summaryrefslogtreecommitdiff
path: root/libavcodec/vp8.c
diff options
context:
space:
mode:
authorVittorio Giovara <vittorio.giovara@gmail.com>2014-03-28 01:13:54 +0100
committerDiego Biurrun <diego@biurrun.de>2014-03-29 16:11:09 +0100
commit53c20f17c78d1d8a0fc2505868f201e69ff59cc5 (patch)
tree7d543871ede0f0adf63d3c5fcaef907275e42758 /libavcodec/vp8.c
parent6adf3bc42e36242d487636786e995149bbb849fe (diff)
vp8: K&R formatting cosmetics
Signed-off-by: Diego Biurrun <diego@biurrun.de>
Diffstat (limited to 'libavcodec/vp8.c')
-rw-r--r--libavcodec/vp8.c949
1 files changed, 516 insertions, 433 deletions
diff --git a/libavcodec/vp8.c b/libavcodec/vp8.c
index 30e24cc893..a1d24cdc1f 100644
--- a/libavcodec/vp8.c
+++ b/libavcodec/vp8.c
@@ -24,12 +24,13 @@
*/
#include "libavutil/imgutils.h"
+
#include "avcodec.h"
#include "internal.h"
-#include "vp8.h"
-#include "vp8data.h"
#include "rectangle.h"
#include "thread.h"
+#include "vp8.h"
+#include "vp8data.h"
#if ARCH_ARM
# include "arm/vp8.h"
@@ -91,7 +92,6 @@ static int vp8_ref_frame(VP8Context *s, VP8Frame *dst, VP8Frame *src)
return 0;
}
-
static void vp8_decode_flush_impl(AVCodecContext *avctx, int free_mem)
{
VP8Context *s = avctx->priv_data;
@@ -124,22 +124,25 @@ static int update_dimensions(VP8Context *s, int width, int height)
return ret;
}
- s->mb_width = (s->avctx->coded_width +15) / 16;
- s->mb_height = (s->avctx->coded_height+15) / 16;
+ s->mb_width = (s->avctx->coded_width + 15) / 16;
+ s->mb_height = (s->avctx->coded_height + 15) / 16;
- s->mb_layout = (avctx->active_thread_type == FF_THREAD_SLICE) && (FFMIN(s->num_coeff_partitions, avctx->thread_count) > 1);
+ s->mb_layout = (avctx->active_thread_type == FF_THREAD_SLICE) &&
+ (FFMIN(s->num_coeff_partitions, avctx->thread_count) > 1);
if (!s->mb_layout) { // Frame threading and one thread
- s->macroblocks_base = av_mallocz((s->mb_width+s->mb_height*2+1)*sizeof(*s->macroblocks));
- s->intra4x4_pred_mode_top = av_mallocz(s->mb_width*4);
- }
- else // Sliced threading
- s->macroblocks_base = av_mallocz((s->mb_width+2)*(s->mb_height+2)*sizeof(*s->macroblocks));
- s->top_nnz = av_mallocz(s->mb_width*sizeof(*s->top_nnz));
- s->top_border = av_mallocz((s->mb_width+1)*sizeof(*s->top_border));
- s->thread_data = av_mallocz(MAX_THREADS*sizeof(VP8ThreadData));
+ s->macroblocks_base = av_mallocz((s->mb_width + s->mb_height * 2 + 1) *
+ sizeof(*s->macroblocks));
+ s->intra4x4_pred_mode_top = av_mallocz(s->mb_width * 4);
+ } else // Sliced threading
+ s->macroblocks_base = av_mallocz((s->mb_width + 2) * (s->mb_height + 2) *
+ sizeof(*s->macroblocks));
+ s->top_nnz = av_mallocz(s->mb_width * sizeof(*s->top_nnz));
+ s->top_border = av_mallocz((s->mb_width + 1) * sizeof(*s->top_border));
+ s->thread_data = av_mallocz(MAX_THREADS * sizeof(VP8ThreadData));
for (i = 0; i < MAX_THREADS; i++) {
- s->thread_data[i].filter_strength = av_mallocz(s->mb_width*sizeof(*s->thread_data[0].filter_strength));
+ s->thread_data[i].filter_strength =
+ av_mallocz(s->mb_width * sizeof(*s->thread_data[0].filter_strength));
#if HAVE_THREADS
pthread_mutex_init(&s->thread_data[i].lock, NULL);
pthread_cond_init(&s->thread_data[i].cond, NULL);
@@ -150,7 +153,7 @@ static int update_dimensions(VP8Context *s, int width, int height)
(!s->intra4x4_pred_mode_top && !s->mb_layout))
return AVERROR(ENOMEM);
- s->macroblocks = s->macroblocks_base + 1;
+ s->macroblocks = s->macroblocks_base + 1;
return 0;
}
@@ -207,13 +210,13 @@ static int setup_partitions(VP8Context *s, const uint8_t *buf, int buf_size)
s->num_coeff_partitions = 1 << vp8_rac_get_uint(&s->c, 2);
- buf += 3*(s->num_coeff_partitions-1);
- buf_size -= 3*(s->num_coeff_partitions-1);
+ buf += 3 * (s->num_coeff_partitions - 1);
+ buf_size -= 3 * (s->num_coeff_partitions - 1);
if (buf_size < 0)
return -1;
- for (i = 0; i < s->num_coeff_partitions-1; i++) {
- int size = AV_RL24(sizes + 3*i);
+ for (i = 0; i < s->num_coeff_partitions - 1; i++) {
+ int size = AV_RL24(sizes + 3 * i);
if (buf_size - size < 0)
return -1;
@@ -246,13 +249,13 @@ static void get_quants(VP8Context *s)
} else
base_qi = yac_qi;
- s->qmat[i].luma_qmul[0] = vp8_dc_qlookup[av_clip_uintp2(base_qi + ydc_delta , 7)];
- s->qmat[i].luma_qmul[1] = vp8_ac_qlookup[av_clip_uintp2(base_qi , 7)];
- s->qmat[i].luma_dc_qmul[0] = 2 * vp8_dc_qlookup[av_clip_uintp2(base_qi + y2dc_delta, 7)];
+ s->qmat[i].luma_qmul[0] = vp8_dc_qlookup[av_clip_uintp2(base_qi + ydc_delta, 7)];
+ s->qmat[i].luma_qmul[1] = vp8_ac_qlookup[av_clip_uintp2(base_qi, 7)];
+ s->qmat[i].luma_dc_qmul[0] = vp8_dc_qlookup[av_clip_uintp2(base_qi + y2dc_delta, 7)] * 2;
/* 101581>>16 is equivalent to 155/100 */
- s->qmat[i].luma_dc_qmul[1] = (101581 * vp8_ac_qlookup[av_clip_uintp2(base_qi + y2ac_delta, 7)]) >> 16;
- s->qmat[i].chroma_qmul[0] = vp8_dc_qlookup[av_clip_uintp2(base_qi + uvdc_delta, 7)];
- s->qmat[i].chroma_qmul[1] = vp8_ac_qlookup[av_clip_uintp2(base_qi + uvac_delta, 7)];
+ s->qmat[i].luma_dc_qmul[1] = vp8_ac_qlookup[av_clip_uintp2(base_qi + y2ac_delta, 7)] * 101581 >> 16;
+ s->qmat[i].chroma_qmul[0] = vp8_dc_qlookup[av_clip_uintp2(base_qi + uvdc_delta, 7)];
+ s->qmat[i].chroma_qmul[1] = vp8_ac_qlookup[av_clip_uintp2(base_qi + uvac_delta, 7)];
s->qmat[i].luma_dc_qmul[1] = FFMAX(s->qmat[i].luma_dc_qmul[1], 8);
s->qmat[i].chroma_qmul[0] = FFMIN(s->qmat[i].chroma_qmul[0], 132);
@@ -317,24 +320,27 @@ static int decode_frame_header(VP8Context *s, const uint8_t *buf, int buf_size)
av_log(s->avctx, AV_LOG_WARNING, "Unknown profile %d\n", s->profile);
if (!s->profile)
- memcpy(s->put_pixels_tab, s->vp8dsp.put_vp8_epel_pixels_tab, sizeof(s->put_pixels_tab));
+ memcpy(s->put_pixels_tab, s->vp8dsp.put_vp8_epel_pixels_tab,
+ sizeof(s->put_pixels_tab));
else // profile 1-3 use bilinear, 4+ aren't defined so whatever
- memcpy(s->put_pixels_tab, s->vp8dsp.put_vp8_bilinear_pixels_tab, sizeof(s->put_pixels_tab));
+ memcpy(s->put_pixels_tab, s->vp8dsp.put_vp8_bilinear_pixels_tab,
+ sizeof(s->put_pixels_tab));
- if (header_size > buf_size - 7*s->keyframe) {
+ if (header_size > buf_size - 7 * s->keyframe) {
av_log(s->avctx, AV_LOG_ERROR, "Header size larger than data provided\n");
return AVERROR_INVALIDDATA;
}
if (s->keyframe) {
if (AV_RL24(buf) != 0x2a019d) {
- av_log(s->avctx, AV_LOG_ERROR, "Invalid start code 0x%x\n", AV_RL24(buf));
+ av_log(s->avctx, AV_LOG_ERROR,
+ "Invalid start code 0x%x\n", AV_RL24(buf));
return AVERROR_INVALIDDATA;
}
- width = AV_RL16(buf+3) & 0x3fff;
- height = AV_RL16(buf+5) & 0x3fff;
- hscale = buf[4] >> 6;
- vscale = buf[6] >> 6;
+ width = AV_RL16(buf + 3) & 0x3fff;
+ height = AV_RL16(buf + 5) & 0x3fff;
+ hscale = buf[4] >> 6;
+ vscale = buf[6] >> 6;
buf += 7;
buf_size -= 7;
@@ -344,11 +350,15 @@ static int decode_frame_header(VP8Context *s, const uint8_t *buf, int buf_size)
s->update_golden = s->update_altref = VP56_FRAME_CURRENT;
for (i = 0; i < 4; i++)
for (j = 0; j < 16; j++)
- memcpy(s->prob->token[i][j], vp8_token_default_probs[i][vp8_coeff_band[j]],
+ memcpy(s->prob->token[i][j],
+ vp8_token_default_probs[i][vp8_coeff_band[j]],
sizeof(s->prob->token[i][j]));
- memcpy(s->prob->pred16x16, vp8_pred16x16_prob_inter, sizeof(s->prob->pred16x16));
- memcpy(s->prob->pred8x8c , vp8_pred8x8c_prob_inter , sizeof(s->prob->pred8x8c));
- memcpy(s->prob->mvc , vp8_mv_default_prob , sizeof(s->prob->mvc));
+ memcpy(s->prob->pred16x16, vp8_pred16x16_prob_inter,
+ sizeof(s->prob->pred16x16));
+ memcpy(s->prob->pred8x8c, vp8_pred8x8c_prob_inter,
+ sizeof(s->prob->pred8x8c));
+ memcpy(s->prob->mvc, vp8_mv_default_prob,
+ sizeof(s->prob->mvc));
memset(&s->segmentation, 0, sizeof(s->segmentation));
memset(&s->lf_delta, 0, sizeof(s->lf_delta));
}
@@ -382,10 +392,9 @@ static int decode_frame_header(VP8Context *s, const uint8_t *buf, int buf_size)
}
if (!s->macroblocks_base || /* first frame */
- width != s->avctx->width || height != s->avctx->height) {
+ width != s->avctx->width || height != s->avctx->height)
if ((ret = update_dimensions(s, width, height)) < 0)
return ret;
- }
get_quants(s);
@@ -405,7 +414,7 @@ static int decode_frame_header(VP8Context *s, const uint8_t *buf, int buf_size)
for (i = 0; i < 4; i++)
for (j = 0; j < 8; j++)
for (k = 0; k < 3; k++)
- for (l = 0; l < NUM_DCT_TOKENS-1; l++)
+ for (l = 0; l < NUM_DCT_TOKENS - 1; l++)
if (vp56_rac_get_prob_branchy(c, vp8_token_update_probs[i][j][k][l])) {
int prob = vp8_rac_get_uint(c, 8);
for (m = 0; vp8_coeff_band_indexes[j][m] >= 0; m++)
@@ -437,7 +446,8 @@ static int decode_frame_header(VP8Context *s, const uint8_t *buf, int buf_size)
return 0;
}
-static av_always_inline void clamp_mv(VP8Context *s, VP56mv *dst, const VP56mv *src)
