* qatar/master:
rtmp: Add support for SWFVerification
api-example: use new video encoding API.
x86: avcodec: Appropriately name files containing only init functions
mpegvideo_mmx_template: drop some commented-out cruft
libavresample: add mix level normalization option
w32pthreads: Add missing #includes to make header compile standalone
rtmp: Gracefully ignore _checkbw errors by tracking them
rtmp: Do not send _checkbw calls as notifications
prores: interlaced ProRes encoding
Conflicts:
doc/examples/decoding_encoding.c
libavcodec/proresenc_kostya.c
libavcodec/w32pthreads.h
libavcodec/x86/Makefile
libavformat/version.h
Merged-by: Michael Niedermayer <michaelni@gmx.at>
... | ... |
@@ -31,10 +31,10 @@ |
31 | 31 |
|
32 | 32 |
#include <math.h> |
33 | 33 |
|
34 |
-#include <libavutil/imgutils.h> |
|
35 | 34 |
#include <libavutil/opt.h> |
36 | 35 |
#include <libavcodec/avcodec.h> |
37 | 36 |
#include <libavutil/audioconvert.h> |
37 |
+#include <libavutil/imgutils.h> |
|
38 | 38 |
#include <libavutil/mathematics.h> |
39 | 39 |
#include <libavutil/samplefmt.h> |
40 | 40 |
|
... | ... |
@@ -315,11 +315,11 @@ static void video_encode_example(const char *filename, int codec_id) |
315 | 315 |
{ |
316 | 316 |
AVCodec *codec; |
317 | 317 |
AVCodecContext *c= NULL; |
318 |
- int i, out_size, x, y, outbuf_size; |
|
318 |
+ int i, ret, x, y, got_output; |
|
319 | 319 |
FILE *f; |
320 | 320 |
AVFrame *picture; |
321 |
- uint8_t *outbuf; |
|
322 |
- int had_output=0; |
|
321 |
+ AVPacket pkt; |
|
322 |
+ uint8_t endcode[] = { 0, 0, 1, 0xb7 }; |
|
323 | 323 |
|
324 | 324 |
printf("Encode video file %s\n", filename); |
325 | 325 |
|
... | ... |
@@ -359,17 +359,25 @@ static void video_encode_example(const char *filename, int codec_id) |
359 | 359 |
exit(1); |
360 | 360 |
} |
361 | 361 |
|
362 |
- /* alloc image and output buffer */ |
|
363 |
- outbuf_size = 100000 + 12*c->width*c->height; |
|
364 |
- outbuf = malloc(outbuf_size); |
|
365 |
- |
|
366 | 362 |
/* the image can be allocated by any means and av_image_alloc() is |
367 | 363 |
* just the most convenient way if av_malloc() is to be used */ |
368 |
- av_image_alloc(picture->data, picture->linesize, |
|
369 |
- c->width, c->height, c->pix_fmt, 1); |
|
364 |
+ ret = av_image_alloc(picture->data, picture->linesize, c->width, c->height, |
|
365 |
+ c->pix_fmt, 32); |
|
366 |
+ if (ret < 0) { |
|
367 |
+ fprintf(stderr, "could not alloc raw picture buffer\n"); |
|
368 |
+ exit(1); |
|
369 |
+ } |
|
370 |
+ |
|
371 |
+ picture->format = c->pix_fmt; |
|
372 |
+ picture->width = c->width; |
|
373 |
+ picture->height = c->height; |
|
370 | 374 |
|
371 | 375 |
/* encode 1 second of video */ |
372 | 376 |
for(i=0;i<25;i++) { |
377 |
+ av_init_packet(&pkt); |
|
378 |
+ pkt.data = NULL; // packet data will be allocated by the encoder |
|
379 |
+ pkt.size = 0; |
|
380 |
+ |
|
373 | 381 |
fflush(stdout); |
374 | 382 |
/* prepare a dummy image */ |
375 | 383 |
/* Y */ |
... | ... |
@@ -387,35 +395,46 @@ static void video_encode_example(const char *filename, int codec_id) |
387 | 387 |
} |
388 | 388 |
} |
389 | 389 |
|
390 |
+ picture->pts = i; |
|
391 |
+ |
|
390 | 392 |
/* encode the image */ |
391 |
- out_size = avcodec_encode_video(c, outbuf, outbuf_size, picture); |
|
392 |
- had_output |= out_size; |
|
393 |
- printf("encoding frame %3d (size=%5d)\n", i, out_size); |
|
394 |
- fwrite(outbuf, 1, out_size, f); |
|
393 |
+ ret = avcodec_encode_video2(c, &pkt, picture, &got_output); |
|
394 |
+ if (ret < 0) { |
|
395 |
+ fprintf(stderr, "error encoding frame\n"); |
|
396 |
+ exit(1); |
|
397 |
+ } |
|
398 |
+ |
|
399 |
+ if (got_output) { |
|
400 |
+ printf("encoding frame %3d (size=%5d)\n", i, pkt.size); |
|
401 |
+ fwrite(pkt.data, 1, pkt.size, f); |
|
402 |
+ av_free_packet(&pkt); |
|
403 |
+ } |
|
395 | 404 |
} |
396 | 405 |
|
397 | 406 |
/* get the delayed frames */ |
398 |
- for(; out_size || !had_output; i++) { |
|
407 |
+ for (got_output = 1; got_output; i++) { |
|
399 | 408 |
fflush(stdout); |
400 | 409 |
|
401 |
- out_size = avcodec_encode_video(c, outbuf, outbuf_size, NULL); |
|
402 |
- had_output |= out_size; |
|
403 |
- printf("write frame %3d (size=%5d)\n", i, out_size); |
|
404 |
- fwrite(outbuf, 1, out_size, f); |
|
410 |
+ ret = avcodec_encode_video2(c, &pkt, NULL, &got_output); |
|
411 |
+ if (ret < 0) { |
|
412 |
+ fprintf(stderr, "error encoding frame\n"); |
|
413 |
+ exit(1); |
|
414 |
+ } |
|
415 |
+ |
|
416 |
+ if (got_output) { |
|
417 |
+ printf("write frame %3d (size=%5d)\n", i, pkt.size); |
|
418 |
+ fwrite(pkt.data, 1, pkt.size, f); |
|
419 |
+ av_free_packet(&pkt); |
|
420 |
+ } |
|
405 | 421 |
} |
406 | 422 |
|
407 | 423 |
/* add sequence end code to have a real mpeg file */ |
408 |
- outbuf[0] = 0x00; |
|
409 |
- outbuf[1] = 0x00; |
|
410 |
- outbuf[2] = 0x01; |
|
411 |
- outbuf[3] = 0xb7; |
|
412 |
- fwrite(outbuf, 1, 4, f); |
|
424 |
+ fwrite(endcode, 1, sizeof(endcode), f); |
|
413 | 425 |
fclose(f); |
414 |
- free(outbuf); |
|
415 | 426 |
|
416 | 427 |
avcodec_close(c); |
417 | 428 |
av_free(c); |
418 |
- av_free(picture->data[0]); |
|
429 |
+ av_freep(&picture->data[0]); |
|
419 | 430 |
av_free(picture); |
420 | 431 |
printf("\n"); |
421 | 432 |
} |
... | ... |
@@ -272,6 +272,12 @@ Name of live stream to subscribe to. By default no value will be sent. |
272 | 272 |
It is only sent if the option is specified or if rtmp_live |
273 | 273 |
is set to live. |
274 | 274 |
|
275 |
+@item rtmp_swfhash |
|
276 |
+SHA256 hash of the decompressed SWF file (32 bytes). |
|
277 |
+ |
|
278 |
+@item rtmp_swfsize |
|
279 |
+Size of the decompressed SWF file, required for SWFVerification. |
|
280 |
+ |
|
275 | 281 |
@item rtmp_swfurl |
276 | 282 |
URL of the SWF player for the media. By default no value will be sent. |
277 | 283 |
|
... | ... |
@@ -406,10 +406,15 @@ static int encode_slice(AVCodecContext *avctx, const AVFrame *pic, |
406 | 406 |
int total_size = 0; |
407 | 407 |
const uint16_t *src; |
408 | 408 |
int slice_width_factor = av_log2(mbs_per_slice); |
409 |
- int num_cblocks, pwidth, linesize, line_offset; |
|
409 |
+ int num_cblocks, pwidth, linesize, line_add; |
|
410 | 410 |
int plane_factor, is_chroma; |
411 | 411 |
uint16_t *qmat; |
412 | 412 |
|
413 |
+ if (ctx->pictures_per_frame == 1) |
|
414 |
+ line_add = 0; |
|
415 |
+ else |
|
416 |
+ line_add = ctx->cur_picture_idx ^ !pic->top_field_first; |
|
417 |
+ |
|
413 | 418 |
if (ctx->force_quant) { |
414 | 419 |
qmat = ctx->quants[0]; |
415 | 420 |
} else if (quant < MAX_STORED_Q) { |
... | ... |
@@ -437,15 +442,14 @@ static int encode_slice(AVCodecContext *avctx, const AVFrame *pic, |
437 | 437 |
pwidth = avctx->width >> 1; |
438 | 438 |
} |
439 | 439 |
|
440 |
- line_offset = ((ctx->cur_picture_idx ^ !pic->top_field_first) & |
|
441 |
- (ctx->pictures_per_frame - 1)) * pic->linesize[i]; |
|
442 | 440 |
linesize = pic->linesize[i] * ctx->pictures_per_frame; |
443 |
- src = (const uint16_t*)(pic->data[i] + yp * linesize + line_offset) + xp; |
|
441 |
+ src = (const uint16_t*)(pic->data[i] + yp * linesize + |
|
442 |
+ line_add * pic->linesize[i]) + xp; |
|
444 | 443 |
|
445 | 444 |
get_slice_data(ctx, src, linesize, xp, yp, |
446 | 445 |
pwidth, avctx->height / ctx->pictures_per_frame, |
447 |
- ctx->blocks[0], ctx->emu_buf, mbs_per_slice, |
|
448 |
- num_cblocks, is_chroma); |
|
446 |
+ ctx->blocks[0], ctx->emu_buf, |
|
447 |
+ mbs_per_slice, num_cblocks, is_chroma); |
|
449 | 448 |
sizes[i] = encode_slice_plane(ctx, pb, src, linesize, |
450 | 449 |
mbs_per_slice, ctx->blocks[0], |
451 | 450 |
num_cblocks, plane_factor, |
... | ... |
@@ -579,8 +583,12 @@ static int find_slice_quant(AVCodecContext *avctx, const AVFrame *pic, |
579 | 579 |
int slice_bits[TRELLIS_WIDTH], slice_score[TRELLIS_WIDTH]; |
580 | 580 |
int overquant; |
581 | 581 |
uint16_t *qmat; |
582 |
- int linesize[4], line_offset; |
|
582 |
+ int linesize[4], line_add; |
|
583 | 583 |
|
584 |
+ if (ctx->pictures_per_frame == 1) |
|
585 |
+ line_add = 0; |
|
586 |
+ else |
|
587 |
+ line_add = ctx->cur_picture_idx ^ !pic->top_field_first; |
|
584 | 588 |
mbs = x + mbs_per_slice; |
585 | 589 |
|
586 | 590 |
for (i = 0; i < ctx->num_planes; i++) { |
... | ... |
@@ -600,15 +608,14 @@ static int find_slice_quant(AVCodecContext *avctx, const AVFrame *pic, |
600 | 600 |
pwidth = avctx->width >> 1; |
601 | 601 |
} |
602 | 602 |
|
603 |
- line_offset = ((ctx->cur_picture_idx ^ !pic->top_field_first) & |
|
604 |
- (ctx->pictures_per_frame - 1)) * pic->linesize[i]; |
|
605 | 603 |
linesize[i] = pic->linesize[i] * ctx->pictures_per_frame; |
606 |
- src = (const uint16_t*)(pic->data[i] + yp * linesize[i] + line_offset) + xp; |
|
604 |
+ src = (const uint16_t*)(pic->data[i] + yp * linesize[i] + |
|
605 |
+ line_add * pic->linesize[i]) + xp; |
|
607 | 606 |
|
608 | 607 |
get_slice_data(ctx, src, linesize[i], xp, yp, |
609 | 608 |
pwidth, avctx->height / ctx->pictures_per_frame, |
610 |
- td->blocks[i], td->emu_buf, mbs_per_slice, |
|
611 |
- num_cblocks[i], is_chroma[i]); |
|
609 |
+ td->blocks[i], td->emu_buf, |
|
610 |
+ mbs_per_slice, num_cblocks[i], is_chroma[i]); |
|
612 | 611 |
} |
613 | 612 |
|
614 | 613 |
for (q = min_quant; q < max_quant + 2; q++) { |
... | ... |
@@ -767,9 +774,8 @@ static int encode_frame(AVCodecContext *avctx, AVPacket *pkt, |
767 | 767 |
bytestream_put_be16 (&buf, avctx->height); |
768 | 768 |
|
769 | 769 |
frame_flags = ctx->chroma_factor << 6; |
770 |
- if (avctx->flags & CODEC_FLAG_INTERLACED_DCT) { |
|
770 |
+ if (avctx->flags & CODEC_FLAG_INTERLACED_DCT) |
|
771 | 771 |
frame_flags |= pic->top_field_first ? 0x04 : 0x08; |
772 |
- } |
|
773 | 772 |
bytestream_put_byte (&buf, frame_flags); |
774 | 773 |
|
775 | 774 |
bytestream_put_byte (&buf, 0); // reserved |
... | ... |
@@ -791,7 +797,9 @@ static int encode_frame(AVCodecContext *avctx, AVPacket *pkt, |
791 | 791 |
} |
792 | 792 |
bytestream_put_be16 (&tmp, buf - orig_buf); // write back frame header size |
793 | 793 |
|
794 |
- for (ctx->cur_picture_idx = 0; ctx->cur_picture_idx < ctx->pictures_per_frame; ++ctx->cur_picture_idx) { |
|
794 |
+ for (ctx->cur_picture_idx = 0; |
|
795 |
+ ctx->cur_picture_idx < ctx->pictures_per_frame; |
|
796 |
+ ctx->cur_picture_idx++) { |
|
795 | 797 |
// picture header |
796 | 798 |
picture_size_pos = buf + 1; |
797 | 799 |
bytestream_put_byte (&buf, 0x40); // picture header size (in bits) |
... | ... |
@@ -845,7 +853,6 @@ static int encode_frame(AVCodecContext *avctx, AVPacket *pkt, |
845 | 845 |
frame_size = buf - orig_buf; |
846 | 846 |
bytestream_put_be32(&orig_buf, frame_size); |
847 | 847 |
|
848 |
- |
|
849 | 848 |
pkt->size = frame_size; |
850 | 849 |
pkt->flags |= AV_PKT_FLAG_KEY; |
851 | 850 |
*got_packet = 1; |
... | ... |
@@ -927,7 +934,8 @@ static av_cold int encode_init(AVCodecContext *avctx) |
927 | 927 |
if (!ctx->force_quant) { |
928 | 928 |
if (!ctx->bits_per_mb) { |
929 | 929 |
for (i = 0; i < NUM_MB_LIMITS - 1; i++) |
930 |
- if (prores_mb_limits[i] >= ctx->mb_width * ctx->mb_height * ctx->pictures_per_frame) |
