51 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2,
52 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5,
53 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3,
57 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 3, 3, 3,
58 3, 3, 3, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6,
59 7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 10, 10, 10, 10,
63 #if CONFIG_H264_DXVA2_HWACCEL
66 #if CONFIG_H264_VAAPI_HWACCEL
69 #if CONFIG_H264_VDA_HWACCEL
83 static const int8_t top[12] = {
86 static const int8_t left[12] = {
92 for (i = 0; i < 4; i++) {
96 "top block unavailable for requested intra4x4 mode %d at %d %d\n",
106 static const int mask[4] = { 0x8000, 0x2000, 0x80, 0x20 };
107 for (i = 0; i < 4; i++)
112 "left block unavailable for requested intra4x4 mode %d at %d %d\n",
136 "out of range intra chroma pred mode at %d %d\n",
145 "top block unavailable for requested intra mode at %d %d\n",
161 "left block unavailable for requested intra mode at %d %d\n",
171 int *dst_length,
int *consumed,
int length)
184 #define STARTCODE_TEST \
185 if (i + 2 < length && src[i + 1] == 0 && src[i + 2] <= 3) { \
186 if (src[i + 2] != 3) { \
192 #if HAVE_FAST_UNALIGNED
193 #define FIND_FIRST_ZERO \
194 if (i > 0 && !src[i]) \
199 for (i = 0; i + 1 < length; i += 9) {
201 (
AV_RN64A(src + i) - 0x0100010001000101ULL)) &
202 0x8000800080008080ULL))
209 for (i = 0; i + 1 < length; i += 5) {
211 (
AV_RN32A(src + i) - 0x01000101U)) &
220 for (i = 0; i + 1 < length; i += 2) {
223 if (i > 0 && src[i - 1] == 0)
229 if (i >= length - 1) {
230 *dst_length = length;
231 *consumed = length + 1;
246 while (si + 2 < length) {
248 if (src[si + 2] > 3) {
249 dst[di++] = src[si++];
250 dst[di++] = src[si++];
251 }
else if (src[si] == 0 && src[si + 1] == 0) {
252 if (src[si + 2] == 3) {
261 dst[di++] = src[si++];
264 dst[di++] = src[si++];
287 for (r = 1; r < 9; r++) {
296 int height,
int y_offset,
int list)
299 int filter_height_up = (raw_my & 3) ? 2 : 0;
300 int filter_height_down = (raw_my & 3) ? 3 : 0;
301 int full_my = (raw_my >> 2) + y_offset;
302 int top = full_my - filter_height_up;
303 int bottom = full_my + filter_height_down +
height;
305 return FFMAX(abs(top), bottom);
309 int height,
int y_offset,
int list0,
310 int list1,
int *nrefs)
327 if (refs[0][ref_n] < 0)
329 refs[0][ref_n] =
FFMAX(refs[0][ref_n], my);
340 if (refs[1][ref_n] < 0)
342 refs[1][ref_n] =
FFMAX(refs[1][ref_n], my);
355 const int mb_xy = h->
mb_xy;
358 int nrefs[2] = { 0 };
361 memset(refs, -1,
sizeof(refs));
381 for (i = 0; i < 4; i++) {
384 int y_offset = (i & 2) << 2;
388 IS_DIR(sub_mb_type, 0, 0),
389 IS_DIR(sub_mb_type, 0, 1),
393 IS_DIR(sub_mb_type, 0, 0),
394 IS_DIR(sub_mb_type, 0, 1),
397 IS_DIR(sub_mb_type, 0, 0),
398 IS_DIR(sub_mb_type, 0, 1),
402 IS_DIR(sub_mb_type, 0, 0),
403 IS_DIR(sub_mb_type, 0, 1),
406 IS_DIR(sub_mb_type, 0, 0),
407 IS_DIR(sub_mb_type, 0, 1),
412 for (j = 0; j < 4; j++) {
413 int sub_y_offset = y_offset + 2 * (j & 2);
415 IS_DIR(sub_mb_type, 0, 0),
416 IS_DIR(sub_mb_type, 0, 1),
423 for (list = h->
list_count - 1; list >= 0; list--)
424 for (ref = 0; ref < 48 && nrefs[list]; ref++) {
425 int row = refs[list][ref];
430 int pic_height = 16 * s->
mb_height >> ref_field_picture;
437 FFMIN((row >> 1) - !(row & 1),
441 FFMIN((row >> 1), pic_height - 1),
445 FFMIN(row * 2 + ref_field,
450 FFMIN(row, pic_height - 1),
454 FFMIN(row, pic_height - 1),
466 int src_x_offset,
int src_y_offset,
469 int pixel_shift,
int chroma_idc)
472 const int mx = h->
mv_cache[list][
scan8[n]][0] + src_x_offset * 8;
474 const int luma_xy = (mx & 3) + ((my & 3) << 2);
475 int offset = ((mx >> 2) << pixel_shift) + (my >> 2) * h->
mb_linesize;
481 const int full_mx = mx >> 2;
482 const int full_my = my >> 2;
483 const int pic_width = 16 * s->
mb_width;
492 if (full_mx < 0 - extra_width ||
493 full_my < 0 - extra_height ||
494 full_mx + 16 > pic_width + extra_width ||
495 full_my + 16 > pic_height + extra_height) {
499 16 + 5, 16 + 5 , full_mx - 2,
500 full_my - 2, pic_width, pic_height);
512 if (chroma_idc == 3 ) {
513 src_cb = pic->
f.