+static av_always_inline
+void clamp_mv(VP8Context *s, VP56mv *dst, const VP56mv *src)
{
dst->x = av_clip(src->x, s->mv_min.x, s->mv_max.x);
dst->y = av_clip(src->y, s->mv_min.y, s->mv_max.y);
@@ -461,13 +471,13 @@ static int read_mv_component(VP56RangeCoder *c, const uint8_t *p)
x += 8;
} else {
// small_mvtree
- const uint8_t *ps = p+2;
+ const uint8_t *ps = p + 2;
bit = vp56_rac_get_prob(c, *ps);
- ps += 1 + 3*bit;
- x += 4*bit;
+ ps += 1 + 3 * bit;
+ x += 4 * bit;
bit = vp56_rac_get_prob(c, *ps);
ps += 1 + bit;
- x += 2*bit;
+ x += 2 * bit;
x += vp56_rac_get_prob(c, *ps);
}
@@ -478,10 +488,10 @@ static av_always_inline
const uint8_t *get_submv_prob(uint32_t left, uint32_t top)
{
if (left == top)
- return vp8_submv_prob[4-!!left];
+ return vp8_submv_prob[4 - !!left];
if (!top)
return vp8_submv_prob[2];
- return vp8_submv_prob[1-!!left];
+ return vp8_submv_prob[1 - !!left];
}
/**
@@ -495,9 +505,8 @@ int decode_splitmvs(VP8Context *s, VP56RangeCoder *c, VP8Macroblock *mb, int lay
int n, num;
VP8Macroblock *top_mb;
VP8Macroblock *left_mb = &mb[-1];
- const uint8_t *mbsplits_left = vp8_mbsplits[left_mb->partitioning],
- *mbsplits_top,
- *mbsplits_cur, *firstidx;
+ const uint8_t *mbsplits_left = vp8_mbsplits[left_mb->partitioning];
+ const uint8_t *mbsplits_top, *mbsplits_cur, *firstidx;
VP56mv *top_mv;
VP56mv *left_mv = left_mb->bmv;
VP56mv *cur_mv = mb->bmv;
@@ -505,23 +514,22 @@ int decode_splitmvs(VP8Context *s, VP56RangeCoder *c, VP8Macroblock *mb, int lay
if (!layout) // layout is inlined, s->mb_layout is not
top_mb = &mb[2];
else
- top_mb = &mb[-s->mb_width-1];
+ top_mb = &mb[-s->mb_width - 1];
mbsplits_top = vp8_mbsplits[top_mb->partitioning];
- top_mv = top_mb->bmv;
+ top_mv = top_mb->bmv;
if (vp56_rac_get_prob_branchy(c, vp8_mbsplit_prob[0])) {
- if (vp56_rac_get_prob_branchy(c, vp8_mbsplit_prob[1])) {
+ if (vp56_rac_get_prob_branchy(c, vp8_mbsplit_prob[1]))
part_idx = VP8_SPLITMVMODE_16x8 + vp56_rac_get_prob(c, vp8_mbsplit_prob[2]);
- } else {
+ else
part_idx = VP8_SPLITMVMODE_8x8;
- }
} else {
part_idx = VP8_SPLITMVMODE_4x4;
}
- num = vp8_mbsplit_count[part_idx];
- mbsplits_cur = vp8_mbsplits[part_idx],
- firstidx = vp8_mbfirstidx[part_idx];
+ num = vp8_mbsplit_count[part_idx];
+ mbsplits_cur = vp8_mbsplits[part_idx],
+ firstidx = vp8_mbfirstidx[part_idx];
mb->partitioning = part_idx;
for (n = 0; n < num; n++) {
@@ -532,7 +540,7 @@ int decode_splitmvs(VP8Context *s, VP56RangeCoder *c, VP8Macroblock *mb, int lay
if (!(k & 3))
left = AV_RN32A(&left_mv[mbsplits_left[k + 3]]);
else
- left = AV_RN32A(&cur_mv[mbsplits_cur[k - 1]]);
+ left = AV_RN32A(&cur_mv[mbsplits_cur[k - 1]]);
if (k <= 3)
above = AV_RN32A(&top_mv[mbsplits_top[k + 12]]);
else
@@ -560,11 +568,12 @@ int decode_splitmvs(VP8Context *s, VP56RangeCoder *c, VP8Macroblock *mb, int lay
}
static av_always_inline
-void decode_mvs(VP8Context *s, VP8Macroblock *mb, int mb_x, int mb_y, int layout)
+void decode_mvs(VP8Context *s, VP8Macroblock *mb,
+ int mb_x, int mb_y, int layout)
{
- VP8Macroblock *mb_edge[3] = { 0 /* top */,
+ VP8Macroblock *mb_edge[3] = { 0 /* top */,
mb - 1 /* left */,
- 0 /* top-left */ };
+ 0 /* top-left */ };
enum { CNT_ZERO, CNT_NEAREST, CNT_NEAR, CNT_SPLITMV };
enum { VP8_EDGE_TOP, VP8_EDGE_LEFT, VP8_EDGE_TOPLEFT };
int idx = CNT_ZERO;
@@ -577,10 +586,9 @@ void decode_mvs(VP8Context *s, VP8Macroblock *mb, int mb_x, int mb_y, int layout
if (!layout) { // layout is inlined (s->mb_layout is not)
mb_edge[0] = mb + 2;
mb_edge[2] = mb + 1;
- }
- else {
- mb_edge[0] = mb - s->mb_width-1;
- mb_edge[2] = mb - s->mb_width-2;
+ } else {
+ mb_edge[0] = mb - s->mb_width - 1;
+ mb_edge[2] = mb - s->mb_width - 2;
}
AV_ZERO32(&near_mv[0]);
@@ -588,24 +596,25 @@ void decode_mvs(VP8Context *s, VP8Macroblock *mb, int mb_x, int mb_y, int layout
AV_ZERO32(&near_mv[2]);
/* Process MB on top, left and top-left */
- #define MV_EDGE_CHECK(n)\
- {\
- VP8Macroblock *edge = mb_edge[n];\
- int edge_ref = edge->ref_frame;\
- if (edge_ref != VP56_FRAME_CURRENT) {\
- uint32_t mv = AV_RN32A(&edge->mv);\
- if (mv) {\
- if (cur_sign_bias != sign_bias[edge_ref]) {\
- /* SWAR negate of the values in mv. */\
- mv = ~mv;\
- mv = ((mv&0x7fff7fff) + 0x00010001) ^ (mv&0x80008000);\
- }\
- if (!n || mv != AV_RN32A(&near_mv[idx]))\
- AV_WN32A(&near_mv[++idx], mv);\
- cnt[idx] += 1 + (n != 2);\
- } else\
- cnt[CNT_ZERO] += 1 + (n != 2);\
- }\
+#define MV_EDGE_CHECK(n) \
+ { \
+ VP8Macroblock *edge = mb_edge[n]; \
+ int edge_ref = edge->ref_frame; \
+ if (edge_ref != VP56_FRAME_CURRENT) { \
+ uint32_t mv = AV_RN32A(&edge->mv); \
+ if (mv) { \
+ if (cur_sign_bias != sign_bias[edge_ref]) { \
+ /* SWAR negate of the values in mv. */ \
+ mv = ~mv; \
+ mv = ((mv & 0x7fff7fff) + \
+ 0x00010001) ^ (mv & 0x80008000); \
+ } \
+ if (!n || mv != AV_RN32A(&near_mv[idx])) \
+ AV_WN32A(&near_mv[++idx], mv); \
+ cnt[idx] += 1 + (n != 2); \
+ } else \
+ cnt[CNT_ZERO] += 1 + (n != 2); \
+ } \
}
MV_EDGE_CHECK(0)
@@ -617,7 +626,8 @@ void decode_mvs(VP8Context *s, VP8Macroblock *mb, int mb_x, int mb_y, int layout
mb->mode = VP8_MVMODE_MV;
/* If we have three distinct MVs, merge first and last if they're the same */
- if (cnt[CNT_SPLITMV] && AV_RN32A(&near_mv[1 + VP8_EDGE_TOP]) == AV_RN32A(&near_mv[1 + VP8_EDGE_TOPLEFT]))
+ if (cnt[CNT_SPLITMV] &&
+ AV_RN32A(&near_mv[1 + VP8_EDGE_TOP]) == AV_RN32A(&near_mv[1 + VP8_EDGE_TOPLEFT]))
cnt[CNT_NEAREST] += 1;
/* Swap near and nearest if necessary */
@@ -628,7 +638,6 @@ void decode_mvs(VP8Context *s, VP8Macroblock *mb, int mb_x, int mb_y, int layout
if (vp56_rac_get_prob_branchy(c, vp8_mode_contexts[cnt[CNT_NEAREST]][1])) {
if (vp56_rac_get_prob_branchy(c, vp8_mode_contexts[cnt[CNT_NEAR]][2])) {
-
/* Choose the best mv out of 0,0 and the nearest mv */
clamp_mv(s, &mb->mv, &near_mv[CNT_ZERO + (cnt[CNT_NEAREST] >= cnt[CNT_ZERO])]);
cnt[CNT_SPLITMV] = ((mb_edge[VP8_EDGE_LEFT]->mode == VP8_MVMODE_SPLIT) +
@@ -637,10 +646,10 @@ void decode_mvs(VP8Context *s, VP8Macroblock *mb, int mb_x, int mb_y, int layout
if (vp56_rac_get_prob_branchy(c, vp8_mode_contexts[cnt[CNT_SPLITMV]][3])) {
mb->mode = VP8_MVMODE_SPLIT;
- mb->mv = mb->bmv[decode_splitmvs(s, c, mb, layout) - 1];
+ mb->mv = mb->bmv[decode_splitmvs(s, c, mb, layout) - 1];
} else {
- mb->mv.y += read_mv_component(c, s->prob->mvc[0]);
- mb->mv.x += read_mv_component(c, s->prob->mvc[1]);
+ mb->mv.y += read_mv_component(c, s->prob->mvc[0]);
+ mb->mv.x += read_mv_component(c, s->prob->mvc[1]);
mb->bmv[0] = mb->mv;
}
} else {
@@ -670,8 +679,8 @@ void decode_intra4x4_modes(VP8Context *s, VP56RangeCoder *c, VP8Macroblock *mb,
}
if (keyframe) {
int x, y;
- uint8_t* top;
- uint8_t* const left = s->intra4x4_pred_mode_left;
+ uint8_t *top;
+ uint8_t *const left = s->intra4x4_pred_mode_left;
if (layout == 1)
top = mb->intra4x4_pred_mode_top;
else
@@ -679,16 +688,17 @@ void decode_intra4x4_modes(VP8Context *s, VP56RangeCoder *c, VP8Macroblock *mb,
for (y = 0; y < 4; y++) {
for (x = 0; x < 4; x++) {
const uint8_t *ctx;
- ctx = vp8_pred4x4_prob_intra[top[x]][left[y]];
+ ctx = vp8_pred4x4_prob_intra[top[x]][left[y]];
*intra4x4 = vp8_rac_get_tree(c, vp8_pred4x4_tree, ctx);
- left[y] = top[x] = *intra4x4;
+ left[y] = top[x] = *intra4x4;
intra4x4++;
}
}
} else {
int i;
for (i = 0; i < 16; i++)
- intra4x4[i] = vp8_rac_get_tree(c, vp8_pred4x4_tree, vp8_pred4x4_prob_inter);
+ intra4x4[i] = vp8_rac_get_tree(c, vp8_pred4x4_tree,
+ vp8_pred4x4_prob_inter);
}
}
@@ -707,7 +717,8 @@ void decode_mb_mode(VP8Context *s, VP8Macroblock *mb, int mb_x, int mb_y,
mb->skip = s->mbskip_enabled ? vp56_rac_get_prob(c, s->prob->mbskip) : 0;
if (s->keyframe) {
- mb->mode = vp8_rac_get_tree(c, vp8_pred16x16_tree_intra, vp8_pred16x16_prob_intra);
+ mb->mode = vp8_rac_get_tree(c, vp8_pred16x16_tree_intra,
+ vp8_pred16x16_prob_intra);
if (mb->mode == MODE_I4x4) {
decode_intra4x4_modes(s, c, mb, mb_x, 1, layout);
@@ -717,19 +728,21 @@ void decode_mb_mode(VP8Context *s, VP8Macroblock *mb, int mb_x, int mb_y,
AV_WN32A(mb->intra4x4_pred_mode_top, modes);
else
AV_WN32A(s->intra4x4_pred_mode_top + 4 * mb_x, modes);
- AV_WN32A( s->intra4x4_pred_mode_left, modes);
+ AV_WN32A(s->intra4x4_pred_mode_left, modes);
}
- mb->chroma_pred_mode = vp8_rac_get_tree(c, vp8_pred8x8c_tree, vp8_pred8x8c_prob_intra);
- mb->ref_frame = VP56_FRAME_CURRENT;
+ mb->chroma_pred_mode = vp8_rac_get_tree(c, vp8_pred8x8c_tree,
+ vp8_pred8x8c_prob_intra);
+ mb->ref_frame = VP56_FRAME_CURRENT;
} else if (vp56_rac_get_prob_branchy(c, s->prob->intra)) {
// inter MB, 16.2
if (vp56_rac_get_prob_branchy(c, s->prob->last))
- mb->ref_frame = vp56_rac_get_prob(c, s->prob->golden) ?
