*/
/**
- * @file h264.c
+ * @file libavcodec/h264.c
* H.264 / AVC / MPEG4 part10 codec.
* @author Michael Niedermayer <michaelni@gmx.at>
*/
+#include "internal.h"
#include "dsputil.h"
#include "avcodec.h"
#include "mpegvideo.h"
#include "h264data.h"
#include "h264_parser.h"
#include "golomb.h"
+#include "mathops.h"
#include "rectangle.h"
+#include "vdpau_internal.h"
#include "cabac.h"
-#ifdef ARCH_X86
-#include "i386/h264_i386.h"
+#if ARCH_X86
+#include "x86/h264_i386.h"
#endif
//#undef NDEBUG
0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6, 7, 7, 7, 7, 7, 7, 8, 8, 8, 8,
};
-static const int left_block_options[4][8]={
+static const uint8_t left_block_options[4][8]={
{0,1,2,3,7,10,8,11},
{2,2,3,3,8,11,8,11},
{0,0,1,1,7,10,7,10},
{0,2,0,2,7,10,7,10}
};
+#define LEVEL_TAB_BITS 8
+static int8_t cavlc_level_tab[7][1<<LEVEL_TAB_BITS][2];
+
static void fill_caches(H264Context *h, int mb_type, int for_deblock){
MpegEncContext * const s = &h->s;
const int mb_xy= h->mb_xy;
int topleft_xy, top_xy, topright_xy, left_xy[2];
int topleft_type, top_type, topright_type, left_type[2];
- const int * left_block;
+ const uint8_t * left_block;
int topleft_partition= -1;
int i;
const int top_pair_xy = pair_xy - s->mb_stride;
const int topleft_pair_xy = top_pair_xy - 1;
const int topright_pair_xy = top_pair_xy + 1;
- const int topleft_mb_frame_flag = !IS_INTERLACED(s->current_picture.mb_type[topleft_pair_xy]);
- const int top_mb_frame_flag = !IS_INTERLACED(s->current_picture.mb_type[top_pair_xy]);
- const int topright_mb_frame_flag = !IS_INTERLACED(s->current_picture.mb_type[topright_pair_xy]);
- const int left_mb_frame_flag = !IS_INTERLACED(s->current_picture.mb_type[pair_xy-1]);
- const int curr_mb_frame_flag = !IS_INTERLACED(mb_type);
+ const int topleft_mb_field_flag = IS_INTERLACED(s->current_picture.mb_type[topleft_pair_xy]);
+ const int top_mb_field_flag = IS_INTERLACED(s->current_picture.mb_type[top_pair_xy]);
+ const int topright_mb_field_flag = IS_INTERLACED(s->current_picture.mb_type[topright_pair_xy]);
+ const int left_mb_field_flag = IS_INTERLACED(s->current_picture.mb_type[pair_xy-1]);
+ const int curr_mb_field_flag = IS_INTERLACED(mb_type);
const int bottom = (s->mb_y & 1);
- tprintf(s->avctx, "fill_caches: curr_mb_frame_flag:%d, left_mb_frame_flag:%d, topleft_mb_frame_flag:%d, top_mb_frame_flag:%d, topright_mb_frame_flag:%d\n", curr_mb_frame_flag, left_mb_frame_flag, topleft_mb_frame_flag, top_mb_frame_flag, topright_mb_frame_flag);
- if (bottom
- ? !curr_mb_frame_flag // bottom macroblock
- : (!curr_mb_frame_flag && !top_mb_frame_flag) // top macroblock
- ) {
+ tprintf(s->avctx, "fill_caches: curr_mb_field_flag:%d, left_mb_field_flag:%d, topleft_mb_field_flag:%d, top_mb_field_flag:%d, topright_mb_field_flag:%d\n", curr_mb_field_flag, left_mb_field_flag, topleft_mb_field_flag, top_mb_field_flag, topright_mb_field_flag);
+
+ if (curr_mb_field_flag && (bottom || top_mb_field_flag)){
top_xy -= s->mb_stride;
}
- if (bottom
- ? !curr_mb_frame_flag // bottom macroblock
- : (!curr_mb_frame_flag && !topleft_mb_frame_flag) // top macroblock
- ) {
+ if (curr_mb_field_flag && (bottom || topleft_mb_field_flag)){
topleft_xy -= s->mb_stride;
- } else if(bottom && curr_mb_frame_flag && !left_mb_frame_flag) {
+ } else if(bottom && !curr_mb_field_flag && left_mb_field_flag) {
topleft_xy += s->mb_stride;
// take top left mv from the middle of the mb, as opposed to all other modes which use the bottom right partition
topleft_partition = 0;
}
- if (bottom
- ? !curr_mb_frame_flag // bottom macroblock
- : (!curr_mb_frame_flag && !topright_mb_frame_flag) // top macroblock
- ) {
+ if (curr_mb_field_flag && (bottom || topright_mb_field_flag)){
topright_xy -= s->mb_stride;
}
- if (left_mb_frame_flag != curr_mb_frame_flag) {
+ if (left_mb_field_flag != curr_mb_field_flag) {
left_xy[1] = left_xy[0] = pair_xy - 1;
- if (curr_mb_frame_flag) {
- if (bottom) {
- left_block = left_block_options[1];
- } else {
- left_block= left_block_options[2];
- }
- } else {
+ if (curr_mb_field_flag) {
left_xy[1] += s->mb_stride;
left_block = left_block_options[3];
+ } else {
+ left_block= left_block_options[2 - bottom];
}
}
}
static const int mask[4]={0x8000,0x2000,0x80,0x20};
for(i=0; i<4; i++){
if(!(h->left_samples_available&mask[i])){
- int status= left[ h->intra4x4_pred_mode_cache[scan8[0] + 8*i] ];
- if(status<0){
- av_log(h->s.avctx, AV_LOG_ERROR, "left block unavailable for requested intra4x4 mode %d at %d %d\n", status, s->mb_x, s->mb_y);
- return -1;
- } else if(status){
- h->intra4x4_pred_mode_cache[scan8[0] + 8*i]= status;
- }
+ int status= left[ h->intra4x4_pred_mode_cache[scan8[0] + 8*i] ];
+ if(status<0){
+ av_log(h->s.avctx, AV_LOG_ERROR, "left block unavailable for requested intra4x4 mode %d at %d %d\n", status, s->mb_x, s->mb_y);
+ return -1;
+ } else if(status){
+ h->intra4x4_pred_mode_cache[scan8[0] + 8*i]= status;
+ }
}
}
}
tprintf(h->s.avctx, "pred_pskip: (%d) (%d) at %2d %2d\n", top_ref, left_ref, h->s.mb_x, h->s.mb_y);
if(top_ref == PART_NOT_AVAILABLE || left_ref == PART_NOT_AVAILABLE
- || (top_ref == 0 && *(uint32_t*)h->mv_cache[0][ scan8[0] - 8 ] == 0)
- || (left_ref == 0 && *(uint32_t*)h->mv_cache[0][ scan8[0] - 1 ] == 0)){
+ || !( top_ref | *(uint32_t*)h->mv_cache[0][ scan8[0] - 8 ])
+ || !(left_ref | *(uint32_t*)h->mv_cache[0][ scan8[0] - 1 ])){
*mx = *my = 0;
return;
}
}
-/**
- * Decodes a network abstraction layer unit.
- * @param consumed is the number of bytes used as input
- * @param length is the length of the array
- * @param dst_length is the number of decoded bytes FIXME here or a decode rbsp tailing?
- * @returns decoded bytes, might be src+1 if no escapes
- */
-static const uint8_t *decode_nal(H264Context *h, const uint8_t *src, int *dst_length, int *consumed, int length){
+const uint8_t *ff_h264_decode_nal(H264Context *h, const uint8_t *src, int *dst_length, int *consumed, int length){
int i, si, di;
uint8_t *dst;
int bufidx;
for(i=0; i<length; i++)
printf("%2X ", src[i]);
#endif
+
+#if HAVE_FAST_UNALIGNED
+# if HAVE_FAST_64BIT
+# define RS 7
+ for(i=0; i+1<length; i+=9){
+ if(!((~*(const uint64_t*)(src+i) & (*(const uint64_t*)(src+i) - 0x0100010001000101ULL)) & 0x8000800080008080ULL))
+# else
+# define RS 3
+ for(i=0; i+1<length; i+=5){
+ if(!((~*(const uint32_t*)(src+i) & (*(const uint32_t*)(src+i) - 0x01000101U)) & 0x80008080U))
+# endif
+ continue;
+ if(i>0 && !src[i]) i--;
+ while(src[i]) i++;
+#else
+# define RS 0
for(i=0; i+1<length; i+=2){
if(src[i]) continue;
if(i>0 && src[i-1]==0) i--;
+#endif
if(i+2<length && src[i+1]==0 && src[i+2]<=3){
if(src[i+2]!=3){
/* startcode, so we must be past the end */
}
break;
}
+ i-= RS;
}
if(i>=length-1){ //no escaped 0
}
bufidx = h->nal_unit_type == NAL_DPC ? 1 : 0; // use second escape buffer for inter data
- h->rbsp_buffer[bufidx]= av_fast_realloc(h->rbsp_buffer[bufidx], &h->rbsp_buffer_size[bufidx], length);
+ h->rbsp_buffer[bufidx]= av_fast_realloc(h->rbsp_buffer[bufidx], &h->rbsp_buffer_size[bufidx], length+FF_INPUT_BUFFER_PADDING_SIZE);
dst= h->rbsp_buffer[bufidx];
if (dst == NULL){
}
//printf("decoding esc\n");
- si=di=0;
- while(si<length){
+ memcpy(dst, src, i);
+ si=di=i;
+ while(si+2<length){
//remove escapes (very rare 1:2^22)
- if(si+2<length && src[si]==0 && src[si+1]==0 && src[si+2]<=3){
+ if(src[si+2]>3){
+ dst[di++]= src[si++];
+ dst[di++]= src[si++];
+ }else if(src[si]==0 && src[si+1]==0){
if(src[si+2]==3){ //escape
dst[di++]= 0;
dst[di++]= 0;
si+=3;
continue;
}else //next start code
- break;
+ goto nsc;
}
dst[di++]= src[si++];
}
+ while(si<length)
+ dst[di++]= src[si++];
+nsc:
+
+ memset(dst+di, 0, FF_INPUT_BUFFER_PADDING_SIZE);
*dst_length= di;
*consumed= si + 1;//+1 for the header
return dst;
}
-/**
- * identifies the exact end of the bitstream
- * @return the length of the trailing, or 0 if damaged
- */
-static int decode_rbsp_trailing(H264Context *h, const uint8_t *src){
+int ff_h264_decode_rbsp_trailing(H264Context *h, const uint8_t *src){
int v= *src;
int r;
const int pic_width = 16*s->mb_width;
const int pic_height = 16*s->mb_height >> MB_FIELD;
- if(!pic->data[0]) //FIXME this is unacceptable, some sensible error concealment must be done for missing reference frames
- return;
-
if(mx&7) extra_width -= 3;
if(my&7) extra_height -= 3;
qpix_op[luma_xy](dest_y + delta, src_y + delta, h->mb_linesize);
}
- if(ENABLE_GRAY && s->flags&CODEC_FLAG_GRAY) return;
+ if(CONFIG_GRAY && s->flags&CODEC_FLAG_GRAY) return;
if(MB_FIELD){
// chroma offset when predicting from a field of opposite parity
prefetch_motion(h, 1);
}
+static av_cold void init_cavlc_level_tab(void){
+ int suffix_length, mask;
+ unsigned int i;
+
+ for(suffix_length=0; suffix_length<7; suffix_length++){
+ for(i=0; i<(1<<LEVEL_TAB_BITS); i++){
+ int prefix= LEVEL_TAB_BITS - av_log2(2*i);
+ int level_code= (prefix<<suffix_length) + (i>>(LEVEL_TAB_BITS-prefix-1-suffix_length)) - (1<<suffix_length);
+
+ mask= -(level_code&1);
+ level_code= (((2+level_code)>>1) ^ mask) - mask;
+ if(prefix + 1 + suffix_length <= LEVEL_TAB_BITS){
+ cavlc_level_tab[suffix_length][i][0]= level_code;
+ cavlc_level_tab[suffix_length][i][1]= prefix + 1 + suffix_length;
+ }else if(prefix + 1 <= LEVEL_TAB_BITS){
+ cavlc_level_tab[suffix_length][i][0]= prefix+100;
+ cavlc_level_tab[suffix_length][i][1]= prefix + 1;
+ }else{
+ cavlc_level_tab[suffix_length][i][0]= LEVEL_TAB_BITS+100;
+ cavlc_level_tab[suffix_length][i][1]= LEVEL_TAB_BITS;
+ }
+ }
+ }
+}
+
static av_cold void decode_init_vlc(void){
static int done = 0;
&run_len [6][0], 1, 1,
&run_bits[6][0], 1, 1,
INIT_VLC_USE_NEW_STATIC);
+
+ init_cavlc_level_tab();
}
}
s->unrestricted_mv=1;
s->decode=1; //FIXME
+ dsputil_init(&s->dsp, s->avctx); // needed so that idct permutation is known early
+
memset(h->pps.scaling_matrix4, 16, 6*16*sizeof(uint8_t));
memset(h->pps.scaling_matrix8, 16, 2*64*sizeof(uint8_t));
}
+/**
+ * Reset SEI values at the beginning of the frame.