data[1] + offset;
519 full_mx - 2, full_my - 2,
520 pic_width, pic_height);
527 src_cr = pic->
f.
data[2] + offset;
533 full_mx - 2, full_my - 2,
534 pic_width, pic_height);
543 ysh = 3 - (chroma_idc == 2 );
547 emu |= (my >> 3) < 0 || (my >> 3) + 8 >= (pic_height >> 1);
550 src_cb = pic->
f.
data[1] + ((mx >> 3) << pixel_shift) +
552 src_cr = pic->
f.
data[2] + ((mx >> 3) << pixel_shift) +
557 9, 8 * chroma_idc + 1, (mx >> 3), (my >> ysh),
558 pic_width >> 1, pic_height >> (chroma_idc == 1 ));
562 height >> (chroma_idc == 1 ),
563 mx & 7, (my << (chroma_idc == 2 )) & 7);
567 9, 8 * chroma_idc + 1, (mx >> 3), (my >> ysh),
568 pic_width >> 1, pic_height >> (chroma_idc == 1 ));
571 chroma_op(dest_cr, src_cr, h->
mb_uvlinesize, height >> (chroma_idc == 1 ),
572 mx & 7, (my << (chroma_idc == 2 )) & 7);
579 int x_offset,
int y_offset,
584 int list0,
int list1,
585 int pixel_shift,
int chroma_idc)
591 dest_y += (2 * x_offset << pixel_shift) + 2 * y_offset * h->
mb_linesize;
592 if (chroma_idc == 3 ) {
593 dest_cb += (2 * x_offset << pixel_shift) + 2 * y_offset * h->
mb_linesize;
594 dest_cr += (2 * x_offset << pixel_shift) + 2 * y_offset * h->
mb_linesize;
595 }
else if (chroma_idc == 2 ) {
596 dest_cb += (x_offset << pixel_shift) + 2 * y_offset * h->
mb_uvlinesize;
597 dest_cr += (x_offset << pixel_shift) + 2 * y_offset * h->
mb_uvlinesize;
599 dest_cb += (x_offset << pixel_shift) + y_offset * h->
mb_uvlinesize;
600 dest_cr += (x_offset << pixel_shift) + y_offset * h->
mb_uvlinesize;
602 x_offset += 8 * s->
mb_x;
608 dest_y, dest_cb, dest_cr, x_offset, y_offset,
609 qpix_op, chroma_op, pixel_shift, chroma_idc);
612 chroma_op = chroma_avg;
618 dest_y, dest_cb, dest_cr, x_offset, y_offset,
619 qpix_op, chroma_op, pixel_shift, chroma_idc);
627 int x_offset,
int y_offset,
634 int list0,
int list1,
635 int pixel_shift,
int chroma_idc)
640 dest_y += (2 * x_offset << pixel_shift) + 2 * y_offset * h->
mb_linesize;
641 if (chroma_idc == 3 ) {
643 chroma_weight_avg = luma_weight_avg;
644 chroma_weight_op = luma_weight_op;
645 dest_cb += (2 * x_offset << pixel_shift) + 2 * y_offset * h->
mb_linesize;
646 dest_cr += (2 * x_offset << pixel_shift) + 2 * y_offset * h->
mb_linesize;
647 }
else if (chroma_idc == 2 ) {
649 dest_cb += (x_offset << pixel_shift) + 2 * y_offset * h->
mb_uvlinesize;
650 dest_cr += (x_offset << pixel_shift) + 2 * y_offset * h->
mb_uvlinesize;
652 chroma_height = height >> 1;
653 dest_cb += (x_offset << pixel_shift) + y_offset * h->
mb_uvlinesize;
654 dest_cr += (x_offset << pixel_shift) + y_offset * h->
