@@ -366,6 +366,8 @@ typedef struct H264SliceContext { | |||||
*/ | */ | ||||
DECLARE_ALIGNED(16, int16_t, mv_cache)[2][5 * 8][2]; | DECLARE_ALIGNED(16, int16_t, mv_cache)[2][5 * 8][2]; | ||||
DECLARE_ALIGNED(8, int8_t, ref_cache)[2][5 * 8]; | DECLARE_ALIGNED(8, int8_t, ref_cache)[2][5 * 8]; | ||||
DECLARE_ALIGNED(8, uint16_t, sub_mb_type)[4]; | |||||
} H264SliceContext; | } H264SliceContext; | ||||
/** | /** | ||||
@@ -438,9 +440,6 @@ typedef struct H264Context { | |||||
int picture_structure; | int picture_structure; | ||||
int first_field; | int first_field; | ||||
DECLARE_ALIGNED(8, uint16_t, sub_mb_type)[4]; | |||||
int direct_spatial_mv_pred; | int direct_spatial_mv_pred; | ||||
int col_parity; | int col_parity; | ||||
int col_fieldoff; | int col_fieldoff; | ||||
@@ -1027,21 +1026,21 @@ static av_always_inline void write_back_motion(H264Context *h, | |||||
if (sl->slice_type_nos == AV_PICTURE_TYPE_B && CABAC(h)) { | if (sl->slice_type_nos == AV_PICTURE_TYPE_B && CABAC(h)) { | ||||
if (IS_8X8(mb_type)) { | if (IS_8X8(mb_type)) { | ||||
uint8_t *direct_table = &h->direct_table[4 * h->mb_xy]; | uint8_t *direct_table = &h->direct_table[4 * h->mb_xy]; | ||||
direct_table[1] = h->sub_mb_type[1] >> 1; | |||||
direct_table[2] = h->sub_mb_type[2] >> 1; | |||||
direct_table[3] = h->sub_mb_type[3] >> 1; | |||||
direct_table[1] = sl->sub_mb_type[1] >> 1; | |||||
direct_table[2] = sl->sub_mb_type[2] >> 1; | |||||
direct_table[3] = sl->sub_mb_type[3] >> 1; | |||||
} | } | ||||
} | } | ||||
} | } | ||||
static av_always_inline int get_dct8x8_allowed(H264Context *h) | |||||
static av_always_inline int get_dct8x8_allowed(H264Context *h, H264SliceContext *sl) | |||||
{ | { | ||||
if (h->sps.direct_8x8_inference_flag) | if (h->sps.direct_8x8_inference_flag) | ||||
return !(AV_RN64A(h->sub_mb_type) & | |||||
return !(AV_RN64A(sl->sub_mb_type) & | |||||
((MB_TYPE_16x8 | MB_TYPE_8x16 | MB_TYPE_8x8) * | ((MB_TYPE_16x8 | MB_TYPE_8x16 | MB_TYPE_8x8) * | ||||
0x0001000100010001ULL)); | 0x0001000100010001ULL)); | ||||
else | else | ||||
return !(AV_RN64A(h->sub_mb_type) & | |||||
return !(AV_RN64A(sl->sub_mb_type) & | |||||
((MB_TYPE_16x8 | MB_TYPE_8x16 | MB_TYPE_8x8 | MB_TYPE_DIRECT2) * | ((MB_TYPE_16x8 | MB_TYPE_8x16 | MB_TYPE_8x8 | MB_TYPE_DIRECT2) * | ||||
0x0001000100010001ULL)); | 0x0001000100010001ULL)); | ||||
} | } | ||||
@@ -2090,32 +2090,32 @@ decode_intra_mb: | |||||
if (sl->slice_type_nos == AV_PICTURE_TYPE_B ) { | if (sl->slice_type_nos == AV_PICTURE_TYPE_B ) { | ||||
for( i = 0; i < 4; i++ ) { | for( i = 0; i < 4; i++ ) { | ||||
h->sub_mb_type[i] = decode_cabac_b_mb_sub_type( h ); | |||||
sub_partition_count[i]= b_sub_mb_type_info[ h->sub_mb_type[i] ].partition_count; | |||||
h->sub_mb_type[i]= b_sub_mb_type_info[ h->sub_mb_type[i] ].type; | |||||
sl->sub_mb_type[i] = decode_cabac_b_mb_sub_type( h ); | |||||
sub_partition_count[i] = b_sub_mb_type_info[sl->sub_mb_type[i]].partition_count; | |||||
