April 2017 - ffmpeg-cvslog - lists.ffmpeg.org

Changelog: add some missing entries
by Anton Khirnov 29 Apr '17

29 Apr '17

ffmpeg | branch: master | Anton Khirnov <anton(a)khirnov.net> | Thu Dec 15 10:23:26 2016 +0100| [1783d7ec03d730c5bd96c07bc5fa7aa566f85c66] | committer: Anton Khirnov Changelog: add some missing entries > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=1783d7ec03d730c5b… --- Changelog | 3 +++ 1 file changed, 3 insertions(+) diff --git a/Changelog b/Changelog index feb151cf3c..ab7f1b6bb0 100644 --- a/Changelog +++ b/Changelog @@ -3,6 +3,9 @@ releases are sorted from youngest to oldest. version <next>: - Support for spherical videos +- Intel QSV-accelerated VP8 and VC-1 decoding +- VAAPI-accelerated VP8 decoding +- VAAPI-accelerated deinterlacing version 12:

1 0

doc/filters: add one lowpass filter example
by Paul B Mahol 29 Apr '17

29 Apr '17

ffmpeg | branch: master | Paul B Mahol <onemda(a)gmail.com> | Sat Apr 29 11:44:21 2017 +0200| [b8b0cece794d8338ec79833c55e77d53888f800f] | committer: Paul B Mahol doc/filters: add one lowpass filter example Signed-off-by: Paul B Mahol <onemda(a)gmail.com> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=b8b0cece794d8338e… --- doc/filters.texi | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/doc/filters.texi b/doc/filters.texi index 2fe7ff7b8c..d49fec1264 100644 --- a/doc/filters.texi +++ b/doc/filters.texi @@ -3002,6 +3002,15 @@ The default is 0.707q and gives a Butterworth response. Specify which channels to filter, by default all available are filtered. @end table +@subsection Examples +@itemize +@item +Lowpass only LFE channel, it LFE is not present it does nothing: +@example +lowpass=c=LFE +@end example +@end itemize + @anchor{pan} @section pan

1 0

avformat/movenc: Explicitly address potential division by zero.
by Lucas Cooper 29 Apr '17

29 Apr '17

ffmpeg | branch: master | Lucas Cooper <bobobo-at-google.com(a)ffmpeg.org> | Thu Apr 27 15:08:29 2017 -0700| [77bc507f6f001b9f5fa75c664106261bd8f2c971] | committer: Michael Niedermayer avformat/movenc: Explicitly address potential division by zero. find_fps attempts to infer framerate from AVCodec's timebase. When this results in a frame rate that isn't explicitly marked as supported in av_timecode_check_frame_rate, find_fps returns the AVStream's avg_frame_rate, which, per avformat.h, _may_ be set (or not). mov_get_mpeg2_xdcam_codec_tag, mov_get_h264_codec_tag and find_compressor attempt to call av_q2d on the return value of find_fps, which in the above case, may result in division by zero and therefore, an undefined frame rate when NaN is converted to int. Signed-off-by: Michael Niedermayer <michael(a)niedermayer.cc> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=77bc507f6f001b9f5… --- libavformat/movenc.c | 15 ++++++++++++--- 1 file changed, 12 insertions(+), 3 deletions(-) diff --git a/libavformat/movenc.c b/libavformat/movenc.c index e6e2313c53..d20d272f89 100644 --- a/libavformat/movenc.c +++ b/libavformat/movenc.c @@ -1321,12 +1321,21 @@ static AVRational find_fps(AVFormatContext *s, AVStream *st) return rate; } +static int defined_frame_rate(AVFormatContext *s, AVStream *st) +{ + AVRational rational_framerate = find_fps(s, st); + int rate = 0; + if (rational_framerate.den != 0) + rate = av_q2d(rational_framerate); + return rate; +} + static int mov_get_mpeg2_xdcam_codec_tag(AVFormatContext *s, MOVTrack *track) { int tag = track->par->codec_tag; int interlaced = track->par->field_order > AV_FIELD_PROGRESSIVE; AVStream *st = track->st; - int rate = av_q2d(find_fps(s, st)); + int rate = defined_frame_rate(s, st); if (!tag) tag = MKTAG('m', '2', 'v', '1'); //fallback tag @@ -1388,7 +1397,7 @@ static int mov_get_h264_codec_tag(AVFormatContext *s, MOVTrack *track) int tag = track->par->codec_tag; int interlaced = track->par->field_order > AV_FIELD_PROGRESSIVE; AVStream *st = track->st; - int rate = av_q2d(find_fps(s, st)); + int rate = defined_frame_rate(s, st); if (!tag) tag = MKTAG('a', 'v', 'c', 'i'); //fallback tag @@ -1850,7 +1859,7 @@ static void find_compressor(char * compressor_name, int len, MOVTrack *track) } else if (track->par->codec_id == AV_CODEC_ID_MPEG2VIDEO && xdcam_res) { int interlaced = track->par->field_order > AV_FIELD_PROGRESSIVE; AVStream *st = track->st; - int rate = av_q2d(find_fps(NULL, st)); + int rate = defined_frame_rate(NULL, st); av_strlcatf(compressor_name, len, "XDCAM"); if (track->par->format == AV_PIX_FMT_YUV422P) { av_strlcatf(compressor_name, len, " HD422");