mb_uvlinesize;
656 x_offset += 8 * s->
mb_x;
659 if (list0 && list1) {
669 dest_y, dest_cb, dest_cr,
670 x_offset, y_offset, qpix_put, chroma_put,
671 pixel_shift, chroma_idc);
673 tmp_y, tmp_cb, tmp_cr,
674 x_offset, y_offset, qpix_put, chroma_put,
675 pixel_shift, chroma_idc);
679 int weight1 = 64 - weight0;
681 height, 5, weight0, weight1, 0);
683 chroma_height, 5, weight0, weight1, 0);
685 chroma_height, 5, weight0, weight1, 0);
687 luma_weight_avg(dest_y, tmp_y, h->
mb_linesize, height,
693 chroma_weight_avg(dest_cb, tmp_cb, h->
mb_uvlinesize, chroma_height,
699 chroma_weight_avg(dest_cr, tmp_cr, h->
mb_uvlinesize, chroma_height,
707 int list = list1 ? 1 : 0;
710 mc_dir_part(h, ref, n, square, height, delta, list,
711 dest_y, dest_cb, dest_cr, x_offset, y_offset,
712 qpix_put, chroma_put, pixel_shift, chroma_idc);
732 int pixel_shift,
int chroma_idc)
742 int off = (mx << pixel_shift) +
746 if (chroma_idc == 3 ) {
750 off = ((mx >> 1) << pixel_shift) +
800 for (i = 0; i < 6; i++) {
802 for (j = 0; j < i; j++)
811 for (q = 0; q < max_qp + 1; q++) {
814 for (x = 0; x < 64; x++)
826 for (i = 0; i < 6; i++) {
828 for (j = 0; j < i; j++)
837 for (q = 0; q < max_qp + 1; q++) {
838 int shift =
div6[q] + 2;
840 for (x = 0; x < 16; x++)
855 for (i = 0; i < 6; i++)
856 for (x = 0; x < 16; x++)
859 for (i = 0; i < 6; i++)
860 for (x = 0; x < 64; x++)
873 row_mb_num * 8 *
sizeof(
uint8_t), fail)
875 big_mb_num * 48 *
sizeof(
uint8_t), fail)
879 big_mb_num *
sizeof(uint16_t), fail)
881 big_mb_num *
sizeof(
uint8_t), fail)
883 16 * row_mb_num *
sizeof(
uint8_t), fail);
885 16 * row_mb_num *
sizeof(
uint8_t), fail);
887 4 * big_mb_num *
sizeof(
uint8_t), fail);
889 big_mb_num *
sizeof(
uint8_t), fail)
896 big_mb_num *
sizeof(uint32_t), fail);
898 big_mb_num *
sizeof(uint32_t), fail);
902 const int b_xy = 4 * x + 4 * y * h->
b_stride;
965 int parse_extradata);
1007 cnt = *(p + 5) & 0x1f;
1009 for (i = 0; i < cnt; i++) {
1015 "Decoding sps %d from avcC failed\n", i);
1022 for (i = 0; i < cnt; i++) {
1028 "Decoding pps %d from avcC failed\n", i);
1096 #define IN_RANGE(a, b, size) (((a) >= (b)) && ((a) < ((b) + (size))))
1104 for (i = 0; i < count; i++) {
1105 assert((
IN_RANGE(from[i], old_base,
sizeof(*old_base)) ||
1117 for (i = 0; i < count; i++) {
1118 if (to[i] && !from[i])
1120 else if (from[i] && !to[i])
1124 memcpy(to[i], from[i], size);
1142 #define copy_fields(to, from, start_field, end_field) \
1143 memcpy(&to->start_field, &from->start_field, \
1144 (char *)&to->end_field - (char *)&to->start_field)
1158 if (dst == src || !