+ *
+ * @param h H.264 context.
+ */
+static void reset_sei(H264Context *h) {
+ h->sei_recovery_frame_cnt = -1;
+ h->sei_dpb_output_delay = 0;
+ h->sei_cpb_removal_delay = -1;
+ h->sei_buffering_period_present = 0;
+}
+
static av_cold int decode_init(AVCodecContext *avctx){
H264Context *h= avctx->priv_data;
MpegEncContext * const s = &h->s;
// set defaults
// s->decode_mb= ff_h263_decode_mb;
s->quarter_sample = 1;
+ if(!avctx->has_b_frames)
s->low_delay= 1;
- if(avctx->codec_id == CODEC_ID_SVQ3)
- avctx->pix_fmt= PIX_FMT_YUVJ420P;
+ if(s->avctx->codec->capabilities&CODEC_CAP_HWACCEL_VDPAU)
+ avctx->pix_fmt= PIX_FMT_VDPAU_H264;
else
- avctx->pix_fmt= PIX_FMT_YUV420P;
+ avctx->pix_fmt= avctx->get_format(avctx, avctx->codec->pix_fmts);
+ avctx->hwaccel = ff_find_hwaccel(avctx->codec->id, avctx->pix_fmt);
decode_init_vlc();
h->thread_context[0] = h;
h->outputed_poc = INT_MIN;
h->prev_poc_msb= 1<<16;
+ reset_sei(h);
+ if(avctx->codec_id == CODEC_ID_H264){
+ if(avctx->ticks_per_frame == 1){
+ s->avctx->time_base.den *=2;
+ }
+ avctx->ticks_per_frame = 2;
+ }
return 0;
}
if(!MB_MBAFF){
*(uint64_t*)(h->top_borders[0][s->mb_x]+ 0)= *(uint64_t*)(src_y + 15*linesize);
*(uint64_t*)(h->top_borders[0][s->mb_x]+ 8)= *(uint64_t*)(src_y +8+15*linesize);
- if(simple || !ENABLE_GRAY || !(s->flags&CODEC_FLAG_GRAY)){
+ if(simple || !CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)){
*(uint64_t*)(h->top_borders[0][s->mb_x]+16)= *(uint64_t*)(src_cb+7*uvlinesize);
*(uint64_t*)(h->top_borders[0][s->mb_x]+24)= *(uint64_t*)(src_cr+7*uvlinesize);
}
}else{
if(!MB_MBAFF){
h->left_border[0]= h->top_borders[0][s->mb_x][15];
- if(simple || !ENABLE_GRAY || !(s->flags&CODEC_FLAG_GRAY)){
+ if(simple || !CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)){
h->left_border[34 ]= h->top_borders[0][s->mb_x][16+7 ];
h->left_border[34+18]= h->top_borders[0][s->mb_x][16+8+7];
}
*(uint64_t*)(h->top_borders[top_idx][s->mb_x]+0)= *(uint64_t*)(src_y + 16*linesize);
*(uint64_t*)(h->top_borders[top_idx][s->mb_x]+8)= *(uint64_t*)(src_y +8+16*linesize);
- if(simple || !ENABLE_GRAY || !(s->flags&CODEC_FLAG_GRAY)){
+ if(simple || !CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)){
h->left_border[uvoffset+34 ]= h->top_borders[top_idx][s->mb_x][16+7];
h->left_border[uvoffset+34+18]= h->top_borders[top_idx][s->mb_x][24+7];
for(i=1; i<9 - skiplast; i++){
}
}
- if(simple || !ENABLE_GRAY || !(s->flags&CODEC_FLAG_GRAY)){
+ if(simple || !CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)){
if(deblock_left){
for(i = !deblock_top; i<8; i++){
XCHG(h->left_border[uvoffset+34 +i*step], src_cb[i*uvlinesize], temp8, xchg);
int i;
int *block_offset = &h->block_offset[0];
const int transform_bypass = !simple && (s->qscale == 0 && h->sps.transform_bypass);
- const int is_h264 = simple || s->codec_id == CODEC_ID_H264;
+ /* is_h264 should always be true if SVQ3 is disabled. */
+ const int is_h264 = !CONFIG_SVQ3_DECODER || simple || s->codec_id == CODEC_ID_H264;
void (*idct_add)(uint8_t *dst, DCTELEM *block, int stride);
void (*idct_dc_add)(uint8_t *dst, DCTELEM *block, int stride);
if(h->deblocking_filter)
xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, 1, simple);
- if(simple || !ENABLE_GRAY || !(s->flags&CODEC_FLAG_GRAY)){
+ if(simple || !CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)){
h->hpc.pred8x8[ h->chroma_pred_mode ](dest_cb, uvlinesize);
h->hpc.pred8x8[ h->chroma_pred_mode ](dest_cr, uvlinesize);
}
idct_dc_add = s->dsp.h264_idct_dc_add;
idct_add = s->dsp.h264_idct_add;
}
- for(i=0; i<16; i++){
- uint8_t * const ptr= dest_y + block_offset[i];
- const int dir= h->intra4x4_pred_mode_cache[ scan8[i] ];
+ for(i=0; i<16; i++){
+ uint8_t * const ptr= dest_y + block_offset[i];
+ const int dir= h->intra4x4_pred_mode_cache[ scan8[i] ];
- if(transform_bypass && h->sps.profile_idc==244 && dir<=1){
- h->hpc.pred4x4_add[dir](ptr, h->mb + i*16, linesize);
- }else{
- uint8_t *topright;
- int nnz, tr;
- if(dir == DIAG_DOWN_LEFT_PRED || dir == VERT_LEFT_PRED){
- const int topright_avail= (h->topright_samples_available<<i)&0x8000;
- assert(mb_y || linesize <= block_offset[i]);
- if(!topright_avail){
- tr= ptr[3 - linesize]*0x01010101;
- topright= (uint8_t*) &tr;
+ if(transform_bypass && h->sps.profile_idc==244 && dir<=1){
+ h->hpc.pred4x4_add[dir](ptr, h->mb + i*16, linesize);
+ }else{
+ uint8_t *topright;
+ int nnz, tr;
+ if(dir == DIAG_DOWN_LEFT_PRED || dir == VERT_LEFT_PRED){
+ const int topright_avail= (h->topright_samples_available<<i)&0x8000;
+ assert(mb_y || linesize <= block_offset[i]);
+ if(!topright_avail){
+ tr= ptr[3 - linesize]*0x01010101;
+ topright= (uint8_t*) &tr;
+ }else
+ topright= ptr + 4 - linesize;
}else
- topright= ptr + 4 - linesize;
- }else
- topright= NULL;
+ topright= NULL;
- h->hpc.pred4x4[ dir ](ptr, topright, linesize);
- nnz = h->non_zero_count_cache[ scan8[i] ];
- if(nnz){
- if(is_h264){
- if(nnz == 1 && h->mb[i*16])
- idct_dc_add(ptr, h->mb + i*16, linesize);
- else
- idct_add (ptr, h->mb + i*16, linesize);
- }else
- svq3_add_idct_c(ptr, h->mb + i*16, linesize, s->qscale, 0);
+ h->hpc.pred4x4[ dir ](ptr, topright, linesize);
+ nnz = h->non_zero_count_cache[ scan8[i] ];
+ if(nnz){
+ if(is_h264){
+ if(nnz == 1 && h->mb[i*16])
+ idct_dc_add(ptr, h->mb + i*16, linesize);
+ else
+ idct_add (ptr, h->mb + i*16, linesize);
+ }else
+ svq3_add_idct_c(ptr, h->mb + i*16, linesize, s->qscale, 0);
+ }
}
}
}
- }
}
}else{
h->hpc.pred16x16[ h->intra16x16_pred_mode ](dest_y , linesize);
}
}
- if((simple || !ENABLE_GRAY || !(s->flags&CODEC_FLAG_GRAY)) && (h->cbp&0x30)){
+ if((simple || !CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)) && (h->cbp&0x30)){
uint8_t *dest[2] = {dest_cb, dest_cr};
if(transform_bypass){
- idct_add = idct_dc_add = s->dsp.add_pixels4;
- }else{
- idct_add = s->dsp.h264_idct_add;
- idct_dc_add = s->dsp.h264_idct_dc_add;
- chroma_dc_dequant_idct_c(h->mb + 16*16, h->chroma_qp[0], h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp[0]][0]);
- chroma_dc_dequant_idct_c(h->mb + 16*16+4*16, h->chroma_qp[1], h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp[1]][0]);
- }
- if(is_h264){
- if(transform_bypass && IS_INTRA(mb_type) && h->sps.profile_idc==244 && (h->chroma_pred_mode==VERT_PRED8x8 || h->chroma_pred_mode==HOR_PRED8x8)){
+ if(IS_INTRA(mb_type) && h->sps.profile_idc==244 && (h->chroma_pred_mode==VERT_PRED8x8 || h->chroma_pred_mode==HOR_PRED8x8)){
h->hpc.pred8x8_add[h->chroma_pred_mode](dest[0], block_offset + 16, h->mb + 16*16, uvlinesize);
h->hpc.pred8x8_add[h->chroma_pred_mode](dest[1], block_offset + 20, h->mb + 20*16, uvlinesize);
}else{
+ idct_add = s->dsp.add_pixels4;
+ for(i=16; i<16+8; i++){
+ if(h->non_zero_count_cache[ scan8[i] ] || h->mb[i*16])
+ idct_add (dest[(i&4)>>2] + block_offset[i], h->mb + i*16, uvlinesize);