1 0

libvpxenc: allow aq-mode 4 (equator360)
by James Zern 29 Apr '17

29 Apr '17

ffmpeg | branch: master | James Zern <jzern(a)google.com> | Thu Apr 13 18:26:44 2017 -0700| [1bee78a0198decb73619115f7e760be6b7d010e5] | committer: James Zern libvpxenc: allow aq-mode 4 (equator360) this was added in 1.6.0 Reviewed-by: Ronald S. Bultje <rsbultje(a)gmail.com> Signed-off-by: James Zern <jzern(a)google.com> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=1bee78a0198decb73… --- doc/encoders.texi | 2 +- libavcodec/libvpxenc.c | 5 +++++ 2 files changed, 6 insertions(+), 1 deletion(-) diff --git a/doc/encoders.texi b/doc/encoders.texi index 7f4754d160..3645b490f9 100644 --- a/doc/encoders.texi +++ b/doc/encoders.texi @@ -1666,7 +1666,7 @@ option to 2. Enable frame parallel decodability features. @item aq-mode Set adaptive quantization mode (0: off (default), 1: variance 2: complexity, 3: -cyclic refresh). +cyclic refresh, 4: equator360). @item colorspace @emph{color-space} Set input color space. The VP9 bitstream supports signaling the following colorspaces: diff --git a/libavcodec/libvpxenc.c b/libavcodec/libvpxenc.c index b8e8d3a87b..64a7459331 100644 --- a/libavcodec/libvpxenc.c +++ b/libavcodec/libvpxenc.c @@ -1142,12 +1142,17 @@ static const AVOption vp9_options[] = { { "tile-columns", "Number of tile columns to use, log2", OFFSET(tile_columns), AV_OPT_TYPE_INT, {.i64 = -1}, -1, 6, VE}, { "tile-rows", "Number of tile rows to use, log2", OFFSET(tile_rows), AV_OPT_TYPE_INT, {.i64 = -1}, -1, 2, VE}, { "frame-parallel", "Enable frame parallel decodability features", OFFSET(frame_parallel), AV_OPT_TYPE_BOOL,{.i64 = -1}, -1, 1, VE}, +#if VPX_ENCODER_ABI_VERSION >= 12 + { "aq-mode", "adaptive quantization mode", OFFSET(aq_mode), AV_OPT_TYPE_INT, {.i64 = -1}, -1, 4, VE, "aq_mode"}, +#else { "aq-mode", "adaptive quantization mode", OFFSET(aq_mode), AV_OPT_TYPE_INT, {.i64 = -1}, -1, 3, VE, "aq_mode"}, +#endif { "none", "Aq not used", 0, AV_OPT_TYPE_CONST, {.i64 = 0}, 0, 0, VE, "aq_mode" }, { "variance", "Variance based Aq", 0, AV_OPT_TYPE_CONST, {.i64 = 1}, 0, 0, VE, "aq_mode" }, { "complexity", "Complexity based Aq", 0, AV_OPT_TYPE_CONST, {.i64 = 2}, 0, 0, VE, "aq_mode" }, { "cyclic", "Cyclic Refresh Aq", 0, AV_OPT_TYPE_CONST, {.i64 = 3}, 0, 0, VE, "aq_mode" }, #if VPX_ENCODER_ABI_VERSION >= 12 + { "equator360", "360 video Aq", 0, AV_OPT_TYPE_CONST, {.i64 = 4}, 0, 0, VE, "aq_mode" }, {"level", "Specify level", OFFSET(level), AV_OPT_TYPE_FLOAT, {.dbl=-1}, -1, 6.2, VE}, #endif #ifdef VPX_CTRL_VP9E_SET_ROW_MT