|
930 |
+ if (prores_mb_limits[i] >= ctx->mb_width * ctx->mb_height * |
|
931 |
+ ctx->pictures_per_frame) |
|
931 | 932 |
break; |
932 | 933 |
ctx->bits_per_mb = ctx->profile_info->br_tab[i]; |
933 | 934 |
} else if (ctx->bits_per_mb < 128) { |
... | ... |
@@ -991,12 +999,15 @@ static av_cold int encode_init(AVCodecContext *avctx) |
991 | 991 |
ctx->frame_size_upper_bound = ctx->pictures_per_frame * |
992 | 992 |
ctx->slices_per_picture * |
993 | 993 |
(2 + 2 * ctx->num_planes + |
994 |
- (mps * ctx->bits_per_mb) / 8) + 200; |
|
994 |
+ (mps * ctx->bits_per_mb) / 8) |
|
995 |
+ + 200; |
|
995 | 996 |
|
996 | 997 |
avctx->codec_tag = ctx->profile_info->tag; |
997 | 998 |
|
998 |
- av_log(avctx, AV_LOG_DEBUG, "profile %d, %d slices/pic, %d pics/frame, %d bits per MB\n", |
|
999 |
- ctx->profile, ctx->slices_per_picture, ctx->pictures_per_frame, ctx->bits_per_mb); |
|
999 |
+ av_log(avctx, AV_LOG_DEBUG, |
|
1000 |
+ "profile %d, %d slices, interlacing: %s, %d bits per MB\n", |
|
1001 |
+ ctx->profile, ctx->slices_per_picture * ctx->pictures_per_frame, |
|
1002 |
+ interlaced ? "yes" : "no", ctx->bits_per_mb); |
|
1000 | 1003 |
av_log(avctx, AV_LOG_DEBUG, "frame size upper bound: %d\n", |
1001 | 1004 |
ctx->frame_size_upper_bound); |
1002 | 1005 |
|
... | ... |
@@ -5,7 +5,7 @@ OBJS-$(CONFIG_XMM_CLOBBER_TEST) += x86/w64xmmtest.o |
5 | 5 |
|
6 | 6 |
MMX-OBJS += x86/dsputil_mmx.o \ |
7 | 7 |
x86/fdct_mmx.o \ |
8 |
- x86/fmtconvert_mmx.o \ |
|
8 |
+ x86/fmtconvert_init.o \ |
|
9 | 9 |
x86/idct_mmx_xvid.o \ |
10 | 10 |
x86/idct_sse2_xvid.o \ |
11 | 11 |
x86/motion_est_mmx.o \ |
... | ... |
@@ -13,15 +13,15 @@ MMX-OBJS += x86/dsputil_mmx.o \ |
13 | 13 |
x86/simple_idct_mmx.o \ |
14 | 14 |
|
15 | 15 |
MMX-OBJS-$(CONFIG_AAC_DECODER) += x86/sbrdsp_init.o |
16 |
-MMX-OBJS-$(CONFIG_AC3DSP) += x86/ac3dsp_mmx.o |
|
16 |
+MMX-OBJS-$(CONFIG_AC3DSP) += x86/ac3dsp_init.o |
|
17 | 17 |
MMX-OBJS-$(CONFIG_CAVS_DECODER) += x86/cavsdsp_mmx.o |
18 | 18 |
MMX-OBJS-$(CONFIG_DNXHD_ENCODER) += x86/dnxhd_mmx.o |
19 | 19 |
MMX-OBJS-$(CONFIG_DWT) += x86/snowdsp_mmx.o \ |
20 | 20 |
x86/dwt.o |
21 | 21 |
MMX-OBJS-$(CONFIG_ENCODERS) += x86/dsputilenc_mmx.o |
22 |
-MMX-OBJS-$(CONFIG_FFT) += x86/fft.o |
|
22 |
+MMX-OBJS-$(CONFIG_FFT) += x86/fft_init.o |
|
23 | 23 |
MMX-OBJS-$(CONFIG_GPL) += x86/idct_mmx.o |
24 |
-MMX-OBJS-$(CONFIG_H264DSP) += x86/h264dsp_mmx.o |
|
24 |
+MMX-OBJS-$(CONFIG_H264DSP) += x86/h264dsp_init.o |
|
25 | 25 |
MMX-OBJS-$(CONFIG_H264PRED) += x86/h264_intrapred_init.o |
26 | 26 |
MMX-OBJS-$(CONFIG_LPC) += x86/lpc_mmx.o |
27 | 27 |
MMX-OBJS-$(CONFIG_MPEGAUDIODSP) += x86/mpegaudiodec_mmx.o |
28 | 28 |
new file mode 100644 |
... | ... |
@@ -0,0 +1,93 @@ |
0 |
+/* |
|
1 |
+ * x86-optimized AC-3 DSP utils |
|
2 |
+ * Copyright (c) 2011 Justin Ruggles |
|
3 |
+ * |
|
4 |
+ * This file is part of FFmpeg. |
|
5 |
+ * |
|
6 |
+ * FFmpeg is free software; you can redistribute it and/or |
|
7 |
+ * modify it under the terms of the GNU Lesser General Public |
|
8 |
+ * License as published by the Free Software Foundation; either |
|
9 |
+ * version 2.1 of the License, or (at your option) any later version. |
|
10 |
+ * |
|
11 |
+ * FFmpeg is distributed in the hope that it will be useful, |
|
12 |
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
13 |
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
14 |
+ * Lesser General Public License for more details. |
|
15 |
+ * |
|
16 |
+ * You should have received a copy of the GNU Lesser General Public |
|
17 |
+ * License along with FFmpeg; if not, write to the Free Software |
|
18 |
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
19 |
+ */ |
|
20 |
+ |
|
21 |
+#include "libavutil/x86/asm.h" |
|
22 |
+#include "dsputil_mmx.h" |
|
23 |
+#include "libavcodec/ac3dsp.h" |
|
24 |
+ |
|
25 |
+extern void ff_ac3_exponent_min_mmx (uint8_t *exp, int num_reuse_blocks, int nb_coefs); |
|
26 |
+extern void ff_ac3_exponent_min_mmxext(uint8_t *exp, int num_reuse_blocks, int nb_coefs); |
|
27 |
+extern void ff_ac3_exponent_min_sse2 (uint8_t *exp, int num_reuse_blocks, int nb_coefs); |
|
28 |
+ |
|
29 |
+extern int ff_ac3_max_msb_abs_int16_mmx (const int16_t *src, int len); |
|
30 |
+extern int ff_ac3_max_msb_abs_int16_mmx2 (const int16_t *src, int len); |
|
31 |
+extern int ff_ac3_max_msb_abs_int16_sse2 (const int16_t *src, int len); |
|
32 |
+extern int ff_ac3_max_msb_abs_int16_ssse3(const int16_t *src, int len); |
|
33 |
+ |
|
34 |
+extern void ff_ac3_lshift_int16_mmx (int16_t *src, unsigned int len, unsigned int shift); |
|
35 |
+extern void ff_ac3_lshift_int16_sse2(int16_t *src, unsigned int len, unsigned int shift); |
|
36 |
+ |
|
37 |
+extern void ff_ac3_rshift_int32_mmx (int32_t *src, unsigned int len, unsigned int shift); |
|
38 |
+extern void ff_ac3_rshift_int32_sse2(int32_t *src, unsigned int len, unsigned int shift); |
|
39 |
+ |
|
40 |
+extern void ff_float_to_fixed24_3dnow(int32_t *dst, const float *src, unsigned int len); |
|
41 |
+extern void ff_float_to_fixed24_sse (int32_t *dst, const float *src, unsigned int len); |
|
42 |
+extern void ff_float_to_fixed24_sse2 (int32_t *dst, const float *src, unsigned int len); |
|
43 |
+ |
|
44 |
+extern int ff_ac3_compute_mantissa_size_sse2(uint16_t mant_cnt[6][16]); |
|
45 |
+ |
|
46 |
+extern void ff_ac3_extract_exponents_3dnow(uint8_t *exp, int32_t *coef, int nb_coefs); |
|
47 |
+extern void ff_ac3_extract_exponents_sse2 (uint8_t *exp, int32_t *coef, int nb_coefs); |
|
48 |
+extern void ff_ac3_extract_exponents_ssse3(uint8_t *exp, int32_t *coef, int nb_coefs); |
|
49 |
+ |
|
50 |
+av_cold void ff_ac3dsp_init_x86(AC3DSPContext *c, int bit_exact) |
|
51 |
+{ |
|
52 |
+#if HAVE_YASM |
|
53 |
+ int mm_flags = av_get_cpu_flags(); |
|
54 |
+ |
|
55 |
+ if (mm_flags & AV_CPU_FLAG_MMX) { |
|
56 |
+ c->ac3_exponent_min = ff_ac3_exponent_min_mmx; |
|
57 |
+ c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_mmx; |
|
58 |
+ c->ac3_lshift_int16 = ff_ac3_lshift_int16_mmx; |
|
59 |
+ c->ac3_rshift_int32 = ff_ac3_rshift_int32_mmx; |
|
60 |
+ } |
|
61 |
+ if (mm_flags & AV_CPU_FLAG_3DNOW && HAVE_AMD3DNOW) { |
|
62 |
+ c->extract_exponents = ff_ac3_extract_exponents_3dnow; |
|
63 |
+ if (!bit_exact) { |
|
64 |
+ c->float_to_fixed24 = ff_float_to_fixed24_3dnow; |
|
65 |
+ } |
|
66 |
+ } |
|
67 |
+ if (mm_flags & AV_CPU_FLAG_MMXEXT && HAVE_MMXEXT) { |
|
68 |
+ c->ac3_exponent_min = ff_ac3_exponent_min_mmxext; |
|
69 |
+ c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_mmx2; |
|
70 |
+ } |
|
71 |
+ if (mm_flags & AV_CPU_FLAG_SSE && HAVE_SSE) { |
|
72 |
+ c->float_to_fixed24 = ff_float_to_fixed24_sse; |
|
73 |
+ } |
|
74 |
+ if (mm_flags & AV_CPU_FLAG_SSE2 && HAVE_SSE) { |
|
75 |
+ c->ac3_exponent_min = ff_ac3_exponent_min_sse2; |
|
76 |
+ c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_sse2; |
|
77 |
+ c->float_to_fixed24 = ff_float_to_fixed24_sse2; |
|
78 |
+ c->compute_mantissa_size = ff_ac3_compute_mantissa_size_sse2; |
|
79 |
+ c->extract_exponents = ff_ac3_extract_exponents_sse2; |
|
80 |
+ if (!(mm_flags & AV_CPU_FLAG_SSE2SLOW)) { |
|
81 |
+ c->ac3_lshift_int16 = ff_ac3_lshift_int16_sse2; |
|
82 |
+ c->ac3_rshift_int32 = ff_ac3_rshift_int32_sse2; |
|
83 |
+ } |
|
84 |
+ } |
|
85 |
+ if (mm_flags & AV_CPU_FLAG_SSSE3 && HAVE_SSSE3) { |
|
86 |
+ c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_ssse3; |
|
87 |
+ if (!(mm_flags & AV_CPU_FLAG_ATOM)) { |
|
88 |
+ c->extract_exponents = ff_ac3_extract_exponents_ssse3; |
|
89 |
+ } |
|
90 |
+ } |
|
91 |
+#endif |
|
92 |
+} |
0 | 93 |
deleted file mode 100644 |
... | ... |
@@ -1,93 +0,0 @@ |
1 |
-/* |
|
2 |
- * x86-optimized AC-3 DSP utils |
|
3 |
- * Copyright (c) 2011 Justin Ruggles |
|
4 |
- * |
|
5 |
- * This file is part of FFmpeg. |
|
6 |
- * |
|
7 |
- * FFmpeg is free software; you can redistribute it and/or |
|
8 |
- * modify it under the terms of the GNU Lesser General Public |
|
9 |
- * License as published by the Free Software Foundation; either |
|
10 |
- * version 2.1 of the License, or (at your option) any later version. |
|
11 |
- * |
|
12 |
- * FFmpeg is distributed in the hope that it will be useful, |
|
13 |
- * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
14 |
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
15 |
- * Lesser General Public License for more details. |
|
16 |
- * |
|
17 |
- * You should have received a copy of the GNU Lesser General Public |
|
18 |
- * License along with FFmpeg; if not, write to the Free Software |
|
19 |
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
20 |
- */ |
|
21 |
- |
|
22 |
-#include "libavutil/x86/asm.h" |
|
23 |
-#include "dsputil_mmx.h" |
|
24 |
-#include "libavcodec/ac3dsp.h" |
|
25 |
- |
|
26 |
-extern void ff_ac3_exponent_min_mmx (uint8_t *exp, int num_reuse_blocks, int nb_coefs); |
|
27 |
-extern void ff_ac3_exponent_min_mmxext(uint8_t *exp, int num_reuse_blocks, int nb_coefs); |
|
28 |
-extern void ff_ac3_exponent_min_sse2 (uint8_t *exp, int num_reuse_blocks, int nb_coefs); |
|
29 |
- |
|
30 |
-extern int ff_ac3_max_msb_abs_int16_mmx (const int16_t *src, int len); |
|
31 |
-extern int ff_ac3_max_msb_abs_int16_mmx2 (const int16_t *src, int len); |
|
32 |
-extern int ff_ac3_max_msb_abs_int16_sse2 (const int16_t *src, int len); |
|
33 |
-extern int ff_ac3_max_msb_abs_int16_ssse3(const int16_t *src, int len); |
|
34 |
- |
|
35 |
-extern void ff_ac3_lshift_int16_mmx (int16_t *src, unsigned int len, unsigned int shift); |
|
36 |
-extern void ff_ac3_lshift_int16_sse2(int16_t *src, unsigned int len, unsigned int shift); |
|
37 |
- |
|
38 |
-extern void ff_ac3_rshift_int32_mmx (int32_t *src, unsigned int len, unsigned int shift); |
|
39 |
-extern void ff_ac3_rshift_int32_sse2(int32_t *src, unsigned int len, unsigned int shift); |
|
40 |
- |
|
41 |
-extern void ff_float_to_fixed24_3dnow(int32_t *dst, const float *src, unsigned int len); |
|
42 |
-extern void ff_float_to_fixed24_sse (int32_t *dst, const float *src, unsigned int len); |
|
43 |
-extern void ff_float_to_fixed24_sse2 (int32_t *dst, const float *src, unsigned int len); |
|
44 |
- |
|
45 |
-extern int ff_ac3_compute_mantissa_size_sse2(uint16_t mant_cnt[6][16]); |
|
46 |
- |
|
47 |
-extern void ff_ac3_extract_exponents_3dnow(uint8_t *exp, int32_t *coef, int nb_coefs); |
|
48 |
-extern void ff_ac3_extract_exponents_sse2 (uint8_t *exp, int32_t *coef, int nb_coefs); |
|
49 |
-extern void ff_ac3_extract_exponents_ssse3(uint8_t *exp, int32_t *coef, int nb_coefs); |
|
50 |
- |
|
51 |
-av_cold void ff_ac3dsp_init_x86(AC3DSPContext *c, int bit_exact) |
|
52 |
-{ |
|
53 |
-#if HAVE_YASM |
|
54 |
- int mm_flags = av_get_cpu_flags(); |
|
55 |
- |
|
56 |
- if (mm_flags & AV_CPU_FLAG_MMX) { |
|
57 |
- c->ac3_exponent_min = ff_ac3_exponent_min_mmx; |
|
58 |
- c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_mmx; |
|
59 |
- c->ac3_lshift_int16 = ff_ac3_lshift_int16_mmx; |
|
60 |
- c->ac3_rshift_int32 = ff_ac3_rshift_int32_mmx; |
|
61 |
- } |
|
62 |
- if (mm_flags & AV_CPU_FLAG_3DNOW && HAVE_AMD3DNOW) { |
|
63 |
- c->extract_exponents = ff_ac3_extract_exponents_3dnow; |
