prev_mi_grid_base = src_cm->prev_mi_grid_base;
dst_cm->prev_mi_grid_visible = src_cm->prev_mi_grid_visible;
dst_cm->last_frame_seg_map = src_cm->last_frame_seg_map;
} else {
dst_cm->prev_mip = src_cm->mip;
dst_cm->prev_mi = src_cm->mi;
dst_cm->prev_mi_grid_base = src_cm->mi_grid_base;
dst_cm->prev_mi_grid_visible = src_cm->mi_grid_visible;
dst_cm->last_frame_seg_map = src_cm->current_frame_seg_map;
}
dst_worker_data->pbi->need_resync = src_worker_data->pbi->need_resync;
vp9_frameworker_unlock_stats(src_worker);
dst_worker_data->pbi->prev_buf =
src_worker_data->pbi->common.show_existing_frame ?
NULL : src_worker_data->pbi->cur_buf;
dst_cm->last_width = !src_cm->show_existing_frame ?
src_cm->width : src_cm->last_width;
dst_cm->last_height = !src_cm->show_existing_frame ?
src_cm->height : src_cm->last_height;
dst_cm->display_width = src_cm->display_width;
dst_cm->display_height = src_cm->display_height;
dst_cm->subsampling_x = src_cm->subsampling_x;
dst_cm->subsampling_y = src_cm->subsampling_y;
dst_cm->last_show_frame = !src_cm->show_existing_frame ?
src_cm->show_frame : src_cm->last_show_frame;
dst_cm->last_frame_type = src_cm->last_frame_type;
dst_cm->frame_type = src_cm->frame_type;
dst_cm->y_dc_delta_q = src_cm->y_dc_delta_q;
dst_cm->uv_dc_delta_q = src_cm->uv_dc_delta_q;
dst_cm->uv_ac_delta_q = src_cm->uv_ac_delta_q;
dst_cm->base_qindex = src_cm->base_qindex;
for (i = 0; i < REF_FRAMES; ++i)
dst_cm->ref_frame_map[i] = src_cm->next_ref_frame_map[i];
memcpy(dst_cm->lf_info.lfthr, src_cm->lf_info.lfthr,
(MAX_LOOP_FILTER + 1) * sizeof(loop_filter_thresh));
dst_cm->lf.last_sharpness_level = src_cm->lf.sharpness_level;
dst_cm->lf.filter_level = src_cm->lf.filter_level;
memcpy(dst_cm->lf.ref_deltas, src_cm->lf.ref_deltas, MAX_REF_LF_DELTAS);
memcpy(dst_cm->lf.mode_deltas, src_cm->lf.mode_deltas, MAX_MODE_LF_DELTAS);
dst_cm->seg = src_cm->seg;
memcpy(dst_cm->frame_contexts, src_cm->frame_contexts,
FRAME_CONTEXTS * sizeof(dst_cm->frame_contexts[0]));
#else
(void) dst_worker;
(void) src_worker;
#endif // CONFIG_MULTITHREAD
}