1 0

avfilter/lavfutils: use image2pipe demuxer on ff_load_image
by Muhammad Faiz 28 Apr '17

28 Apr '17

ffmpeg | branch: master | Muhammad Faiz <mfcc64(a)gmail.com> | Tue Apr 25 14:33:24 2017 +0700| [e061826eb2a5706dd75cf42f2fb7603c34c3ea0f] | committer: Muhammad Faiz avfilter/lavfutils: use image2pipe demuxer on ff_load_image allow protocols other than file to be used for example, use data protocol to embed a file in script Signed-off-by: Muhammad Faiz <mfcc64(a)gmail.com> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=e061826eb2a5706dd… --- libavfilter/lavfutils.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libavfilter/lavfutils.c b/libavfilter/lavfutils.c index 706badf63d..35878b3d50 100644 --- a/libavfilter/lavfutils.c +++ b/libavfilter/lavfutils.c @@ -38,7 +38,7 @@ int ff_load_image(uint8_t *data[4], int linesize[4], av_register_all(); - iformat = av_find_input_format("image2"); + iformat = av_find_input_format("image2pipe"); if ((ret = avformat_open_input(&format_ctx, filename, iformat, NULL)) < 0) { av_log(log_ctx, AV_LOG_ERROR, "Failed to open input file '%s'\n", filename);

1 0

avformat/hlsenc: fix CID 1405135
by Steven Liu 28 Apr '17

28 Apr '17

ffmpeg | branch: master | Steven Liu <lq(a)chinaffmpeg.org> | Fri Apr 28 10:18:44 2017 +0800| [cbfd44a9229cdead17c2857469ed1340615b3f36] | committer: Steven Liu avformat/hlsenc: fix CID 1405135 Fixes Coverity CID: 1405135 Reviewed-by: Rodger Combs <rodger.combs(a)gmail.com> Signed-off-by: Steven Liu <lq(a)chinaffmpeg.org> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=cbfd44a9229cdead1… --- libavformat/hlsenc.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/libavformat/hlsenc.c b/libavformat/hlsenc.c index 3ec0f330fb..221089c1be 100644 --- a/libavformat/hlsenc.c +++ b/libavformat/hlsenc.c @@ -394,11 +394,11 @@ static int do_encrypt(AVFormatContext *s) av_strlcat(hls->key_basename, ".key", len); if (hls->key_url) { - strncpy(hls->key_file, hls->key_url, sizeof(hls->key_file)); - strncpy(hls->key_uri, hls->key_url, sizeof(hls->key_uri)); + av_strlcpy(hls->key_file, hls->key_url, sizeof(hls->key_file)); + av_strlcpy(hls->key_uri, hls->key_url, sizeof(hls->key_uri)); } else { - strncpy(hls->key_file, hls->key_basename, sizeof(hls->key_file)); - strncpy(hls->key_uri, hls->key_basename, sizeof(hls->key_uri)); + av_strlcpy(hls->key_file, hls->key_basename, sizeof(hls->key_file)); + av_strlcpy(hls->key_uri, hls->key_basename, sizeof(hls->key_uri)); } if (!*hls->iv_string) {