s1->context_initialized)
1205 memcpy(&h->
s + 1, &h1->s + 1,
1215 for (i = 0; i < 2; i++) {
1245 copy_fields(h, h1, dequant4_buffer, dequant4_coeff);
1247 for (i = 0; i < 6; i++)
1249 (h1->dequant4_coeff[i] - h1->dequant4_buffer[0]);
1251 for (i = 0; i < 6; i++)
1253 (h1->dequant8_coeff[i] - h1->dequant8_buffer[0]);
1309 for (i = 0; i < 16; i++) {
1313 for (i = 0; i < 16; i++) {
1364 int i, pics, out_of_order, out_idx;
1365 int invalid = 0, cnt = 0;
1490 invalid += out->
poc == INT_MIN;
1493 cnt + invalid == MAX_DELAYED_PIC_COUNT && cnt > 0) {
1502 invalid = MAX_DELAYED_PIC_COUNT;
1506 for (i = 1; i < MAX_DELAYED_PIC_COUNT &&
1525 if (invalid + cnt < MAX_DELAYED_PIC_COUNT) {
1546 sizeof(*h->
last_pocs) * (MAX_DELAYED_PIC_COUNT - 1));
1575 int linesize,
int uvlinesize,
1586 src_cb -= uvlinesize;
1587 src_cr -= uvlinesize;
1593 AV_COPY128(top_border, src_y + 15 * linesize);
1595 AV_COPY128(top_border + 16, src_y + 15 * linesize + 16);
1599 AV_COPY128(top_border + 32, src_cb + 15 * uvlinesize);
1600 AV_COPY128(top_border + 48, src_cb + 15 * uvlinesize + 16);
1601 AV_COPY128(top_border + 64, src_cr + 15 * uvlinesize);
1602 AV_COPY128(top_border + 80, src_cr + 15 * uvlinesize + 16);
1604 AV_COPY128(top_border + 16, src_cb + 15 * uvlinesize);
1605 AV_COPY128(top_border + 32, src_cr + 15 * uvlinesize);
1607 }
else if (chroma422) {
1609 AV_COPY128(top_border + 32, src_cb + 15 * uvlinesize);
1610 AV_COPY128(top_border + 48, src_cr + 15 * uvlinesize);
1612 AV_COPY64(top_border + 16, src_cb + 15 * uvlinesize);
1613 AV_COPY64(top_border + 24, src_cr + 15 * uvlinesize);
1617 AV_COPY128(top_border + 32, src_cb + 7 * uvlinesize);
1618 AV_COPY128(top_border + 48, src_cr + 7 * uvlinesize);
1620 AV_COPY64(top_border + 16, src_cb + 7 * uvlinesize);
1621 AV_COPY64(top_border + 24, src_cr + 7 * uvlinesize);
1635 AV_COPY128(top_border, src_y + 16 * linesize);
1637 AV_COPY128(top_border + 16, src_y + 16 * linesize + 16);
1642 AV_COPY128(top_border + 32, src_cb + 16 * linesize);
1643 AV_COPY128(top_border + 48, src_cb + 16 * linesize + 16);
1644 AV_COPY128(top_border + 64, src_cr + 16 * linesize);
1645 AV_COPY128(top_border + 80, src_cr + 16 * linesize + 16);
1647 AV_COPY128(top_border + 16, src_cb + 16 * linesize);
1648 AV_COPY128(top_border + 32, src_cr + 16 * linesize);
1650 }
else if (chroma422) {
1652 AV_COPY128(top_border + 32, src_cb + 16 * uvlinesize);
1653 AV_COPY128(top_border + 48, src_cr + 16 * uvlinesize);
1655 AV_COPY64(top_border + 16, src_cb + 16 * uvlinesize);
1656 AV_COPY64(top_border + 24, src_cr + 16 * uvlinesize);
1660 AV_COPY128(top_border + 32, src_cb + 8 * uvlinesize);
1661 AV_COPY128(top_border + 48, src_cr + 8 * uvlinesize);
1663 AV_COPY64(top_border + 16, src_cb + 8 * uvlinesize);
1664 AV_COPY64(top_border + 24, src_cr + 8 * uvlinesize);
1672 int linesize,
int uvlinesize,
1673 int xchg,
int chroma444,
1674 int simple,
int pixel_shift)
1677 int deblock_topleft;
1696 deblock_topleft = (s->
mb_x > 0);
1700 src_y -= linesize + 1 + pixel_shift;
1701 src_cb -= uvlinesize + 1 + pixel_shift;
1702 src_cr -= uvlinesize + 1 + pixel_shift;
1707 #define XCHG(a, b, xchg) \
1708 if (pixel_shift) { \
1710 AV_SWAP64(b + 0, a + 0); \
1711 AV_SWAP64(b + 8, a + 8); \
1721 if (deblock_topleft) {
1722 XCHG(top_border_m1 + (8 << pixel_shift),
1723 src_y - (7 << pixel_shift), 1);
1725 XCHG(top_border + (0 << pixel_shift), src_y + (1 << pixel_shift), xchg);
1726 XCHG(top_border + (8 << pixel_shift), src_y + (9 << pixel_shift), 1);
1729 src_y + (17 << pixel_shift), 1);
1734 if (deblock_topleft) {
1735 XCHG(top_border_m1 + (24 << pixel_shift), src_cb - (7 << pixel_shift), 1);
1736 XCHG(top_border_m1 + (40 << pixel_shift), src_cr - (7 << pixel_shift), 1);
1738 XCHG(top_border + (16 << pixel_shift), src_cb + (1 << pixel_shift), xchg);
1739 XCHG(top_border + (24 << pixel_shift), src_cb + (9 << pixel_shift), 1);
1740 XCHG(top_border + (32 << pixel_shift), src_cr + (1 << pixel_shift), xchg);
1741 XCHG(top_border + (40 << pixel_shift), src_cr + (9 << pixel_shift), 1);
1743 XCHG(h->
top_borders[top_idx][s->
mb_x + 1] + (16 << pixel_shift), src_cb + (17 << pixel_shift), 1);
1744 XCHG(h->
top_borders[top_idx][s->
mb_x + 1] + (32 << pixel_shift), src_cr + (17 << pixel_shift), 1);
1748 if (deblock_topleft) {
1749 XCHG(top_border_m1 + (16 << pixel_shift), src_cb - (7 << pixel_shift), 1);
1750 XCHG(top_border_m1 + (24 << pixel_shift), src_cr - (7 << pixel_shift), 1);
1752 XCHG(top_border + (16 << pixel_shift), src_cb + 1 + pixel_shift, 1);
1753 XCHG(top_border + (24 << pixel_shift), src_cr + 1 + pixel_shift, 1);
1762 if (high_bit_depth) {
1769 int index,
int value)
1771 if (high_bit_depth) {
1778 int mb_type,
int is_h264,
1780 int transform_bypass,
1791 block_offset += 16 * p;
1795 if (transform_bypass) {
1802 for (i = 0; i < 16; i += 4) {
1803 uint8_t *
const ptr = dest_y + block_offset[i];
1806 h->
hpc.