|
64 |
- if (!bit_exact) { |
|
65 |
- c->float_to_fixed24 = ff_float_to_fixed24_3dnow; |
|
66 |
- } |
|
67 |
- } |
|
68 |
- if (mm_flags & AV_CPU_FLAG_MMXEXT && HAVE_MMXEXT) { |
|
69 |
- c->ac3_exponent_min = ff_ac3_exponent_min_mmxext; |
|
70 |
- c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_mmx2; |
|
71 |
- } |
|
72 |
- if (mm_flags & AV_CPU_FLAG_SSE && HAVE_SSE) { |
|
73 |
- c->float_to_fixed24 = ff_float_to_fixed24_sse; |
|
74 |
- } |
|
75 |
- if (mm_flags & AV_CPU_FLAG_SSE2 && HAVE_SSE) { |
|
76 |
- c->ac3_exponent_min = ff_ac3_exponent_min_sse2; |
|
77 |
- c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_sse2; |
|
78 |
- c->float_to_fixed24 = ff_float_to_fixed24_sse2; |
|
79 |
- c->compute_mantissa_size = ff_ac3_compute_mantissa_size_sse2; |
|
80 |
- c->extract_exponents = ff_ac3_extract_exponents_sse2; |
|
81 |
- if (!(mm_flags & AV_CPU_FLAG_SSE2SLOW)) { |
|
82 |
- c->ac3_lshift_int16 = ff_ac3_lshift_int16_sse2; |
|
83 |
- c->ac3_rshift_int32 = ff_ac3_rshift_int32_sse2; |
|
84 |
- } |
|
85 |
- } |
|
86 |
- if (mm_flags & AV_CPU_FLAG_SSSE3 && HAVE_SSSE3) { |
|
87 |
- c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_ssse3; |
|
88 |
- if (!(mm_flags & AV_CPU_FLAG_ATOM)) { |
|
89 |
- c->extract_exponents = ff_ac3_extract_exponents_ssse3; |
|
90 |
- } |
|
91 |
- } |
|
92 |
-#endif |
|
93 |
-} |
94 | 1 |
deleted file mode 100644 |
... | ... |
@@ -1,72 +0,0 @@ |
1 |
-/* |
|
2 |
- * This file is part of FFmpeg. |
|
3 |
- * |
|
4 |
- * FFmpeg is free software; you can redistribute it and/or |
|
5 |
- * modify it under the terms of the GNU Lesser General Public |
|
6 |
- * License as published by the Free Software Foundation; either |
|
7 |
- * version 2.1 of the License, or (at your option) any later version. |
|
8 |
- * |
|
9 |
- * FFmpeg is distributed in the hope that it will be useful, |
|
10 |
- * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
11 |
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
12 |
- * Lesser General Public License for more details. |
|
13 |
- * |
|
14 |
- * You should have received a copy of the GNU Lesser General Public |
|
15 |
- * License along with FFmpeg; if not, write to the Free Software |
|
16 |
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
17 |
- */ |
|
18 |
- |
|
19 |
-#include "libavutil/cpu.h" |
|
20 |
-#include "libavcodec/dsputil.h" |
|
21 |
-#include "libavcodec/dct.h" |
|
22 |
-#include "fft.h" |
|
23 |
- |
|
24 |
-av_cold void ff_fft_init_mmx(FFTContext *s) |
|
25 |
-{ |
|
26 |
-#if HAVE_YASM |
|
27 |
- int has_vectors = av_get_cpu_flags(); |
|
28 |
-#if ARCH_X86_32 |
|
29 |
- if (has_vectors & AV_CPU_FLAG_3DNOW && HAVE_AMD3DNOW) { |
|
30 |
- /* 3DNow! for K6-2/3 */ |
|
31 |
- s->imdct_calc = ff_imdct_calc_3dnow; |
|
32 |
- s->imdct_half = ff_imdct_half_3dnow; |
|
33 |
- s->fft_calc = ff_fft_calc_3dnow; |
|
34 |
- } |
|
35 |
- if (has_vectors & AV_CPU_FLAG_3DNOWEXT && HAVE_AMD3DNOWEXT) { |
|
36 |
- /* 3DNowEx for K7 */ |
|
37 |
- s->imdct_calc = ff_imdct_calc_3dnowext; |
|
38 |
- s->imdct_half = ff_imdct_half_3dnowext; |
|
39 |
- s->fft_calc = ff_fft_calc_3dnowext; |
|
40 |
- } |
|
41 |
-#endif |
|
42 |
- if (has_vectors & AV_CPU_FLAG_SSE && HAVE_SSE) { |
|
43 |
- /* SSE for P3/P4/K8 */ |
|
44 |
- s->imdct_calc = ff_imdct_calc_sse; |
|
45 |
- s->imdct_half = ff_imdct_half_sse; |
|
46 |
- s->fft_permute = ff_fft_permute_sse; |
|
47 |
- s->fft_calc = ff_fft_calc_sse; |
|
48 |
- s->fft_permutation = FF_FFT_PERM_SWAP_LSBS; |
|
49 |
- } |
|
50 |
- if (has_vectors & AV_CPU_FLAG_AVX && HAVE_AVX && s->nbits >= 5) { |
|
51 |
- /* AVX for SB */ |
|
52 |
- s->imdct_half = ff_imdct_half_avx; |
|
53 |
- s->fft_calc = ff_fft_calc_avx; |
|
54 |
- s->fft_permutation = FF_FFT_PERM_AVX; |
|
55 |
- } |
|
56 |
-#endif |
|
57 |
-} |
|
58 |
- |
|
59 |
-#if CONFIG_DCT |
|
60 |
-av_cold void ff_dct_init_mmx(DCTContext *s) |
|
61 |
-{ |
|
62 |
-#if HAVE_YASM |
|
63 |
- int has_vectors = av_get_cpu_flags(); |
|
64 |
- if (has_vectors & AV_CPU_FLAG_SSE && HAVE_SSE) |
|
65 |
- s->dct32 = ff_dct32_float_sse; |
|
66 |
- if (has_vectors & AV_CPU_FLAG_SSE2 && HAVE_SSE) |
|
67 |
- s->dct32 = ff_dct32_float_sse2; |
|
68 |
- if (has_vectors & AV_CPU_FLAG_AVX && HAVE_AVX) |
|
69 |
- s->dct32 = ff_dct32_float_avx; |
|
70 |
-#endif |
|
71 |
-} |
|
72 |
-#endif |
73 | 1 |
new file mode 100644 |
... | ... |
@@ -0,0 +1,72 @@ |
0 |
+/* |
|
1 |
+ * This file is part of FFmpeg. |
|
2 |
+ * |
|
3 |
+ * FFmpeg is free software; you can redistribute it and/or |
|
4 |
+ * modify it under the terms of the GNU Lesser General Public |
|
5 |
+ * License as published by the Free Software Foundation; either |
|
6 |
+ * version 2.1 of the License, or (at your option) any later version. |
|
7 |
+ * |
|
8 |
+ * FFmpeg is distributed in the hope that it will be useful, |
|
9 |
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
10 |
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
11 |
+ * Lesser General Public License for more details. |
|
12 |
+ * |
|
13 |
+ * You should have received a copy of the GNU Lesser General Public |
|
14 |
+ * License along with FFmpeg; if not, write to the Free Software |
|
15 |
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
16 |
+ */ |
|
17 |
+ |
|
18 |
+#include "libavutil/cpu.h" |
|
19 |
+#include "libavcodec/dsputil.h" |
|
20 |
+#include "libavcodec/dct.h" |
|
21 |
+#include "fft.h" |
|
22 |
+ |
|
23 |
+av_cold void ff_fft_init_mmx(FFTContext *s) |
|
24 |
+{ |
|
25 |
+#if HAVE_YASM |
|
26 |
+ int has_vectors = av_get_cpu_flags(); |
|
27 |
+#if ARCH_X86_32 |
|
28 |
+ if (has_vectors & AV_CPU_FLAG_3DNOW && HAVE_AMD3DNOW) { |
|
29 |
+ /* 3DNow! for K6-2/3 */ |
|
30 |
+ s->imdct_calc = ff_imdct_calc_3dnow; |
|
31 |
+ s->imdct_half = ff_imdct_half_3dnow; |
|
32 |
+ s->fft_calc = ff_fft_calc_3dnow; |
|
33 |
+ } |
|
34 |
+ if (has_vectors & AV_CPU_FLAG_3DNOWEXT && HAVE_AMD3DNOWEXT) { |
|
35 |
+ /* 3DNowEx for K7 */ |
|
36 |
+ s->imdct_calc = ff_imdct_calc_3dnowext; |
|
37 |
+ s->imdct_half = ff_imdct_half_3dnowext; |
|
38 |
+ s->fft_calc = ff_fft_calc_3dnowext; |
|
39 |
+ } |
|
40 |
+#endif |
|
41 |
+ if (has_vectors & AV_CPU_FLAG_SSE && HAVE_SSE) { |
|
42 |
+ /* SSE for P3/P4/K8 */ |
|
43 |
+ s->imdct_calc = ff_imdct_calc_sse; |
|
44 |
+ s->imdct_half = ff_imdct_half_sse; |
|
45 |
+ s->fft_permute = ff_fft_permute_sse; |
|
46 |
+ s->fft_calc = ff_fft_calc_sse; |
|
47 |
+ s->fft_permutation = FF_FFT_PERM_SWAP_LSBS; |
|
48 |
+ } |
|
49 |
+ if (has_vectors & AV_CPU_FLAG_AVX && HAVE_AVX && s->nbits >= 5) { |
|
50 |
+ /* AVX for SB */ |
|
51 |
+ s->imdct_half = ff_imdct_half_avx; |
|
52 |
+ s->fft_calc = ff_fft_calc_avx; |
|
53 |
+ s->fft_permutation = FF_FFT_PERM_AVX; |
|
54 |
+ } |
|
55 |
+#endif |
|
56 |
+} |
|
57 |
+ |
|
58 |
+#if CONFIG_DCT |
|
59 |
+av_cold void ff_dct_init_mmx(DCTContext *s) |
|
60 |
+{ |
|
61 |
+#if HAVE_YASM |
|
62 |
+ int has_vectors = av_get_cpu_flags(); |
|
63 |
+ if (has_vectors & AV_CPU_FLAG_SSE && HAVE_SSE) |
|
64 |
+ s->dct32 = ff_dct32_float_sse; |
|
65 |
+ if (has_vectors & AV_CPU_FLAG_SSE2 && HAVE_SSE) |
|
66 |
+ s->dct32 = ff_dct32_float_sse2; |
|
67 |
+ if (has_vectors & AV_CPU_FLAG_AVX && HAVE_AVX) |
|
68 |
+ s->dct32 = ff_dct32_float_avx; |
|
69 |
+#endif |
|
70 |
+} |
|
71 |
+#endif |
0 | 72 |
new file mode 100644 |
... | ... |
@@ -0,0 +1,147 @@ |
0 |
+/* |
|
1 |
+ * Format Conversion Utils |
|
2 |
+ * Copyright (c) 2000, 2001 Fabrice Bellard |
|
3 |
+ * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at> |
|
4 |
+ * |
|
5 |
+ * This file is part of FFmpeg. |
|
6 |
+ * |
|
7 |
+ * FFmpeg is free software; you can redistribute it and/or |
|
8 |
+ * modify it under the terms of the GNU Lesser General Public |
|
9 |
+ * License as published by the Free Software Foundation; either |
|
10 |
+ * version 2.1 of the License, or (at your option) any later version. |
|
11 |
+ * |
|
12 |
+ * FFmpeg is distributed in the hope that it will be useful, |
|
13 |
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
14 |
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
15 |
+ * Lesser General Public License for more details. |
|
16 |
+ * |
|
17 |
+ * You should have received a copy of the GNU Lesser General Public |
|
18 |
+ * License along with FFmpeg; if not, write to the Free Software |
|
19 |
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
20 |
+ * |
|
21 |
+ * MMX optimization by Nick Kurshev <nickols_k@mail.ru> |
|
22 |
+ */ |
|
23 |
+ |
|
24 |
+#include "libavutil/cpu.h" |
|
25 |
+#include "libavutil/x86/asm.h" |
|
26 |
+#include "libavcodec/fmtconvert.h" |
|
27 |
+#include "libavcodec/dsputil.h" |
|
28 |
+ |
|
29 |
+#if HAVE_YASM |
|
30 |
+ |
|
31 |
+void ff_int32_to_float_fmul_scalar_sse (float *dst, const int *src, float mul, int len); |
|
32 |
+void ff_int32_to_float_fmul_scalar_sse2(float *dst, const int *src, float mul, int len); |
|
33 |
+ |
|
34 |
+void ff_float_to_int16_3dnow(int16_t *dst, const float *src, long len); |
|
35 |
+void ff_float_to_int16_sse (int16_t *dst, const float *src, long len); |
|
36 |
+void ff_float_to_int16_sse2 (int16_t *dst, const float *src, long len); |
|
37 |
+ |
|
38 |
+void ff_float_to_int16_step_3dnow(int16_t *dst, const float *src, long len, long step); |
|
39 |
+void ff_float_to_int16_step_sse (int16_t *dst, const float *src, long len, long step); |
|
40 |
+void ff_float_to_int16_step_sse2 (int16_t *dst, const float *src, long len, long step); |
|
41 |
+ |
|
42 |
+void ff_float_to_int16_interleave2_3dnow(int16_t *dst, const float **src, long len); |
|
43 |
+void ff_float_to_int16_interleave2_sse (int16_t *dst, const float **src, long len); |
|
44 |
+void ff_float_to_int16_interleave2_sse2 (int16_t *dst, const float **src, long len); |
|
45 |
+ |
|
46 |
+void ff_float_to_int16_interleave6_sse(int16_t *dst, const float **src, int len); |
|
47 |
+void ff_float_to_int16_interleave6_3dnow(int16_t *dst, const float **src, int len); |
|
48 |
+void ff_float_to_int16_interleave6_3dnowext(int16_t *dst, const float **src, int len); |
|
49 |
+ |
|
50 |
+#define ff_float_to_int16_interleave6_sse2 ff_float_to_int16_interleave6_sse |
|
51 |
+ |
|
52 |
+#define FLOAT_TO_INT16_INTERLEAVE(cpu) \ |
|
53 |
+/* gcc pessimizes register allocation if this is in the same function as float_to_int16_interleave_sse2*/\ |
|
54 |
+static av_noinline void float_to_int16_interleave_misc_##cpu(int16_t *dst, const float **src, long len, int channels){\ |
|
55 |
+ int c;\ |
|
56 |
+ for(c=0; c<channels; c++){\ |
|
57 |
+ ff_float_to_int16_step_##cpu(dst+c, src[c], len, channels);\ |
|
58 |
+ }\ |
|
59 |
+}\ |
|
60 |
+\ |
|
61 |
+static void float_to_int16_interleave_##cpu(int16_t *dst, const float **src, long len, int channels){\ |
|
62 |
+ if(channels==1)\ |
|
63 |
+ ff_float_to_int16_##cpu(dst, src[0], len);\ |
|
64 |
+ else if(channels==2){\ |
|
65 |
+ ff_float_to_int16_interleave2_##cpu(dst, src, len);\ |
|
66 |
+ }else if(channels==6){\ |
|
67 |
+ ff_float_to_int16_interleave6_##cpu(dst, src, len);\ |
|
68 |
+ }else\ |
|
69 |
+ float_to_int16_interleave_misc_##cpu(dst, src, len, channels);\ |