+ }
+ }
+ }else{
+ chroma_dc_dequant_idct_c(h->mb + 16*16, h->chroma_qp[0], h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp[0]][0]);
+ chroma_dc_dequant_idct_c(h->mb + 16*16+4*16, h->chroma_qp[1], h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp[1]][0]);
+ if(is_h264){
+ idct_add = s->dsp.h264_idct_add;
+ idct_dc_add = s->dsp.h264_idct_dc_add;
for(i=16; i<16+8; i++){
if(h->non_zero_count_cache[ scan8[i] ])
idct_add (dest[(i&4)>>2] + block_offset[i], h->mb + i*16, uvlinesize);
else if(h->mb[i*16])
idct_dc_add(dest[(i&4)>>2] + block_offset[i], h->mb + i*16, uvlinesize);
}
- }
- }else{
- for(i=16; i<16+8; i++){
- if(h->non_zero_count_cache[ scan8[i] ] || h->mb[i*16]){
- uint8_t * const ptr= dest[(i&4)>>2] + block_offset[i];
- svq3_add_idct_c(ptr, h->mb + i*16, uvlinesize, chroma_qp[s->qscale + 12] - 12, 2);
+ }else{
+ for(i=16; i<16+8; i++){
+ if(h->non_zero_count_cache[ scan8[i] ] || h->mb[i*16]){
+ uint8_t * const ptr= dest[(i&4)>>2] + block_offset[i];
+ svq3_add_idct_c(ptr, h->mb + i*16, uvlinesize, chroma_qp[s->qscale + 12] - 12, 2);
+ }
}
}
}
}
}
+ if(h->cbp || IS_INTRA(mb_type))
+ s->dsp.clear_blocks(h->mb);
+
if(h->deblocking_filter) {
backup_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, simple);
fill_caches(h, mb_type, 1); //FIXME don't fill stuff which isn't used by filter_mb
MpegEncContext * const s = &h->s;
const int mb_xy= h->mb_xy;
const int mb_type= s->current_picture.mb_type[mb_xy];
- int is_complex = ENABLE_SMALL || h->is_complex || IS_INTRA_PCM(mb_type) || s->qscale == 0;
-
- if(ENABLE_H264_ENCODER && !s->decode)
- return;
+ int is_complex = CONFIG_SMALL || h->is_complex || IS_INTRA_PCM(mb_type) || s->qscale == 0;
if (is_complex)
hl_decode_mb_complex(h);
int pred= h->curr_pic_num;
for(index=0; ; index++){
- unsigned int reordering_of_pic_nums_idc= get_ue_golomb(&s->gb);
+ unsigned int reordering_of_pic_nums_idc= get_ue_golomb_31(&s->gb);
unsigned int pic_id;
int i;
Picture *ref = NULL;
chroma_def = 1<<h->chroma_log2_weight_denom;
for(list=0; list<2; list++){
+ h->luma_weight_flag[list] = 0;
+ h->chroma_weight_flag[list] = 0;
for(i=0; i<h->ref_count[list]; i++){
int luma_weight_flag, chroma_weight_flag;
h->luma_weight[list][i]= get_se_golomb(&s->gb);
h->luma_offset[list][i]= get_se_golomb(&s->gb);
if( h->luma_weight[list][i] != luma_def
- || h->luma_offset[list][i] != 0)
+ || h->luma_offset[list][i] != 0) {
h->use_weight= 1;
+ h->luma_weight_flag[list]= 1;
+ }
}else{
h->luma_weight[list][i]= luma_def;
h->luma_offset[list][i]= 0;
h->chroma_weight[list][i][j]= get_se_golomb(&s->gb);
h->chroma_offset[list][i][j]= get_se_golomb(&s->gb);
if( h->chroma_weight[list][i][j] != chroma_def
- || h->chroma_offset[list][i][j] != 0)
+ || h->chroma_offset[list][i][j] != 0) {
h->use_weight_chroma= 1;
+ h->chroma_weight_flag[list]= 1;
+ }
}
}else{
int j;
static void implicit_weight_table(H264Context *h){
MpegEncContext * const s = &h->s;
- int ref0, ref1;
+ int ref0, ref1, i;
int cur_poc = s->current_picture_ptr->poc;
+ for (i = 0; i < 2; i++) {
+ h->luma_weight_flag[i] = 0;
+ h->chroma_weight_flag[i] = 0;
+ }
+
if( h->ref_count[0] == 1 && h->ref_count[1] == 1
&& h->ref_list[0][0].poc + h->ref_list[1][0].poc == 2*cur_poc){
h->use_weight= 0;
if(h->s.current_picture_ptr)
h->s.current_picture_ptr->reference= 0;
h->s.first_field= 0;
+ reset_sei(h);
ff_mpeg_flush(avctx);
}
MpegEncContext * const s = &h->s;
int i, j;
int current_ref_assigned=0;
- Picture *pic;
+ Picture *av_uninit(pic);
if((s->avctx->debug&FF_DEBUG_MMCO) && mmco_count==0)
av_log(h->s.avctx, AV_LOG_DEBUG, "no mmco here\n");
for(i=0; i<mmco_count; i++){
- int structure, frame_num;
+ int structure, av_uninit(frame_num);
if(s->avctx->debug&FF_DEBUG_MMCO)
av_log(h->s.avctx, AV_LOG_DEBUG, "mmco:%d %d %d\n", h->mmco[i].opcode, h->mmco[i].short_pic_num, h->mmco[i].long_arg);
}else{
if(get_bits1(gb)){ // adaptive_ref_pic_marking_mode_flag
for(i= 0; i<MAX_MMCO_COUNT; i++) {
- MMCOOpcode opcode= get_ue_golomb(gb);
+ MMCOOpcode opcode= get_ue_golomb_31(gb);
h->mmco[i].opcode= opcode;
if(opcode==MMCO_SHORT2UNUSED || opcode==MMCO_SHORT2LONG){
}*/
}
if(opcode==MMCO_SHORT2LONG || opcode==MMCO_LONG2UNUSED || opcode==MMCO_LONG || opcode==MMCO_SET_MAX_LONG){
- unsigned int long_arg= get_ue_golomb(gb);
+ unsigned int long_arg= get_ue_golomb_31(gb);
if(long_arg >= 32 || (long_arg >= 16 && !(opcode == MMCO_LONG2UNUSED && FIELD_PICTURE))){
av_log(h->s.avctx, AV_LOG_ERROR, "illegal long ref in memory management control operation %d\n", opcode);
return -1;
}
}
if(s->dsp.h264_idct8_add == ff_h264_idct8_add_c){
- memcpy(h->zigzag_scan8x8, zigzag_scan8x8, 64*sizeof(uint8_t));
+ memcpy(h->zigzag_scan8x8, ff_zigzag_direct, 64*sizeof(uint8_t));
memcpy(h->zigzag_scan8x8_cavlc, zigzag_scan8x8_cavlc, 64*sizeof(uint8_t));
memcpy(h->field_scan8x8, field_scan8x8, 64*sizeof(uint8_t));
memcpy(h->field_scan8x8_cavlc, field_scan8x8_cavlc, 64*sizeof(uint8_t));
}else{
for(i=0; i<64; i++){
#define T(x) (x>>3) | ((x&7)<<3)
- h->zigzag_scan8x8[i] = T(zigzag_scan8x8[i]);
+ h->zigzag_scan8x8[i] = T(ff_zigzag_direct[i]);
h->zigzag_scan8x8_cavlc[i] = T(zigzag_scan8x8_cavlc[i]);
h->field_scan8x8[i] = T(field_scan8x8[i]);
h->field_scan8x8_cavlc[i] = T(field_scan8x8_cavlc[i]);
}
if(h->sps.transform_bypass){ //FIXME same ugly
h->zigzag_scan_q0 = zigzag_scan;
- h->zigzag_scan8x8_q0 = zigzag_scan8x8;
+ h->zigzag_scan8x8_q0 = ff_zigzag_direct;
h->zigzag_scan8x8_cavlc_q0 = zigzag_scan8x8_cavlc;
h->field_scan_q0 = field_scan;
h->field_scan8x8_q0 = field_scan8x8;
s->current_picture_ptr= NULL;
}
- slice_type= get_ue_golomb(&s->gb);
+ slice_type= get_ue_golomb_31(&s->gb);
if(slice_type > 9){
av_log(h->s.avctx, AV_LOG_ERROR, "slice type too large (%d) at %d %d\n", h->slice_type, s->mb_x, s->mb_y);
return -1;
return -1;
}
if(!h0->pps_buffers[pps_id]) {
- av_log(h->s.avctx, AV_LOG_ERROR, "non-existing PPS referenced\n");
+ av_log(h->s.avctx, AV_LOG_ERROR, "non-existing PPS %u referenced\n", pps_id);
return -1;
}
h->pps= *h0->pps_buffers[pps_id];
if(!h0->sps_buffers[h->pps.sps_id]) {
- av_log(h->s.avctx, AV_LOG_ERROR, "non-existing SPS referenced\n");
+ av_log(h->s.avctx, AV_LOG_ERROR, "non-existing SPS %u referenced\n", h->pps.sps_id);
return -1;
}
h->sps = *h0->sps_buffers[h->pps.sps_id];
s->avctx->sample_aspect_ratio.den = 1;
if(h->sps.timing_info_present_flag){
- s->avctx->time_base= (AVRational){h->sps.num_units_in_tick * 2, h->sps.time_scale};
+ s->avctx->time_base= (AVRational){h->sps.num_units_in_tick, h->sps.time_scale};
if(h->x264_build > 0 && h->x264_build < 44)
s->avctx->time_base.den *= 2;
av_reduce(&s->avctx->time_base.num, &s->avctx->time_base.den,
while(h->frame_num != h->prev_frame_num &&
h->frame_num != (h->prev_frame_num+1)%(1<<h->sps.log2_max_frame_num)){