pred8x8l_add[dir](ptr, h->
mb + (i * 16 + p * 256 << pixel_shift), linesize);
1812 if (nnz == 1 &&
dctcoef_get(h->
mb, pixel_shift, i * 16 + p * 256))
1813 idct_dc_add(ptr, h->
mb + (i * 16 + p * 256 << pixel_shift), linesize);
1815 idct_add(ptr, h->
mb + (i * 16 + p * 256 << pixel_shift), linesize);
1820 if (transform_bypass) {
1827 for (i = 0; i < 16; i++) {
1828 uint8_t *
const ptr = dest_y + block_offset[i];
1832 h->
hpc.
pred4x4_add[dir](ptr, h->
mb + (i * 16 + p * 256 << pixel_shift), linesize);
1839 assert(s->
mb_y || linesize <= block_offset[i]);
1840 if (!topright_avail) {
1842 tr_high = ((uint16_t *)ptr)[3 - linesize / 2] * 0x0001000100010001ULL;
1843 topright = (
uint8_t *)&tr_high;
1845 tr = ptr[3 - linesize] * 0x01010101u;
1849 topright = ptr + (4 << pixel_shift) - linesize;
1853 h->
hpc.
pred4x4[dir](ptr, topright, linesize);
1857 if (nnz == 1 &&
dctcoef_get(h->
mb, pixel_shift, i * 16 + p * 256))
1858 idct_dc_add(ptr, h->
mb + (i * 16 + p * 256 << pixel_shift), linesize);
1860 idct_add(ptr, h->
mb + (i * 16 + p * 256 << pixel_shift), linesize);
1872 if (!transform_bypass)
1877 static const uint8_t dc_mapping[16] = {
1878 0 * 16, 1 * 16, 4 * 16, 5 * 16,
1879 2 * 16, 3 * 16, 6 * 16, 7 * 16,
1880 8 * 16, 9 * 16, 12 * 16, 13 * 16,
1881 10 * 16, 11 * 16, 14 * 16, 15 * 16 };
1882 for (i = 0; i < 16; i++)
1884 pixel_shift, dc_mapping[i],
1896 int is_h264,
int simple,
1897 int transform_bypass,
1906 block_offset += 16 * p;
1910 if (transform_bypass) {
1915 h->
mb + (p * 256 << pixel_shift),
1918 for (i = 0; i < 16; i++)
1922 h->
mb + (i * 16 + p * 256 << pixel_shift),
1927 h->
mb + (p * 256 << pixel_shift),
1931 }
else if (h->
cbp & 15) {
1932 if (transform_bypass) {
1933 const int di =
IS_8x8DCT(mb_type) ? 4 : 1;
1936 for (i = 0; i < 16; i += di)
1939 h->
mb + (i * 16 + p * 256 << pixel_shift),
1944 h->
mb + (p * 256 << pixel_shift),
1949 h->
mb + (p * 256 << pixel_shift),
1955 for (i = 0; i < 16; i++)
1958 uint8_t *
const ptr = dest_y + block_offset[i];
1981 const int mb_xy = h->
mb_xy;
1987 hl_decode_mb_444_complex(h);
1989 hl_decode_mb_444_simple_8(h);
1990 }
else if (is_complex) {
1991 hl_decode_mb_complex(h);
1993 hl_decode_mb_simple_16(h);
1995 hl_decode_mb_simple_8(h);
2002 int luma_def, chroma_def;
2012 for (list = 0; list < 2; list++) {
2015 for (i = 0; i < h->
ref_count[list]; i++) {
2016 int luma_weight_flag, chroma_weight_flag;
2019 if (luma_weight_flag) {
2034 if (chroma_weight_flag) {
2036 for (j = 0; j < 2; j++) {
2047 for (j = 0; j < 2; j++) {
2069 int ref0, ref1, i, cur_poc, ref_start, ref_count0, ref_count1;
2071 for (i = 0; i < 2; i++) {
2103 for (ref0 = ref_start; ref0 < ref_count0; ref0++) {
2105 for (ref1 = ref_start; ref1 < ref_count1; ref1++) {
2109 int td = av_clip(poc1 - poc0, -128, 127);
2111 int tb = av_clip(cur_poc - poc0, -128, 127);
2112 int tx = (16384 + (
FFABS(td) >> 1)) / td;
2113 int dist_scale_factor = (tb * tx + 32) >> 8;