- VP56_FRAME_GOLDEN2 /* altref */ : VP56_FRAME_GOLDEN;
+ mb->ref_frame =
+ vp56_rac_get_prob(c, s->prob->golden) ? VP56_FRAME_GOLDEN2 /* altref */
+ : VP56_FRAME_GOLDEN;
else
mb->ref_frame = VP56_FRAME_PREVIOUS;
- s->ref_count[mb->ref_frame-1]++;
+ s->ref_count[mb->ref_frame - 1]++;
// motion vectors, 16.3
decode_mvs(s, mb, mb_x, mb_y, layout);
@@ -740,26 +753,29 @@ void decode_mb_mode(VP8Context *s, VP8Macroblock *mb, int mb_x, int mb_y,
if (mb->mode == MODE_I4x4)
decode_intra4x4_modes(s, c, mb, mb_x, 0, layout);
- mb->chroma_pred_mode = vp8_rac_get_tree(c, vp8_pred8x8c_tree, s->prob->pred8x8c);
- mb->ref_frame = VP56_FRAME_CURRENT;
- mb->partitioning = VP8_SPLITMVMODE_NONE;
+ mb->chroma_pred_mode = vp8_rac_get_tree(c, vp8_pred8x8c_tree,
+ s->prob->pred8x8c);
+ mb->ref_frame = VP56_FRAME_CURRENT;
+ mb->partitioning = VP8_SPLITMVMODE_NONE;
AV_ZERO32(&mb->bmv[0]);
}
}
#ifndef decode_block_coeffs_internal
/**
- * @param r arithmetic bitstream reader context
+ * @param r arithmetic bitstream reader context
* @param block destination for block coefficients
* @param probs probabilities to use when reading trees from the bitstream
- * @param i initial coeff index, 0 unless a separate DC block is coded
- * @param qmul array holding the dc/ac dequant factor at position 0/1
+ * @param i initial coeff index, 0 unless a separate DC block is coded
+ * @param qmul array holding the dc/ac dequant factor at position 0/1
+ *
* @return 0 if no coeffs were decoded
* otherwise, the index of the last coeff decoded plus one
*/
static int decode_block_coeffs_internal(VP56RangeCoder *r, int16_t block[16],
- uint8_t probs[16][3][NUM_DCT_TOKENS-1],
- int i, uint8_t *token_prob, int16_t qmul[2])
+ uint8_t probs[16][3][NUM_DCT_TOKENS - 1],
+ int i, uint8_t *token_prob,
+ int16_t qmul[2])
{
VP56RangeCoder c = *r;
goto skip_eob;
@@ -778,7 +794,7 @@ skip_eob:
if (!vp56_rac_get_prob_branchy(&c, token_prob[2])) { // DCT_1
coeff = 1;
- token_prob = probs[i+1][1];
+ token_prob = probs[i + 1][1];
} else {
if (!vp56_rac_get_prob_branchy(&c, token_prob[3])) { // DCT 2,3,4
coeff = vp56_rac_get_prob_branchy(&c, token_prob[4]);
@@ -789,21 +805,21 @@ skip_eob:
// DCT_CAT*
if (!vp56_rac_get_prob_branchy(&c, token_prob[6])) {
if (!vp56_rac_get_prob_branchy(&c, token_prob[7])) { // DCT_CAT1
- coeff = 5 + vp56_rac_get_prob(&c, vp8_dct_cat1_prob[0]);
+ coeff = 5 + vp56_rac_get_prob(&c, vp8_dct_cat1_prob[0]);
} else { // DCT_CAT2
coeff = 7;
coeff += vp56_rac_get_prob(&c, vp8_dct_cat2_prob[0]) << 1;
coeff += vp56_rac_get_prob(&c, vp8_dct_cat2_prob[1]);
}
} else { // DCT_CAT3 and up
- int a = vp56_rac_get_prob(&c, token_prob[8]);
- int b = vp56_rac_get_prob(&c, token_prob[9+a]);
- int cat = (a<<1) + b;
- coeff = 3 + (8<<cat);
+ int a = vp56_rac_get_prob(&c, token_prob[8]);
+ int b = vp56_rac_get_prob(&c, token_prob[9 + a]);
+ int cat = (a << 1) + b;
+ coeff = 3 + (8 << cat);
coeff += vp8_rac_get_coeff(&c, ff_vp8_dct_cat_prob[cat]);
}
}
- token_prob = probs[i+1][2];
+ token_prob = probs[i + 1][2];
}
block[zigzag_scan[i]] = (vp8_rac_get(&c) ? -coeff : coeff) * qmul[!!i];
} while (++i < 16);
@@ -814,19 +830,20 @@ skip_eob:
#endif
/**
- * @param c arithmetic bitstream reader context
- * @param block destination for block coefficients
- * @param probs probabilities to use when reading trees from the bitstream
- * @param i initial coeff index, 0 unless a separate DC block is coded
+ * @param c arithmetic bitstream reader context
+ * @param block destination for block coefficients
+ * @param probs probabilities to use when reading trees from the bitstream
+ * @param i initial coeff index, 0 unless a separate DC block is coded
* @param zero_nhood the initial prediction context for number of surrounding
* all-zero blocks (only left/top, so 0-2)
- * @param qmul array holding the dc/ac dequant factor at position 0/1
+ * @param qmul array holding the dc/ac dequant factor at position 0/1
+ *
* @return 0 if no coeffs were decoded
* otherwise, the index of the last coeff decoded plus one
*/
static av_always_inline
int decode_block_coeffs(VP56RangeCoder *c, int16_t block[16],
- uint8_t probs[16][3][NUM_DCT_TOKENS-1],
+ uint8_t probs[16][3][NUM_DCT_TOKENS - 1],
int i, int zero_nhood, int16_t qmul[2])
{
uint8_t *token_prob = probs[i][zero_nhood];
@@ -836,8 +853,8 @@ int decode_block_coeffs(VP56RangeCoder *c, int16_t block[16],
}
static av_always_inline
-void decode_mb_coeffs(VP8Context *s, VP8ThreadData *td, VP56RangeCoder *c, VP8Macroblock *mb,
- uint8_t t_nnz[9], uint8_t l_nnz[9])
+void decode_mb_coeffs(VP8Context *s, VP8ThreadData *td, VP56RangeCoder *c,
+ VP8Macroblock *mb, uint8_t t_nnz[9], uint8_t l_nnz[9])
{
int i, x, y, luma_start = 0, luma_ctx = 3;
int nnz_pred, nnz, nnz_total = 0;
@@ -848,28 +865,31 @@ void decode_mb_coeffs(VP8Context *s, VP8ThreadData *td, VP56RangeCoder *c, VP8Ma
nnz_pred = t_nnz[8] + l_nnz[8];
// decode DC values and do hadamard
- nnz = decode_block_coeffs(c, td->block_dc, s->prob->token[1], 0, nnz_pred,
- s->qmat[segment].luma_dc_qmul);
+ nnz = decode_block_coeffs(c, td->block_dc, s->prob->token[1], 0,
+ nnz_pred, s->qmat[segment].luma_dc_qmul);
l_nnz[8] = t_nnz[8] = !!nnz;
if (nnz) {
nnz_total += nnz;
- block_dc = 1;
+ block_dc = 1;
if (nnz == 1)
s->vp8dsp.vp8_luma_dc_wht_dc(td->block, td->block_dc);
else
s->vp8dsp.vp8_luma_dc_wht(td->block, td->block_dc);
}
luma_start = 1;
- luma_ctx = 0;
+ luma_ctx = 0;
}
// luma blocks
for (y = 0; y < 4; y++)
for (x = 0; x < 4; x++) {
nnz_pred = l_nnz[y] + t_nnz[x];
- nnz = decode_block_coeffs(c, td->block[y][x], s->prob->token[luma_ctx], luma_start,
- nnz_pred, s->qmat[segment].luma_qmul);
- // nnz+block_dc may be one more than the actual last index, but we don't care
+ nnz = decode_block_coeffs(c, td->block[y][x],
+ s->prob->token[luma_ctx],
+ luma_start, nnz_pred,
+ s->qmat[segment].luma_qmul);
+ /* nnz+block_dc may be one more than the actual last index,
+ * but we don't care */
td->non_zero_count_cache[y][x] = nnz + block_dc;
t_nnz[x] = l_nnz[y] = !!nnz;
nnz_total += nnz;
@@ -881,12 +901,14 @@ void decode_mb_coeffs(VP8Context *s, VP8ThreadData *td, VP56RangeCoder *c, VP8Ma
for (i = 4; i < 6; i++)
for (y = 0; y < 2; y++)
for (x = 0; x < 2; x++) {
- nnz_pred = l_nnz[i+2*y] + t_nnz[i+2*x];
- nnz = decode_block_coeffs(c, td->block[i][(y<<1)+x], s->prob->token[2], 0,
- nnz_pred, s->qmat[segment].chroma_qmul);
- td->non_zero_count_cache[i][(y<<1)+x] = nnz;
- t_nnz[i+2*x] = l_nnz[i+2*y] = !!nnz;
- nnz_total += nnz;
+ nnz_pred = l_nnz[i + 2 * y] + t_nnz[i + 2 * x];
+ nnz = decode_block_coeffs(c, td->block[i][(y << 1) + x],
+ s->prob->token[2],
+ 0, nnz_pred,
+ s->qmat[segment].chroma_qmul);
+ td->non_zero_count_cache[i][(y << 1) + x] = nnz;
+ t_nnz[i + 2 * x] = l_nnz[i + 2 * y] = !!nnz;
+ nnz_total += nnz;
}
// if there were no coded coeffs despite the macroblock not being marked skip,
@@ -897,65 +919,67 @@ void decode_mb_coeffs(VP8Context *s, VP8ThreadData *td, VP56RangeCoder *c, VP8Ma
}
static av_always_inline
-void backup_mb_border(uint8_t *top_border, uint8_t *src_y, uint8_t *src_cb, uint8_t *src_cr,
+void backup_mb_border(uint8_t *top_border, uint8_t *src_y,
+ uint8_t *src_cb, uint8_t *src_cr,
int linesize, int uvlinesize, int simple)
{
- AV_COPY128(top_border, src_y + 15*linesize);
+ AV_COPY128(top_border, src_y + 15 * linesize);
if (!simple) {
- AV_COPY64(top_border+16, src_cb + 7*uvlinesize);
- AV_COPY64(top_border+24, src_cr + 7*uvlinesize);
+ AV_COPY64(top_border + 16, src_cb + 7 * uvlinesize);
+ AV_COPY64(top_border + 24, src_cr + 7 * uvlinesize);
}
}
static av_always_inline
-void xchg_mb_border(uint8_t *top_border, uint8_t *src_y, uint8_t *src_cb, uint8_t *src_cr,
- int linesize, int uvlinesize, int mb_x, int mb_y, int mb_width,
- int simple, int xchg)
+void xchg_mb_border(uint8_t *top_border, uint8_t *src_y, uint8_t *src_cb,
+ uint8_t *src_cr, int linesize, int uvlinesize, int mb_x,
+ int mb_y, int mb_width, int simple, int xchg)
{
- uint8_t *top_border_m1 = top_border-32; // for TL prediction
- src_y -= linesize;
+ uint8_t *top_border_m1 = top_border - 32; // for TL prediction
+ src_y -= linesize;
src_cb -= uvlinesize;
src_cr -= uvlinesize;
-#define XCHG(a,b,xchg) do { \
- if (xchg) AV_SWAP64(b,a); \
- else AV_COPY64(b,a); \
+#define XCHG(a, b, xchg) \
+ do { \
+ if (xchg) \
+ AV_SWAP64(b, a); \
+ else \
+ AV_COPY64(b, a); \