av_log(NULL, AV_LOG_DEBUG, "Frame num gap %d %d\n", h->frame_num, h->prev_frame_num);
- frame_start(h);
+ if (frame_start(h) < 0)
+ return -1;
h->prev_frame_num++;
h->prev_frame_num %= 1<<h->sps.log2_max_frame_num;
s->current_picture_ptr->frame_num= h->prev_frame_num;
pred_weight_table(h);
else if(h->pps.weighted_bipred_idc==2 && h->slice_type_nos== FF_B_TYPE)
implicit_weight_table(h);
- else
+ else {
h->use_weight = 0;
+ for (i = 0; i < 2; i++) {
+ h->luma_weight_flag[i] = 0;
+ h->chroma_weight_flag[i] = 0;
+ }
+ }
if(h->nal_ref_idc)
decode_ref_pic_marking(h0, &s->gb);
direct_ref_list_init(h);
if( h->slice_type_nos != FF_I_TYPE && h->pps.cabac ){
- tmp = get_ue_golomb(&s->gb);
+ tmp = get_ue_golomb_31(&s->gb);
if(tmp > 2){
av_log(s->avctx, AV_LOG_ERROR, "cabac_init_idc overflow\n");
return -1;
h->slice_alpha_c0_offset = 0;
h->slice_beta_offset = 0;
if( h->pps.deblocking_filter_parameters_present ) {
- tmp= get_ue_golomb(&s->gb);
+ tmp= get_ue_golomb_31(&s->gb);
if(tmp > 2){
av_log(s->avctx, AV_LOG_ERROR, "deblocking_filter_idc %u out of range\n", tmp);
return -1;
h->emu_edge_width= (s->flags&CODEC_FLAG_EMU_EDGE) ? 0 : 16;
h->emu_edge_height= (FRAME_MBAFF || FIELD_PICTURE) ? 0 : h->emu_edge_width;
+ s->avctx->refs= h->sps.ref_frame_count;
+
if(s->avctx->debug&FF_DEBUG_PICT_INFO){
av_log(h->s.avctx, AV_LOG_DEBUG, "slice:%d %s mb:%d %c%s%s pps:%u frame:%d poc:%d/%d ref:%d/%d qp:%d loop:%d:%d:%d weight:%d%s %s\n",
h->slice_num,
}
static inline int get_dct8x8_allowed(H264Context *h){
- int i;
- for(i=0; i<4; i++){
- if(!IS_SUB_8X8(h->sub_mb_type[i])
- || (!h->sps.direct_8x8_inference_flag && IS_DIRECT(h->sub_mb_type[i])))
- return 0;
- }
- return 1;
+ if(h->sps.direct_8x8_inference_flag)
+ return !(*(uint64_t*)h->sub_mb_type & ((MB_TYPE_16x8|MB_TYPE_8x16|MB_TYPE_8x8 )*0x0001000100010001ULL));
+ else
+ return !(*(uint64_t*)h->sub_mb_type & ((MB_TYPE_16x8|MB_TYPE_8x16|MB_TYPE_8x8|MB_TYPE_DIRECT2)*0x0001000100010001ULL));
}
/**
level[2] = 1-((i&1)<<1);
if(trailing_ones<total_coeff) {
- int level_code, mask;
+ int mask, prefix;
int suffix_length = total_coeff > 10 && trailing_ones < 3;
- int prefix= get_level_prefix(gb);
+ int bitsi= show_bits(gb, LEVEL_TAB_BITS);
+ int level_code= cavlc_level_tab[suffix_length][bitsi][0];
+
+ skip_bits(gb, cavlc_level_tab[suffix_length][bitsi][1]);
+ if(level_code >= 100){
+ prefix= level_code - 100;
+ if(prefix == LEVEL_TAB_BITS)
+ prefix += get_level_prefix(gb);
+
+ //first coefficient has suffix_length equal to 0 or 1
+ if(prefix<14){ //FIXME try to build a large unified VLC table for all this
+ if(suffix_length)
+ level_code= (prefix<<1) + get_bits1(gb); //part
+ else
+ level_code= prefix; //part
+ }else if(prefix==14){
+ if(suffix_length)
+ level_code= (prefix<<1) + get_bits1(gb); //part
+ else
+ level_code= prefix + get_bits(gb, 4); //part
+ }else{
+ level_code= 30 + get_bits(gb, prefix-3); //part
+ if(prefix>=16)
+ level_code += (1<<(prefix-3))-4096;
+ }
- //first coefficient has suffix_length equal to 0 or 1
- if(prefix<14){ //FIXME try to build a large unified VLC table for all this
- if(suffix_length)
- level_code= (prefix<<suffix_length) + get_bits(gb, suffix_length); //part
- else
- level_code= (prefix<<suffix_length); //part
- }else if(prefix==14){
- if(suffix_length)
- level_code= (prefix<<suffix_length) + get_bits(gb, suffix_length); //part
- else
- level_code= prefix + get_bits(gb, 4); //part
- }else{
- level_code= (15<<suffix_length) + get_bits(gb, prefix-3); //part
- if(suffix_length==0) level_code+=15; //FIXME doesn't make (much)sense
- if(prefix>=16)
- level_code += (1<<(prefix-3))-4096;
- }
+ if(trailing_ones < 3) level_code += 2;
- if(trailing_ones < 3) level_code += 2;
+ suffix_length = 2;
+ mask= -(level_code&1);
+ level[trailing_ones]= (((2+level_code)>>1) ^ mask) - mask;
+ }else{
+ if(trailing_ones < 3) level_code += (level_code>>31)|1;
- suffix_length = 1;
- if(level_code > 5)
- suffix_length++;
- mask= -(level_code&1);
- level[trailing_ones]= (((2+level_code)>>1) ^ mask) - mask;
+ suffix_length = 1;
+ if(level_code + 3U > 6U)
+ suffix_length++;
+ level[trailing_ones]= level_code;
+ }
//remaining coefficients have suffix_length > 0
for(i=trailing_ones+1;i<total_coeff;i++) {
- static const int suffix_limit[7] = {0,5,11,23,47,95,INT_MAX };
- prefix = get_level_prefix(gb);
- if(prefix<15){
- level_code = (prefix<<suffix_length) + get_bits(gb, suffix_length);
- }else{
- level_code = (15<<suffix_length) + get_bits(gb, prefix-3);
- if(prefix>=16)
- level_code += (1<<(prefix-3))-4096;
+ static const unsigned int suffix_limit[7] = {0,3,6,12,24,48,INT_MAX };
+ int bitsi= show_bits(gb, LEVEL_TAB_BITS);
+ level_code= cavlc_level_tab[suffix_length][bitsi][0];
+
+ skip_bits(gb, cavlc_level_tab[suffix_length][bitsi][1]);
+ if(level_code >= 100){
+ prefix= level_code - 100;
+ if(prefix == LEVEL_TAB_BITS){
+ prefix += get_level_prefix(gb);
+ }
+ if(prefix<15){
+ level_code = (prefix<<suffix_length) + get_bits(gb, suffix_length);
+ }else{
+ level_code = (15<<suffix_length) + get_bits(gb, prefix-3);
+ if(prefix>=16)
+ level_code += (1<<(prefix-3))-4096;
+ }
+ mask= -(level_code&1);
+ level_code= (((2+level_code)>>1) ^ mask) - mask;
}
- mask= -(level_code&1);
- level[i]= (((2+level_code)>>1) ^ mask) - mask;
- if(level_code > suffix_limit[suffix_length])
+ level[i]= level_code;
+
+ if(suffix_limit[suffix_length] + level_code > 2U*suffix_limit[suffix_length])
suffix_length++;
}
}
mb_xy = h->mb_xy = s->mb_x + s->mb_y*s->mb_stride;
- s->dsp.clear_blocks(h->mb); //FIXME avoid if already clear (move after skip handlong?
-
tprintf(s->avctx, "pic:%d mb:%d/%d\n", h->frame_num, s->mb_x, s->mb_y);
cbp = 0; /* avoid warning. FIXME: find a solution without slowing
down the code */
return -1;
}
if(CHROMA){
- pred_mode= check_intra_pred_mode(h, get_ue_golomb(&s->gb));
+ pred_mode= check_intra_pred_mode(h, get_ue_golomb_31(&s->gb));
if(pred_mode < 0)
return -1;
h->chroma_pred_mode= pred_mode;
if(h->slice_type_nos == FF_B_TYPE){
for(i=0; i<4; i++){
- h->sub_mb_type[i]= get_ue_golomb(&s->gb);
+ h->sub_mb_type[i]= get_ue_golomb_31(&s->gb);
if(h->sub_mb_type[i] >=13){
av_log(h->s.avctx, AV_LOG_ERROR, "B sub_mb_type %u out of range at %d %d\n", h->sub_mb_type[i], s->mb_x, s->mb_y);
return -1;
}else{
assert(h->slice_type_nos == FF_P_TYPE); //FIXME SP correct ?
for(i=0; i<4; i++){
- h->sub_mb_type[i]= get_ue_golomb(&s->gb);
+ h->sub_mb_type[i]= get_ue_golomb_31(&s->gb);
if(h->sub_mb_type[i] >=4){
av_log(h->s.avctx, AV_LOG_ERROR, "P sub_mb_type %u out of range at %d %d\n", h->sub_mb_type[i], s->mb_x, s->mb_y);
return -1;
for(i=0; i<4; i++){
if(IS_DIRECT(h->sub_mb_type[i])) continue;
if(IS_DIR(h->sub_mb_type[i], 0, list)){
- unsigned int tmp = get_te0_golomb(&s->gb, ref_count); //FIXME init to 0 before and skip?