2114 if (dist_scale_factor >= -64 && dist_scale_factor <= 128)
2115 w = 64 - dist_scale_factor;
2200 int abs_frame_num, expected_delta_per_poc_cycle, expectedpoc;
2211 expected_delta_per_poc_cycle = 0;
2216 if (abs_frame_num > 0) {
2220 expectedpoc = poc_cycle_cnt * expected_delta_per_poc_cycle;
2221 for (i = 0; i <= frame_num_in_poc_cycle; i++)
2229 field_poc[0] = expectedpoc + h->
delta_poc[0];
2259 for (i = 0; i < 16; i++) {
2260 #define T(x) (x >> 2) | ((x << 2) & 0xF)
2265 for (i = 0; i < 64; i++) {
2266 #define T(x) (x >> 3) | ((x & 7) << 3)
2319 "hardware accelerator failed to decode picture\n");
2365 "Failed to allocate scratch buffers\n");
2422 "Reenabling low delay requires a codec flush.\n");
2436 "VDPAU decoding does not support video colorspace.\n");
2548 "Could not allocate memory for h264\n");
2595 unsigned int first_mb_in_slice;
2596 unsigned int pps_id;
2597 int num_ref_idx_active_override_flag, max_refs, ret;
2598 unsigned int slice_type, tmp, i, j;
2599 int default_ref_list_done = 0;
2600 int last_pic_structure, last_pic_droppable;
2601 int needs_reinit = 0;
2615 if (first_mb_in_slice == 0) {
2632 if (slice_type > 9) {
2634 "slice type too large (%d) at %d %d\n",
2638 if (slice_type > 4) {
2647 default_ref_list_done = 1;
2668 "non-existing PPS %u referenced\n",
2676 "non-existing SPS %u referenced\n",
2761 "h264_slice_header_init() failed\n");
2769 "Cannot (re-)initialize context during parallel decoding.\n");
2779 "h264_slice_header_init() failed\n");
2812 "Changing field mode (%d -> %d) between slices is not allowed\n",
2819 "unset current_picture_ptr on %d. slice\n",
2830 if (unwrap_prev_frame_num > h->
frame_num)
2831 unwrap_prev_frame_num -= max_frame_num;
2835 if (unwrap_prev_frame_num < 0)
2836 unwrap_prev_frame_num += max_frame_num;
2862 if (!last_pic_droppable && last_pic_structure !=
PICT_FRAME) {
2872 if (!last_pic_droppable && last_pic_structure !=
PICT_FRAME) {
2883 "Invalid field mode combination %d/%d\n",
2888 }
else if (last_pic_droppable != s->
droppable) {
2890 "Cannot combine reference and non-reference fields in the same frame\n");
2992 first_mb_in_slice >= s->
mb_num) {
3039 num_ref_idx_active_override_flag =
get_bits1(&s->
gb);
3041 if (num_ref_idx_active_override_flag) {
3070 if (!default_ref_list_done)
3099 for (i = 0; i < 2; i++) {
3162 "deblocking_filter_idc %u out of range\n", tmp);
3177 "deblocking filter parameters %d %d out of range\n",
3202 "Cannot parallelize deblocking type 1, decoding such frames in sequential order\n");
3207 "Deblocking switched inside frame.\n");
3223 "Too many slices, increase MAX_SLICES and recompile\n");
3226 for (j = 0; j < 2; j++) {
3229 for (i = 0; i < 16; i++) {
3249 for (i = 0; i < 16; i++)
3250 ref2frm[i + 2] = 4 * id_list[i] +
3253 ref2frm[18 + 1] = -1;
3254 for (i = 16; i < 48; i++)
3255 ref2frm[i + 4] = 4 * id_list[(i - 16) >> 1] +
3268 "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",
3310 int mb_type,
int top_xy,