} while (0)
- XCHG(top_border_m1+8, src_y-8, xchg);
- XCHG(top_border, src_y, xchg);
- XCHG(top_border+8, src_y+8, 1);
- if (mb_x < mb_width-1)
- XCHG(top_border+32, src_y+16, 1);
+ XCHG(top_border_m1 + 8, src_y - 8, xchg);
+ XCHG(top_border, src_y, xchg);
+ XCHG(top_border + 8, src_y + 8, 1);
+ if (mb_x < mb_width - 1)
+ XCHG(top_border + 32, src_y + 16, 1);
// only copy chroma for normal loop filter
// or to initialize the top row to 127
if (!simple || !mb_y) {
- XCHG(top_border_m1+16, src_cb-8, xchg);
- XCHG(top_border_m1+24, src_cr-8, xchg);
- XCHG(top_border+16, src_cb, 1);
- XCHG(top_border+24, src_cr, 1);
+ XCHG(top_border_m1 + 16, src_cb - 8, xchg);
+ XCHG(top_border_m1 + 24, src_cr - 8, xchg);
+ XCHG(top_border + 16, src_cb, 1);
+ XCHG(top_border + 24, src_cr, 1);
}
}
static av_always_inline
int check_dc_pred8x8_mode(int mode, int mb_x, int mb_y)
{
- if (!mb_x) {
+ if (!mb_x)
return mb_y ? TOP_DC_PRED8x8 : DC_128_PRED8x8;
- } else {
+ else
return mb_y ? mode : LEFT_DC_PRED8x8;
- }
}
static av_always_inline
int check_tm_pred8x8_mode(int mode, int mb_x, int mb_y)
{
- if (!mb_x) {
+ if (!mb_x)
return mb_y ? VERT_PRED8x8 : DC_129_PRED8x8;
- } else {
+ else
return mb_y ? mode : HOR_PRED8x8;
- }
}
static av_always_inline
@@ -968,7 +992,7 @@ int check_intra_pred8x8_mode_emuedge(int mode, int mb_x, int mb_y)
return !mb_y ? DC_127_PRED8x8 : mode;
case HOR_PRED8x8:
return !mb_x ? DC_129_PRED8x8 : mode;
- case PLANE_PRED8x8 /*TM*/:
+ case PLANE_PRED8x8: /* TM */
return check_tm_pred8x8_mode(mode, mb_x, mb_y);
}
return mode;
@@ -1007,7 +1031,8 @@ int check_intra_pred4x4_mode_emuedge(int mode, int mb_x, int mb_y, int *copy_buf
return !mb_x ? DC_129_PRED : mode;
case TM_VP8_PRED:
return check_tm_pred4x4_mode(mode, mb_x, mb_y);
- case DC_PRED: // 4x4 DC doesn't use the same "H.264-style" exceptions as 16x16/8x8 DC
+ case DC_PRED: /* 4x4 DC doesn't use the same "H.264-style" exceptions
+ * as 16x16/8x8 DC */
case DIAG_DOWN_RIGHT_PRED:
case VERT_RIGHT_PRED:
case HOR_DOWN_PRED:
@@ -1025,10 +1050,10 @@ void intra_predict(VP8Context *s, VP8ThreadData *td, uint8_t *dst[3],
int x, y, mode, nnz;
uint32_t tr;
- // for the first row, we need to run xchg_mb_border to init the top edge to 127
- // otherwise, skip it if we aren't going to deblock
+ /* for the first row, we need to run xchg_mb_border to init the top edge
+ * to 127 otherwise, skip it if we aren't going to deblock */
if (mb_y && (s->deblock_filter || !mb_y) && td->thread_nr == 0)
- xchg_mb_border(s->top_border[mb_x+1], dst[0], dst[1], dst[2],
+ xchg_mb_border(s->top_border[mb_x + 1], dst[0], dst[1], dst[2],
s->linesize, s->uvlinesize, mb_x, mb_y, s->mb_width,
s->filter.simple, 1);
@@ -1046,10 +1071,9 @@ void intra_predict(VP8Context *s, VP8ThreadData *td, uint8_t *dst[3],
// if we're on the right edge of the frame, said edge is extended
// from the top macroblock
- if (mb_y &&
- mb_x == s->mb_width-1) {
- tr = tr_right[-1]*0x01010101u;
- tr_right = (uint8_t *)&tr;
+ if (mb_y && mb_x == s->mb_width - 1) {
+ tr = tr_right[-1] * 0x01010101u;
+ tr_right = (uint8_t *) &tr;
}
if (mb->skip)
@@ -1059,27 +1083,29 @@ void intra_predict(VP8Context *s, VP8ThreadData *td, uint8_t *dst[3],
uint8_t *topright = ptr + 4 - s->linesize;
for (x = 0; x < 4; x++) {
int copy = 0, linesize = s->linesize;
- uint8_t *dst = ptr+4*x;
- DECLARE_ALIGNED(4, uint8_t, copy_dst)[5*8];
+ uint8_t *dst = ptr + 4 * x;
+ DECLARE_ALIGNED(4, uint8_t, copy_dst)[5 * 8];
if ((y == 0 || x == 3) && mb_y == 0) {
topright = tr_top;
} else if (x == 3)
topright = tr_right;
- mode = check_intra_pred4x4_mode_emuedge(intra4x4[x], mb_x + x, mb_y + y, &copy);
+ mode = check_intra_pred4x4_mode_emuedge(intra4x4[x],
+ mb_x + x, mb_y + y,
+ &copy);
if (copy) {
- dst = copy_dst + 12;
+ dst = copy_dst + 12;
linesize = 8;
if (!(mb_y + y)) {
copy_dst[3] = 127U;
- AV_WN32A(copy_dst+4, 127U * 0x01010101U);
+ AV_WN32A(copy_dst + 4, 127U * 0x01010101U);
} else {
- AV_COPY32(copy_dst+4, ptr+4*x-s->linesize);
+ AV_COPY32(copy_dst + 4, ptr + 4 * x - s->linesize);
if (!(mb_x + x)) {
copy_dst[3] = 129U;
} else {
- copy_dst[3] = ptr[4*x-s->linesize-1];
+ copy_dst[3] = ptr[4 * x - s->linesize - 1];
}
}
if (!(mb_x + x)) {
@@ -1088,31 +1114,33 @@ void intra_predict(VP8Context *s, VP8ThreadData *td, uint8_t *dst[3],
copy_dst[27] =
copy_dst[35] = 129U;
} else {
- copy_dst[11] = ptr[4*x -1];
- copy_dst[19] = ptr[4*x+s->linesize -1];
- copy_dst[27] = ptr[4*x+s->linesize*2-1];
- copy_dst[35] = ptr[4*x+s->linesize*3-1];
+ copy_dst[11] = ptr[4 * x - 1];
+ copy_dst[19] = ptr[4 * x + s->linesize - 1];
+ copy_dst[27] = ptr[4 * x + s->linesize * 2 - 1];
+ copy_dst[35] = ptr[4 * x + s->linesize * 3 - 1];
}
}
s->hpc.pred4x4[mode](dst, topright, linesize);
if (copy) {
- AV_COPY32(ptr+4*x , copy_dst+12);
- AV_COPY32(ptr+4*x+s->linesize , copy_dst+20);
- AV_COPY32(ptr+4*x+s->linesize*2, copy_dst+28);
- AV_COPY32(ptr+4*x+s->linesize*3, copy_dst+36);
+ AV_COPY32(ptr + 4 * x, copy_dst + 12);
+ AV_COPY32(ptr + 4 * x + s->linesize, copy_dst + 20);
+ AV_COPY32(ptr + 4 * x + s->linesize * 2, copy_dst + 28);
+ AV_COPY32(ptr + 4 * x + s->linesize * 3, copy_dst + 36);
}
nnz = td->non_zero_count_cache[y][x];
if (nnz) {
if (nnz == 1)
- s->vp8dsp.vp8_idct_dc_add(ptr+4*x, td->block[y][x], s->linesize);
+ s->vp8dsp.vp8_idct_dc_add(ptr + 4 * x,
+ td->block[y][x], s->linesize);
else
- s->vp8dsp.vp8_idct_add(ptr+4*x, td->block[y][x], s->linesize);
+ s->vp8dsp.vp8_idct_add(ptr + 4 * x,
+ td->block[y][x], s->linesize);
}
topright += 4;
}
- ptr += 4*s->linesize;
+ ptr += 4 * s->linesize;
intra4x4 += 4;
}
}
@@ -1122,7 +1150,7 @@ void intra_predict(VP8Context *s, VP8ThreadData *td, uint8_t *dst[3],
s->hpc.pred8x8[mode](dst[2], s->uvlinesize);
if (mb_y && (s->deblock_filter || !mb_y) && td->thread_nr == 0)
- xchg_mb_border(s->top_border[mb_x+1], dst[0], dst[1], dst[2],
+ xchg_mb_border(s->top_border[mb_x + 1], dst[0], dst[1], dst[2],
s->linesize, s->uvlinesize, mb_x, mb_y, s->mb_width,
s->filter.simple, 0);
}
@@ -1137,18 +1165,18 @@ static const uint8_t subpel_idx[3][8] = {
/**
* luma MC function
*
- * @param s VP8 decoding context
- * @param dst target buffer for block data at block position
- * @param ref reference picture buffer at origin (0, 0)
- * @param mv motion vector (relative to block position) to get pixel data from
- * @param x_off horizontal position of block from origin (0, 0)
- * @param y_off vertical position of block from origin (0, 0)
- * @param block_w width of block (16, 8 or 4)
- * @param block_h height of block (always same as block_w)
- * @param width width of src/dst plane data
- * @param height height of src/dst plane data
+ * @param s VP8 decoding context
+ * @param dst target buffer for block data at block position
+ * @param ref reference picture buffer at origin (0, 0)
+ * @param mv motion vector (relative to block position) to get pixel data from
+ * @param x_off horizontal position of block from origin (0, 0)
+ * @param y_off vertical position of block from origin (0, 0)
+ * @param block_w width of block (16, 8 or 4)
+ * @param block_h height of block (always same as block_w)
+ * @param width width of src/dst plane data
+ * @param height height of src/dst plane data
* @param linesize size of a single line of plane data, including padding
- * @param mc_func motion compensation function pointers (bilinear or sixtap MC)
+ * @param mc_func motion compensation function pointers (bilinear or sixtap MC)
*/
static av_always_inline
void vp8_mc_luma(VP8Context *s, VP8ThreadData *td, uint8_t *dst,
@@ -1162,8 +1190,8 @@ void vp8_mc_luma(VP8Context *s, VP8ThreadData *td, uint8_t *dst,
if (AV_RN32A(mv)) {
int src_linesize = linesize;
- int mx = (mv->x << 1)&7, mx_idx = subpel_idx[0][mx];
- int my = (mv->y << 1)&7, my_idx = subpel_idx[0][my];
+ int mx = (mv->x << 1) & 7, mx_idx = subpel_idx[0][mx];
+ int my = (mv->y << 1) & 7, my_idx = subpel_idx[0][my];
x_off += mv->x >> 2;
y_off += mv->y >> 2;
@@ -1176,46 +1204,50 @@ void vp8_mc_luma(VP8Context *s, VP8ThreadData *td, uint8_t *dst,
s->vdsp.emulated_edge_mc(td->edge_emu_buffer,
src - my_idx * linesize - mx_idx,
EDGE_EMU_LINESIZE, linesize,
- block_w + subpel_idx[1][mx], block_h + subpel_idx[1][my],
- x_off - mx_idx, y_off - my_idx, width, height);
+ block_w + subpel_idx[1][mx],
+ block_h + subpel_idx[1][my],
+ x_off - mx_idx, y_off - my_idx,