- if(tmp>=ref_count){
- av_log(h->s.avctx, AV_LOG_ERROR, "ref %u overflow\n", tmp);
- return -1;
+ unsigned int tmp;
+ if(ref_count == 1){
+ tmp= 0;
+ }else if(ref_count == 2){
+ tmp= get_bits1(&s->gb)^1;
+ }else{
+ tmp= get_ue_golomb_31(&s->gb);
+ if(tmp>=ref_count){
+ av_log(h->s.avctx, AV_LOG_ERROR, "ref %u overflow\n", tmp);
+ return -1;
+ }
}
ref[list][i]= tmp;
}else{
for(list=0; list<h->list_count; list++){
unsigned int val;
if(IS_DIR(mb_type, 0, list)){
- val= get_te0_golomb(&s->gb, h->ref_count[list]);
- if(val >= h->ref_count[list]){
- av_log(h->s.avctx, AV_LOG_ERROR, "ref %u overflow\n", val);
- return -1;
+ if(h->ref_count[list]==1){
+ val= 0;
+ }else if(h->ref_count[list]==2){
+ val= get_bits1(&s->gb)^1;
+ }else{
+ val= get_ue_golomb_31(&s->gb);
+ if(val >= h->ref_count[list]){
+ av_log(h->s.avctx, AV_LOG_ERROR, "ref %u overflow\n", val);
+ return -1;
+ }
}
}else
val= LIST_NOT_USED&0xFF;
for(i=0; i<2; i++){
unsigned int val;
if(IS_DIR(mb_type, i, list)){
- val= get_te0_golomb(&s->gb, h->ref_count[list]);
- if(val >= h->ref_count[list]){
- av_log(h->s.avctx, AV_LOG_ERROR, "ref %u overflow\n", val);
- return -1;
+ if(h->ref_count[list] == 1){
+ val= 0;
+ }else if(h->ref_count[list] == 2){
+ val= get_bits1(&s->gb)^1;
+ }else{
+ val= get_ue_golomb_31(&s->gb);
+ if(val >= h->ref_count[list]){
+ av_log(h->s.avctx, AV_LOG_ERROR, "ref %u overflow\n", val);
+ return -1;
+ }
}
}else
val= LIST_NOT_USED&0xFF;
for(i=0; i<2; i++){
unsigned int val;
if(IS_DIR(mb_type, i, list)){ //FIXME optimize
- val= get_te0_golomb(&s->gb, h->ref_count[list]);
- if(val >= h->ref_count[list]){
- av_log(h->s.avctx, AV_LOG_ERROR, "ref %u overflow\n", val);
- return -1;
+ if(h->ref_count[list]==1){
+ val= 0;
+ }else if(h->ref_count[list]==2){
+ val= get_bits1(&s->gb)^1;
+ }else{
+ val= get_ue_golomb_31(&s->gb);
+ if(val >= h->ref_count[list]){
+ av_log(h->s.avctx, AV_LOG_ERROR, "ref %u overflow\n", val);
+ return -1;
+ }
}
}else
val= LIST_NOT_USED&0xFF;
return mb_type;
}
-static int decode_cabac_mb_type( H264Context *h ) {
+static int decode_cabac_mb_type_b( H264Context *h ) {
MpegEncContext * const s = &h->s;
- if( h->slice_type_nos == FF_I_TYPE ) {
- return decode_cabac_intra_mb_type(h, 3, 1);
- } else if( h->slice_type_nos == FF_P_TYPE ) {
- if( get_cabac_noinline( &h->cabac, &h->cabac_state[14] ) == 0 ) {
- /* P-type */
- if( get_cabac_noinline( &h->cabac, &h->cabac_state[15] ) == 0 ) {
- /* P_L0_D16x16, P_8x8 */
- return 3 * get_cabac_noinline( &h->cabac, &h->cabac_state[16] );
- } else {
- /* P_L0_D8x16, P_L0_D16x8 */
- return 2 - get_cabac_noinline( &h->cabac, &h->cabac_state[17] );
- }
- } else {
- return decode_cabac_intra_mb_type(h, 17, 0) + 5;
- }
- } else {
const int mba_xy = h->left_mb_xy[0];
const int mbb_xy = h->top_mb_xy;
int ctx = 0;
bits= ( bits<<1 ) | get_cabac_noinline( &h->cabac, &h->cabac_state[27+5] );
return bits - 4; /* B_L0_Bi_* through B_Bi_Bi_* */
- }
}
static int decode_cabac_mb_skip( H264Context *h, int mb_x, int mb_y ) {
return 1 + get_cabac_noinline( &h->cabac, &h->cabac_state[77 + ctx] );
}
static int decode_cabac_mb_dqp( H264Context *h) {
- int ctx = 0;
+ int ctx= h->last_qscale_diff != 0;
int val = 0;
- if( h->last_qscale_diff != 0 )
- ctx++;
-
while( get_cabac_noinline( &h->cabac, &h->cabac_state[60 + ctx] ) ) {
- if( ctx < 2 )
- ctx = 2;
- else
- ctx = 3;
+ ctx= 2+(ctx>>1);
val++;
if(val > 102) //prevent infinite loop
return INT_MIN;
}
if( val&0x01 )
- return (val + 1)/2;
+ return (val + 1)>>1 ;
else
- return -(val + 1)/2;
+ return -((val + 1)>>1);
}
static int decode_cabac_p_mb_sub_type( H264Context *h ) {
if( get_cabac( &h->cabac, &h->cabac_state[21] ) )
while( get_cabac( &h->cabac, &h->cabac_state[54+ctx] ) ) {
ref++;
- if( ctx < 4 )
- ctx = 4;
- else
- ctx = 5;
+ ctx = (ctx>>2)+4;
if(ref >= 32 /*h->ref_list[list]*/){
- av_log(h->s.avctx, AV_LOG_ERROR, "overflow in decode_cabac_mb_ref\n");
- return 0; //FIXME we should return -1 and check the return everywhere
+ return -1;
}
}
return ref;
int amvd = abs( h->mvd_cache[list][scan8[n] - 1][l] ) +
abs( h->mvd_cache[list][scan8[n] - 8][l] );
int ctxbase = (l == 0) ? 40 : 47;
- int ctx, mvd;
-
- if( amvd < 3 )
- ctx = 0;
- else if( amvd > 32 )
- ctx = 2;
- else
- ctx = 1;
+ int mvd;
+ int ctx = (amvd>2) + (amvd>32);
if(!get_cabac(&h->cabac, &h->cabac_state[ctxbase+ctx]))
return 0;
nzb = (h-> top_cbp>>(6+idx))&0x01;
}
} else {
- if( cat == 4 ) {
- nza = h->non_zero_count_cache[scan8[16+idx] - 1];
- nzb = h->non_zero_count_cache[scan8[16+idx] - 8];
- } else {
- assert(cat == 1 || cat == 2);
- nza = h->non_zero_count_cache[scan8[idx] - 1];
- nzb = h->non_zero_count_cache[scan8[idx] - 8];
- }
+ assert(cat == 1 || cat == 2 || cat == 4);
+ nza = h->non_zero_count_cache[scan8[idx] - 1];
+ nzb = h->non_zero_count_cache[scan8[idx] - 8];
}
if( nza > 0 )
uint8_t *last_coeff_ctx_base;
uint8_t *abs_level_m1_ctx_base;
-#ifndef ARCH_X86
+#if !ARCH_X86
#define CABAC_ON_STACK
#endif
#ifdef CABAC_ON_STACK
* 1-> AC 16x16 n = luma4x4idx
* 2-> Luma4x4 n = luma4x4idx
* 3-> DC Chroma n = iCbCr
- * 4-> AC Chroma n = 4 * iCbCr + chroma4x4idx
+ * 4-> AC Chroma n = 16 + 4 * iCbCr + chroma4x4idx
* 5-> Luma8x8 n = 4 * luma8x8idx
*/
/* read coded block flag */
if( is_dc || cat != 5 ) {
if( get_cabac( CC, &h->cabac_state[85 + get_cabac_cbf_ctx( h, cat, n, is_dc ) ] ) == 0 ) {
- if( !is_dc ) {
- if( cat == 4 )
- h->non_zero_count_cache[scan8[16+n]] = 0;
- else
- h->non_zero_count_cache[scan8[n]] = 0;
- }
+ if( !is_dc )
+ h->non_zero_count_cache[scan8[n]] = 0;
#ifdef CABAC_ON_STACK
h->cabac.range = cc.range ;
index[coeff_count++] = last;\
}
const uint8_t *sig_off = significant_coeff_flag_offset_8x8[MB_FIELD];
-#if defined(ARCH_X86) && defined(HAVE_7REGS) && defined(HAVE_EBX_AVAILABLE) && !defined(BROKEN_RELOCATIONS)
+#if ARCH_X86 && HAVE_7REGS && HAVE_EBX_AVAILABLE && !defined(BROKEN_RELOCATIONS)
coeff_count= decode_significance_8x8_x86(CC, significant_coeff_ctx_base, index, sig_off);
} else {
coeff_count= decode_significance_x86(CC, max_coeff, significant_coeff_ctx_base, index);
} else {
if( cat == 5 )
fill_rectangle(&h->non_zero_count_cache[scan8[n]], 2, 2, 8, coeff_count, 1);
- else if( cat == 4 )
- h->non_zero_count_cache[scan8[16+n]] = coeff_count;
else {
- assert( cat == 1 || cat == 2 );
+ assert( cat == 1 || cat == 2 || cat == 4 );
h->non_zero_count_cache[scan8[n]] = coeff_count;
}
}
}
-#ifndef CONFIG_SMALL
+#if !CONFIG_SMALL
static void decode_cabac_residual_dc( H264Context *h, DCTELEM *block, int cat, int n, const uint8_t *scantable, const uint32_t *qmul, int max_coeff ) {
decode_cabac_residual_internal(h, block, cat, n, scantable, qmul, max_coeff, 1);
}
#endif
static void decode_cabac_residual( H264Context *h, DCTELEM *block, int cat, int n, const uint8_t *scantable, const uint32_t *qmul, int max_coeff ) {
-#ifdef CONFIG_SMALL
+#if CONFIG_SMALL
decode_cabac_residual_internal(h, block, cat, n, scantable, qmul, max_coeff, cat == 0 || cat == 3);
#else
if( cat == 0 || cat == 3 ) decode_cabac_residual_dc(h, block, cat, n, scantable, qmul, max_coeff);
if(FRAME_MBAFF){
const int pair_xy = s->mb_x + (s->mb_y & ~1)*s->mb_stride;
const int top_pair_xy = pair_xy - s->mb_stride;
- const int top_mb_frame_flag = !IS_INTERLACED(s->current_picture.mb_type[top_pair_xy]);
- const int left_mb_frame_flag = !IS_INTERLACED(s->current_picture.mb_type[pair_xy-1]);
- const int curr_mb_frame_flag = !MB_FIELD;
+ const int top_mb_field_flag = IS_INTERLACED(s->current_picture.mb_type[top_pair_xy]);
+ const int left_mb_field_flag = IS_INTERLACED(s->current_picture.mb_type[pair_xy-1]);
+ const int curr_mb_field_flag = MB_FIELD;
const int bottom = (s->mb_y & 1);
- if (bottom
- ? !curr_mb_frame_flag // bottom macroblock
- : (!curr_mb_frame_flag && !top_mb_frame_flag) // top macroblock
- ) {
+
+ if (curr_mb_field_flag && (bottom || top_mb_field_flag)){
h->top_mb_xy -= s->mb_stride;
}
- if (left_mb_frame_flag != curr_mb_frame_flag) {
+ if (!left_mb_field_flag == curr_mb_field_flag) {
h->left_mb_xy[0] = pair_xy - 1;
}
} else if (FIELD_PICTURE) {
mb_xy = h->mb_xy = s->mb_x + s->mb_y*s->mb_stride;
- s->dsp.clear_blocks(h->mb); //FIXME avoid if already clear (move after skip handlong?)