|
70 |
+} |
|
71 |
+ |
|
72 |
+FLOAT_TO_INT16_INTERLEAVE(3dnow) |
|
73 |
+FLOAT_TO_INT16_INTERLEAVE(sse) |
|
74 |
+FLOAT_TO_INT16_INTERLEAVE(sse2) |
|
75 |
+ |
|
76 |
+static void float_to_int16_interleave_3dnowext(int16_t *dst, const float **src, |
|
77 |
+ long len, int channels) |
|
78 |
+{ |
|
79 |
+ if(channels==6) |
|
80 |
+ ff_float_to_int16_interleave6_3dnowext(dst, src, len); |
|
81 |
+ else |
|
82 |
+ float_to_int16_interleave_3dnow(dst, src, len, channels); |
|
83 |
+} |
|
84 |
+ |
|
85 |
+void ff_float_interleave2_mmx(float *dst, const float **src, unsigned int len); |
|
86 |
+void ff_float_interleave2_sse(float *dst, const float **src, unsigned int len); |
|
87 |
+ |
|
88 |
+void ff_float_interleave6_mmx(float *dst, const float **src, unsigned int len); |
|
89 |
+void ff_float_interleave6_sse(float *dst, const float **src, unsigned int len); |
|
90 |
+ |
|
91 |
+static void float_interleave_mmx(float *dst, const float **src, |
|
92 |
+ unsigned int len, int channels) |
|
93 |
+{ |
|
94 |
+ if (channels == 2) { |
|
95 |
+ ff_float_interleave2_mmx(dst, src, len); |
|
96 |
+ } else if (channels == 6) |
|
97 |
+ ff_float_interleave6_mmx(dst, src, len); |
|
98 |
+ else |
|
99 |
+ ff_float_interleave_c(dst, src, len, channels); |
|
100 |
+} |
|
101 |
+ |
|
102 |
+static void float_interleave_sse(float *dst, const float **src, |
|
103 |
+ unsigned int len, int channels) |
|
104 |
+{ |
|
105 |
+ if (channels == 2) { |
|
106 |
+ ff_float_interleave2_sse(dst, src, len); |
|
107 |
+ } else if (channels == 6) |
|
108 |
+ ff_float_interleave6_sse(dst, src, len); |
|
109 |
+ else |
|
110 |
+ ff_float_interleave_c(dst, src, len, channels); |
|
111 |
+} |
|
112 |
+#endif |
|
113 |
+ |
|
114 |
+void ff_fmt_convert_init_x86(FmtConvertContext *c, AVCodecContext *avctx) |
|
115 |
+{ |
|
116 |
+#if HAVE_YASM |
|
117 |
+ int mm_flags = av_get_cpu_flags(); |
|
118 |
+ |
|
119 |
+ if (mm_flags & AV_CPU_FLAG_MMX) { |
|
120 |
+ c->float_interleave = float_interleave_mmx; |
|
121 |
+ |
|
122 |
+ if (HAVE_AMD3DNOW && mm_flags & AV_CPU_FLAG_3DNOW) { |
|
123 |
+ if(!(avctx->flags & CODEC_FLAG_BITEXACT)){ |
|
124 |
+ c->float_to_int16 = ff_float_to_int16_3dnow; |
|
125 |
+ c->float_to_int16_interleave = float_to_int16_interleave_3dnow; |
|
126 |
+ } |
|
127 |
+ } |
|
128 |
+ if (HAVE_AMD3DNOWEXT && mm_flags & AV_CPU_FLAG_3DNOWEXT) { |
|
129 |
+ if(!(avctx->flags & CODEC_FLAG_BITEXACT)){ |
|
130 |
+ c->float_to_int16_interleave = float_to_int16_interleave_3dnowext; |
|
131 |
+ } |
|
132 |
+ } |
|
133 |
+ if (HAVE_SSE && mm_flags & AV_CPU_FLAG_SSE) { |
|
134 |
+ c->int32_to_float_fmul_scalar = ff_int32_to_float_fmul_scalar_sse; |
|
135 |
+ c->float_to_int16 = ff_float_to_int16_sse; |
|
136 |
+ c->float_to_int16_interleave = float_to_int16_interleave_sse; |
|
137 |
+ c->float_interleave = float_interleave_sse; |
|
138 |
+ } |
|
139 |
+ if (HAVE_SSE && mm_flags & AV_CPU_FLAG_SSE2) { |
|
140 |
+ c->int32_to_float_fmul_scalar = ff_int32_to_float_fmul_scalar_sse2; |
|
141 |
+ c->float_to_int16 = ff_float_to_int16_sse2; |
|
142 |
+ c->float_to_int16_interleave = float_to_int16_interleave_sse2; |
|
143 |
+ } |
|
144 |
+ } |
|
145 |
+#endif |
|
146 |
+} |
0 | 147 |
deleted file mode 100644 |
... | ... |
@@ -1,147 +0,0 @@ |
1 |
-/* |
|
2 |
- * Format Conversion Utils |
|
3 |
- * Copyright (c) 2000, 2001 Fabrice Bellard |
|
4 |
- * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at> |
|
5 |
- * |
|
6 |
- * This file is part of FFmpeg. |
|
7 |
- * |
|
8 |
- * FFmpeg is free software; you can redistribute it and/or |
|
9 |
- * modify it under the terms of the GNU Lesser General Public |
|
10 |
- * License as published by the Free Software Foundation; either |
|
11 |
- * version 2.1 of the License, or (at your option) any later version. |
|
12 |
- * |
|
13 |
- * FFmpeg is distributed in the hope that it will be useful, |
|
14 |
- * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
15 |
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
16 |
- * Lesser General Public License for more details. |
|
17 |
- * |
|
18 |
- * You should have received a copy of the GNU Lesser General Public |
|
19 |
- * License along with FFmpeg; if not, write to the Free Software |
|
20 |
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
21 |
- * |
|
22 |
- * MMX optimization by Nick Kurshev <nickols_k@mail.ru> |
|
23 |
- */ |
|
24 |
- |
|
25 |
-#include "libavutil/cpu.h" |
|
26 |
-#include "libavutil/x86/asm.h" |
|
27 |
-#include "libavcodec/fmtconvert.h" |
|
28 |
-#include "libavcodec/dsputil.h" |
|
29 |
- |
|
30 |
-#if HAVE_YASM |
|
31 |
- |
|
32 |
-void ff_int32_to_float_fmul_scalar_sse (float *dst, const int *src, float mul, int len); |
|
33 |
-void ff_int32_to_float_fmul_scalar_sse2(float *dst, const int *src, float mul, int len); |
|
34 |
- |
|
35 |
-void ff_float_to_int16_3dnow(int16_t *dst, const float *src, long len); |
|
36 |
-void ff_float_to_int16_sse (int16_t *dst, const float *src, long len); |
|
37 |
-void ff_float_to_int16_sse2 (int16_t *dst, const float *src, long len); |
|
38 |
- |
|
39 |
-void ff_float_to_int16_step_3dnow(int16_t *dst, const float *src, long len, long step); |
|
40 |
-void ff_float_to_int16_step_sse (int16_t *dst, const float *src, long len, long step); |
|
41 |
-void ff_float_to_int16_step_sse2 (int16_t *dst, const float *src, long len, long step); |
|
42 |
- |
|
43 |
-void ff_float_to_int16_interleave2_3dnow(int16_t *dst, const float **src, long len); |
|
44 |
-void ff_float_to_int16_interleave2_sse (int16_t *dst, const float **src, long len); |
|
45 |
-void ff_float_to_int16_interleave2_sse2 (int16_t *dst, const float **src, long len); |
|
46 |
- |
|
47 |
-void ff_float_to_int16_interleave6_sse(int16_t *dst, const float **src, int len); |
|
48 |
-void ff_float_to_int16_interleave6_3dnow(int16_t *dst, const float **src, int len); |
|
49 |
-void ff_float_to_int16_interleave6_3dnowext(int16_t *dst, const float **src, int len); |
|
50 |
- |
|
51 |
-#define ff_float_to_int16_interleave6_sse2 ff_float_to_int16_interleave6_sse |
|
52 |
- |
|
53 |
-#define FLOAT_TO_INT16_INTERLEAVE(cpu) \ |
|
54 |
-/* gcc pessimizes register allocation if this is in the same function as float_to_int16_interleave_sse2*/\ |
|
55 |
-static av_noinline void float_to_int16_interleave_misc_##cpu(int16_t *dst, const float **src, long len, int channels){\ |
|
56 |
- int c;\ |
|
57 |
- for(c=0; c<channels; c++){\ |
|
58 |
- ff_float_to_int16_step_##cpu(dst+c, src[c], len, channels);\ |
|
59 |
- }\ |
|
60 |
-}\ |
|
61 |
-\ |
|
62 |
-static void float_to_int16_interleave_##cpu(int16_t *dst, const float **src, long len, int channels){\ |
|
63 |
- if(channels==1)\ |
|
64 |
- ff_float_to_int16_##cpu(dst, src[0], len);\ |
|
65 |
- else if(channels==2){\ |
|
66 |
- ff_float_to_int16_interleave2_##cpu(dst, src, len);\ |
|
67 |
- }else if(channels==6){\ |
|
68 |
- ff_float_to_int16_interleave6_##cpu(dst, src, len);\ |
|
69 |
- }else\ |
|
70 |
- float_to_int16_interleave_misc_##cpu(dst, src, len, channels);\ |
|
71 |
-} |
|
72 |
- |
|
73 |
-FLOAT_TO_INT16_INTERLEAVE(3dnow) |
|
74 |
-FLOAT_TO_INT16_INTERLEAVE(sse) |
|
75 |
-FLOAT_TO_INT16_INTERLEAVE(sse2) |
|
76 |
- |
|
77 |
-static void float_to_int16_interleave_3dnowext(int16_t *dst, const float **src, |
|
78 |
- long len, int channels) |
|
79 |
-{ |
|
80 |
- if(channels==6) |
|
81 |
- ff_float_to_int16_interleave6_3dnowext(dst, src, len); |
|
82 |
- else |
|
83 |
- float_to_int16_interleave_3dnow(dst, src, len, channels); |
|
84 |
-} |
|
85 |
- |
|
86 |
-void ff_float_interleave2_mmx(float *dst, const float **src, unsigned int len); |
|
87 |
-void ff_float_interleave2_sse(float *dst, const float **src, unsigned int len); |
|
88 |
- |
|
89 |
-void ff_float_interleave6_mmx(float *dst, const float **src, unsigned int len); |
|
90 |
-void ff_float_interleave6_sse(float *dst, const float **src, unsigned int len); |
|
91 |
- |
|
92 |
-static void float_interleave_mmx(float *dst, const float **src, |
|
93 |
- unsigned int len, int channels) |
|
94 |
-{ |
|
95 |
- if (channels == 2) { |
|
96 |
- ff_float_interleave2_mmx(dst, src, len); |
|
97 |
- } else if (channels == 6) |
|
98 |
- ff_float_interleave6_mmx(dst, src, len); |
|
99 |
- else |
|
100 |
- ff_float_interleave_c(dst, src, len, channels); |
|
101 |
-} |
|
102 |
- |
|
103 |
-static void float_interleave_sse(float *dst, const float **src, |
|
104 |
- unsigned int len, int channels) |
|
105 |
-{ |
|
106 |
- if (channels == 2) { |
|
107 |
- ff_float_interleave2_sse(dst, src, len); |
|
108 |
- } else if (channels == 6) |
|
109 |
- ff_float_interleave6_sse(dst, src, len); |
|
110 |
- else |
|
111 |
- ff_float_interleave_c(dst, src, len, channels); |
|
112 |
-} |
|
113 |
-#endif |
|
114 |
- |
|
115 |
-void ff_fmt_convert_init_x86(FmtConvertContext *c, AVCodecContext *avctx) |
|
116 |
-{ |
|
117 |
-#if HAVE_YASM |
|
118 |
- int mm_flags = av_get_cpu_flags(); |
|
119 |
- |
|
120 |
- if (mm_flags & AV_CPU_FLAG_MMX) { |
|
121 |
- c->float_interleave = float_interleave_mmx; |
|
122 |
- |
|
123 |
- if (HAVE_AMD3DNOW && mm_flags & AV_CPU_FLAG_3DNOW) { |
|
124 |
- if(!(avctx->flags & CODEC_FLAG_BITEXACT)){ |
|
125 |
- c->float_to_int16 = ff_float_to_int16_3dnow; |
|
126 |
- c->float_to_int16_interleave = float_to_int16_interleave_3dnow; |
|
127 |
- } |
|
128 |
- } |
|
129 |
- if (HAVE_AMD3DNOWEXT && mm_flags & AV_CPU_FLAG_3DNOWEXT) { |
|
130 |
- if(!(avctx->flags & CODEC_FLAG_BITEXACT)){ |
|
131 |
- c->float_to_int16_interleave = float_to_int16_interleave_3dnowext; |
|
132 |
- } |
|
133 |
- } |
|
134 |
- if (HAVE_SSE && mm_flags & AV_CPU_FLAG_SSE) { |
|
135 |
- c->int32_to_float_fmul_scalar = ff_int32_to_float_fmul_scalar_sse; |
|
136 |
- c->float_to_int16 = ff_float_to_int16_sse; |
|
137 |
- c->float_to_int16_interleave = float_to_int16_interleave_sse; |
|
138 |
- c->float_interleave = float_interleave_sse; |
|
139 |
- } |
|
140 |
- if (HAVE_SSE && mm_flags & AV_CPU_FLAG_SSE2) { |
|
141 |
- c->int32_to_float_fmul_scalar = ff_int32_to_float_fmul_scalar_sse2; |
|
142 |
- c->float_to_int16 = ff_float_to_int16_sse2; |
|
143 |
- c->float_to_int16_interleave = float_to_int16_interleave_sse2; |
|
144 |
- } |
|
145 |
- } |
|
146 |
-#endif |
|
147 |
-} |
148 | 1 |
new file mode 100644 |
... | ... |
@@ -0,0 +1,385 @@ |
0 |
+/* |
|
1 |
+ * Copyright (c) 2004-2005 Michael Niedermayer, Loren Merritt |
|
2 |
+ * |
|
3 |
+ * This file is part of FFmpeg. |
|
4 |
+ * |
|
5 |
+ * FFmpeg is free software; you can redistribute it and/or |
|
6 |
+ * modify it under the terms of the GNU Lesser General Public |
|
7 |
+ * License as published by the Free Software Foundation; either |
|
8 |
+ * version 2.1 of the License, or (at your option) any later version. |
|
9 |
+ * |
|
10 |
+ * FFmpeg is distributed in the hope that it will be useful, |
|
11 |
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
12 |
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
13 |
+ * Lesser General Public License for more details. |
|
14 |
+ * |
|
15 |
+ * You should have received a copy of the GNU Lesser General Public |