+ width, height);
src = td->edge_emu_buffer + mx_idx + EDGE_EMU_LINESIZE * my_idx;
src_linesize = EDGE_EMU_LINESIZE;
}
mc_func[my_idx][mx_idx](dst, linesize, src, src_linesize, block_h, mx, my);
} else {
ff_thread_await_progress(ref, (3 + y_off + block_h) >> 4, 0);
- mc_func[0][0](dst, linesize, src + y_off * linesize + x_off, linesize, block_h, 0, 0);
+ mc_func[0][0](dst, linesize, src + y_off * linesize + x_off,
+ linesize, block_h, 0, 0);
}
}
/**
* chroma MC function
*
- * @param s VP8 decoding context
- * @param dst1 target buffer for block data at block position (U plane)
- * @param dst2 target buffer for block data at block position (V plane)
- * @param ref reference picture buffer at origin (0, 0)
- * @param mv motion vector (relative to block position) to get pixel data from
- * @param x_off horizontal position of block from origin (0, 0)
- * @param y_off vertical position of block from origin (0, 0)
- * @param block_w width of block (16, 8 or 4)
- * @param block_h height of block (always same as block_w)
- * @param width width of src/dst plane data
- * @param height height of src/dst plane data
+ * @param s VP8 decoding context
+ * @param dst1 target buffer for block data at block position (U plane)
+ * @param dst2 target buffer for block data at block position (V plane)
+ * @param ref reference picture buffer at origin (0, 0)
+ * @param mv motion vector (relative to block position) to get pixel data from
+ * @param x_off horizontal position of block from origin (0, 0)
+ * @param y_off vertical position of block from origin (0, 0)
+ * @param block_w width of block (16, 8 or 4)
+ * @param block_h height of block (always same as block_w)
+ * @param width width of src/dst plane data
+ * @param height height of src/dst plane data
* @param linesize size of a single line of plane data, including padding
- * @param mc_func motion compensation function pointers (bilinear or sixtap MC)
+ * @param mc_func motion compensation function pointers (bilinear or sixtap MC)
*/
static av_always_inline
-void vp8_mc_chroma(VP8Context *s, VP8ThreadData *td, uint8_t *dst1, uint8_t *dst2,
- ThreadFrame *ref, const VP56mv *mv, int x_off, int y_off,
- int block_w, int block_h, int width, int height, ptrdiff_t linesize,
+void vp8_mc_chroma(VP8Context *s, VP8ThreadData *td, uint8_t *dst1,
+ uint8_t *dst2, ThreadFrame *ref, const VP56mv *mv,
+ int x_off, int y_off, int block_w, int block_h,
+ int width, int height, ptrdiff_t linesize,
vp8_mc_func mc_func[3][3])
{
uint8_t *src1 = ref->f->data[1], *src2 = ref->f->data[2];
if (AV_RN32A(mv)) {
- int mx = mv->x&7, mx_idx = subpel_idx[0][mx];
- int my = mv->y&7, my_idx = subpel_idx[0][my];
+ int mx = mv->x & 7, mx_idx = subpel_idx[0][mx];
+ int my = mv->y & 7, my_idx = subpel_idx[0][my];
x_off += mv->x >> 3;
y_off += mv->y >> 3;
@@ -1239,7 +1271,7 @@ void vp8_mc_chroma(VP8Context *s, VP8ThreadData *td, uint8_t *dst1, uint8_t *dst
EDGE_EMU_LINESIZE, linesize,
block_w + subpel_idx[1][mx], block_h + subpel_idx[1][my],
x_off - mx_idx, y_off - my_idx, width, height);
- src2 = td->edge_emu_buffer + mx_idx + EDGE_EMU_LINESIZE* my_idx;
+ src2 = td->edge_emu_buffer + mx_idx + EDGE_EMU_LINESIZE * my_idx;
mc_func[my_idx][mx_idx](dst2, linesize, src2, EDGE_EMU_LINESIZE, block_h, mx, my);
} else {
mc_func[my_idx][mx_idx](dst1, linesize, src1, linesize, block_h, mx, my);
@@ -1255,8 +1287,7 @@ void vp8_mc_chroma(VP8Context *s, VP8ThreadData *td, uint8_t *dst1, uint8_t *dst
static av_always_inline
void vp8_mc_part(VP8Context *s, VP8ThreadData *td, uint8_t *dst[3],
ThreadFrame *ref_frame, int x_off, int y_off,
- int bx_off, int by_off,
- int block_w, int block_h,
+ int bx_off, int by_off, int block_w, int block_h,
int width, int height, VP56mv *mv)
{
VP56mv uvmv = *mv;
@@ -1272,10 +1303,14 @@ void vp8_mc_part(VP8Context *s, VP8ThreadData *td, uint8_t *dst[3],
uvmv.x &= ~7;
uvmv.y &= ~7;
}
- x_off >>= 1; y_off >>= 1;
- bx_off >>= 1; by_off >>= 1;
- width >>= 1; height >>= 1;
- block_w >>= 1; block_h >>= 1;
+ x_off >>= 1;
+ y_off >>= 1;
+ bx_off >>= 1;
+ by_off >>= 1;
+ width >>= 1;
+ height >>= 1;
+ block_w >>= 1;
+ block_h >>= 1;
vp8_mc_chroma(s, td, dst[1] + by_off * s->uvlinesize + bx_off,
dst[2] + by_off * s->uvlinesize + bx_off, ref_frame,
&uvmv, x_off + bx_off, y_off + by_off,
@@ -1284,22 +1319,24 @@ void vp8_mc_part(VP8Context *s, VP8ThreadData *td, uint8_t *dst[3],
}
/* Fetch pixels for estimated mv 4 macroblocks ahead.
- * Optimized for 64-byte cache lines. Inspired by ffh264 prefetch_motion. */
-static av_always_inline void prefetch_motion(VP8Context *s, VP8Macroblock *mb, int mb_x, int mb_y, int mb_xy, int ref)
+ * Optimized for 64-byte cache lines. Inspired by ffh264 prefetch_motion. */
+static av_always_inline
+void prefetch_motion(VP8Context *s, VP8Macroblock *mb, int mb_x, int mb_y,
+ int mb_xy, int ref)
{
/* Don't prefetch refs that haven't been used very often this frame. */
- if (s->ref_count[ref-1] > (mb_xy >> 5)) {
+ if (s->ref_count[ref - 1] > (mb_xy >> 5)) {
int x_off = mb_x << 4, y_off = mb_y << 4;
- int mx = (mb->mv.x>>2) + x_off + 8;
- int my = (mb->mv.y>>2) + y_off;
- uint8_t **src= s->framep[ref]->tf.f->data;
- int off= mx + (my + (mb_x&3)*4)*s->linesize + 64;
+ int mx = (mb->mv.x >> 2) + x_off + 8;
+ int my = (mb->mv.y >> 2) + y_off;
+ uint8_t **src = s->framep[ref]->tf.f->data;
+ int off = mx + (my + (mb_x & 3) * 4) * s->linesize + 64;
/* For threading, a ff_thread_await_progress here might be useful, but
* it actually slows down the decoder. Since a bad prefetch doesn't
* generate bad decoder output, we don't run it here. */
- s->vdsp.prefetch(src[0]+off, s->linesize, 4);
- off= (mx>>1) + ((my>>1) + (mb_x&7))*s->uvlinesize + 64;
- s->vdsp.prefetch(src[1]+off, src[2]-src[1], 2);
+ s->vdsp.prefetch(src[0] + off, s->linesize, 4);
+ off = (mx >> 1) + ((my >> 1) + (mb_x & 7)) * s->uvlinesize + 64;
+ s->vdsp.prefetch(src[1] + off, src[2] - src[1], 2);
}
}
@@ -1311,7 +1348,7 @@ void inter_predict(VP8Context *s, VP8ThreadData *td, uint8_t *dst[3],
VP8Macroblock *mb, int mb_x, int mb_y)
{
int x_off = mb_x << 4, y_off = mb_y << 4;
- int width = 16*s->mb_width, height = 16*s->mb_height;
+ int width = 16 * s->mb_width, height = 16 * s->mb_height;
ThreadFrame *ref = &s->framep[mb->ref_frame]->tf;
VP56mv *bmv = mb->bmv;
@@ -1327,35 +1364,38 @@ void inter_predict(VP8Context *s, VP8ThreadData *td, uint8_t *dst[3],
/* Y */
for (y = 0; y < 4; y++) {
for (x = 0; x < 4; x++) {
- vp8_mc_luma(s, td, dst[0] + 4*y*s->linesize + x*4,
- ref, &bmv[4*y + x],
- 4*x + x_off, 4*y + y_off, 4, 4,
+ vp8_mc_luma(s, td, dst[0] + 4 * y * s->linesize + x * 4,
+ ref, &bmv[4 * y + x],
+ 4 * x + x_off, 4 * y + y_off, 4, 4,
width, height, s->linesize,
s->put_pixels_tab[2]);
}
}
/* U/V */
- x_off >>= 1; y_off >>= 1; width >>= 1; height >>= 1;
+ x_off >>= 1;
+ y_off >>= 1;
+ width >>= 1;
+ height >>= 1;
for (y = 0; y < 2; y++) {
for (x = 0; x < 2; x++) {
- uvmv.x = mb->bmv[ 2*y * 4 + 2*x ].x +
- mb->bmv[ 2*y * 4 + 2*x+1].x +
- mb->bmv[(2*y+1) * 4 + 2*x ].x +
- mb->bmv[(2*y+1) * 4 + 2*x+1].x;
- uvmv.y = mb->bmv[ 2*y * 4 + 2*x ].y +
- mb->bmv[ 2*y * 4 + 2*x+1].y +
- mb->bmv[(2*y+1) * 4 + 2*x ].y +
- mb->bmv[(2*y+1) * 4 + 2*x+1].y;
- uvmv.x = (uvmv.x + 2 + (uvmv.x >> (INT_BIT-1))) >> 2;
- uvmv.y = (uvmv.y + 2 + (uvmv.y >> (INT_BIT-1))) >> 2;
+ uvmv.x = mb->bmv[2 * y * 4 + 2 * x ].x +
+ mb->bmv[2 * y * 4 + 2 * x + 1].x +
+ mb->bmv[(2 * y + 1) * 4 + 2 * x ].x +
+ mb->bmv[(2 * y + 1) * 4 + 2 * x + 1].x;
+ uvmv.y = mb->bmv[2 * y * 4 + 2 * x ].y +
+ mb->bmv[2 * y * 4 + 2 * x + 1].y +
+ mb->bmv[(2 * y + 1) * 4 + 2 * x ].y +
+ mb->bmv[(2 * y + 1) * 4 + 2 * x + 1].y;
+ uvmv.x = (uvmv.x + 2 + (uvmv.x >> (INT_BIT - 1))) >> 2;
+ uvmv.y = (uvmv.y + 2 + (uvmv.y >> (INT_BIT - 1))) >> 2;
if (s->profile == 3) {
uvmv.x &= ~7;
uvmv.y &= ~7;
}
- vp8_mc_chroma(s, td, dst[1] + 4*y*s->uvlinesize + x*4,
- dst[2] + 4*y*s->uvlinesize + x*4, ref, &uvmv,
- 4*x + x_off, 4*y + y_off, 4, 4,
+ vp8_mc_chroma(s, td, dst[1] + 4 * y * s->uvlinesize + x * 4,
+ dst[2] + 4 * y * s->uvlinesize + x * 4, ref,
+ &uvmv, 4 * x + x_off, 4 * y + y_off, 4, 4,
width, height, s->uvlinesize,
s->put_pixels_tab[2]);
}
@@ -1387,8 +1427,8 @@ void inter_predict(VP8Context *s, VP8ThreadData *td, uint8_t *dst[3],
}
}
-static av_always_inline void idct_mb(VP8Context *s, VP8ThreadData *td,