-
tprintf(s->avctx, "pic:%d mb:%d/%d\n", h->frame_num, s->mb_x, s->mb_y);
if( h->slice_type_nos != FF_I_TYPE ) {
int skip;
if( FRAME_MBAFF && (s->mb_y&1)==0 ){
s->current_picture.mb_type[mb_xy] = MB_TYPE_SKIP;
h->next_mb_skipped = decode_cabac_mb_skip( h, s->mb_x, s->mb_y+1 );
- if(h->next_mb_skipped)
- predict_field_decoding_flag(h);
- else
+ if(!h->next_mb_skipped)
h->mb_mbaff = h->mb_field_decoding_flag = decode_cabac_field_decoding_flag(h);
}
h->prev_mb_skipped = 0;
compute_mb_neighbors(h);
- mb_type = decode_cabac_mb_type( h );
- assert(mb_type >= 0);
if( h->slice_type_nos == FF_B_TYPE ) {
+ mb_type = decode_cabac_mb_type_b( h );
if( mb_type < 23 ){
partition_count= b_mb_type_info[mb_type].partition_count;
mb_type= b_mb_type_info[mb_type].type;
goto decode_intra_mb;
}
} else if( h->slice_type_nos == FF_P_TYPE ) {
- if( mb_type < 5) {
+ if( get_cabac_noinline( &h->cabac, &h->cabac_state[14] ) == 0 ) {
+ /* P-type */
+ if( get_cabac_noinline( &h->cabac, &h->cabac_state[15] ) == 0 ) {
+ /* P_L0_D16x16, P_8x8 */
+ mb_type= 3 * get_cabac_noinline( &h->cabac, &h->cabac_state[16] );
+ } else {
+ /* P_L0_D8x16, P_L0_D16x8 */
+ mb_type= 2 - get_cabac_noinline( &h->cabac, &h->cabac_state[17] );
+ }
partition_count= p_mb_type_info[mb_type].partition_count;
mb_type= p_mb_type_info[mb_type].type;
} else {
- mb_type -= 5;
+ mb_type= decode_cabac_intra_mb_type(h, 17, 0);
goto decode_intra_mb;
}
} else {
+ mb_type= decode_cabac_intra_mb_type(h, 3, 1);
if(h->slice_type == FF_SI_TYPE && mb_type)
mb_type--;
assert(h->slice_type_nos == FF_I_TYPE);
for( i = 0; i < 4; i++ ) {
if(IS_DIRECT(h->sub_mb_type[i])) continue;
if(IS_DIR(h->sub_mb_type[i], 0, list)){
- if( h->ref_count[list] > 1 )
+ if( h->ref_count[list] > 1 ){
ref[list][i] = decode_cabac_mb_ref( h, list, 4*i );
- else
+ if(ref[list][i] >= (unsigned)h->ref_count[list]){
+ av_log(s->avctx, AV_LOG_ERROR, "Reference %d >= %d\n", ref[list][i], h->ref_count[list]);
+ return -1;
+ }
+ }else
ref[list][i] = 0;
} else {
ref[list][i] = -1;
if(IS_16X16(mb_type)){
for(list=0; list<h->list_count; list++){
if(IS_DIR(mb_type, 0, list)){
- const int ref = h->ref_count[list] > 1 ? decode_cabac_mb_ref( h, list, 0 ) : 0;
+ int ref;
+ if(h->ref_count[list] > 1){
+ ref= decode_cabac_mb_ref(h, list, 0);
+ if(ref >= (unsigned)h->ref_count[list]){
+ av_log(s->avctx, AV_LOG_ERROR, "Reference %d >= %d\n", ref, h->ref_count[list]);
+ return -1;
+ }
+ }else
+ ref=0;
fill_rectangle(&h->ref_cache[list][ scan8[0] ], 4, 4, 8, ref, 1);
}else
fill_rectangle(&h->ref_cache[list][ scan8[0] ], 4, 4, 8, (uint8_t)LIST_NOT_USED, 1); //FIXME factorize and the other fill_rect below too
for(list=0; list<h->list_count; list++){
for(i=0; i<2; i++){
if(IS_DIR(mb_type, i, list)){
- const int ref= h->ref_count[list] > 1 ? decode_cabac_mb_ref( h, list, 8*i ) : 0;
+ int ref;
+ if(h->ref_count[list] > 1){
+ ref= decode_cabac_mb_ref( h, list, 8*i );
+ if(ref >= (unsigned)h->ref_count[list]){
+ av_log(s->avctx, AV_LOG_ERROR, "Reference %d >= %d\n", ref, h->ref_count[list]);
+ return -1;
+ }
+ }else
+ ref=0;
fill_rectangle(&h->ref_cache[list][ scan8[0] + 16*i ], 4, 2, 8, ref, 1);
}else
fill_rectangle(&h->ref_cache[list][ scan8[0] + 16*i ], 4, 2, 8, (LIST_NOT_USED&0xFF), 1);
for(list=0; list<h->list_count; list++){
for(i=0; i<2; i++){
if(IS_DIR(mb_type, i, list)){ //FIXME optimize
- const int ref= h->ref_count[list] > 1 ? decode_cabac_mb_ref( h, list, 4*i ) : 0;
+ int ref;
+ if(h->ref_count[list] > 1){
+ ref= decode_cabac_mb_ref( h, list, 4*i );
+ if(ref >= (unsigned)h->ref_count[list]){
+ av_log(s->avctx, AV_LOG_ERROR, "Reference %d >= %d\n", ref, h->ref_count[list]);
+ return -1;
+ }
+ }else
+ ref=0;
fill_rectangle(&h->ref_cache[list][ scan8[0] + 2*i ], 2, 4, 8, ref, 1);
}else
fill_rectangle(&h->ref_cache[list][ scan8[0] + 2*i ], 2, 4, 8, (LIST_NOT_USED&0xFF), 1);
for( i = 0; i < 4; i++ ) {
const int index = 16 + 4 * c + i;
//av_log( s->avctx, AV_LOG_ERROR, "INTRA C%d-AC %d\n",c, index - 16 );
- decode_cabac_residual(h, h->mb + 16*index, 4, index - 16, scan + 1, qmul, 15);
+ decode_cabac_residual(h, h->mb + 16*index, 4, index, scan + 1, qmul, 15);
}
}
} else {
static void filter_mb_edgev( H264Context *h, uint8_t *pix, int stride, int16_t bS[4], int qp ) {
- int i, d;
const int index_a = qp + h->slice_alpha_c0_offset;
const int alpha = (alpha_table+52)[index_a];
const int beta = (beta_table+52)[qp + h->slice_beta_offset];
if( bS[0] < 4 ) {
int8_t tc[4];
- for(i=0; i<4; i++)
- tc[i] = bS[i] ? (tc0_table+52)[index_a][bS[i] - 1] : -1;
+ tc[0] = (tc0_table+52)[index_a][bS[0]];
+ tc[1] = (tc0_table+52)[index_a][bS[1]];
+ tc[2] = (tc0_table+52)[index_a][bS[2]];
+ tc[3] = (tc0_table+52)[index_a][bS[3]];
h->s.dsp.h264_h_loop_filter_luma(pix, stride, alpha, beta, tc);
} else {
- /* 16px edge length, because bS=4 is triggered by being at
- * the edge of an intra MB, so all 4 bS are the same */
- for( d = 0; d < 16; d++ ) {
- const int p0 = pix[-1];
- const int p1 = pix[-2];
- const int p2 = pix[-3];
-
- const int q0 = pix[0];
- const int q1 = pix[1];
- const int q2 = pix[2];
-
- if( FFABS( p0 - q0 ) < alpha &&
- FFABS( p1 - p0 ) < beta &&
- FFABS( q1 - q0 ) < beta ) {
-
- if(FFABS( p0 - q0 ) < (( alpha >> 2 ) + 2 )){
- if( FFABS( p2 - p0 ) < beta)
- {
- const int p3 = pix[-4];
- /* p0', p1', p2' */
- pix[-1] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3;
- pix[-2] = ( p2 + p1 + p0 + q0 + 2 ) >> 2;
- pix[-3] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3;
- } else {
- /* p0' */
- pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
- }
- if( FFABS( q2 - q0 ) < beta)
- {
- const int q3 = pix[3];
- /* q0', q1', q2' */
- pix[0] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3;
- pix[1] = ( p0 + q0 + q1 + q2 + 2 ) >> 2;
- pix[2] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3;
- } else {
- /* q0' */
- pix[0] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
- }
- }else{
- /* p0', q0' */
- pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
- pix[ 0] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
- }
- tprintf(h->s.avctx, "filter_mb_edgev i:%d d:%d\n# bS:4 -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x]\n", i, d, p2, p1, p0, q0, q1, q2, pix[-2], pix[-1], pix[0], pix[1]);
- }
- pix += stride;
- }
+ h->s.dsp.h264_h_loop_filter_luma_intra(pix, stride, alpha, beta);
}
}
static void filter_mb_edgecv( H264Context *h, uint8_t *pix, int stride, int16_t bS[4], int qp ) {
- int i;
const int index_a = qp + h->slice_alpha_c0_offset;
const int alpha = (alpha_table+52)[index_a];
const int beta = (beta_table+52)[qp + h->slice_beta_offset];
if( bS[0] < 4 ) {
int8_t tc[4];
- for(i=0; i<4; i++)
- tc[i] = bS[i] ? (tc0_table+52)[index_a][bS[i] - 1] + 1 : 0;
+ tc[0] = (tc0_table+52)[index_a][bS[0]]+1;
+ tc[1] = (tc0_table+52)[index_a][bS[1]]+1;
+ tc[2] = (tc0_table+52)[index_a][bS[2]]+1;
+ tc[3] = (tc0_table+52)[index_a][bS[3]]+1;
h->s.dsp.h264_h_loop_filter_chroma(pix, stride, alpha, beta, tc);
} else {
h->s.dsp.h264_h_loop_filter_chroma_intra(pix, stride, alpha, beta);
beta = (beta_table+52)[qp[qp_index] + h->slice_beta_offset];
if( bS[bS_index] < 4 ) {
- const int tc0 = (tc0_table+52)[index_a][bS[bS_index] - 1];
+ const int tc0 = (tc0_table+52)[index_a][bS[bS_index]];
const int p0 = pix[-1];
const int p1 = pix[-2];
const int p2 = pix[-3];
beta = (beta_table+52)[qp[qp_index] + h->slice_beta_offset];
if( bS[bS_index] < 4 ) {
- const int tc = (tc0_table+52)[index_a][bS[bS_index] - 1] + 1;
+ const int tc = (tc0_table+52)[index_a][bS[bS_index]] + 1;
const int p0 = pix[-1];
const int p1 = pix[-2];
const int q0 = pix[0];
}
static void filter_mb_edgeh( H264Context *h, uint8_t *pix, int stride, int16_t bS[4], int qp ) {
- int i, d;
const int index_a = qp + h->slice_alpha_c0_offset;
const int alpha = (alpha_table+52)[index_a];
const int beta = (beta_table+52)[qp + h->slice_beta_offset];
- const int pix_next = stride;
if( bS[0] < 4 ) {
int8_t tc[4];