|
16 |
+ * License along with FFmpeg; if not, write to the Free Software |
|
17 |
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
18 |
+ */ |
|
19 |
+ |
|
20 |
+#include "libavutil/cpu.h" |
|
21 |
+#include "libavutil/x86/asm.h" |
|
22 |
+#include "libavcodec/h264dsp.h" |
|
23 |
+#include "dsputil_mmx.h" |
|
24 |
+ |
|
25 |
+/***********************************/ |
|
26 |
+/* IDCT */ |
|
27 |
+#define IDCT_ADD_FUNC(NUM, DEPTH, OPT) \ |
|
28 |
+void ff_h264_idct ## NUM ## _add_ ## DEPTH ## _ ## OPT(uint8_t *dst, \ |
|
29 |
+ int16_t *block, \ |
|
30 |
+ int stride); |
|
31 |
+ |
|
32 |
+IDCT_ADD_FUNC(, 8, mmx) |
|
33 |
+IDCT_ADD_FUNC(, 10, sse2) |
|
34 |
+IDCT_ADD_FUNC(_dc, 8, mmx2) |
|
35 |
+IDCT_ADD_FUNC(_dc, 10, mmx2) |
|
36 |
+IDCT_ADD_FUNC(8_dc, 8, mmx2) |
|
37 |
+IDCT_ADD_FUNC(8_dc, 10, sse2) |
|
38 |
+IDCT_ADD_FUNC(8, 8, mmx) |
|
39 |
+IDCT_ADD_FUNC(8, 8, sse2) |
|
40 |
+IDCT_ADD_FUNC(8, 10, sse2) |
|
41 |
+#if HAVE_AVX |
|
42 |
+IDCT_ADD_FUNC(, 10, avx) |
|
43 |
+IDCT_ADD_FUNC(8_dc, 10, avx) |
|
44 |
+IDCT_ADD_FUNC(8, 10, avx) |
|
45 |
+#endif |
|
46 |
+ |
|
47 |
+ |
|
48 |
+#define IDCT_ADD_REP_FUNC(NUM, REP, DEPTH, OPT) \ |
|
49 |
+void ff_h264_idct ## NUM ## _add ## REP ## _ ## DEPTH ## _ ## OPT \ |
|
50 |
+ (uint8_t *dst, const int *block_offset, \ |
|
51 |
+ DCTELEM *block, int stride, const uint8_t nnzc[6 * 8]); |
|
52 |
+ |
|
53 |
+IDCT_ADD_REP_FUNC(8, 4, 8, mmx) |
|
54 |
+IDCT_ADD_REP_FUNC(8, 4, 8, mmx2) |
|
55 |
+IDCT_ADD_REP_FUNC(8, 4, 8, sse2) |
|
56 |
+IDCT_ADD_REP_FUNC(8, 4, 10, sse2) |
|
57 |
+IDCT_ADD_REP_FUNC(8, 4, 10, avx) |
|
58 |
+IDCT_ADD_REP_FUNC(, 16, 8, mmx) |
|
59 |
+IDCT_ADD_REP_FUNC(, 16, 8, mmx2) |
|
60 |
+IDCT_ADD_REP_FUNC(, 16, 8, sse2) |
|
61 |
+IDCT_ADD_REP_FUNC(, 16, 10, sse2) |
|
62 |
+IDCT_ADD_REP_FUNC(, 16intra, 8, mmx) |
|
63 |
+IDCT_ADD_REP_FUNC(, 16intra, 8, mmx2) |
|
64 |
+IDCT_ADD_REP_FUNC(, 16intra, 8, sse2) |
|
65 |
+IDCT_ADD_REP_FUNC(, 16intra, 10, sse2) |
|
66 |
+#if HAVE_AVX |
|
67 |
+IDCT_ADD_REP_FUNC(, 16, 10, avx) |
|
68 |
+IDCT_ADD_REP_FUNC(, 16intra, 10, avx) |
|
69 |
+#endif |
|
70 |
+ |
|
71 |
+ |
|
72 |
+#define IDCT_ADD_REP_FUNC2(NUM, REP, DEPTH, OPT) \ |
|
73 |
+void ff_h264_idct ## NUM ## _add ## REP ## _ ## DEPTH ## _ ## OPT \ |
|
74 |
+ (uint8_t **dst, const int *block_offset, \ |
|
75 |
+ DCTELEM *block, int stride, const uint8_t nnzc[6 * 8]); |
|
76 |
+ |
|
77 |
+IDCT_ADD_REP_FUNC2(, 8, 8, mmx) |
|
78 |
+IDCT_ADD_REP_FUNC2(, 8, 8, mmx2) |
|
79 |
+IDCT_ADD_REP_FUNC2(, 8, 8, sse2) |
|
80 |
+IDCT_ADD_REP_FUNC2(, 8, 10, sse2) |
|
81 |
+#if HAVE_AVX |
|
82 |
+IDCT_ADD_REP_FUNC2(, 8, 10, avx) |
|
83 |
+#endif |
|
84 |
+ |
|
85 |
+void ff_h264_luma_dc_dequant_idct_mmx(DCTELEM *output, DCTELEM *input, int qmul); |
|
86 |
+void ff_h264_luma_dc_dequant_idct_sse2(DCTELEM *output, DCTELEM *input, int qmul); |
|
87 |
+ |
|
88 |
+/***********************************/ |
|
89 |
+/* deblocking */ |
|
90 |
+ |
|
91 |
+void ff_h264_loop_filter_strength_mmx2(int16_t bS[2][4][4], uint8_t nnz[40], |
|
92 |
+ int8_t ref[2][40], int16_t mv[2][40][2], |
|
93 |
+ int bidir, int edges, int step, |
|
94 |
+ int mask_mv0, int mask_mv1, int field); |
|
95 |
+ |
|
96 |
+#define LF_FUNC(DIR, TYPE, DEPTH, OPT) \ |
|
97 |
+void ff_deblock_ ## DIR ## _ ## TYPE ## _ ## DEPTH ## _ ## OPT(uint8_t *pix, \ |
|
98 |
+ int stride, \ |
|
99 |
+ int alpha, \ |
|
100 |
+ int beta, \ |
|
101 |
+ int8_t *tc0); |
|
102 |
+#define LF_IFUNC(DIR, TYPE, DEPTH, OPT) \ |
|
103 |
+void ff_deblock_ ## DIR ## _ ## TYPE ## _ ## DEPTH ## _ ## OPT(uint8_t *pix, \ |
|
104 |
+ int stride, \ |
|
105 |
+ int alpha, \ |
|
106 |
+ int beta); |
|
107 |
+ |
|
108 |
+#define LF_FUNCS(type, depth) \ |
|
109 |
+LF_FUNC(h, chroma, depth, mmx2) \ |
|
110 |
+LF_IFUNC(h, chroma_intra, depth, mmx2) \ |
|
111 |
+LF_FUNC(v, chroma, depth, mmx2) \ |
|
112 |
+LF_IFUNC(v, chroma_intra, depth, mmx2) \ |
|
113 |
+LF_FUNC(h, luma, depth, mmx2) \ |
|
114 |
+LF_IFUNC(h, luma_intra, depth, mmx2) \ |
|
115 |
+LF_FUNC(h, luma, depth, sse2) \ |
|
116 |
+LF_IFUNC(h, luma_intra, depth, sse2) \ |
|
117 |
+LF_FUNC(v, luma, depth, sse2) \ |
|
118 |
+LF_IFUNC(v, luma_intra, depth, sse2) \ |
|
119 |
+LF_FUNC(h, chroma, depth, sse2) \ |
|
120 |
+LF_IFUNC(h, chroma_intra, depth, sse2) \ |
|
121 |
+LF_FUNC(v, chroma, depth, sse2) \ |
|
122 |
+LF_IFUNC(v, chroma_intra, depth, sse2) \ |
|
123 |
+LF_FUNC(h, luma, depth, avx) \ |
|
124 |
+LF_IFUNC(h, luma_intra, depth, avx) \ |
|
125 |
+LF_FUNC(v, luma, depth, avx) \ |
|
126 |
+LF_IFUNC(v, luma_intra, depth, avx) \ |
|
127 |
+LF_FUNC(h, chroma, depth, avx) \ |
|
128 |
+LF_IFUNC(h, chroma_intra, depth, avx) \ |
|
129 |
+LF_FUNC(v, chroma, depth, avx) \ |
|
130 |
+LF_IFUNC(v, chroma_intra, depth, avx) |
|
131 |
+ |
|
132 |
+LF_FUNCS(uint8_t, 8) |
|
133 |
+LF_FUNCS(uint16_t, 10) |
|
134 |
+ |
|
135 |
+#if ARCH_X86_32 && HAVE_YASM |
|
136 |
+LF_FUNC(v8, luma, 8, mmx2) |
|
137 |
+static void ff_deblock_v_luma_8_mmx2(uint8_t *pix, int stride, int alpha, |
|
138 |
+ int beta, int8_t *tc0) |
|
139 |
+{ |
|
140 |
+ if ((tc0[0] & tc0[1]) >= 0) |
|
141 |
+ ff_deblock_v8_luma_8_mmx2(pix + 0, stride, alpha, beta, tc0); |
|
142 |
+ if ((tc0[2] & tc0[3]) >= 0) |
|
143 |
+ ff_deblock_v8_luma_8_mmx2(pix + 8, stride, alpha, beta, tc0 + 2); |
|
144 |
+} |
|
145 |
+ |
|
146 |
+LF_IFUNC(v8, luma_intra, 8, mmx2) |
|
147 |
+static void ff_deblock_v_luma_intra_8_mmx2(uint8_t *pix, int stride, |
|
148 |
+ int alpha, int beta) |
|
149 |
+{ |
|
150 |
+ ff_deblock_v8_luma_intra_8_mmx2(pix + 0, stride, alpha, beta); |
|
151 |
+ ff_deblock_v8_luma_intra_8_mmx2(pix + 8, stride, alpha, beta); |
|
152 |
+} |
|
153 |
+#endif /* ARCH_X86_32 */ |
|
154 |
+ |
|
155 |
+LF_FUNC(v, luma, 10, mmx2) |
|
156 |
+LF_IFUNC(v, luma_intra, 10, mmx2) |
|
157 |
+ |
|
158 |
+/***********************************/ |
|
159 |
+/* weighted prediction */ |
|
160 |
+ |
|
161 |
+#define H264_WEIGHT(W, OPT) \ |
|
162 |
+void ff_h264_weight_ ## W ## _ ## OPT(uint8_t *dst, int stride, \ |
|
163 |
+ int height, int log2_denom, \ |
|
164 |
+ int weight, int offset); |
|
165 |
+ |
|
166 |
+#define H264_BIWEIGHT(W, OPT) \ |
|
167 |
+void ff_h264_biweight_ ## W ## _ ## OPT(uint8_t *dst, uint8_t *src, \ |
|
168 |
+ int stride, int height, \ |
|
169 |
+ int log2_denom, int weightd, \ |
|
170 |
+ int weights, int offset); |
|
171 |
+ |
|
172 |
+#define H264_BIWEIGHT_MMX(W) \ |
|
173 |
+ H264_WEIGHT(W, mmx2) \ |
|
174 |
+ H264_BIWEIGHT(W, mmx2) |
|
175 |
+ |
|
176 |
+#define H264_BIWEIGHT_MMX_SSE(W) \ |
|
177 |
+ H264_BIWEIGHT_MMX(W) \ |
|
178 |
+ H264_WEIGHT(W, sse2) \ |
|
179 |
+ H264_BIWEIGHT(W, sse2) \ |
|
180 |
+ H264_BIWEIGHT(W, ssse3) |
|
181 |
+ |
|
182 |
+H264_BIWEIGHT_MMX_SSE(16) |
|
183 |
+H264_BIWEIGHT_MMX_SSE(8) |
|
184 |
+H264_BIWEIGHT_MMX(4) |
|
185 |
+ |
|
186 |
+#define H264_WEIGHT_10(W, DEPTH, OPT) \ |
|
187 |
+void ff_h264_weight_ ## W ## _ ## DEPTH ## _ ## OPT(uint8_t *dst, \ |
|
188 |
+ int stride, \ |
|
189 |
+ int height, \ |
|
190 |
+ int log2_denom, \ |
|
191 |
+ int weight, \ |
|
192 |
+ int offset); |
|
193 |
+ |
|
194 |
+#define H264_BIWEIGHT_10(W, DEPTH, OPT) \ |
|
195 |
+void ff_h264_biweight_ ## W ## _ ## DEPTH ## _ ## OPT(uint8_t *dst, \ |
|
196 |
+ uint8_t *src, \ |
|
197 |
+ int stride, \ |
|
198 |
+ int height, \ |
|
199 |
+ int log2_denom, \ |
|
200 |
+ int weightd, \ |
|
201 |
+ int weights, \ |
|
202 |
+ int offset); |
|
203 |
+ |
|
204 |
+#define H264_BIWEIGHT_10_SSE(W, DEPTH) \ |
|
205 |
+ H264_WEIGHT_10(W, DEPTH, sse2) \ |
|
206 |
+ H264_WEIGHT_10(W, DEPTH, sse4) \ |
|
207 |
+ H264_BIWEIGHT_10(W, DEPTH, sse2) \ |
|
208 |
+ H264_BIWEIGHT_10(W, DEPTH, sse4) |
|
209 |
+ |
|
210 |
+H264_BIWEIGHT_10_SSE(16, 10) |
|
211 |
+H264_BIWEIGHT_10_SSE(8, 10) |
|
212 |
+H264_BIWEIGHT_10_SSE(4, 10) |
|
213 |
+ |
|
214 |
+void ff_h264dsp_init_x86(H264DSPContext *c, const int bit_depth, |
|
215 |
+ const int chroma_format_idc) |
|
216 |
+{ |
|
217 |
+#if HAVE_YASM |
|
218 |
+ int mm_flags = av_get_cpu_flags(); |
|
219 |
+ |
|
220 |
+ if (chroma_format_idc == 1 && mm_flags & AV_CPU_FLAG_MMXEXT) |
|
221 |
+ c->h264_loop_filter_strength = ff_h264_loop_filter_strength_mmx2; |
|
222 |
+ |
|
223 |
+ if (bit_depth == 8) { |
|
224 |
+ if (mm_flags & AV_CPU_FLAG_MMX) { |
|
225 |
+ c->h264_idct_dc_add = |
|
226 |
+ c->h264_idct_add = ff_h264_idct_add_8_mmx; |
|
227 |
+ c->h264_idct8_dc_add = |
|
228 |
+ c->h264_idct8_add = ff_h264_idct8_add_8_mmx; |
|
229 |
+ |
|
230 |
+ c->h264_idct_add16 = ff_h264_idct_add16_8_mmx; |
|
231 |
+ c->h264_idct8_add4 = ff_h264_idct8_add4_8_mmx; |
|
232 |
+ if (chroma_format_idc == 1) |
|
233 |
+ c->h264_idct_add8 = ff_h264_idct_add8_8_mmx; |
|
234 |
+ c->h264_idct_add16intra = ff_h264_idct_add16intra_8_mmx; |
|
235 |
+ if (mm_flags & AV_CPU_FLAG_CMOV) |
|
236 |
+ c->h264_luma_dc_dequant_idct = ff_h264_luma_dc_dequant_idct_mmx; |
|
237 |
+ |
|
238 |
+ if (mm_flags & AV_CPU_FLAG_MMXEXT) { |
|
239 |
+ c->h264_idct_dc_add = ff_h264_idct_dc_add_8_mmx2; |
|
240 |
+ c->h264_idct8_dc_add = ff_h264_idct8_dc_add_8_mmx2; |
|
241 |
+ c->h264_idct_add16 = ff_h264_idct_add16_8_mmx2; |
|
242 |
+ c->h264_idct8_add4 = ff_h264_idct8_add4_8_mmx2; |
|
243 |
+ if (chroma_format_idc == 1) |
|
244 |
+ c->h264_idct_add8 = ff_h264_idct_add8_8_mmx2; |
|
245 |
+ c->h264_idct_add16intra = ff_h264_idct_add16intra_8_mmx2; |
|
246 |
+ |
|
247 |
+ c->h264_v_loop_filter_chroma = ff_deblock_v_chroma_8_mmx2; |
|
248 |
+ c->h264_v_loop_filter_chroma_intra = ff_deblock_v_chroma_intra_8_mmx2; |
|
249 |
+ if (chroma_format_idc == 1) { |
|
250 |
+ c->h264_h_loop_filter_chroma = ff_deblock_h_chroma_8_mmx2; |
|
251 |
+ c->h264_h_loop_filter_chroma_intra = ff_deblock_h_chroma_intra_8_mmx2; |
|
252 |
+ } |
|
253 |
+#if ARCH_X86_32 |
|
254 |
+ c->h264_v_loop_filter_luma = ff_deblock_v_luma_8_mmx2; |
|
255 |
+ c->h264_h_loop_filter_luma = ff_deblock_h_luma_8_mmx2; |
|
256 |
+ c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_8_mmx2; |
|
257 |
+ c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_8_mmx2; |
|
258 |
+#endif /* ARCH_X86_32 */ |
|
259 |
+ c->weight_h264_pixels_tab[0] = ff_h264_weight_16_mmx2; |
|
260 |
+ c->weight_h264_pixels_tab[1] = ff_h264_weight_8_mmx2; |
|
261 |
+ c->weight_h264_pixels_tab[2] = ff_h264_weight_4_mmx2; |
|
262 |
+ |
|
263 |
+ c->biweight_h264_pixels_tab[0] = ff_h264_biweight_16_mmx2; |
|
264 |
+ c->biweight_h264_pixels_tab[1] = ff_h264_biweight_8_mmx2; |
|
265 |
+ c->biweight_h264_pixels_tab[2] = ff_h264_biweight_4_mmx2; |
|
266 |
+ |
|
267 |
+ if (mm_flags & AV_CPU_FLAG_SSE2) { |
|
268 |
+ c->h264_idct8_add = ff_h264_idct8_add_8_sse2; |
|
269 |
+ |
|
270 |
+ c->h264_idct_add16 = ff_h264_idct_add16_8_sse2; |
|
271 |
+ c->h264_idct8_add4 = ff_h264_idct8_add4_8_sse2; |
|
272 |
+ if (chroma_format_idc == 1) |
|
273 |
+ c->h264_idct_add8 = ff_h264_idct_add8_8_sse2; |
|
274 |
+ c->h264_idct_add16intra = ff_h264_idct_add16intra_8_sse2; |
|
275 |
+ c->h264_luma_dc_dequant_idct = ff_h264_luma_dc_dequant_idct_sse2; |
|
276 |
+ |
|
277 |
+ c->weight_h264_pixels_tab[0] = ff_h264_weight_16_sse2; |