3313 int left_type[LEFT_MBS],
3314 int mb_xy,
int list)
3321 const int b_xy = h->
mb2b_xy[top_xy] + 3 * b_stride;
3322 const int b8_xy = 4 * top_xy + 2;
3325 ref_cache[0 - 1 * 8] =
3327 ref_cache[2 - 1 * 8] =
3337 const int b8_xy = 4 * left_xy[
LTOP] + 1;
3345 ref_cache[-1 + 16] =
3354 ref_cache[-1 + 16] =
3372 uint32_t ref01 = (
pack16to32(ref2frm[list][ref[0]], ref2frm[list][ref[1]]) & 0x00FF00FF) * 0x0101;
3373 uint32_t ref23 = (
pack16to32(ref2frm[list][ref[2]], ref2frm[list][ref[3]]) & 0x00FF00FF) * 0x0101;
3374 AV_WN32A(&ref_cache[0 * 8], ref01);
3375 AV_WN32A(&ref_cache[1 * 8], ref01);
3376 AV_WN32A(&ref_cache[2 * 8], ref23);
3377 AV_WN32A(&ref_cache[3 * 8], ref23);
3382 AV_COPY128(mv_dst + 8 * 0, mv_src + 0 * b_stride);
3383 AV_COPY128(mv_dst + 8 * 1, mv_src + 1 * b_stride);
3384 AV_COPY128(mv_dst + 8 * 2, mv_src + 2 * b_stride);
3385 AV_COPY128(mv_dst + 8 * 3, mv_src + 3 * b_stride);
3396 const int mb_xy = h->
mb_xy;
3407 left_xy[
LBOT] = left_xy[
LTOP] = mb_xy - 1;
3412 if (left_mb_field_flag != curr_mb_field_flag)
3415 if (curr_mb_field_flag)
3418 if (left_mb_field_flag != curr_mb_field_flag)
3432 if (qp <= qp_thresh &&
3433 (left_xy[
LTOP] < 0 ||
3439 if ((left_xy[
LTOP] < 0 ||
3454 left_type[
LTOP] = left_type[
LBOT] = 0;
3459 left_type[
LTOP] = left_type[
LBOT] = 0;
3469 top_type, left_type, mb_xy, 0);
3472 top_type, left_type, mb_xy, 1);
3476 AV_COPY32(&nnz_cache[4 + 8 * 1], &nnz[0]);
3477 AV_COPY32(&nnz_cache[4 + 8 * 2], &nnz[4]);
3478 AV_COPY32(&nnz_cache[4 + 8 * 3], &nnz[8]);
3479 AV_COPY32(&nnz_cache[4 + 8 * 4], &nnz[12]);
3484 AV_COPY32(&nnz_cache[4 + 8 * 0], &nnz[3 * 4]);
3487 if (left_type[
LTOP]) {
3489 nnz_cache[3 + 8 * 1] = nnz[3 + 0 * 4];
3490 nnz_cache[3 + 8 * 2] = nnz[3 + 1 * 4];
3491 nnz_cache[3 + 8 * 3] = nnz[3 + 2 * 4];
3492 nnz_cache[3 + 8 * 4] = nnz[3 + 3 * 4];
3499 nnz_cache[4 + 8 * 0] =
3500 nnz_cache[5 + 8 * 0] = (h->
cbp_table[top_xy] & 0x4000) >> 12;
3501 nnz_cache[6 + 8 * 0] =
3502 nnz_cache[7 + 8 * 0] = (h->
cbp_table[top_xy] & 0x8000) >> 12;
3505 nnz_cache[3 + 8 * 1] =
3506 nnz_cache[3 + 8 * 2] = (h->
cbp_table[left_xy[
LTOP]] & 0x2000) >> 12;
3509 nnz_cache[3 + 8 * 3] =
3510 nnz_cache[3 + 8 * 4] = (h->
cbp_table[left_xy[
LBOT]] & 0x8000) >> 12;
3514 nnz_cache[
scan8[0]] =
3515 nnz_cache[scan8[1]] =
3516 nnz_cache[scan8[2]] =
3517 nnz_cache[scan8[3]] = (h->
cbp & 0x1000) >> 12;
3519 nnz_cache[scan8[0 + 4]] =
3520 nnz_cache[scan8[1 + 4]] =
3521 nnz_cache[scan8[2 + 4]] =
3522 nnz_cache[scan8[3 + 4]] = (h->
cbp & 0x2000) >> 12;
3524 nnz_cache[scan8[0 + 8]] =
3525 nnz_cache[scan8[1 + 8]] =
3526 nnz_cache[scan8[2 + 8]] =
3527 nnz_cache[scan8[3 + 8]] = (h->
cbp & 0x4000) >> 12;
3529 nnz_cache[scan8[0 + 12]] =
3530 nnz_cache[scan8[1 + 12]] =
3531 nnz_cache[scan8[2 + 12]] =
3532 nnz_cache[scan8[3 + 12]] = (h->
cbp & 0x8000) >> 12;
3542 uint8_t *dest_y, *dest_cb, *dest_cr;
3543 int linesize, uvlinesize, mb_x, mb_y;
3550 for (mb_x = start_x; mb_x < end_x; mb_x++)