- for(i=0; i<4; i++)
- tc[i] = bS[i] ? (tc0_table+52)[index_a][bS[i] - 1] : -1;
+ tc[0] = (tc0_table+52)[index_a][bS[0]];
+ tc[1] = (tc0_table+52)[index_a][bS[1]];
+ tc[2] = (tc0_table+52)[index_a][bS[2]];
+ tc[3] = (tc0_table+52)[index_a][bS[3]];
h->s.dsp.h264_v_loop_filter_luma(pix, stride, alpha, beta, tc);
} else {
- /* 16px edge length, see filter_mb_edgev */
- for( d = 0; d < 16; d++ ) {
- const int p0 = pix[-1*pix_next];
- const int p1 = pix[-2*pix_next];
- const int p2 = pix[-3*pix_next];
- const int q0 = pix[0];
- const int q1 = pix[1*pix_next];
- const int q2 = pix[2*pix_next];
-
- if( FFABS( p0 - q0 ) < alpha &&
- FFABS( p1 - p0 ) < beta &&
- FFABS( q1 - q0 ) < beta ) {
-
- const int p3 = pix[-4*pix_next];
- const int q3 = pix[ 3*pix_next];
-
- if(FFABS( p0 - q0 ) < (( alpha >> 2 ) + 2 )){
- if( FFABS( p2 - p0 ) < beta) {
- /* p0', p1', p2' */
- pix[-1*pix_next] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3;
- pix[-2*pix_next] = ( p2 + p1 + p0 + q0 + 2 ) >> 2;
- pix[-3*pix_next] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3;
- } else {
- /* p0' */
- pix[-1*pix_next] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
- }
- if( FFABS( q2 - q0 ) < beta) {
- /* q0', q1', q2' */
- pix[0*pix_next] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3;
- pix[1*pix_next] = ( p0 + q0 + q1 + q2 + 2 ) >> 2;
- pix[2*pix_next] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3;
- } else {
- /* q0' */
- pix[0*pix_next] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
- }
- }else{
- /* p0', q0' */
- pix[-1*pix_next] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
- pix[ 0*pix_next] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
- }
- tprintf(h->s.avctx, "filter_mb_edgeh i:%d d:%d, qp:%d, indexA:%d, alpha:%d, beta:%d\n# bS:%d -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x]\n", i, d, qp, index_a, alpha, beta, bS[i], p2, p1, p0, q0, q1, q2, pix[-2*pix_next], pix[-pix_next], pix[0], pix[pix_next]);
- }
- pix++;
- }
+ h->s.dsp.h264_v_loop_filter_luma_intra(pix, stride, alpha, beta);
}
}
static void filter_mb_edgech( H264Context *h, uint8_t *pix, int stride, int16_t bS[4], int qp ) {
- int i;
const int index_a = qp + h->slice_alpha_c0_offset;
const int alpha = (alpha_table+52)[index_a];
const int beta = (beta_table+52)[qp + h->slice_beta_offset];
if( bS[0] < 4 ) {
int8_t tc[4];
- for(i=0; i<4; i++)
- tc[i] = bS[i] ? (tc0_table+52)[index_a][bS[i] - 1] + 1 : 0;
+ tc[0] = (tc0_table+52)[index_a][bS[0]]+1;
+ tc[1] = (tc0_table+52)[index_a][bS[1]]+1;
+ tc[2] = (tc0_table+52)[index_a][bS[2]]+1;
+ tc[3] = (tc0_table+52)[index_a][bS[3]]+1;
h->s.dsp.h264_v_loop_filter_chroma(pix, stride, alpha, beta, tc);
} else {
h->s.dsp.h264_v_loop_filter_chroma_intra(pix, stride, alpha, beta);
}
-static void av_always_inline filter_mb_dir(H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr, unsigned int linesize, unsigned int uvlinesize, int mb_xy, int mb_type, int mvy_limit, int first_vertical_edge_done, int dir) {
+static av_always_inline void filter_mb_dir(H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr, unsigned int linesize, unsigned int uvlinesize, int mb_xy, int mb_type, int mvy_limit, int first_vertical_edge_done, int dir) {
MpegEncContext * const s = &h->s;
int edge;
const int mbm_xy = dir == 0 ? mb_xy -1 : h->top_mb_xy;
const int mb_type = s->current_picture.mb_type[mb_xy];
const int mvy_limit = IS_INTERLACED(mb_type) ? 2 : 4;
int first_vertical_edge_done = 0;
- int dir;
+ av_unused int dir;
//for sufficiently low qp, filtering wouldn't do anything
//this is a conservative estimate: could also check beta_offset and more accurate chroma_qp
int qp = s->current_picture.qscale_table[mb_xy];
if(qp <= qp_thresh
&& (mb_x == 0 || ((qp + s->current_picture.qscale_table[mb_xy-1] + 1)>>1) <= qp_thresh)
- && (mb_y == 0 || ((qp + s->current_picture.qscale_table[h->top_mb_xy] + 1)>>1) <= qp_thresh)){
+ && (h->top_mb_xy < 0 || ((qp + s->current_picture.qscale_table[h->top_mb_xy] + 1)>>1) <= qp_thresh)){
return;
}
}
filter_mb_mbaff_edgecv( h, &img_cr[0], uvlinesize, bS, rqp );
}
-#ifdef CONFIG_SMALL
+#if CONFIG_SMALL
for( dir = 0; dir < 2; dir++ )
filter_mb_dir(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, mb_xy, mb_type, mvy_limit, dir ? 0 : first_vertical_edge_done, dir);
#else
s->mb_skip_run= -1;
h->is_complex = FRAME_MBAFF || s->picture_structure != PICT_FRAME || s->codec_id != CODEC_ID_H264 ||
- (ENABLE_GRAY && (s->flags&CODEC_FLAG_GRAY)) || (ENABLE_H264_ENCODER && s->encoding);
+ (CONFIG_GRAY && (s->flags&CODEC_FLAG_GRAY));
if( h->pps.cabac ) {
int i;
if( ret >= 0 && FRAME_MBAFF ) { //FIXME optimal? or let mb_decode decode 16x32 ?
s->mb_y++;
- if(ret>=0) ret = decode_mb_cabac(h);
+ ret = decode_mb_cabac(h);
if(ret>=0) hl_decode_mb(h);
s->mb_y--;
static int decode_picture_timing(H264Context *h){
MpegEncContext * const s = &h->s;
if(h->sps.nal_hrd_parameters_present_flag || h->sps.vcl_hrd_parameters_present_flag){
- skip_bits(&s->gb, h->sps.cpb_removal_delay_length); /* cpb_removal_delay */
- skip_bits(&s->gb, h->sps.dpb_output_delay_length); /* dpb_output_delay */
+ h->sei_cpb_removal_delay = get_bits(&s->gb, h->sps.cpb_removal_delay_length);
+ h->sei_dpb_output_delay = get_bits(&s->gb, h->sps.dpb_output_delay_length);
}
if(h->sps.pic_struct_present_flag){
unsigned int i, num_clock_ts;
return 0;
}
-static int decode_sei(H264Context *h){
+static int decode_recovery_point(H264Context *h){
+ MpegEncContext * const s = &h->s;
+
+ h->sei_recovery_frame_cnt = get_ue_golomb(&s->gb);
+ skip_bits(&s->gb, 4); /* 1b exact_match_flag, 1b broken_link_flag, 2b changing_slice_group_idc */
+
+ return 0;
+}
+
+static int decode_buffering_period(H264Context *h){
+ MpegEncContext * const s = &h->s;
+ unsigned int sps_id;
+ int sched_sel_idx;
+ SPS *sps;
+
+ sps_id = get_ue_golomb_31(&s->gb);
+ if(sps_id > 31 || !h->sps_buffers[sps_id]) {
+ av_log(h->s.avctx, AV_LOG_ERROR, "non-existing SPS %d referenced in buffering period\n", sps_id);
+ return -1;
+ }
+ sps = h->sps_buffers[sps_id];
+
+ // NOTE: This is really so duplicated in the standard... See H.264, D.1.1
+ if (sps->nal_hrd_parameters_present_flag) {
+ for (sched_sel_idx = 0; sched_sel_idx < sps->cpb_cnt; sched_sel_idx++) {
+ h->initial_cpb_removal_delay[sched_sel_idx] = get_bits(&s->gb, sps->initial_cpb_removal_delay_length);
+ skip_bits(&s->gb, sps->initial_cpb_removal_delay_length); // initial_cpb_removal_delay_offset
+ }
+ }
+ if (sps->vcl_hrd_parameters_present_flag) {
+ for (sched_sel_idx = 0; sched_sel_idx < sps->cpb_cnt; sched_sel_idx++) {
+ h->initial_cpb_removal_delay[sched_sel_idx] = get_bits(&s->gb, sps->initial_cpb_removal_delay_length);
+ skip_bits(&s->gb, sps->initial_cpb_removal_delay_length); // initial_cpb_removal_delay_offset
+ }
+ }
+
+ h->sei_buffering_period_present = 1;
+ return 0;
+}
+
+int ff_h264_decode_sei(H264Context *h){
MpegEncContext * const s = &h->s;
while(get_bits_count(&s->gb) + 16 < s->gb.size_in_bits){
}while(get_bits(&s->gb, 8) == 255);
switch(type){
- case 1: // Picture timing SEI
+ case SEI_TYPE_PIC_TIMING: // Picture timing SEI
if(decode_picture_timing(h) < 0)
return -1;
break;
- case 5:
+ case SEI_TYPE_USER_DATA_UNREGISTERED:
if(decode_unregistered_user_data(h, size) < 0)
return -1;
break;
+ case SEI_TYPE_RECOVERY_POINT:
+ if(decode_recovery_point(h) < 0)
+ return -1;
+ break;
+ case SEI_BUFFERING_PERIOD:
+ if(decode_buffering_period(h) < 0)
+ return -1;
+ break;
default:
skip_bits(&s->gb, 8*size);
}
static inline int decode_hrd_parameters(H264Context *h, SPS *sps){
MpegEncContext * const s = &h->s;
int cpb_count, i;
- cpb_count = get_ue_golomb(&s->gb) + 1;
+ cpb_count = get_ue_golomb_31(&s->gb) + 1;
if(cpb_count > 32U){
av_log(h->s.avctx, AV_LOG_ERROR, "cpb_count %d invalid\n", cpb_count);
get_ue_golomb(&s->gb); /* cpb_size_value_minus1 */
get_bits1(&s->gb); /* cbr_flag */
}
- get_bits(&s->gb, 5); /* initial_cpb_removal_delay_length_minus1 */