|
278 |
+ c->weight_h264_pixels_tab[1] = ff_h264_weight_8_sse2; |
|
279 |
+ |
|
280 |
+ c->biweight_h264_pixels_tab[0] = ff_h264_biweight_16_sse2; |
|
281 |
+ c->biweight_h264_pixels_tab[1] = ff_h264_biweight_8_sse2; |
|
282 |
+ |
|
283 |
+#if HAVE_ALIGNED_STACK |
|
284 |
+ c->h264_v_loop_filter_luma = ff_deblock_v_luma_8_sse2; |
|
285 |
+ c->h264_h_loop_filter_luma = ff_deblock_h_luma_8_sse2; |
|
286 |
+ c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_8_sse2; |
|
287 |
+ c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_8_sse2; |
|
288 |
+#endif /* HAVE_ALIGNED_STACK */ |
|
289 |
+ } |
|
290 |
+ if (mm_flags & AV_CPU_FLAG_SSSE3) { |
|
291 |
+ c->biweight_h264_pixels_tab[0] = ff_h264_biweight_16_ssse3; |
|
292 |
+ c->biweight_h264_pixels_tab[1] = ff_h264_biweight_8_ssse3; |
|
293 |
+ } |
|
294 |
+ if (HAVE_AVX && mm_flags & AV_CPU_FLAG_AVX) { |
|
295 |
+#if HAVE_ALIGNED_STACK |
|
296 |
+ c->h264_v_loop_filter_luma = ff_deblock_v_luma_8_avx; |
|
297 |
+ c->h264_h_loop_filter_luma = ff_deblock_h_luma_8_avx; |
|
298 |
+ c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_8_avx; |
|
299 |
+ c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_8_avx; |
|
300 |
+#endif /* HAVE_ALIGNED_STACK */ |
|
301 |
+ } |
|
302 |
+ } |
|
303 |
+ } |
|
304 |
+ } else if (bit_depth == 10) { |
|
305 |
+ if (mm_flags & AV_CPU_FLAG_MMX) { |
|
306 |
+ if (mm_flags & AV_CPU_FLAG_MMXEXT) { |
|
307 |
+#if ARCH_X86_32 |
|
308 |
+ c->h264_v_loop_filter_chroma = ff_deblock_v_chroma_10_mmx2; |
|
309 |
+ c->h264_v_loop_filter_chroma_intra = ff_deblock_v_chroma_intra_10_mmx2; |
|
310 |
+ c->h264_v_loop_filter_luma = ff_deblock_v_luma_10_mmx2; |
|
311 |
+ c->h264_h_loop_filter_luma = ff_deblock_h_luma_10_mmx2; |
|
312 |
+ c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_10_mmx2; |
|
313 |
+ c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_10_mmx2; |
|
314 |
+#endif /* ARCH_X86_32 */ |
|
315 |
+ c->h264_idct_dc_add = ff_h264_idct_dc_add_10_mmx2; |
|
316 |
+ if (mm_flags & AV_CPU_FLAG_SSE2) { |
|
317 |
+ c->h264_idct_add = ff_h264_idct_add_10_sse2; |
|
318 |
+ c->h264_idct8_dc_add = ff_h264_idct8_dc_add_10_sse2; |
|
319 |
+ |
|
320 |
+ c->h264_idct_add16 = ff_h264_idct_add16_10_sse2; |
|
321 |
+ if (chroma_format_idc == 1) |
|
322 |
+ c->h264_idct_add8 = ff_h264_idct_add8_10_sse2; |
|
323 |
+ c->h264_idct_add16intra = ff_h264_idct_add16intra_10_sse2; |
|
324 |
+#if HAVE_ALIGNED_STACK |
|
325 |
+ c->h264_idct8_add = ff_h264_idct8_add_10_sse2; |
|
326 |
+ c->h264_idct8_add4 = ff_h264_idct8_add4_10_sse2; |
|
327 |
+#endif /* HAVE_ALIGNED_STACK */ |
|
328 |
+ |
|
329 |
+ c->weight_h264_pixels_tab[0] = ff_h264_weight_16_10_sse2; |
|
330 |
+ c->weight_h264_pixels_tab[1] = ff_h264_weight_8_10_sse2; |
|
331 |
+ c->weight_h264_pixels_tab[2] = ff_h264_weight_4_10_sse2; |
|
332 |
+ |
|
333 |
+ c->biweight_h264_pixels_tab[0] = ff_h264_biweight_16_10_sse2; |
|
334 |
+ c->biweight_h264_pixels_tab[1] = ff_h264_biweight_8_10_sse2; |
|
335 |
+ c->biweight_h264_pixels_tab[2] = ff_h264_biweight_4_10_sse2; |
|
336 |
+ |
|
337 |
+ c->h264_v_loop_filter_chroma = ff_deblock_v_chroma_10_sse2; |
|
338 |
+ c->h264_v_loop_filter_chroma_intra = ff_deblock_v_chroma_intra_10_sse2; |
|
339 |
+#if HAVE_ALIGNED_STACK |
|
340 |
+ c->h264_v_loop_filter_luma = ff_deblock_v_luma_10_sse2; |
|
341 |
+ c->h264_h_loop_filter_luma = ff_deblock_h_luma_10_sse2; |
|
342 |
+ c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_10_sse2; |
|
343 |
+ c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_10_sse2; |
|
344 |
+#endif /* HAVE_ALIGNED_STACK */ |
|
345 |
+ } |
|
346 |
+ if (mm_flags & AV_CPU_FLAG_SSE4) { |
|
347 |
+ c->weight_h264_pixels_tab[0] = ff_h264_weight_16_10_sse4; |
|
348 |
+ c->weight_h264_pixels_tab[1] = ff_h264_weight_8_10_sse4; |
|
349 |
+ c->weight_h264_pixels_tab[2] = ff_h264_weight_4_10_sse4; |
|
350 |
+ |
|
351 |
+ c->biweight_h264_pixels_tab[0] = ff_h264_biweight_16_10_sse4; |
|
352 |
+ c->biweight_h264_pixels_tab[1] = ff_h264_biweight_8_10_sse4; |
|
353 |
+ c->biweight_h264_pixels_tab[2] = ff_h264_biweight_4_10_sse4; |
|
354 |
+ } |
|
355 |
+#if HAVE_AVX |
|
356 |
+ if (mm_flags & AV_CPU_FLAG_AVX) { |
|
357 |
+ c->h264_idct_dc_add = |
|
358 |
+ c->h264_idct_add = ff_h264_idct_add_10_avx; |
|
359 |
+ c->h264_idct8_dc_add = ff_h264_idct8_dc_add_10_avx; |
|
360 |
+ |
|
361 |
+ c->h264_idct_add16 = ff_h264_idct_add16_10_avx; |
|
362 |
+ if (chroma_format_idc == 1) |
|
363 |
+ c->h264_idct_add8 = ff_h264_idct_add8_10_avx; |
|
364 |
+ c->h264_idct_add16intra = ff_h264_idct_add16intra_10_avx; |
|
365 |
+#if HAVE_ALIGNED_STACK |
|
366 |
+ c->h264_idct8_add = ff_h264_idct8_add_10_avx; |
|
367 |
+ c->h264_idct8_add4 = ff_h264_idct8_add4_10_avx; |
|
368 |
+#endif /* HAVE_ALIGNED_STACK */ |
|
369 |
+ |
|
370 |
+ c->h264_v_loop_filter_chroma = ff_deblock_v_chroma_10_avx; |
|
371 |
+ c->h264_v_loop_filter_chroma_intra = ff_deblock_v_chroma_intra_10_avx; |
|
372 |
+#if HAVE_ALIGNED_STACK |
|
373 |
+ c->h264_v_loop_filter_luma = ff_deblock_v_luma_10_avx; |
|
374 |
+ c->h264_h_loop_filter_luma = ff_deblock_h_luma_10_avx; |
|
375 |
+ c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_10_avx; |
|
376 |
+ c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_10_avx; |
|
377 |
+#endif /* HAVE_ALIGNED_STACK */ |
|
378 |
+ } |
|
379 |
+#endif /* HAVE_AVX */ |
|
380 |
+ } |
|
381 |
+ } |
|
382 |
+ } |
|
383 |
+#endif /* HAVE_YASM */ |
|
384 |
+} |
0 | 385 |
deleted file mode 100644 |
... | ... |
@@ -1,385 +0,0 @@ |
1 |
-/* |
|
2 |
- * Copyright (c) 2004-2005 Michael Niedermayer, Loren Merritt |
|
3 |
- * |
|
4 |
- * This file is part of FFmpeg. |
|
5 |
- * |
|
6 |
- * FFmpeg is free software; you can redistribute it and/or |
|
7 |
- * modify it under the terms of the GNU Lesser General Public |
|
8 |
- * License as published by the Free Software Foundation; either |
|
9 |
- * version 2.1 of the License, or (at your option) any later version. |
|
10 |
- * |
|
11 |
- * FFmpeg is distributed in the hope that it will be useful, |
|
12 |
- * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
13 |
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
14 |
- * Lesser General Public License for more details. |
|
15 |
- * |
|
16 |
- * You should have received a copy of the GNU Lesser General Public |
|
17 |
- * License along with FFmpeg; if not, write to the Free Software |
|
18 |
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
19 |
- */ |
|
20 |
- |
|
21 |
-#include "libavutil/cpu.h" |
|
22 |
-#include "libavutil/x86/asm.h" |
|
23 |
-#include "libavcodec/h264dsp.h" |
|
24 |
-#include "dsputil_mmx.h" |
|
25 |
- |
|
26 |
-/***********************************/ |
|
27 |
-/* IDCT */ |
|
28 |
-#define IDCT_ADD_FUNC(NUM, DEPTH, OPT) \ |
|
29 |
-void ff_h264_idct ## NUM ## _add_ ## DEPTH ## _ ## OPT(uint8_t *dst, \ |
|
30 |
- int16_t *block, \ |
|
31 |
- int stride); |
|
32 |
- |
|
33 |
-IDCT_ADD_FUNC(, 8, mmx) |
|
34 |
-IDCT_ADD_FUNC(, 10, sse2) |
|
35 |
-IDCT_ADD_FUNC(_dc, 8, mmx2) |
|
36 |
-IDCT_ADD_FUNC(_dc, 10, mmx2) |
|
37 |
-IDCT_ADD_FUNC(8_dc, 8, mmx2) |
|
38 |
-IDCT_ADD_FUNC(8_dc, 10, sse2) |
|
39 |
-IDCT_ADD_FUNC(8, 8, mmx) |
|
40 |
-IDCT_ADD_FUNC(8, 8, sse2) |
|
41 |
-IDCT_ADD_FUNC(8, 10, sse2) |
|
42 |
-#if HAVE_AVX |
|
43 |
-IDCT_ADD_FUNC(, 10, avx) |
|
44 |
-IDCT_ADD_FUNC(8_dc, 10, avx) |
|
45 |
-IDCT_ADD_FUNC(8, 10, avx) |
|
46 |
-#endif |
|
47 |
- |
|
48 |
- |
|
49 |
-#define IDCT_ADD_REP_FUNC(NUM, REP, DEPTH, OPT) \ |
|
50 |
-void ff_h264_idct ## NUM ## _add ## REP ## _ ## DEPTH ## _ ## OPT \ |
|
51 |
- (uint8_t *dst, const int *block_offset, \ |
|
52 |
- DCTELEM *block, int stride, const uint8_t nnzc[6 * 8]); |
|
53 |
- |
|
54 |
-IDCT_ADD_REP_FUNC(8, 4, 8, mmx) |
|
55 |
-IDCT_ADD_REP_FUNC(8, 4, 8, mmx2) |
|
56 |
-IDCT_ADD_REP_FUNC(8, 4, 8, sse2) |
|
57 |
-IDCT_ADD_REP_FUNC(8, 4, 10, sse2) |
|
58 |
-IDCT_ADD_REP_FUNC(8, 4, 10, avx) |
|
59 |
-IDCT_ADD_REP_FUNC(, 16, 8, mmx) |
|
60 |
-IDCT_ADD_REP_FUNC(, 16, 8, mmx2) |
|
61 |
-IDCT_ADD_REP_FUNC(, 16, 8, sse2) |
|
62 |
-IDCT_ADD_REP_FUNC(, 16, 10, sse2) |
|
63 |
-IDCT_ADD_REP_FUNC(, 16intra, 8, mmx) |
|
64 |
-IDCT_ADD_REP_FUNC(, 16intra, 8, mmx2) |
|
65 |
-IDCT_ADD_REP_FUNC(, 16intra, 8, sse2) |
|
66 |
-IDCT_ADD_REP_FUNC(, 16intra, 10, sse2) |
|
67 |
-#if HAVE_AVX |
|
68 |
-IDCT_ADD_REP_FUNC(, 16, 10, avx) |
|
69 |
-IDCT_ADD_REP_FUNC(, 16intra, 10, avx) |
|
70 |
-#endif |
|
71 |
- |
|
72 |
- |
|
73 |
-#define IDCT_ADD_REP_FUNC2(NUM, REP, DEPTH, OPT) \ |
|
74 |
-void ff_h264_idct ## NUM ## _add ## REP ## _ ## DEPTH ## _ ## OPT \ |
|
75 |
- (uint8_t **dst, const int *block_offset, \ |
|
76 |
- DCTELEM *block, int stride, const uint8_t nnzc[6 * 8]); |
|
77 |
- |
|
78 |
-IDCT_ADD_REP_FUNC2(, 8, 8, mmx) |
|
79 |
-IDCT_ADD_REP_FUNC2(, 8, 8, mmx2) |
|
80 |
-IDCT_ADD_REP_FUNC2(, 8, 8, sse2) |
|
81 |
-IDCT_ADD_REP_FUNC2(, 8, 10, sse2) |
|
82 |
-#if HAVE_AVX |
|
83 |
-IDCT_ADD_REP_FUNC2(, 8, 10, avx) |
|
84 |
-#endif |
|
85 |
- |
|
86 |
-void ff_h264_luma_dc_dequant_idct_mmx(DCTELEM *output, DCTELEM *input, int qmul); |
|
87 |
-void ff_h264_luma_dc_dequant_idct_sse2(DCTELEM *output, DCTELEM *input, int qmul); |
|
88 |
- |
|
89 |
-/***********************************/ |
|
90 |
-/* deblocking */ |
|
91 |
- |
|
92 |
-void ff_h264_loop_filter_strength_mmx2(int16_t bS[2][4][4], uint8_t nnz[40], |
|
93 |
- int8_t ref[2][40], int16_t mv[2][40][2], |
|
94 |
- int bidir, int edges, int step, |
|
95 |
- int mask_mv0, int mask_mv1, int field); |
|
96 |
- |
|
97 |
-#define LF_FUNC(DIR, TYPE, DEPTH, OPT) \ |
|
98 |
-void ff_deblock_ ## DIR ## _ ## TYPE ## _ ## DEPTH ## _ ## OPT(uint8_t *pix, \ |
|
99 |
- int stride, \ |
|
100 |
- int alpha, \ |
|
101 |
- int beta, \ |
|
102 |
- int8_t *tc0); |
|
103 |
-#define LF_IFUNC(DIR, TYPE, DEPTH, OPT) \ |
|
104 |
-void ff_deblock_ ## DIR ## _ ## TYPE ## _ ## DEPTH ## _ ## OPT(uint8_t *pix, \ |
|
105 |
- int stride, \ |
|
106 |
- int alpha, \ |
|
107 |
- int beta); |
|
108 |
- |
|
109 |
-#define LF_FUNCS(type, depth) \ |
|
110 |
-LF_FUNC(h, chroma, depth, mmx2) \ |
|
111 |
-LF_IFUNC(h, chroma_intra, depth, mmx2) \ |
|
112 |
-LF_FUNC(v, chroma, depth, mmx2) \ |
|
113 |
-LF_IFUNC(v, chroma_intra, depth, mmx2) \ |
|
114 |
-LF_FUNC(h, luma, depth, mmx2) \ |
|
115 |
-LF_IFUNC(h, luma_intra, depth, mmx2) \ |
|
116 |
-LF_FUNC(h, luma, depth, sse2) \ |
|
117 |
-LF_IFUNC(h, luma_intra, depth, sse2) \ |
|
118 |
-LF_FUNC(v, luma, depth, sse2) \ |
|
119 |
-LF_IFUNC(v, luma_intra, depth, sse2) \ |
|
120 |
-LF_FUNC(h, chroma, depth, sse2) \ |
|
121 |
-LF_IFUNC(h, chroma_intra, depth, sse2) \ |
|
122 |
-LF_FUNC(v, chroma, depth, sse2) \ |
|
123 |
-LF_IFUNC(v, chroma_intra, depth, sse2) \ |
|
124 |
-LF_FUNC(h, luma, depth, avx) \ |
|
125 |
-LF_IFUNC(h, luma_intra, depth, avx) \ |
|
126 |
-LF_FUNC(v, luma, depth, avx) \ |
|
127 |
-LF_IFUNC(v, luma_intra, depth, avx) \ |
|
128 |
-LF_FUNC(h, chroma, depth, avx) \ |
|
129 |
-LF_IFUNC(h, chroma_intra, depth, avx) \ |
|
130 |
-LF_FUNC(v, chroma, depth, avx) \ |
|
131 |
-LF_IFUNC(v, chroma_intra, depth, avx) |
|
132 |
- |
|
133 |
-LF_FUNCS(uint8_t, 8) |