sl->sub_mb_type[i] = b_sub_mb_type_info[sl->sub_mb_type[i]].type; | |||||
} | } | ||||
if( IS_DIRECT(h->sub_mb_type[0] | h->sub_mb_type[1] | | |||||
h->sub_mb_type[2] | h->sub_mb_type[3]) ) { | |||||
if (IS_DIRECT(sl->sub_mb_type[0] | sl->sub_mb_type[1] | | |||||
sl->sub_mb_type[2] | sl->sub_mb_type[3])) { | |||||
ff_h264_pred_direct_motion(h, sl, &mb_type); | ff_h264_pred_direct_motion(h, sl, &mb_type); | ||||
sl->ref_cache[0][scan8[4]] = | sl->ref_cache[0][scan8[4]] = | ||||
sl->ref_cache[1][scan8[4]] = | sl->ref_cache[1][scan8[4]] = | ||||
sl->ref_cache[0][scan8[12]] = | sl->ref_cache[0][scan8[12]] = | ||||
sl->ref_cache[1][scan8[12]] = PART_NOT_AVAILABLE; | sl->ref_cache[1][scan8[12]] = PART_NOT_AVAILABLE; | ||||
for( i = 0; i < 4; i++ ) | for( i = 0; i < 4; i++ ) | ||||
fill_rectangle( &h->direct_cache[scan8[4*i]], 2, 2, 8, (h->sub_mb_type[i]>>1)&0xFF, 1 ); | |||||
fill_rectangle(&h->direct_cache[scan8[4*i]], 2, 2, 8, (sl->sub_mb_type[i] >> 1) & 0xFF, 1); | |||||
} | } | ||||
} else { | } else { | ||||
for( i = 0; i < 4; i++ ) { | for( i = 0; i < 4; i++ ) { | ||||
h->sub_mb_type[i] = decode_cabac_p_mb_sub_type( h ); | |||||
sub_partition_count[i]= p_sub_mb_type_info[ h->sub_mb_type[i] ].partition_count; | |||||
h->sub_mb_type[i]= p_sub_mb_type_info[ h->sub_mb_type[i] ].type; | |||||
sl->sub_mb_type[i] = decode_cabac_p_mb_sub_type(h); | |||||
sub_partition_count[i] = p_sub_mb_type_info[sl->sub_mb_type[i]].partition_count; | |||||
sl->sub_mb_type[i] = p_sub_mb_type_info[sl->sub_mb_type[i]].type; | |||||
} | } | ||||
} | } | ||||
for( list = 0; list < h->list_count; list++ ) { | for( list = 0; list < h->list_count; list++ ) { | ||||
for( i = 0; i < 4; i++ ) { | 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(IS_DIRECT(sl->sub_mb_type[i])) continue; | |||||
if(IS_DIR(sl->sub_mb_type[i], 0, list)){ | |||||
int rc = h->ref_count[list] << MB_MBAFF(h); | int rc = h->ref_count[list] << MB_MBAFF(h); | ||||
if (rc > 1) { | if (rc > 1) { | ||||
ref[list][i] = decode_cabac_mb_ref(h, sl, list, 4 * i); | ref[list][i] = decode_cabac_mb_ref(h, sl, list, 4 * i); | ||||
@@ -2134,18 +2134,18 @@ decode_intra_mb: | |||||
} | } | ||||
if(dct8x8_allowed) | if(dct8x8_allowed) | ||||
dct8x8_allowed = get_dct8x8_allowed(h); | |||||
dct8x8_allowed = get_dct8x8_allowed(h, sl); | |||||
for(list=0; list<h->list_count; list++){ | for(list=0; list<h->list_count; list++){ | ||||
for(i=0; i<4; i++){ | for(i=0; i<4; i++){ | ||||
sl->ref_cache[list][scan8[4 * i]] = sl->ref_cache[list][scan8[4 * i] + 1]; | sl->ref_cache[list][scan8[4 * i]] = sl->ref_cache[list][scan8[4 * i] + 1]; | ||||
if(IS_DIRECT(h->sub_mb_type[i])){ | |||||
if(IS_DIRECT(sl->sub_mb_type[i])){ | |||||
fill_rectangle(h->mvd_cache[list][scan8[4*i]], 2, 2, 8, 0, 2); | fill_rectangle(h->mvd_cache[list][scan8[4*i]], 2, 2, 8, 0, 2); | ||||
continue; | continue; | ||||
} | } | ||||
if(IS_DIR(h->sub_mb_type[i], 0, list) && !IS_DIRECT(h->sub_mb_type[i])){ | |||||