- uint8_t *dst[3], VP8Macroblock *mb)
+static av_always_inline
+void idct_mb(VP8Context *s, VP8ThreadData *td, uint8_t *dst[3], VP8Macroblock *mb)
{
int x, y, ch;
@@ -1397,12 +1437,16 @@ static av_always_inline void idct_mb(VP8Context *s, VP8ThreadData *td,
for (y = 0; y < 4; y++) {
uint32_t nnz4 = AV_RL32(td->non_zero_count_cache[y]);
if (nnz4) {
- if (nnz4&~0x01010101) {
+ if (nnz4 & ~0x01010101) {
for (x = 0; x < 4; x++) {
- if ((uint8_t)nnz4 == 1)
- s->vp8dsp.vp8_idct_dc_add(y_dst+4*x, td->block[y][x], s->linesize);
- else if((uint8_t)nnz4 > 1)
- s->vp8dsp.vp8_idct_add(y_dst+4*x, td->block[y][x], s->linesize);
+ if ((uint8_t) nnz4 == 1)
+ s->vp8dsp.vp8_idct_dc_add(y_dst + 4 * x,
+ td->block[y][x],
+ s->linesize);
+ else if ((uint8_t) nnz4 > 1)
+ s->vp8dsp.vp8_idct_add(y_dst + 4 * x,
+ td->block[y][x],
+ s->linesize);
nnz4 >>= 8;
if (!nnz4)
break;
@@ -1411,36 +1455,42 @@ static av_always_inline void idct_mb(VP8Context *s, VP8ThreadData *td,
s->vp8dsp.vp8_idct_dc_add4y(y_dst, td->block[y], s->linesize);
}
}
- y_dst += 4*s->linesize;
+ y_dst += 4 * s->linesize;
}
}
for (ch = 0; ch < 2; ch++) {
- uint32_t nnz4 = AV_RL32(td->non_zero_count_cache[4+ch]);
+ uint32_t nnz4 = AV_RL32(td->non_zero_count_cache[4 + ch]);
if (nnz4) {
- uint8_t *ch_dst = dst[1+ch];
- if (nnz4&~0x01010101) {
+ uint8_t *ch_dst = dst[1 + ch];
+ if (nnz4 & ~0x01010101) {
for (y = 0; y < 2; y++) {
for (x = 0; x < 2; x++) {
- if ((uint8_t)nnz4 == 1)
- s->vp8dsp.vp8_idct_dc_add(ch_dst+4*x, td->block[4+ch][(y<<1)+x], s->uvlinesize);
- else if((uint8_t)nnz4 > 1)
- s->vp8dsp.vp8_idct_add(ch_dst+4*x, td->block[4+ch][(y<<1)+x], s->uvlinesize);
+ if ((uint8_t) nnz4 == 1)
+ s->vp8dsp.vp8_idct_dc_add(ch_dst + 4 * x,
+ td->block[4 + ch][(y << 1) + x],
+ s->uvlinesize);
+ else if ((uint8_t) nnz4 > 1)
+ s->vp8dsp.vp8_idct_add(ch_dst + 4 * x,
+ td->block[4 + ch][(y << 1) + x],
+ s->uvlinesize);
nnz4 >>= 8;
if (!nnz4)
goto chroma_idct_end;
}
- ch_dst += 4*s->uvlinesize;
+ ch_dst += 4 * s->uvlinesize;
}
} else {
- s->vp8dsp.vp8_idct_dc_add4uv(ch_dst, td->block[4+ch], s->uvlinesize);
+ s->vp8dsp.vp8_idct_dc_add4uv(ch_dst, td->block[4 + ch], s->uvlinesize);
}
}
-chroma_idct_end: ;
+chroma_idct_end:
+ ;
}
}
-static av_always_inline void filter_level_for_mb(VP8Context *s, VP8Macroblock *mb, VP8FilterStrength *f )
+static av_always_inline
+void filter_level_for_mb(VP8Context *s, VP8Macroblock *mb, VP8FilterStrength *f)
{
int interior_limit, filter_level;
@@ -1467,10 +1517,13 @@ static av_always_inline void filter_level_for_mb(VP8Context *s, VP8Macroblock *m
f->filter_level = filter_level;
f->inner_limit = interior_limit;
- f->inner_filter = !mb->skip || mb->mode == MODE_I4x4 || mb->mode == VP8_MVMODE_SPLIT;
+ f->inner_filter = !mb->skip || mb->mode == MODE_I4x4 ||
+ mb->mode == VP8_MVMODE_SPLIT;
}
-static av_always_inline void filter_mb(VP8Context *s, uint8_t *dst[3], VP8FilterStrength *f, int mb_x, int mb_y)
+static av_always_inline
+void filter_mb(VP8Context *s, uint8_t *dst[3], VP8FilterStrength *f,
+ int mb_x, int mb_y)
{
int mbedge_lim, bedge_lim, hev_thresh;
int filter_level = f->filter_level;
@@ -1492,82 +1545,84 @@ static av_always_inline void filter_mb(VP8Context *s, uint8_t *dst[3], VP8Filter
if (!filter_level)
return;
- bedge_lim = 2*filter_level + inner_limit;
+ bedge_lim = 2 * filter_level + inner_limit;
mbedge_lim = bedge_lim + 4;
hev_thresh = hev_thresh_lut[s->keyframe][filter_level];
if (mb_x) {
- s->vp8dsp.vp8_h_loop_filter16y(dst[0], linesize,
+ s->vp8dsp.vp8_h_loop_filter16y(dst[0], linesize,
mbedge_lim, inner_limit, hev_thresh);
- s->vp8dsp.vp8_h_loop_filter8uv(dst[1], dst[2], uvlinesize,
+ s->vp8dsp.vp8_h_loop_filter8uv(dst[1], dst[2], uvlinesize,
mbedge_lim, inner_limit, hev_thresh);
}
if (inner_filter) {
- s->vp8dsp.vp8_h_loop_filter16y_inner(dst[0]+ 4, linesize, bedge_lim,
+ s->vp8dsp.vp8_h_loop_filter16y_inner(dst[0] + 4, linesize, bedge_lim,
inner_limit, hev_thresh);
- s->vp8dsp.vp8_h_loop_filter16y_inner(dst[0]+ 8, linesize, bedge_lim,
+ s->vp8dsp.vp8_h_loop_filter16y_inner(dst[0] + 8, linesize, bedge_lim,
inner_limit, hev_thresh);
- s->vp8dsp.vp8_h_loop_filter16y_inner(dst[0]+12, linesize, bedge_lim,
+ s->vp8dsp.vp8_h_loop_filter16y_inner(dst[0] + 12, linesize, bedge_lim,
inner_limit, hev_thresh);
- s->vp8dsp.vp8_h_loop_filter8uv_inner(dst[1] + 4, dst[2] + 4,
- uvlinesize, bedge_lim,
+ s->vp8dsp.vp8_h_loop_filter8uv_inner(dst[1] + 4, dst[2] + 4,
+ uvlinesize, bedge_lim,
inner_limit, hev_thresh);
}
if (mb_y) {
- s->vp8dsp.vp8_v_loop_filter16y(dst[0], linesize,
+ s->vp8dsp.vp8_v_loop_filter16y(dst[0], linesize,
mbedge_lim, inner_limit, hev_thresh);
- s->vp8dsp.vp8_v_loop_filter8uv(dst[1], dst[2], uvlinesize,
+ s->vp8dsp.vp8_v_loop_filter8uv(dst[1], dst[2], uvlinesize,
mbedge_lim, inner_limit, hev_thresh);
}
if (inner_filter) {
- s->vp8dsp.vp8_v_loop_filter16y_inner(dst[0]+ 4*linesize,
- linesize, bedge_lim,
+ s->vp8dsp.vp8_v_loop_filter16y_inner(dst[0] + 4 * linesize,
+ linesize, bedge_lim,
inner_limit, hev_thresh);
- s->vp8dsp.vp8_v_loop_filter16y_inner(dst[0]+ 8*linesize,
- linesize, bedge_lim,
+ s->vp8dsp.vp8_v_loop_filter16y_inner(dst[0] + 8 * linesize,
+ linesize, bedge_lim,
inner_limit, hev_thresh);
- s->vp8dsp.vp8_v_loop_filter16y_inner(dst[0]+12*linesize,
- linesize, bedge_lim,
+ s->vp8dsp.vp8_v_loop_filter16y_inner(dst[0] + 12 * linesize,
+ linesize, bedge_lim,
inner_limit, hev_thresh);
- s->vp8dsp.vp8_v_loop_filter8uv_inner(dst[1] + 4 * uvlinesize,
- dst[2] + 4 * uvlinesize,
- uvlinesize, bedge_lim,
+ s->vp8dsp.vp8_v_loop_filter8uv_inner(dst[1] + 4 * uvlinesize,
+ dst[2] + 4 * uvlinesize,
+ uvlinesize, bedge_lim,
inner_limit, hev_thresh);
}
}
-static av_always_inline void filter_mb_simple(VP8Context *s, uint8_t *dst, VP8FilterStrength *f, int mb_x, int mb_y)
+static av_always_inline
+void filter_mb_simple(VP8Context *s, uint8_t *dst, VP8FilterStrength *f,
+ int mb_x, int mb_y)
{
int mbedge_lim, bedge_lim;
int filter_level = f->filter_level;
- int inner_limit = f->inner_limit;
+ int inner_limit = f->inner_limit;
int inner_filter = f->inner_filter;
- int linesize = s->linesize;
+ int linesize = s->linesize;
if (!filter_level)
return;
- bedge_lim = 2*filter_level + inner_limit;
+ bedge_lim = 2 * filter_level + inner_limit;
mbedge_lim = bedge_lim + 4;
if (mb_x)
s->vp8dsp.vp8_h_loop_filter_simple(dst, linesize, mbedge_lim);
if (inner_filter) {
- s->vp8dsp.vp8_h_loop_filter_simple(dst+ 4, linesize, bedge_lim);
- s->vp8dsp.vp8_h_loop_filter_simple(dst+ 8, linesize, bedge_lim);
- s->vp8dsp.vp8_h_loop_filter_simple(dst+12, linesize, bedge_lim);
+ s->vp8dsp.vp8_h_loop_filter_simple(dst + 4, linesize, bedge_lim);
+ s->vp8dsp.vp8_h_loop_filter_simple(dst + 8, linesize, bedge_lim);
+ s->vp8dsp.vp8_h_loop_filter_simple(dst + 12, linesize, bedge_lim);
}
if (mb_y)
s->vp8dsp.vp8_v_loop_filter_simple(dst, linesize, mbedge_lim);
if (inner_filter) {
- s->vp8dsp.vp8_v_loop_filter_simple(dst+ 4*linesize, linesize, bedge_lim);
- s->vp8dsp.vp8_v_loop_filter_simple(dst+ 8*linesize, linesize, bedge_lim);
- s->vp8dsp.vp8_v_loop_filter_simple(dst+12*linesize, linesize, bedge_lim);
+ s->vp8dsp.vp8_v_loop_filter_simple(dst + 4 * linesize, linesize, bedge_lim);
+ s->vp8dsp.vp8_v_loop_filter_simple(dst + 8 * linesize, linesize, bedge_lim);
+ s->vp8dsp.vp8_v_loop_filter_simple(dst + 12 * linesize, linesize, bedge_lim);
}
}
@@ -1581,16 +1636,18 @@ static void vp8_decode_mv_mb_modes(AVCodecContext *avctx, VP8Frame *curframe,
s->mv_min.y = -MARGIN;
s->mv_max.y = ((s->mb_height - 1) << 6) + MARGIN;
for (mb_y = 0; mb_y < s->mb_height; mb_y++) {
- VP8Macroblock *mb = s->macroblocks_base + ((s->mb_width+1)*(mb_y + 1) + 1);
- int mb_xy = mb_y*s->mb_width;
+ VP8Macroblock *mb = s->macroblocks_base +
+ ((s->mb_width + 1) * (mb_y + 1) + 1);
+ int mb_xy = mb_y * s->mb_width;
- AV_WN32A(s->intra4x4_pred_mode_left, DC_PRED*0x01010101);
+ AV_WN32A(s->intra4x4_pred_mode_left, DC_PRED * 0x01010101);
s->mv_min.x = -MARGIN;
s->mv_max.x = ((s->mb_width - 1) << 6) + MARGIN;
for (mb_x = 0; mb_x < s->mb_width; mb_x++, mb_xy++, mb++) {
if (mb_y == 0)
- AV_WN32A((mb-s->mb_width-1)->intra4x4_pred_mode_top, DC_PRED*0x01010101);
+ AV_WN32A((mb - s->mb_width - 1)->intra4x4_pred_mode_top,
+ DC_PRED * 0x01010101);
decode_mb_mode(s, mb, mb_x, mb_y, curframe->seg_map->data + mb_xy,
prev_frame && prev_frame->seg_map ?