|
134 |
-LF_FUNCS(uint16_t, 10) |
|
135 |
- |
|
136 |
-#if ARCH_X86_32 && HAVE_YASM |
|
137 |
-LF_FUNC(v8, luma, 8, mmx2) |
|
138 |
-static void ff_deblock_v_luma_8_mmx2(uint8_t *pix, int stride, int alpha, |
|
139 |
- int beta, int8_t *tc0) |
|
140 |
-{ |
|
141 |
- if ((tc0[0] & tc0[1]) >= 0) |
|
142 |
- ff_deblock_v8_luma_8_mmx2(pix + 0, stride, alpha, beta, tc0); |
|
143 |
- if ((tc0[2] & tc0[3]) >= 0) |
|
144 |
- ff_deblock_v8_luma_8_mmx2(pix + 8, stride, alpha, beta, tc0 + 2); |
|
145 |
-} |
|
146 |
- |
|
147 |
-LF_IFUNC(v8, luma_intra, 8, mmx2) |
|
148 |
-static void ff_deblock_v_luma_intra_8_mmx2(uint8_t *pix, int stride, |
|
149 |
- int alpha, int beta) |
|
150 |
-{ |
|
151 |
- ff_deblock_v8_luma_intra_8_mmx2(pix + 0, stride, alpha, beta); |
|
152 |
- ff_deblock_v8_luma_intra_8_mmx2(pix + 8, stride, alpha, beta); |
|
153 |
-} |
|
154 |
-#endif /* ARCH_X86_32 */ |
|
155 |
- |
|
156 |
-LF_FUNC(v, luma, 10, mmx2) |
|
157 |
-LF_IFUNC(v, luma_intra, 10, mmx2) |
|
158 |
- |
|
159 |
-/***********************************/ |
|
160 |
-/* weighted prediction */ |
|
161 |
- |
|
162 |
-#define H264_WEIGHT(W, OPT) \ |
|
163 |
-void ff_h264_weight_ ## W ## _ ## OPT(uint8_t *dst, int stride, \ |
|
164 |
- int height, int log2_denom, \ |
|
165 |
- int weight, int offset); |
|
166 |
- |
|
167 |
-#define H264_BIWEIGHT(W, OPT) \ |
|
168 |
-void ff_h264_biweight_ ## W ## _ ## OPT(uint8_t *dst, uint8_t *src, \ |
|
169 |
- int stride, int height, \ |
|
170 |
- int log2_denom, int weightd, \ |
|
171 |
- int weights, int offset); |
|
172 |
- |
|
173 |
-#define H264_BIWEIGHT_MMX(W) \ |
|
174 |
- H264_WEIGHT(W, mmx2) \ |
|
175 |
- H264_BIWEIGHT(W, mmx2) |
|
176 |
- |
|
177 |
-#define H264_BIWEIGHT_MMX_SSE(W) \ |
|
178 |
- H264_BIWEIGHT_MMX(W) \ |
|
179 |
- H264_WEIGHT(W, sse2) \ |
|
180 |
- H264_BIWEIGHT(W, sse2) \ |
|
181 |
- H264_BIWEIGHT(W, ssse3) |
|
182 |
- |
|
183 |
-H264_BIWEIGHT_MMX_SSE(16) |
|
184 |
-H264_BIWEIGHT_MMX_SSE(8) |
|
185 |
-H264_BIWEIGHT_MMX(4) |
|
186 |
- |
|
187 |
-#define H264_WEIGHT_10(W, DEPTH, OPT) \ |
|
188 |
-void ff_h264_weight_ ## W ## _ ## DEPTH ## _ ## OPT(uint8_t *dst, \ |
|
189 |
- int stride, \ |
|
190 |
- int height, \ |
|
191 |
- int log2_denom, \ |
|
192 |
- int weight, \ |
|
193 |
- int offset); |
|
194 |
- |
|
195 |
-#define H264_BIWEIGHT_10(W, DEPTH, OPT) \ |
|
196 |
-void ff_h264_biweight_ ## W ## _ ## DEPTH ## _ ## OPT(uint8_t *dst, \ |
|
197 |
- uint8_t *src, \ |
|
198 |
- int stride, \ |
|
199 |
- int height, \ |
|
200 |
- int log2_denom, \ |
|
201 |
- int weightd, \ |
|
202 |
- int weights, \ |
|
203 |
- int offset); |
|
204 |
- |
|
205 |
-#define H264_BIWEIGHT_10_SSE(W, DEPTH) \ |
|
206 |
- H264_WEIGHT_10(W, DEPTH, sse2) \ |
|
207 |
- H264_WEIGHT_10(W, DEPTH, sse4) \ |
|
208 |
- H264_BIWEIGHT_10(W, DEPTH, sse2) \ |
|
209 |
- H264_BIWEIGHT_10(W, DEPTH, sse4) |
|
210 |
- |
|
211 |
-H264_BIWEIGHT_10_SSE(16, 10) |
|
212 |
-H264_BIWEIGHT_10_SSE(8, 10) |
|
213 |
-H264_BIWEIGHT_10_SSE(4, 10) |
|
214 |
- |
|
215 |
-void ff_h264dsp_init_x86(H264DSPContext *c, const int bit_depth, |
|
216 |
- const int chroma_format_idc) |
|
217 |
-{ |
|
218 |
-#if HAVE_YASM |
|
219 |
- int mm_flags = av_get_cpu_flags(); |
|
220 |
- |
|
221 |
- if (chroma_format_idc == 1 && mm_flags & AV_CPU_FLAG_MMXEXT) |
|
222 |
- c->h264_loop_filter_strength = ff_h264_loop_filter_strength_mmx2; |
|
223 |
- |
|
224 |
- if (bit_depth == 8) { |
|
225 |
- if (mm_flags & AV_CPU_FLAG_MMX) { |
|
226 |
- c->h264_idct_dc_add = |
|
227 |
- c->h264_idct_add = ff_h264_idct_add_8_mmx; |
|
228 |
- c->h264_idct8_dc_add = |
|
229 |
- c->h264_idct8_add = ff_h264_idct8_add_8_mmx; |
|
230 |
- |
|
231 |
- c->h264_idct_add16 = ff_h264_idct_add16_8_mmx; |
|
232 |
- c->h264_idct8_add4 = ff_h264_idct8_add4_8_mmx; |
|
233 |
- if (chroma_format_idc == 1) |
|
234 |
- c->h264_idct_add8 = ff_h264_idct_add8_8_mmx; |
|
235 |
- c->h264_idct_add16intra = ff_h264_idct_add16intra_8_mmx; |
|
236 |
- if (mm_flags & AV_CPU_FLAG_CMOV) |
|
237 |
- c->h264_luma_dc_dequant_idct = ff_h264_luma_dc_dequant_idct_mmx; |
|
238 |
- |
|
239 |
- if (mm_flags & AV_CPU_FLAG_MMXEXT) { |
|
240 |
- c->h264_idct_dc_add = ff_h264_idct_dc_add_8_mmx2; |
|
241 |
- c->h264_idct8_dc_add = ff_h264_idct8_dc_add_8_mmx2; |
|
242 |
- c->h264_idct_add16 = ff_h264_idct_add16_8_mmx2; |
|
243 |
- c->h264_idct8_add4 = ff_h264_idct8_add4_8_mmx2; |
|
244 |
- if (chroma_format_idc == 1) |
|
245 |
- c->h264_idct_add8 = ff_h264_idct_add8_8_mmx2; |
|
246 |
- c->h264_idct_add16intra = ff_h264_idct_add16intra_8_mmx2; |
|
247 |
- |
|
248 |
- c->h264_v_loop_filter_chroma = ff_deblock_v_chroma_8_mmx2; |
|
249 |
- c->h264_v_loop_filter_chroma_intra = ff_deblock_v_chroma_intra_8_mmx2; |
|
250 |
- if (chroma_format_idc == 1) { |
|
251 |
- c->h264_h_loop_filter_chroma = ff_deblock_h_chroma_8_mmx2; |
|
252 |
- c->h264_h_loop_filter_chroma_intra = ff_deblock_h_chroma_intra_8_mmx2; |
|
253 |
- } |
|
254 |
-#if ARCH_X86_32 |
|
255 |
- c->h264_v_loop_filter_luma = ff_deblock_v_luma_8_mmx2; |
|
256 |
- c->h264_h_loop_filter_luma = ff_deblock_h_luma_8_mmx2; |
|
257 |
- c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_8_mmx2; |
|
258 |
- c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_8_mmx2; |
|
259 |
-#endif /* ARCH_X86_32 */ |
|
260 |
- c->weight_h264_pixels_tab[0] = ff_h264_weight_16_mmx2; |
|
261 |
- c->weight_h264_pixels_tab[1] = ff_h264_weight_8_mmx2; |
|
262 |
- c->weight_h264_pixels_tab[2] = ff_h264_weight_4_mmx2; |
|
263 |
- |
|
264 |
- c->biweight_h264_pixels_tab[0] = ff_h264_biweight_16_mmx2; |
|
265 |
- c->biweight_h264_pixels_tab[1] = ff_h264_biweight_8_mmx2; |
|
266 |
- c->biweight_h264_pixels_tab[2] = ff_h264_biweight_4_mmx2; |
|
267 |
- |
|
268 |
- if (mm_flags & AV_CPU_FLAG_SSE2) { |
|
269 |
- c->h264_idct8_add = ff_h264_idct8_add_8_sse2; |
|
270 |
- |
|
271 |
- c->h264_idct_add16 = ff_h264_idct_add16_8_sse2; |
|
272 |
- c->h264_idct8_add4 = ff_h264_idct8_add4_8_sse2; |
|
273 |
- if (chroma_format_idc == 1) |
|
274 |
- c->h264_idct_add8 = ff_h264_idct_add8_8_sse2; |
|
275 |
- c->h264_idct_add16intra = ff_h264_idct_add16intra_8_sse2; |
|
276 |
- c->h264_luma_dc_dequant_idct = ff_h264_luma_dc_dequant_idct_sse2; |
|
277 |
- |
|
278 |
- c->weight_h264_pixels_tab[0] = ff_h264_weight_16_sse2; |
|
279 |
- c->weight_h264_pixels_tab[1] = ff_h264_weight_8_sse2; |
|
280 |
- |
|
281 |
- c->biweight_h264_pixels_tab[0] = ff_h264_biweight_16_sse2; |
|
282 |
- c->biweight_h264_pixels_tab[1] = ff_h264_biweight_8_sse2; |
|
283 |
- |
|
284 |
-#if HAVE_ALIGNED_STACK |
|
285 |
- c->h264_v_loop_filter_luma = ff_deblock_v_luma_8_sse2; |
|
286 |
- c->h264_h_loop_filter_luma = ff_deblock_h_luma_8_sse2; |
|
287 |
- c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_8_sse2; |
|
288 |
- c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_8_sse2; |
|
289 |
-#endif /* HAVE_ALIGNED_STACK */ |
|
290 |
- } |
|
291 |
- if (mm_flags & AV_CPU_FLAG_SSSE3) { |
|
292 |
- c->biweight_h264_pixels_tab[0] = ff_h264_biweight_16_ssse3; |
|
293 |
- c->biweight_h264_pixels_tab[1] = ff_h264_biweight_8_ssse3; |
|
294 |
- } |
|
295 |
- if (HAVE_AVX && mm_flags & AV_CPU_FLAG_AVX) { |
|
296 |
-#if HAVE_ALIGNED_STACK |
|
297 |
- c->h264_v_loop_filter_luma = ff_deblock_v_luma_8_avx; |
|
298 |
- c->h264_h_loop_filter_luma = ff_deblock_h_luma_8_avx; |
|
299 |
- c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_8_avx; |
|
300 |
- c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_8_avx; |
|
301 |
-#endif /* HAVE_ALIGNED_STACK */ |
|
302 |
- } |
|
303 |
- } |
|
304 |
- } |
|
305 |
- } else if (bit_depth == 10) { |
|
306 |
- if (mm_flags & AV_CPU_FLAG_MMX) { |
|
307 |
- if (mm_flags & AV_CPU_FLAG_MMXEXT) { |
|
308 |
-#if ARCH_X86_32 |
|
309 |
- c->h264_v_loop_filter_chroma = ff_deblock_v_chroma_10_mmx2; |
|
310 |
- c->h264_v_loop_filter_chroma_intra = ff_deblock_v_chroma_intra_10_mmx2; |
|
311 |
- c->h264_v_loop_filter_luma = ff_deblock_v_luma_10_mmx2; |
|
312 |
- c->h264_h_loop_filter_luma = ff_deblock_h_luma_10_mmx2; |
|
313 |
- c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_10_mmx2; |
|
314 |
- c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_10_mmx2; |
|
315 |
-#endif /* ARCH_X86_32 */ |
|
316 |
- c->h264_idct_dc_add = ff_h264_idct_dc_add_10_mmx2; |
|
317 |
- if (mm_flags & AV_CPU_FLAG_SSE2) { |
|
318 |
- c->h264_idct_add = ff_h264_idct_add_10_sse2; |
|
319 |
- c->h264_idct8_dc_add = ff_h264_idct8_dc_add_10_sse2; |
|
320 |
- |
|
321 |
- c->h264_idct_add16 = ff_h264_idct_add16_10_sse2; |
|
322 |
- if (chroma_format_idc == 1) |
|
323 |
- c->h264_idct_add8 = ff_h264_idct_add8_10_sse2; |
|
324 |
- c->h264_idct_add16intra = ff_h264_idct_add16intra_10_sse2; |
|
325 |
-#if HAVE_ALIGNED_STACK |
|
326 |
- c->h264_idct8_add = ff_h264_idct8_add_10_sse2; |
|
327 |
- c->h264_idct8_add4 = ff_h264_idct8_add4_10_sse2; |
|
328 |
-#endif /* HAVE_ALIGNED_STACK */ |
|
329 |
- |
|
330 |
- c->weight_h264_pixels_tab[0] = ff_h264_weight_16_10_sse2; |
|
331 |
- c->weight_h264_pixels_tab[1] = ff_h264_weight_8_10_sse2; |
|
332 |
- c->weight_h264_pixels_tab[2] = ff_h264_weight_4_10_sse2; |
|
333 |
- |
|
334 |
- c->biweight_h264_pixels_tab[0] = ff_h264_biweight_16_10_sse2; |
|
335 |
- c->biweight_h264_pixels_tab[1] = ff_h264_biweight_8_10_sse2; |
|
336 |
- c->biweight_h264_pixels_tab[2] = ff_h264_biweight_4_10_sse2; |
|
337 |
- |
|
338 |
- c->h264_v_loop_filter_chroma = ff_deblock_v_chroma_10_sse2; |
|
339 |
- c->h264_v_loop_filter_chroma_intra = ff_deblock_v_chroma_intra_10_sse2; |
|
340 |
-#if HAVE_ALIGNED_STACK |
|
341 |
- c->h264_v_loop_filter_luma = ff_deblock_v_luma_10_sse2; |
|
342 |
- c->h264_h_loop_filter_luma = ff_deblock_h_luma_10_sse2; |
|
343 |
- c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_10_sse2; |
|
344 |
- c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_10_sse2; |
|
345 |
-#endif /* HAVE_ALIGNED_STACK */ |
|
346 |
- } |
|
347 |
- if (mm_flags & AV_CPU_FLAG_SSE4) { |
|
348 |
- c->weight_h264_pixels_tab[0] = ff_h264_weight_16_10_sse4; |
|
349 |
- c->weight_h264_pixels_tab[1] = ff_h264_weight_8_10_sse4; |
|
350 |
- c->weight_h264_pixels_tab[2] = ff_h264_weight_4_10_sse4; |
|
351 |
- |
|
352 |
- c->biweight_h264_pixels_tab[0] = ff_h264_biweight_16_10_sse4; |
|
353 |
- c->biweight_h264_pixels_tab[1] = ff_h264_biweight_8_10_sse4; |
|
354 |
- c->biweight_h264_pixels_tab[2] = ff_h264_biweight_4_10_sse4; |
|
355 |
- } |
|
356 |
-#if HAVE_AVX |
|
357 |
- if (mm_flags & AV_CPU_FLAG_AVX) { |
|
358 |
- c->h264_idct_dc_add = |
|
359 |
- c->h264_idct_add = ff_h264_idct_add_10_avx; |
|
360 |
- c->h264_idct8_dc_add = ff_h264_idct8_dc_add_10_avx; |
|
361 |
- |
|
362 |
- c->h264_idct_add16 = ff_h264_idct_add16_10_avx; |
|
363 |
- if (chroma_format_idc == 1) |
|
364 |
- c->h264_idct_add8 = ff_h264_idct_add8_10_avx; |
|
365 |
- c->h264_idct_add16intra = ff_h264_idct_add16intra_10_avx; |
|
366 |
-#if HAVE_ALIGNED_STACK |
|
367 |
- c->h264_idct8_add = ff_h264_idct8_add_10_avx; |
|
368 |
- c->h264_idct8_add4 = ff_h264_idct8_add4_10_avx; |
|
369 |
-#endif /* HAVE_ALIGNED_STACK */ |
|
370 |
- |
|
371 |