+ sps->initial_cpb_removal_delay_length = get_bits(&s->gb, 5) + 1;
sps->cpb_removal_delay_length = get_bits(&s->gb, 5) + 1;
sps->dpb_output_delay_length = get_bits(&s->gb, 5) + 1;
sps->time_offset_length = get_bits(&s->gb, 5);
+ sps->cpb_cnt = cpb_count;
return 0;
}
const uint8_t *jvt_list, const uint8_t *fallback_list){
MpegEncContext * const s = &h->s;
int i, last = 8, next = 8;
- const uint8_t *scan = size == 16 ? zigzag_scan : zigzag_scan8x8;
+ const uint8_t *scan = size == 16 ? zigzag_scan : ff_zigzag_direct;
if(!get_bits1(&s->gb)) /* matrix not written, we use the predicted one */
memcpy(factors, fallback_list, size*sizeof(uint8_t));
else
}
}
-static inline int decode_seq_parameter_set(H264Context *h){
+int ff_h264_decode_seq_parameter_set(H264Context *h){
MpegEncContext * const s = &h->s;
int profile_idc, level_idc;
unsigned int sps_id;
get_bits1(&s->gb); //constraint_set3_flag
get_bits(&s->gb, 4); // reserved
level_idc= get_bits(&s->gb, 8);
- sps_id= get_ue_golomb(&s->gb);
+ sps_id= get_ue_golomb_31(&s->gb);
if(sps_id >= MAX_SPS_COUNT) {
av_log(h->s.avctx, AV_LOG_ERROR, "sps_id (%d) out of range\n", sps_id);
sps->scaling_matrix_present = 0;
if(sps->profile_idc >= 100){ //high profile
- sps->chroma_format_idc= get_ue_golomb(&s->gb);
+ sps->chroma_format_idc= get_ue_golomb_31(&s->gb);
if(sps->chroma_format_idc == 3)
- get_bits1(&s->gb); //residual_color_transform_flag
- get_ue_golomb(&s->gb); //bit_depth_luma_minus8
- get_ue_golomb(&s->gb); //bit_depth_chroma_minus8
+ sps->residual_color_transform_flag = get_bits1(&s->gb);
+ sps->bit_depth_luma = get_ue_golomb(&s->gb) + 8;
+ sps->bit_depth_chroma = get_ue_golomb(&s->gb) + 8;
sps->transform_bypass = get_bits1(&s->gb);
decode_scaling_matrices(h, sps, NULL, 1, sps->scaling_matrix4, sps->scaling_matrix8);
}else{
}
sps->log2_max_frame_num= get_ue_golomb(&s->gb) + 4;
- sps->poc_type= get_ue_golomb(&s->gb);
+ sps->poc_type= get_ue_golomb_31(&s->gb);
if(sps->poc_type == 0){ //FIXME #define
sps->log2_max_poc_lsb= get_ue_golomb(&s->gb) + 4;
goto fail;
}
- sps->ref_frame_count= get_ue_golomb(&s->gb);
+ sps->ref_frame_count= get_ue_golomb_31(&s->gb);
if(sps->ref_frame_count > MAX_PICTURE_COUNT-2 || sps->ref_frame_count >= 32U){
av_log(h->s.avctx, AV_LOG_ERROR, "too many reference frames\n");
goto fail;
((const char*[]){"Gray","420","422","444"})[sps->chroma_format_idc]
);
}
+
av_free(h->sps_buffers[sps_id]);
h->sps_buffers[sps_id]= sps;
+ h->sps = *sps;
return 0;
fail:
av_free(sps);
pps->chroma_qp_table[t][i] = chroma_qp[av_clip(i + index, 0, 51)];
}
-static inline int decode_picture_parameter_set(H264Context *h, int bit_length){
+int ff_h264_decode_picture_parameter_set(H264Context *h, int bit_length){
MpegEncContext * const s = &h->s;
unsigned int pps_id= get_ue_golomb(&s->gb);
PPS *pps;
pps= av_mallocz(sizeof(PPS));
if(pps == NULL)
return -1;
- pps->sps_id= get_ue_golomb(&s->gb);
+ pps->sps_id= get_ue_golomb_31(&s->gb);
if((unsigned)pps->sps_id>=MAX_SPS_COUNT || h->sps_buffers[pps->sps_id] == NULL){
av_log(h->s.avctx, AV_LOG_ERROR, "sps_id out of range\n");
goto fail;
H264Context *hx;
int i;
+ if (s->avctx->hwaccel)
+ return;
+ if(s->avctx->codec->capabilities&CODEC_CAP_HWACCEL_VDPAU)
+ return;
if(context_count == 1) {
decode_slice(avctx, &h);
} else {
h->current_slice = 0;
if (!s->first_field)
s->current_picture_ptr= NULL;
+ reset_sei(h);
}
for(;;){
hx = h->thread_context[context_count];
- ptr= decode_nal(hx, buf + buf_index, &dst_length, &consumed, h->is_avc ? nalsize : buf_size - buf_index);
+ ptr= ff_h264_decode_nal(hx, buf + buf_index, &dst_length, &consumed, h->is_avc ? nalsize : buf_size - buf_index);
if (ptr==NULL || dst_length < 0){
return -1;
}
while(ptr[dst_length - 1] == 0 && dst_length > 0)
dst_length--;
- bit_length= !dst_length ? 0 : (8*dst_length - decode_rbsp_trailing(h, ptr + dst_length - 1));
+ bit_length= !dst_length ? 0 : (8*dst_length - ff_h264_decode_rbsp_trailing(h, ptr + dst_length - 1));
if(s->avctx->debug&FF_DEBUG_STARTCODE){
av_log(h->s.avctx, AV_LOG_DEBUG, "NAL %d at %d/%d length %d\n", hx->nal_unit_type, buf_index, buf_size, dst_length);
}
if (h->is_avc && (nalsize != consumed)){
- av_log(h->s.avctx, AV_LOG_ERROR, "AVC: Consumed only %d bytes instead of %d\n", consumed, nalsize);
+ int i, debug_level = AV_LOG_DEBUG;
+ for (i = consumed; i < nalsize; i++)
+ if (buf[buf_index+i])
+ debug_level = AV_LOG_ERROR;
+ av_log(h->s.avctx, debug_level, "AVC: Consumed only %d bytes instead of %d\n", consumed, nalsize);
consumed= nalsize;
}
if((err = decode_slice_header(hx, h)))
break;
- s->current_picture_ptr->key_frame|= (hx->nal_unit_type == NAL_IDR_SLICE);
+ if (s->avctx->hwaccel && h->current_slice == 1) {
+ if (s->avctx->hwaccel->start_frame(s->avctx, NULL, 0) < 0)
+ return -1;
+ }
+
+ s->current_picture_ptr->key_frame |=
+ (hx->nal_unit_type == NAL_IDR_SLICE) ||
+ (h->sei_recovery_frame_cnt >= 0);
if(hx->redundant_pic_count==0 && hx->s.hurry_up < 5
&& (avctx->skip_frame < AVDISCARD_NONREF || hx->nal_ref_idc)
&& (avctx->skip_frame < AVDISCARD_BIDIR || hx->slice_type_nos!=FF_B_TYPE)
&& (avctx->skip_frame < AVDISCARD_NONKEY || hx->slice_type_nos==FF_I_TYPE)
- && avctx->skip_frame < AVDISCARD_ALL)
- context_count++;
+ && avctx->skip_frame < AVDISCARD_ALL){
+ if(avctx->hwaccel) {
+ if (avctx->hwaccel->decode_slice(avctx, &buf[buf_index - consumed], consumed) < 0)
+ return -1;
+ }else
+ if(CONFIG_H264_VDPAU_DECODER && s->avctx->codec->capabilities&CODEC_CAP_HWACCEL_VDPAU){
+ static const uint8_t start_code[] = {0x00, 0x00, 0x01};
+ ff_vdpau_add_data_chunk(s, start_code, sizeof(start_code));
+ ff_vdpau_add_data_chunk(s, &buf[buf_index - consumed], consumed );
+ }else
+ context_count++;
+ }
break;
case NAL_DPA:
init_get_bits(&hx->s.gb, ptr, bit_length);
break;
case NAL_SEI:
init_get_bits(&s->gb, ptr, bit_length);
- decode_sei(h);
+ ff_h264_decode_sei(h);
break;
case NAL_SPS:
init_get_bits(&s->gb, ptr, bit_length);
- decode_seq_parameter_set(h);
+ ff_h264_decode_seq_parameter_set(h);
if(s->flags& CODEC_FLAG_LOW_DELAY)
s->low_delay=1;
case NAL_PPS:
init_get_bits(&s->gb, ptr, bit_length);
- decode_picture_parameter_set(h, bit_length);
+ ff_h264_decode_picture_parameter_set(h, bit_length);
break;
case NAL_AUD:
s->current_picture_ptr->qscale_type= FF_QSCALE_TYPE_H264;
s->current_picture_ptr->pict_type= s->pict_type;
+ if (CONFIG_H264_VDPAU_DECODER && s->avctx->codec->capabilities&CODEC_CAP_HWACCEL_VDPAU)
+ ff_vdpau_h264_set_reference_frames(s);
+
if(!s->dropable) {
execute_ref_pic_marking(h, h->mmco, h->mmco_index);
h->prev_poc_msb= h->poc_msb;
h->prev_frame_num_offset= h->frame_num_offset;
h->prev_frame_num= h->frame_num;
+ if (avctx->hwaccel) {
+ if (avctx->hwaccel->end_frame(avctx) < 0)
+ av_log(avctx, AV_LOG_ERROR, "hardware accelerator failed to decode picture\n");
+ }
+
+ if (CONFIG_H264_VDPAU_DECODER && s->avctx->codec->capabilities&CODEC_CAP_HWACCEL_VDPAU)
+ ff_vdpau_h264_picture_complete(s);
+
/*
* FIXME: Error handling code does not seem to support interlaced
* when slices span multiple rows
return -1;
}
- out= decode_nal(&h, nal, &out_length, &consumed, nal_length);
+ out= ff_h264_decode_nal(&h, nal, &out_length, &consumed, nal_length);
STOP_TIMER("NAL")
/*CODEC_CAP_DRAW_HORIZ_BAND |*/ CODEC_CAP_DR1 | CODEC_CAP_DELAY,
.flush= flush_dpb,
.long_name = NULL_IF_CONFIG_SMALL("H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
+ .pix_fmts= ff_hwaccel_pixfmt_list_420,
};
+#if CONFIG_H264_VDPAU_DECODER
+AVCodec h264_vdpau_decoder = {
+ "h264_vdpau",
+ CODEC_TYPE_VIDEO,
+ CODEC_ID_H264,
+ sizeof(H264Context),
+ decode_init,
+ NULL,
+ decode_end,
+ decode_frame,
+ CODEC_CAP_DR1 | CODEC_CAP_DELAY | CODEC_CAP_HWACCEL_VDPAU,
+ .flush= flush_dpb,
+ .long_name = NULL_IF_CONFIG_SMALL("H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10 (VDPAU acceleration)"),
+};
+#endif
+
+#if CONFIG_SVQ3_DECODER
#include "svq3.c"
+#endif
projects / ffmpeg / blobdiff