3551 for (mb_y = end_mb_y -
FRAME_MBAFF; mb_y <= end_mb_y; mb_y++) {
3565 ((mb_x << pixel_shift) + mb_y * s->
linesize) * 16;
3567 (mb_x << pixel_shift) * (8 <<
CHROMA444) +
3570 (mb_x << pixel_shift) * (8 <<
CHROMA444) +
3595 linesize, uvlinesize);
3598 dest_cr, linesize, uvlinesize);
3632 if ((top + height) >= pic_height)
3633 height += deblock_border;
3634 top -= deblock_border;
3640 height =
FFMIN(height, pic_height - top);
3641 if (top < h->emu_edge_height) {
3661 int lf_x_start = s->
mb_x;
3706 if (s->
mb_x >= lf_x_start)
3712 "error while decoding MB %d %d, bytestream (%td)\n",
3722 s->
mb_x = lf_x_start = 0;
3737 if (s->
mb_x > lf_x_start)
3761 "error while decoding MB %d %d\n", s->
mb_x, s->
mb_y);
3769 s->
mb_x = lf_x_start = 0;
3804 if (s->
mb_x > lf_x_start)
3834 "Input contains more MB rows than the frame height.\n");
3841 if (context_count == 1) {
3844 for (i = 1; i < context_count; i++) {
3851 NULL, context_count,
sizeof(
void *));
3859 for (i = 1; i < context_count; i++)
3867 int parse_extradata)
3876 int nals_needed = 0;
3887 for (; pass <= 1; pass++) {
3890 next_avc = h->
is_avc ? 0 : buf_size;
3900 if (buf_index >= next_avc) {
3905 nalsize = (nalsize << 8) | buf[buf_index++];
3906 if (nalsize <= 0 || nalsize > buf_size - buf_index) {
3908 "AVC: nal size %d\n", nalsize);
3911 next_avc = buf_index + nalsize;
3914 for (; buf_index + 3 < next_avc; buf_index++)
3916 if (buf[buf_index] == 0 &&
3917 buf[buf_index + 1] == 0 &&
3918 buf[buf_index + 2] == 1)
3921 if (buf_index + 3 >= buf_size) {
3922 buf_index = buf_size;
3927 if (buf_index >= next_avc)
3934 &consumed, next_avc - buf_index);
3935 if (ptr ==
NULL || dst_length < 0) {
3939 i = buf_index + consumed;
3941 buf[i] == 0x00 && buf[i + 1] == 0x00 &&
3942 buf[i + 2] == 0x01 && buf[i + 3] == 0xE0)
3946 while (dst_length > 0 && ptr[dst_length - 1] == 0)
3948 bit_length = !dst_length ? 0
3954 "NAL %d at %d/%d length %d\n",
3957 if (h->
is_avc && (nalsize != consumed) && nalsize)
3959 "AVC: Consumed only %d bytes instead of %d\n",
3962 buf_index += consumed;
3973 nals_needed = nal_index;
3980 nals_needed = nal_index;
4006 "Invalid mix of idr and non-idr slices\n");
4046 &buf[buf_index - consumed],
4051 static const uint8_t start_code[] = {
4054 sizeof(start_code));
4064 "Decoding in chunks is not supported for "
4065 "partitioned slices.\n");
4111 h->
is_avc && (nalsize != consumed) && nalsize) {
4113 "SPS decoding failure, trying again with the complete NAL\n");
4150 }
else if (err == 1) {
4183 if (pos + 10 > buf_size)
4193 int buf_size = avpkt->
size;
4207 if (buf_size == 0) {
4270 assert(pict->
data[0] || !*got_frame);
4338 #if CONFIG_H264_VDPAU_DECODER
4339 AVCodec ff_h264_vdpau_decoder = {
4340 .
name =
"h264_vdpau",
4347 .capabilities = CODEC_CAP_DR1 | CODEC_CAP_DELAY | CODEC_CAP_HWACCEL_VDPAU,
4349 .long_name =
NULL_IF_CONFIG_SMALL(
"H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10 (VDPAU acceleration)"),