const int sub_mb_type= h->sub_mb_type[i]; | |||||
if(IS_DIR(sl->sub_mb_type[i], 0, list) && !IS_DIRECT(sl->sub_mb_type[i])){ | |||||
const int sub_mb_type= sl->sub_mb_type[i]; | |||||
const int block_width= (sub_mb_type & (MB_TYPE_16x16|MB_TYPE_16x8)) ? 2 : 1; | const int block_width= (sub_mb_type & (MB_TYPE_16x16|MB_TYPE_16x8)) ? 2 : 1; | ||||
for(j=0; j<sub_partition_count[i]; j++){ | for(j=0; j<sub_partition_count[i]; j++){ | ||||
int mpx, mpy; | int mpx, mpy; | ||||
@@ -836,15 +836,15 @@ decode_intra_mb: | |||||
if (sl->slice_type_nos == AV_PICTURE_TYPE_B) { | if (sl->slice_type_nos == AV_PICTURE_TYPE_B) { | ||||
for(i=0; i<4; i++){ | for(i=0; i<4; i++){ | ||||
h->sub_mb_type[i]= get_ue_golomb_31(&h->gb); | |||||
if(h->sub_mb_type[i] >=13){ | |||||
av_log(h->avctx, AV_LOG_ERROR, "B sub_mb_type %u out of range at %d %d\n", h->sub_mb_type[i], h->mb_x, h->mb_y); | |||||
sl->sub_mb_type[i]= get_ue_golomb_31(&h->gb); | |||||
if(sl->sub_mb_type[i] >=13){ | |||||
av_log(h->avctx, AV_LOG_ERROR, "B sub_mb_type %u out of range at %d %d\n", sl->sub_mb_type[i], h->mb_x, h->mb_y); | |||||
return -1; | return -1; | ||||
} | } | ||||
sub_partition_count[i]= b_sub_mb_type_info[ h->sub_mb_type[i] ].partition_count; | |||||
h->sub_mb_type[i]= b_sub_mb_type_info[ h->sub_mb_type[i] ].type; | |||||
sub_partition_count[i]= b_sub_mb_type_info[ sl->sub_mb_type[i] ].partition_count; | |||||
sl->sub_mb_type[i]= b_sub_mb_type_info[ sl->sub_mb_type[i] ].type; | |||||
} | } | ||||
if( IS_DIRECT(h->sub_mb_type[0]|h->sub_mb_type[1]|h->sub_mb_type[2]|h->sub_mb_type[3])) { | |||||
if( IS_DIRECT(sl->sub_mb_type[0]|sl->sub_mb_type[1]|sl->sub_mb_type[2]|sl->sub_mb_type[3])) { | |||||
ff_h264_pred_direct_motion(h, sl, &mb_type); | ff_h264_pred_direct_motion(h, sl, &mb_type); | ||||
sl->ref_cache[0][scan8[4]] = | sl->ref_cache[0][scan8[4]] = | ||||
sl->ref_cache[1][scan8[4]] = | sl->ref_cache[1][scan8[4]] = | ||||
@@ -854,21 +854,21 @@ decode_intra_mb: | |||||
}else{ | }else{ | ||||
assert(sl->slice_type_nos == AV_PICTURE_TYPE_P); //FIXME SP correct ? | assert(sl->slice_type_nos == AV_PICTURE_TYPE_P); //FIXME SP correct ? | ||||
for(i=0; i<4; i++){ | for(i=0; i<4; i++){ | ||||
h->sub_mb_type[i]= get_ue_golomb_31(&h->gb); | |||||
if(h->sub_mb_type[i] >=4){ | |||||
av_log(h->avctx, AV_LOG_ERROR, "P sub_mb_type %u out of range at %d %d\n", h->sub_mb_type[i], h->mb_x, h->mb_y); | |||||
sl->sub_mb_type[i]= get_ue_golomb_31(&h->gb); | |||||
if(sl->sub_mb_type[i] >=4){ | |||||
av_log(h->avctx, AV_LOG_ERROR, "P sub_mb_type %u out of range at %d %d\n", sl->sub_mb_type[i], h->mb_x, h->mb_y); | |||||
return -1; | return -1; | ||||
} | } | ||||
sub_partition_count[i]= p_sub_mb_type_info[ h->sub_mb_type[i] ].partition_count; | |||||
h->sub_mb_type[i]= p_sub_mb_type_info[ h->sub_mb_type[i] ].type; | |||||
sub_partition_count[i]= p_sub_mb_type_info[ sl->sub_mb_type[i] ].partition_count; | |||||