prev_frame->seg_map->data + mb_xy : NULL, 1);
@@ -1603,37 +1660,40 @@ static void vp8_decode_mv_mb_modes(AVCodecContext *avctx, VP8Frame *curframe,
}
#if HAVE_THREADS
-#define check_thread_pos(td, otd, mb_x_check, mb_y_check)\
- do {\
- int tmp = (mb_y_check << 16) | (mb_x_check & 0xFFFF);\
- if (otd->thread_mb_pos < tmp) {\
- pthread_mutex_lock(&otd->lock);\
- td->wait_mb_pos = tmp;\
- do {\
- if (otd->thread_mb_pos >= tmp)\
- break;\
- pthread_cond_wait(&otd->cond, &otd->lock);\
- } while (1);\
- td->wait_mb_pos = INT_MAX;\
- pthread_mutex_unlock(&otd->lock);\
- }\
- } while(0);
-
-#define update_pos(td, mb_y, mb_x)\
- do {\
- int pos = (mb_y << 16) | (mb_x & 0xFFFF);\
- int sliced_threading = (avctx->active_thread_type == FF_THREAD_SLICE) && (num_jobs > 1);\
- int is_null = (next_td == NULL) || (prev_td == NULL);\
- int pos_check = (is_null) ? 1 :\
- (next_td != td && pos >= next_td->wait_mb_pos) ||\
- (prev_td != td && pos >= prev_td->wait_mb_pos);\
- td->thread_mb_pos = pos;\
- if (sliced_threading && pos_check) {\
- pthread_mutex_lock(&td->lock);\
- pthread_cond_broadcast(&td->cond);\
- pthread_mutex_unlock(&td->lock);\
- }\
- } while(0);
+#define check_thread_pos(td, otd, mb_x_check, mb_y_check) \
+ do { \
+ int tmp = (mb_y_check << 16) | (mb_x_check & 0xFFFF); \
+ if (otd->thread_mb_pos < tmp) { \
+ pthread_mutex_lock(&otd->lock); \
+ td->wait_mb_pos = tmp; \
+ do { \
+ if (otd->thread_mb_pos >= tmp) \
+ break; \
+ pthread_cond_wait(&otd->cond, &otd->lock); \
+ } while (1); \
+ td->wait_mb_pos = INT_MAX; \
+ pthread_mutex_unlock(&otd->lock); \
+ } \
+ } while (0);
+
+#define update_pos(td, mb_y, mb_x) \
+ do { \
+ int pos = (mb_y << 16) | (mb_x & 0xFFFF); \
+ int sliced_threading = (avctx->active_thread_type == FF_THREAD_SLICE) && \
+ (num_jobs > 1); \
+ int is_null = (next_td == NULL) || (prev_td == NULL); \
+ int pos_check = (is_null) ? 1 \
+ : (next_td != td && \
+ pos >= next_td->wait_mb_pos) || \
+ (prev_td != td && \
+ pos >= prev_td->wait_mb_pos); \
+ td->thread_mb_pos = pos; \
+ if (sliced_threading && pos_check) { \
+ pthread_mutex_lock(&td->lock); \
+ pthread_cond_broadcast(&td->cond); \
+ pthread_mutex_unlock(&td->lock); \
+ } \
+ } while (0);
#else
#define check_thread_pos(td, otd, mb_x_check, mb_y_check)
#define update_pos(td, mb_y, mb_x)
@@ -1644,51 +1704,58 @@ static void vp8_decode_mb_row_no_filter(AVCodecContext *avctx, void *tdata,
{
VP8Context *s = avctx->priv_data;
VP8ThreadData *prev_td, *next_td, *td = &s->thread_data[threadnr];
- int mb_y = td->thread_mb_pos>>16;
- int mb_x, mb_xy = mb_y*s->mb_width;
+ int mb_y = td->thread_mb_pos >> 16;
+ int mb_x, mb_xy = mb_y * s->mb_width;
int num_jobs = s->num_jobs;
VP8Frame *curframe = s->curframe, *prev_frame = s->prev_frame;
- VP56RangeCoder *c = &s->coeff_partition[mb_y & (s->num_coeff_partitions-1)];
+ VP56RangeCoder *c = &s->coeff_partition[mb_y & (s->num_coeff_partitions - 1)];
VP8Macroblock *mb;
uint8_t *dst[3] = {
- curframe->tf.f->data[0] + 16*mb_y*s->linesize,
- curframe->tf.f->data[1] + 8*mb_y*s->uvlinesize,
- curframe->tf.f->data[2] + 8*mb_y*s->uvlinesize
+ curframe->tf.f->data[0] + 16 * mb_y * s->linesize,
+ curframe->tf.f->data[1] + 8 * mb_y * s->uvlinesize,
+ curframe->tf.f->data[2] + 8 * mb_y * s->uvlinesize
};
- if (mb_y == 0) prev_td = td;
- else prev_td = &s->thread_data[(jobnr + num_jobs - 1)%num_jobs];
- if (mb_y == s->mb_height-1) next_td = td;
- else next_td = &s->thread_data[(jobnr + 1)%num_jobs];
+ if (mb_y == 0)
+ prev_td = td;
+ else
+ prev_td = &s->thread_data[(jobnr + num_jobs - 1) % num_jobs];
+ if (mb_y == s->mb_height - 1)
+ next_td = td;
+ else
+ next_td = &s->thread_data[(jobnr + 1) % num_jobs];
if (s->mb_layout == 1)
- mb = s->macroblocks_base + ((s->mb_width+1)*(mb_y + 1) + 1);
+ mb = s->macroblocks_base + ((s->mb_width + 1) * (mb_y + 1) + 1);
else {
// Make sure the previous frame has read its segmentation map,
// if we re-use the same map.
if (prev_frame && s->segmentation.enabled &&
!s->segmentation.update_map)
ff_thread_await_progress(&prev_frame->tf, mb_y, 0);
- mb = s->macroblocks + (s->mb_height - mb_y - 1)*2;
+ mb = s->macroblocks + (s->mb_height - mb_y - 1) * 2;
memset(mb - 1, 0, sizeof(*mb)); // zero left macroblock
- AV_WN32A(s->intra4x4_pred_mode_left, DC_PRED*0x01010101);
+ AV_WN32A(s->intra4x4_pred_mode_left, DC_PRED * 0x01010101);
}
memset(td->left_nnz, 0, sizeof(td->left_nnz));
s->mv_min.x = -MARGIN;
- s->mv_max.x = ((s->mb_width - 1) << 6) + MARGIN;
+ s->mv_max.x = ((s->mb_width - 1) << 6) + MARGIN;
for (mb_x = 0; mb_x < s->mb_width; mb_x++, mb_xy++, mb++) {
// Wait for previous thread to read mb_x+2, and reach mb_y-1.
if (prev_td != td) {
if (threadnr != 0) {
- check_thread_pos(td, prev_td, mb_x+1, mb_y-1);
+ check_thread_pos(td, prev_td, mb_x + 1, mb_y - 1);
} else {
- check_thread_pos(td, prev_td, (s->mb_width+3) + (mb_x+1), mb_y-1);
+ check_thread_pos(td, prev_td,
+ (s->mb_width + 3) + (mb_x + 1), mb_y - 1);
}
}
- s->vdsp.prefetch(dst[0] + (mb_x&3)*4*s->linesize + 64, s->linesize, 4);
- s->vdsp.prefetch(dst[1] + (mb_x&7)*s->uvlinesize + 64, dst[2] - dst[1], 2);
+ s->vdsp.prefetch(dst[0] + (mb_x & 3) * 4 * s->linesize + 64,
+ s->linesize, 4);
+ s->vdsp.prefetch(dst[1] + (mb_x & 7) * s->uvlinesize + 64,
+ dst[2] - dst[1], 2);
if (!s->mb_layout)
decode_mb_mode(s, mb, mb_x, mb_y, curframe->seg_map->data + mb_xy,
@@ -1713,7 +1780,8 @@ static void vp8_decode_mb_row_no_filter(AVCodecContext *avctx, void *tdata,
AV_ZERO64(td->left_nnz);
AV_WN64(s->top_nnz[mb_x], 0); // array of 9, so unaligned
- // Reset DC block predictors if they would exist if the mb had coefficients
+ /* Reset DC block predictors if they would exist
+ * if the mb had coefficients */
if (mb->mode != MODE_I4x4 && mb->mode != VP8_MVMODE_SPLIT) {
td->left_nnz[8] = 0;
s->top_nnz[mb_x][8] = 0;
@@ -1723,23 +1791,25 @@ static void vp8_decode_mb_row_no_filter(AVCodecContext *avctx, void *tdata,
if (s->deblock_filter)
filter_level_for_mb(s, mb, &td->filter_strength[mb_x]);
- if (s->deblock_filter && num_jobs != 1 && threadnr == num_jobs-1) {
+ if (s->deblock_filter && num_jobs != 1 && threadnr == num_jobs - 1) {
if (s->filter.simple)
- backup_mb_border(s->top_border[mb_x+1], dst[0], NULL, NULL, s->linesize, 0, 1);
+ backup_mb_border(s->top_border[mb_x + 1], dst[0],
+ NULL, NULL, s->linesize, 0, 1);
else
- backup_mb_border(s->top_border[mb_x+1], dst[0], dst[1], dst[2], s->linesize, s->uvlinesize, 0);
+ backup_mb_border(s->top_border[mb_x + 1], dst[0],
+ dst[1], dst[2], s->linesize, s->uvlinesize, 0);
}
prefetch_motion(s, mb, mb_x, mb_y, mb_xy, VP56_FRAME_GOLDEN2);
- dst[0] += 16;
- dst[1] += 8;
- dst[2] += 8;
+ dst[0] += 16;
+ dst[1] += 8;
+ dst[2] += 8;
s->mv_min.x -= 64;
s->mv_max.x -= 64;
- if (mb_x == s->mb_width+1) {
- update_pos(td, mb_y, s->mb_width+3);
+ if (mb_x == s->mb_width + 1) {
+ update_pos(td, mb_y, s->mb_width + 3);
} else {
update_pos(td, mb_y, mb_x);
}
@@ -1751,41 +1821,46 @@ static void vp8_filter_mb_row(AVCodecContext *avctx, void *tdata,
{
VP8Context *s = avctx->priv_data;
VP8ThreadData *td = &s->thread_data[threadnr];
- int mb_x, mb_y = td->thread_mb_pos>>16, num_jobs = s->num_jobs;
+ int mb_x, mb_y = td->thread_mb_pos >> 16, num_jobs = s->num_jobs;
AVFrame *curframe = s->curframe->tf.f;
VP8Macroblock *mb;
VP8ThreadData *prev_td, *next_td;
uint8_t *dst[3] = {
- curframe->data[0] + 16*mb_y*s->linesize,
- curframe->data[1] + 8*mb_y*s->uvlinesize,
- curframe->data[2] + 8*mb_y*s->uvlinesize
+ curframe->data[0] + 16 * mb_y * s->linesize,
+ curframe->data[1] + 8 * mb_y * s->uvlinesize,
+ curframe->data[2] + 8 * mb_y * s->uvlinesize
};
if (s->mb_layout == 1)
- mb = s->macroblocks_base + ((s->mb_width+1)*(mb_y + 1) + 1);
+ mb = s->macroblocks_base + ((s->mb_width + 1) * (mb_y + 1) + 1);
else
- mb = s->macroblocks + (s->mb_height - mb_y - 1)*2;
+ mb = s->macroblocks + (s->mb_height - mb_y - 1) * 2;
- if (mb_y == 0) prev_td = td;
- else prev_td = &s->thread_data[(jobnr + num_jobs - 1)%num_jobs];
- if (mb_y == s->mb_height-1) next_td = td;
- else next_td = &s->thread_data[(jobnr + 1)%num_jobs];
+ if (mb_y == 0)
+ prev_td = td;
+ else
+ prev_td = &s->thread_data[(jobnr + num_jobs - 1) % num_jobs];
+ if (mb_y == s->mb_height - 1)
+ next_td = td;
+ else
+ next_td = &s->thread_data[(jobnr + 1) % num_jobs];
for (mb_x = 0; mb_x < s->mb_width; mb_x++, mb++) {
VP8FilterStrength *f = &td->filter_strength[mb_x];