- c->h264_v_loop_filter_chroma = ff_deblock_v_chroma_10_avx; |
|
372 |
- c->h264_v_loop_filter_chroma_intra = ff_deblock_v_chroma_intra_10_avx; |
|
373 |
-#if HAVE_ALIGNED_STACK |
|
374 |
- c->h264_v_loop_filter_luma = ff_deblock_v_luma_10_avx; |
|
375 |
- c->h264_h_loop_filter_luma = ff_deblock_h_luma_10_avx; |
|
376 |
- c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_10_avx; |
|
377 |
- c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_10_avx; |
|
378 |
-#endif /* HAVE_ALIGNED_STACK */ |
|
379 |
- } |
|
380 |
-#endif /* HAVE_AVX */ |
|
381 |
- } |
|
382 |
- } |
|
383 |
- } |
|
384 |
-#endif /* HAVE_YASM */ |
|
385 |
-} |
... | ... |
@@ -360,13 +360,5 @@ static int RENAME(dct_quantize)(MpegEncContext *s, |
360 | 360 |
block[0x3E] = temp_block[0x3E]; block[0x3F] = temp_block[0x3F]; |
361 | 361 |
} |
362 | 362 |
end: |
363 |
-/* |
|
364 |
- for(i=0; i<last_non_zero_p1; i++) |
|
365 |
- { |
|
366 |
- int j= zigzag_direct_noperm[i]; |
|
367 |
- block[block_permute_op(j)]= temp_block[j]; |
|
368 |
- } |
|
369 |
-*/ |
|
370 |
- |
|
371 | 363 |
return last_non_zero_p1 - 1; |
372 | 364 |
} |
... | ... |
@@ -33,15 +33,6 @@ |
33 | 33 |
#define HMAC_OPAD_VAL 0x5C |
34 | 34 |
|
35 | 35 |
/** |
36 |
- * A non-zero transaction id requires the server to send back |
|
37 |
- * a _result or _error response. |
|
38 |
- * Setting it to 0 marks the message as a notification not |
|
39 |
- * requiring feedback. |
|
40 |
- */ |
|
41 |
- |
|
42 |
-#define RTMP_NOTIFICATION 0 |
|
43 |
- |
|
44 |
-/** |
|
45 | 36 |
* emulated Flash client version - 9.0.124.2 on Linux |
46 | 37 |
* @{ |
47 | 38 |
*/ |
... | ... |
@@ -91,7 +91,11 @@ typedef struct RTMPContext { |
91 | 91 |
int nb_invokes; ///< keeps track of invoke messages |
92 | 92 |
char* tcurl; ///< url of the target stream |
93 | 93 |
char* flashver; ///< version of the flash plugin |
94 |
+ char* swfhash; ///< SHA256 hash of the decompressed SWF file (32 bytes) |
|
95 |
+ int swfhash_len; ///< length of the SHA256 hash |
|
96 |
+ int swfsize; ///< size of the decompressed SWF file |
|
94 | 97 |
char* swfurl; ///< url of the swf player |
98 |
+ char swfverification[42]; ///< hash of the SWF verification |
|
95 | 99 |
char* pageurl; ///< url of the web page |
96 | 100 |
char* subscribe; ///< name of live stream to subscribe |
97 | 101 |
int server_bw; ///< server bandwidth |
... | ... |
@@ -593,6 +597,27 @@ static int gen_pong(URLContext *s, RTMPContext *rt, RTMPPacket *ppkt) |
593 | 593 |
} |
594 | 594 |
|
595 | 595 |
/** |
596 |
+ * Generate SWF verification message and send it to the server. |
|
597 |
+ */ |
|
598 |
+static int gen_swf_verification(URLContext *s, RTMPContext *rt) |
|
599 |
+{ |
|
600 |
+ RTMPPacket pkt; |
|
601 |
+ uint8_t *p; |
|
602 |
+ int ret; |
|
603 |
+ |
|
604 |
+ av_log(s, AV_LOG_DEBUG, "Sending SWF verification...\n"); |
|
605 |
+ if ((ret = ff_rtmp_packet_create(&pkt, RTMP_NETWORK_CHANNEL, RTMP_PT_PING, |
|
606 |
+ 0, 44)) < 0) |
|
607 |
+ return ret; |
|
608 |
+ |
|
609 |
+ p = pkt.data; |
|
610 |
+ bytestream_put_be16(&p, 27); |
|
611 |
+ memcpy(p, rt->swfverification, 42); |
|
612 |
+ |
|
613 |
+ return rtmp_send_packet(rt, &pkt, 0); |
|
614 |
+} |
|
615 |
+ |
|
616 |
+/** |
|
596 | 617 |
* Generate server bandwidth message and send it to the server. |
597 | 618 |
*/ |
598 | 619 |
static int gen_server_bw(URLContext *s, RTMPContext *rt) |
... | ... |
@@ -626,10 +651,10 @@ static int gen_check_bw(URLContext *s, RTMPContext *rt) |
626 | 626 |
|
627 | 627 |
p = pkt.data; |
628 | 628 |
ff_amf_write_string(&p, "_checkbw"); |
629 |
- ff_amf_write_number(&p, RTMP_NOTIFICATION); |
|
629 |
+ ff_amf_write_number(&p, ++rt->nb_invokes); |
|
630 | 630 |
ff_amf_write_null(&p); |
631 | 631 |
|
632 |
- return rtmp_send_packet(rt, &pkt, 0); |
|
632 |
+ return rtmp_send_packet(rt, &pkt, 1); |
|
633 | 633 |
} |
634 | 634 |
|
635 | 635 |
/** |
... | ... |
@@ -776,6 +801,30 @@ static int rtmp_validate_digest(uint8_t *buf, int off) |
776 | 776 |
return 0; |
777 | 777 |
} |
778 | 778 |
|
779 |
+static int rtmp_calc_swf_verification(URLContext *s, RTMPContext *rt, |
|
780 |
+ uint8_t *buf) |
|
781 |
+{ |
|
782 |
+ uint8_t *p; |
|
783 |
+ int ret; |
|
784 |
+ |
|
785 |
+ if (rt->swfhash_len != 32) { |
|
786 |
+ av_log(s, AV_LOG_ERROR, |
|
787 |
+ "Hash of the decompressed SWF file is not 32 bytes long.\n"); |
|
788 |
+ return AVERROR(EINVAL); |
|
789 |
+ } |
|
790 |
+ |
|
791 |
+ p = &rt->swfverification[0]; |
|
792 |
+ bytestream_put_byte(&p, 1); |
|
793 |
+ bytestream_put_byte(&p, 1); |
|
794 |
+ bytestream_put_be32(&p, rt->swfsize); |
|
795 |
+ bytestream_put_be32(&p, rt->swfsize); |
|
796 |
+ |
|
797 |
+ if ((ret = ff_rtmp_calc_digest(rt->swfhash, 32, 0, buf, 32, p)) < 0) |
|
798 |
+ return ret; |
|
799 |
+ |
|
800 |
+ return 0; |
|
801 |
+} |
|
802 |
+ |
|
779 | 803 |
/** |
780 | 804 |
* Perform handshake with the server by means of exchanging pseudorandom data |
781 | 805 |
* signed with HMAC-SHA2 digest. |
... | ... |
@@ -866,6 +915,14 @@ static int rtmp_handshake(URLContext *s, RTMPContext *rt) |
866 | 866 |
} |
867 | 867 |
} |
868 | 868 |
|
869 |
+ /* Generate SWFVerification token (SHA256 HMAC hash of decompressed SWF, |
|
870 |
+ * key are the last 32 bytes of the server handshake. */ |
|
871 |
+ if (rt->swfsize) { |
|
872 |
+ if ((ret = rtmp_calc_swf_verification(s, rt, serverdata + 1 + |
|
873 |
+ RTMP_HANDSHAKE_PACKET_SIZE - 32)) < 0) |
|
874 |
+ return ret; |
|
875 |
+ } |
|
876 |
+ |
|
869 | 877 |
ret = ff_rtmp_calc_digest(tosend + 1 + client_pos, 32, 0, |
870 | 878 |
rtmp_server_key, sizeof(rtmp_server_key), |
871 | 879 |
digest); |
... | ... |
@@ -1001,6 +1058,13 @@ static int handle_ping(URLContext *s, RTMPPacket *pkt) |
1001 | 1001 |
if (t == 6) { |
1002 | 1002 |
if ((ret = gen_pong(s, rt, pkt)) < 0) |
1003 | 1003 |
return ret; |
1004 |
+ } else if (t == 26) { |
|
1005 |
+ if (rt->swfsize) { |
|
1006 |
+ if ((ret = gen_swf_verification(s, rt)) < 0) |
|
1007 |
+ return ret; |
|
1008 |
+ } else { |
|
1009 |
+ av_log(s, AV_LOG_WARNING, "Ignoring SWFVerification request.\n"); |
|
1010 |
+ } |
|
1004 | 1011 |
} |
1005 | 1012 |
|
1006 | 1013 |
return 0; |
... | ... |
@@ -1055,15 +1119,27 @@ static int handle_server_bw(URLContext *s, RTMPPacket *pkt) |
1055 | 1055 |
static int handle_invoke_error(URLContext *s, RTMPPacket *pkt) |
1056 | 1056 |
{ |
1057 | 1057 |
const uint8_t *data_end = pkt->data + pkt->data_size; |
1058 |
+ char *tracked_method = NULL; |
|
1059 |
+ int level = AV_LOG_ERROR; |
|
1058 | 1060 |
uint8_t tmpstr[256]; |
1061 |
+ int ret; |
|
1062 |
+ |
|
1063 |
+ if ((ret = find_tracked_method(s, pkt, 9, &tracked_method)) < 0) |
|
1064 |
+ return ret; |
|
1059 | 1065 |
|
1060 | 1066 |
if (!ff_amf_get_field_value(pkt->data + 9, data_end, |
1061 | 1067 |
"description", tmpstr, sizeof(tmpstr))) { |
1062 |
- av_log(s, AV_LOG_ERROR, "Server error: %s\n", tmpstr); |
|
1063 |
- return -1; |
|
1068 |
+ if (tracked_method && !strcmp(tracked_method, "_checkbw")) { |
|
1069 |
+ /* Ignore _checkbw errors. */ |
|
1070 |
+ level = AV_LOG_WARNING; |
|
1071 |
+ ret = 0; |
|
1072 |
+ } else |
|
1073 |
+ ret = -1; |
|
1074 |
+ av_log(s, level, "Server error: %s\n", tmpstr); |
|
1064 | 1075 |
} |
1065 | 1076 |
|
1066 |
- return 0; |
|
1077 |
+ av_free(tracked_method); |
|
1078 |
+ return ret; |
|
1067 | 1079 |
} |
1068 | 1080 |
|
1069 | 1081 |
static int handle_invoke_result(URLContext *s, RTMPPacket *pkt) |
... | ... |
@@ -1705,6 +1781,8 @@ static const AVOption rtmp_options[] = { |
1705 | 1705 |
{"rtmp_pageurl", "URL of the web page in which the media was embedded. By default no value will be sent.", OFFSET(pageurl), AV_OPT_TYPE_STRING, {.str = NULL }, 0, 0, DEC}, |
1706 | 1706 |
{"rtmp_playpath", "Stream identifier to play or to publish", OFFSET(playpath), AV_OPT_TYPE_STRING, {.str = NULL }, 0, 0, DEC|ENC}, |
1707 | 1707 |
{"rtmp_subscribe", "Name of live stream to subscribe to. Defaults to rtmp_playpath.", OFFSET(subscribe), AV_OPT_TYPE_STRING, {.str = NULL }, 0, 0, DEC}, |
1708 |
+ {"rtmp_swfhash", "SHA256 hash of the decompressed SWF file (32 bytes).", OFFSET(swfhash), AV_OPT_TYPE_BINARY, .flags = DEC}, |
|
1709 |
+ {"rtmp_swfsize", "Size of the decompressed SWF file, required for SWFVerification.", OFFSET(swfsize), AV_OPT_TYPE_INT, {0}, 0, INT_MAX, DEC}, |
|
1708 | 1710 |
{"rtmp_swfurl", "URL of the SWF player. By default no value will be sent", OFFSET(swfurl), AV_OPT_TYPE_STRING, {.str = NULL }, 0, 0, DEC|ENC}, |
1709 | 1711 |
{"rtmp_tcurl", "URL of the target stream. Defaults to proto://host[:port]/app.", OFFSET(tcurl), AV_OPT_TYPE_STRING, {.str = NULL }, 0, 0, DEC|ENC}, |
1710 | 1712 |
{ NULL }, |
... | ... |
@@ -31,7 +31,7 @@ |
31 | 31 |
|
32 | 32 |
#define LIBAVFORMAT_VERSION_MAJOR 54 |
33 | 33 |
#define LIBAVFORMAT_VERSION_MINOR 23 |
34 |
-#define LIBAVFORMAT_VERSION_MICRO 100 |
|
34 |
+#define LIBAVFORMAT_VERSION_MICRO 101 |
|
35 | 35 |
|
36 | 36 |
#define LIBAVFORMAT_VERSION_INT AV_VERSION_INT(LIBAVFORMAT_VERSION_MAJOR, \ |
37 | 37 |
LIBAVFORMAT_VERSION_MINOR, \ |
... | ... |
@@ -335,7 +335,9 @@ int ff_audio_mix_init(AVAudioResampleContext *avr) |
335 | 335 |
avr->out_channel_layout, |
336 | 336 |
avr->center_mix_level, |
337 | 337 |
avr->surround_mix_level, |
338 |
- avr->lfe_mix_level, 1, matrix_dbl, |
|
338 |
+ avr->lfe_mix_level, |
|
339 |
+ avr->normalize_mix_level, |
|
340 |
+ matrix_dbl, |
|
339 | 341 |
avr->in_channels, |
340 | 342 |
avr->matrix_encoding); |
341 | 343 |
if (ret < 0) { |
... | ... |
@@ -45,6 +45,7 @@ struct AVAudioResampleContext { |
45 | 45 |
double center_mix_level; /**< center mix level */ |
46 | 46 |
double surround_mix_level; /**< surround mix level */ |
47 | 47 |
double lfe_mix_level; /**< lfe mix level */ |
48 |
+ int normalize_mix_level; /**< enable mix level normalization */ |
|
48 | 49 |
int force_resampling; /**< force resampling */ |
49 | 50 |
int filter_size; /**< length of each FIR filter in the resampling filterbank relative to the cutoff frequency */ |
50 | 51 |
int phase_shift; /**< log2 of the number of entries in the resampling polyphase filterbank */ |
... | ... |
@@ -47,6 +47,7 @@ static const AVOption options[] = { |
47 | 47 |
{ "center_mix_level", "Center Mix Level", OFFSET(center_mix_level), AV_OPT_TYPE_DOUBLE, { M_SQRT1_2 }, -32.0, 32.0, PARAM }, |
48 | 48 |
{ "surround_mix_level", "Surround Mix Level", OFFSET(surround_mix_level), AV_OPT_TYPE_DOUBLE, { M_SQRT1_2 }, -32.0, 32.0, PARAM }, |
49 | 49 |
{ "lfe_mix_level", "LFE Mix Level", OFFSET(lfe_mix_level), AV_OPT_TYPE_DOUBLE, { 0.0 }, -32.0, 32.0, PARAM }, |
50 |
+ { "normalize_mix_level", "Normalize Mix Level", OFFSET(normalize_mix_level), AV_OPT_TYPE_INT, { 1 }, 0, 1, PARAM }, |
|
50 | 51 |
{ "force_resampling", "Force Resampling", OFFSET(force_resampling), AV_OPT_TYPE_INT, { 0 }, 0, 1, PARAM }, |
51 | 52 |
{ "filter_size", "Resampling Filter Size", OFFSET(filter_size), AV_OPT_TYPE_INT, { 16 }, 0, 32, /* ??? */ PARAM }, |
52 | 53 |
{ "phase_shift", "Resampling Phase Shift", OFFSET(phase_shift), AV_OPT_TYPE_INT, { 10 }, 0, 30, /* ??? */ PARAM }, |