sl->sub_mb_type[i]= p_sub_mb_type_info[ sl->sub_mb_type[i] ].type; | |||||
} | } | ||||
} | } | ||||
for(list=0; list<h->list_count; list++){ | for(list=0; list<h->list_count; list++){ | ||||
int ref_count = IS_REF0(mb_type) ? 1 : h->ref_count[list] << MB_MBAFF(h); | int ref_count = IS_REF0(mb_type) ? 1 : h->ref_count[list] << MB_MBAFF(h); | ||||
for(i=0; i<4; i++){ | 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(IS_DIRECT(sl->sub_mb_type[i])) continue; | |||||
if(IS_DIR(sl->sub_mb_type[i], 0, list)){ | |||||
unsigned int tmp; | unsigned int tmp; | ||||
if(ref_count == 1){ | if(ref_count == 1){ | ||||
tmp= 0; | tmp= 0; | ||||
@@ -890,19 +890,19 @@ decode_intra_mb: | |||||
} | } | ||||
if(dct8x8_allowed) | if(dct8x8_allowed) | ||||
dct8x8_allowed = get_dct8x8_allowed(h); | |||||
dct8x8_allowed = get_dct8x8_allowed(h, sl); | |||||
for(list=0; list<h->list_count; list++){ | for(list=0; list<h->list_count; list++){ | ||||
for(i=0; i<4; i++){ | for(i=0; i<4; i++){ | ||||
if(IS_DIRECT(h->sub_mb_type[i])) { | |||||
if(IS_DIRECT(sl->sub_mb_type[i])) { | |||||
sl->ref_cache[list][ scan8[4*i] ] = sl->ref_cache[list][ scan8[4*i]+1 ]; | sl->ref_cache[list][ scan8[4*i] ] = sl->ref_cache[list][ scan8[4*i]+1 ]; | ||||
continue; | continue; | ||||
} | } | ||||
sl->ref_cache[list][ scan8[4*i] ]=sl->ref_cache[list][ scan8[4*i]+1 ]= | sl->ref_cache[list][ scan8[4*i] ]=sl->ref_cache[list][ scan8[4*i]+1 ]= | ||||
sl->ref_cache[list][ scan8[4*i]+8 ]=sl->ref_cache[list][ scan8[4*i]+9 ]= ref[list][i]; | sl->ref_cache[list][ scan8[4*i]+8 ]=sl->ref_cache[list][ scan8[4*i]+9 ]= ref[list][i]; | ||||
if(IS_DIR(h->sub_mb_type[i], 0, list)){ | |||||
const int sub_mb_type= h->sub_mb_type[i]; | |||||
if(IS_DIR(sl->sub_mb_type[i], 0, list)){ | |||||
const int sub_mb_type= sl->sub_mb_type[i]; | |||||
const int block_width= (sub_mb_type & (MB_TYPE_16x16|MB_TYPE_16x8)) ? 2 : 1; | const int block_width= (sub_mb_type & (MB_TYPE_16x16|MB_TYPE_16x8)) ? 2 : 1; | ||||
for(j=0; j<sub_partition_count[i]; j++){ | for(j=0; j<sub_partition_count[i]; j++){ | ||||
int mx, my; | int mx, my; | ||||
@@ -337,9 +337,9 @@ single_col: | |||||
int xy4 = x8 * 3 + y8 * b4_stride; | int xy4 = x8 * 3 + y8 * b4_stride; | ||||
int a, b; | int a, b; | ||||
if (is_b8x8 && !IS_DIRECT(h->sub_mb_type[i8])) | |||||
if (is_b8x8 && !IS_DIRECT(sl->sub_mb_type[i8])) | |||||
continue; | continue; | ||||
h->sub_mb_type[i8] = sub_mb_type; | |||||
sl->sub_mb_type[i8] = sub_mb_type; | |||||
fill_rectangle(&sl->ref_cache[0][scan8[i8 * 4]], 2, 2, 8, | fill_rectangle(&sl->ref_cache[0][scan8[i8 * 4]], 2, 2, 8, | ||||
(uint8_t)ref[0], 1); | (uint8_t)ref[0], 1); | ||||
@@ -401,9 +401,9 @@ single_col: | |||||
const int x8 = i8 & 1; | const int x8 = i8 & 1; | ||||
const int y8 = i8 >> 1; | const int y8 = i8 >> 1; | ||||
if (is_b8x8 && !IS_DIRECT(h->sub_mb_type[i8])) | |||||
if (is_b8x8 && !IS_DIRECT(sl->sub_mb_type[i8])) | |||||
continue; | continue; | ||||