- if (prev_td != td) {
- check_thread_pos(td, prev_td, (mb_x+1) + (s->mb_width+3), mb_y-1);
- }
+ if (prev_td != td)
+ check_thread_pos(td, prev_td,
+ (mb_x + 1) + (s->mb_width + 3), mb_y - 1);
if (next_td != td)
- if (next_td != &s->thread_data[0]) {
- check_thread_pos(td, next_td, mb_x+1, mb_y+1);
- }
+ if (next_td != &s->thread_data[0])
+ check_thread_pos(td, next_td, mb_x + 1, mb_y + 1);
if (num_jobs == 1) {
if (s->filter.simple)
- backup_mb_border(s->top_border[mb_x+1], dst[0], NULL, NULL, s->linesize, 0, 1);
+ backup_mb_border(s->top_border[mb_x + 1], dst[0],
+ NULL, NULL, s->linesize, 0, 1);
else
- backup_mb_border(s->top_border[mb_x+1], dst[0], dst[1], dst[2], s->linesize, s->uvlinesize, 0);
+ backup_mb_border(s->top_border[mb_x + 1], dst[0],
+ dst[1], dst[2], s->linesize, s->uvlinesize, 0);
}
if (s->filter.simple)
@@ -1796,7 +1871,7 @@ static void vp8_filter_mb_row(AVCodecContext *avctx, void *tdata,
dst[1] += 8;
dst[2] += 8;
- update_pos(td, mb_y, (s->mb_width+3) + mb_x);
+ update_pos(td, mb_y, (s->mb_width + 3) + mb_x);
}
}
@@ -1808,10 +1883,12 @@ static int vp8_decode_mb_row_sliced(AVCodecContext *avctx, void *tdata,
VP8ThreadData *next_td = NULL, *prev_td = NULL;
VP8Frame *curframe = s->curframe;
int mb_y, num_jobs = s->num_jobs;
+
td->thread_nr = threadnr;
for (mb_y = jobnr; mb_y < s->mb_height; mb_y += num_jobs) {
- if (mb_y >= s->mb_height) break;
- td->thread_mb_pos = mb_y<<16;
+ if (mb_y >= s->mb_height)
+ break;
+ td->thread_mb_pos = mb_y << 16;
vp8_decode_mb_row_no_filter(avctx, tdata, jobnr, threadnr);
if (s->deblock_filter)
vp8_filter_mb_row(avctx, tdata, jobnr, threadnr);
@@ -1840,11 +1917,12 @@ int ff_vp8_decode_frame(AVCodecContext *avctx, void *data, int *got_frame,
prev_frame = s->framep[VP56_FRAME_CURRENT];
- referenced = s->update_last || s->update_golden == VP56_FRAME_CURRENT
- || s->update_altref == VP56_FRAME_CURRENT;
+ referenced = s->update_last || s->update_golden == VP56_FRAME_CURRENT ||
+ s->update_altref == VP56_FRAME_CURRENT;
- skip_thresh = !referenced ? AVDISCARD_NONREF :
- !s->keyframe ? AVDISCARD_NONKEY : AVDISCARD_ALL;
+ skip_thresh = !referenced ? AVDISCARD_NONREF
+ : !s->keyframe ? AVDISCARD_NONKEY
+ : AVDISCARD_ALL;
if (avctx->skip_frame >= skip_thresh) {
s->invisible = 1;
@@ -1858,7 +1936,7 @@ int ff_vp8_decode_frame(AVCodecContext *avctx, void *data, int *got_frame,
if (s->frames[i].tf.f->data[0] &&
&s->frames[i] != prev_frame &&
&s->frames[i] != s->framep[VP56_FRAME_PREVIOUS] &&
- &s->frames[i] != s->framep[VP56_FRAME_GOLDEN] &&
+ &s->frames[i] != s->framep[VP56_FRAME_GOLDEN] &&
&s->frames[i] != s->framep[VP56_FRAME_GOLDEN2])
vp8_release_frame(s, &s->frames[i]);
@@ -1866,7 +1944,7 @@ int ff_vp8_decode_frame(AVCodecContext *avctx, void *data, int *got_frame,
for (i = 0; i < 5; i++)
if (&s->frames[i] != prev_frame &&
&s->frames[i] != s->framep[VP56_FRAME_PREVIOUS] &&
- &s->frames[i] != s->framep[VP56_FRAME_GOLDEN] &&
+ &s->frames[i] != s->framep[VP56_FRAME_GOLDEN] &&
&s->frames[i] != s->framep[VP56_FRAME_GOLDEN2]) {
curframe = s->framep[VP56_FRAME_CURRENT] = &s->frames[i];
break;
@@ -1878,57 +1956,61 @@ int ff_vp8_decode_frame(AVCodecContext *avctx, void *data, int *got_frame,
if (curframe->tf.f->data[0])
vp8_release_frame(s, curframe);
- // Given that arithmetic probabilities are updated every frame, it's quite likely
- // that the values we have on a random interframe are complete junk if we didn't
- // start decode on a keyframe. So just don't display anything rather than junk.
+ /* Given that arithmetic probabilities are updated every frame, it's quite
+ * likely that the values we have on a random interframe are complete
+ * junk if we didn't start decode on a keyframe. So just don't display
+ * anything rather than junk. */
if (!s->keyframe && (!s->framep[VP56_FRAME_PREVIOUS] ||
- !s->framep[VP56_FRAME_GOLDEN] ||
+ !s->framep[VP56_FRAME_GOLDEN] ||
!s->framep[VP56_FRAME_GOLDEN2])) {
- av_log(avctx, AV_LOG_WARNING, "Discarding interframe without a prior keyframe!\n");
+ av_log(avctx, AV_LOG_WARNING,
+ "Discarding interframe without a prior keyframe!\n");
ret = AVERROR_INVALIDDATA;
goto err;
}
curframe->tf.f->key_frame = s->keyframe;
- curframe->tf.f->pict_type = s->keyframe ? AV_PICTURE_TYPE_I : AV_PICTURE_TYPE_P;
+ curframe->tf.f->pict_type = s->keyframe ? AV_PICTURE_TYPE_I
+ : AV_PICTURE_TYPE_P;
if ((ret = vp8_alloc_frame(s, curframe, referenced))) {
av_log(avctx, AV_LOG_ERROR, "get_buffer() failed!\n");
goto err;
}
// check if golden and altref are swapped
- if (s->update_altref != VP56_FRAME_NONE) {
- s->next_framep[VP56_FRAME_GOLDEN2] = s->framep[s->update_altref];
- } else {
- s->next_framep[VP56_FRAME_GOLDEN2] = s->framep[VP56_FRAME_GOLDEN2];
- }
- if (s->update_golden != VP56_FRAME_NONE) {
- s->next_framep[VP56_FRAME_GOLDEN] = s->framep[s->update_golden];
- } else {
- s->next_framep[VP56_FRAME_GOLDEN] = s->framep[VP56_FRAME_GOLDEN];
- }
- if (s->update_last) {
+ if (s->update_altref != VP56_FRAME_NONE)
+ s->next_framep[VP56_FRAME_GOLDEN2] = s->framep[s->update_altref];
+ else
+ s->next_framep[VP56_FRAME_GOLDEN2] = s->framep[VP56_FRAME_GOLDEN2];
+
+ if (s->update_golden != VP56_FRAME_NONE)
+ s->next_framep[VP56_FRAME_GOLDEN] = s->framep[s->update_golden];
+ else
+ s->next_framep[VP56_FRAME_GOLDEN] = s->framep[VP56_FRAME_GOLDEN];
+
+ if (s->update_last)
s->next_framep[VP56_FRAME_PREVIOUS] = curframe;
- } else {
+ else
s->next_framep[VP56_FRAME_PREVIOUS] = s->framep[VP56_FRAME_PREVIOUS];
- }
- s->next_framep[VP56_FRAME_CURRENT] = curframe;
+
+ s->next_framep[VP56_FRAME_CURRENT] = curframe;
ff_thread_finish_setup(avctx);
s->linesize = curframe->tf.f->linesize[0];
s->uvlinesize = curframe->tf.f->linesize[1];
- memset(s->top_nnz, 0, s->mb_width*sizeof(*s->top_nnz));
- /* Zero macroblock structures for top/top-left prediction from outside the frame. */
+ memset(s->top_nnz, 0, s->mb_width * sizeof(*s->top_nnz));
+ /* Zero macroblock structures for top/top-left prediction
+ * from outside the frame. */
if (!s->mb_layout)
- memset(s->macroblocks + s->mb_height*2 - 1, 0, (s->mb_width+1)*sizeof(*s->macroblocks));
+ memset(s->macroblocks + s->mb_height * 2 - 1, 0,
+ (s->mb_width + 1) * sizeof(*s->macroblocks));
if (!s->mb_layout && s->keyframe)
- memset(s->intra4x4_pred_mode_top, DC_PRED, s->mb_width*4);
+ memset(s->intra4x4_pred_mode_top, DC_PRED, s->mb_width * 4);
memset(s->ref_count, 0, sizeof(s->ref_count));
-
if (s->mb_layout == 1) {
// Make sure the previous frame has read its segmentation map,
// if we re-use the same map.
@@ -1949,9 +2031,10 @@ int ff_vp8_decode_frame(AVCodecContext *avctx, void *data, int *got_frame,
s->mv_max.y = ((s->mb_height - 1) << 6) + MARGIN;
for (i = 0; i < MAX_THREADS; i++) {
s->thread_data[i].thread_mb_pos = 0;
- s->thread_data[i].wait_mb_pos = INT_MAX;
+ s->thread_data[i].wait_mb_pos = INT_MAX;
}
- avctx->execute2(avctx, vp8_decode_mb_row_sliced, s->thread_data, NULL, num_jobs);
+ avctx->execute2(avctx, vp8_decode_mb_row_sliced,
+ s->thread_data, NULL, num_jobs);
ff_thread_report_progress(&curframe->tf, INT_MAX, 0);
memcpy(&s->framep[0], &s->next_framep[0], sizeof(s->framep[0]) * 4);
@@ -1965,7 +2048,7 @@ skip_decode:
if (!s->invisible) {
if ((ret = av_frame_ref(data, curframe->tf.f)) < 0)
return ret;
- *got_frame = 1;
+ *got_frame = 1;
}
return avpkt->size;
@@ -2033,10 +2116,10 @@ static av_cold int vp8_decode_init_thread_copy(AVCodecContext *avctx)
return 0;
}
-#define REBASE(pic) \
- pic ? pic - &s_src->frames[0] + &s->frames[0] : NULL
+#define REBASE(pic) pic ? pic - &s_src->frames[0] + &s->frames[0] : NULL
-static int vp8_decode_update_thread_context(AVCodecContext *dst, const AVCodecContext *src)
+static int vp8_decode_update_thread_context(AVCodecContext *dst,
+ const AVCodecContext *src)
{
VP8Context *s = dst->priv_data, *s_src = src->priv_data;
int i;
@@ -2048,9 +2131,9 @@ static int vp8_decode_update_thread_context(AVCodecContext *dst, const AVCodecCo
s->mb_height = s_src->mb_height;
}
- s->prob[0] = s_src->prob[!s_src->update_probabilities];
+ s->prob[0] = s_src->prob[!s_src->update_probabilities];
s->segmentation = s_src->segmentation;
- s->lf_delta = s_src->lf_delta;
+ s->lf_delta = s_src->lf_delta;
memcpy(s->sign_bias, s_src->sign_bias, sizeof(s->sign_bias));
for (i = 0; i < FF_ARRAY_ELEMS(s_src->frames); i++) {
generated by cgit v1.2.3 (git 2.39.1) at 2024-05-18 11:32:49 +0000