h->sub_mb_type[i8] = sub_mb_type; | |||||
sl->sub_mb_type[i8] = sub_mb_type; | |||||
fill_rectangle(&sl->mv_cache[0][scan8[i8 * 4]], 2, 2, 8, mv[0], 4); | fill_rectangle(&sl->mv_cache[0][scan8[i8 * 4]], 2, 2, 8, mv[0], 4); | ||||
fill_rectangle(&sl->mv_cache[1][scan8[i8 * 4]], 2, 2, 8, mv[1], 4); | fill_rectangle(&sl->mv_cache[1][scan8[i8 * 4]], 2, 2, 8, mv[1], 4); | ||||
@@ -445,7 +445,7 @@ single_col: | |||||
} | } | ||||
} | } | ||||
if (!(m & 3)) | if (!(m & 3)) | ||||
h->sub_mb_type[i8] += MB_TYPE_16x16 - MB_TYPE_8x8; | |||||
sl->sub_mb_type[i8] += MB_TYPE_16x16 - MB_TYPE_8x8; | |||||
n += m; | n += m; | ||||
} | } | ||||
} | } | ||||
@@ -575,9 +575,9 @@ single_col: | |||||
int ref0, scale; | int ref0, scale; | ||||
const int16_t (*l1mv)[2] = l1mv0; | const int16_t (*l1mv)[2] = l1mv0; | ||||
if (is_b8x8 && !IS_DIRECT(h->sub_mb_type[i8])) | |||||
if (is_b8x8 && !IS_DIRECT(sl->sub_mb_type[i8])) | |||||
continue; | continue; | ||||
h->sub_mb_type[i8] = sub_mb_type; | |||||
sl->sub_mb_type[i8] = sub_mb_type; | |||||
fill_rectangle(&sl->ref_cache[1][scan8[i8 * 4]], 2, 2, 8, 0, 1); | fill_rectangle(&sl->ref_cache[1][scan8[i8 * 4]], 2, 2, 8, 0, 1); | ||||
if (IS_INTRA(mb_type_col[y8])) { | if (IS_INTRA(mb_type_col[y8])) { | ||||
@@ -643,9 +643,9 @@ single_col: | |||||
int ref0, scale; | int ref0, scale; | ||||
const int16_t (*l1mv)[2] = l1mv0; | const int16_t (*l1mv)[2] = l1mv0; | ||||
if (is_b8x8 && !IS_DIRECT(h->sub_mb_type[i8])) | |||||
if (is_b8x8 && !IS_DIRECT(sl->sub_mb_type[i8])) | |||||
continue; | continue; | ||||
h->sub_mb_type[i8] = sub_mb_type; | |||||
sl->sub_mb_type[i8] = sub_mb_type; | |||||
fill_rectangle(&sl->ref_cache[1][scan8[i8 * 4]], 2, 2, 8, 0, 1); | fill_rectangle(&sl->ref_cache[1][scan8[i8 * 4]], 2, 2, 8, 0, 1); | ||||
if (IS_INTRA(mb_type_col[0])) { | if (IS_INTRA(mb_type_col[0])) { | ||||
fill_rectangle(&sl->ref_cache[0][scan8[i8 * 4]], 2, 2, 8, 0, 1); | fill_rectangle(&sl->ref_cache[0][scan8[i8 * 4]], 2, 2, 8, 0, 1); | ||||
@@ -123,7 +123,7 @@ static void await_references(H264Context *h, H264SliceContext *sl) | |||||
assert(IS_8X8(mb_type)); | assert(IS_8X8(mb_type)); | ||||
for (i = 0; i < 4; i++) { | for (i = 0; i < 4; i++) { | ||||
const int sub_mb_type = h->sub_mb_type[i]; | |||||
const int sub_mb_type = sl->sub_mb_type[i]; | |||||
const int n = 4 * i; | const int n = 4 * i; | ||||
int y_offset = (i & 2) << 2; | int y_offset = (i & 2) << 2; | ||||
@@ -109,7 +109,7 @@ static void MCFUNC(hl_motion)(H264Context *h, H264SliceContext *sl, | |||||
assert(IS_8X8(mb_type)); | assert(IS_8X8(mb_type)); | ||||
for (i = 0; i < 4; i++) { | for (i = 0; i < 4; i++) { | ||||
const int sub_mb_type = h->sub_mb_type[i]; | |||||
const int sub_mb_type = sl->sub_mb_type[i]; | |||||
const int n = 4 * i; | const int n = 4 * i; | ||||
int x_offset = (i & 1) << 2; | int x_offset = (i & 1) << 2; | ||||
int y_offset = (i & 2) << 1; | int y_offset = (i & 2) << 1; | ||||