1 0

avformat/hlsenc: hold old key info when append list
by Steven Liu 28 Apr '17

28 Apr '17

ffmpeg | branch: master | Steven Liu <lq(a)chinaffmpeg.org> | Fri Apr 28 10:17:40 2017 +0800| [363e4f0810d4085bbee3dced41a2de2d2c135dca] | committer: Steven Liu avformat/hlsenc: hold old key info when append list fix ticket id: #6353 Signed-off-by: Steven Liu <lq(a)chinaffmpeg.org> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=363e4f0810d4085bb… --- libavformat/hlsenc.c | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) diff --git a/libavformat/hlsenc.c b/libavformat/hlsenc.c index 27c8e3355d..3ec0f330fb 100644 --- a/libavformat/hlsenc.c +++ b/libavformat/hlsenc.c @@ -810,6 +810,7 @@ static int parse_playlist(AVFormatContext *s, const char *url) int64_t new_start_pos; char line[1024]; const char *ptr; + const char *end; if ((ret = ffio_open_whitelist(&in, url, AVIO_FLAG_READ, &s->interrupt_callback, NULL, @@ -842,6 +843,29 @@ static int parse_playlist(AVFormatContext *s, const char *url) } else if (av_strstart(line, "#EXTINF:", &ptr)) { is_segment = 1; hls->duration = atof(ptr); + } else if (av_stristart(line, "#EXT-X-KEY:", &ptr)) { + ptr = av_stristr(line, "URI=\""); + if (ptr) { + ptr += strlen("URI=\""); + end = av_stristr(ptr, ","); + if (end) { + av_strlcpy(hls->key_uri, ptr, end - ptr); + } else { + av_strlcpy(hls->key_uri, ptr, sizeof(hls->key_uri)); + } + } + + ptr = av_stristr(line, "IV=0x"); + if (ptr) { + ptr += strlen("IV=0x"); + end = av_stristr(ptr, ","); + if (end) { + av_strlcpy(hls->iv_string, ptr, end - ptr); + } else { + av_strlcpy(hls->iv_string, ptr, sizeof(hls->iv_string)); + } + } + } else if (av_strstart(line, "#", NULL)) { continue; } else if (line[0]) {

1 0

opus_pvq: use function pointers for recursion
by Rostislav Pehlivanov 27 Apr '17

27 Apr '17

ffmpeg | branch: master | Rostislav Pehlivanov <atomnuker(a)gmail.com> | Thu Apr 27 05:50:39 2017 +0100| [f7542d7e35e9bb9aff7829c15c78b9b467627482] | committer: Rostislav Pehlivanov opus_pvq: use function pointers for recursion Signed-off-by: Rostislav Pehlivanov <atomnuker(a)gmail.com> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=f7542d7e35e9bb9af… --- libavcodec/opus_pvq.c | 62 +++++++++++++++++++++------------------------------ 1 file changed, 25 insertions(+), 37 deletions(-) diff --git a/libavcodec/opus_pvq.c b/libavcodec/opus_pvq.c index d4c2bcbdd6..fa349c47da 100644 --- a/libavcodec/opus_pvq.c +++ b/libavcodec/opus_pvq.c @@ -477,6 +477,14 @@ static void celt_stereo_ms_decouple(float *X, float *Y, int N) } } +#define QUANT_FN(name) uint32_t (*name)(CeltFrame *f, OpusRangeCoder *rc, \ + const int band, float *X, float *Y, \ + int N, int b, uint32_t blocks, \ + float *lowband, int duration, \ + float *lowband_out, int level, \ + float gain, float *lowband_scratch, \ + int fill) + static av_always_inline uint32_t quant_band_template(CeltFrame *f, OpusRangeCoder *rc, const int band, float *X, float *Y, int N, int b, uint32_t blocks, float *lowband, int duration, float *lowband_out, @@ -497,6 +505,7 @@ static av_always_inline uint32_t quant_band_template(CeltFrame *f, OpusRangeCode float mid = 0, side = 0; int longblocks = (B0 == 1); uint32_t cm = 0; + QUANT_FN(rec) = quant ? ff_celt_encode_band : ff_celt_decode_band; if (N == 1) { float *x = X; @@ -693,9 +702,8 @@ static av_always_inline uint32_t quant_band_template(CeltFrame *f, OpusRangeCode sign = 1 - 2 * sign; /* We use orig_fill here because we want to fold the side, but if itheta==16384, we'll have cleared the low bits of fill. */ - cm = ff_celt_decode_band(f, rc, band, x2, NULL, N, mbits, blocks, - lowband, duration, lowband_out, level, gain, - lowband_scratch, orig_fill); + cm = rec(f, rc, band, x2, NULL, N, mbits, blocks, lowband, duration, + lowband_out, level, gain, lowband_scratch, orig_fill); /* We don't split N=2 bands, so cm is either 1 or 0 (for a fold-collapse), and there's no need to worry about mixing with the other channel. */ y2[0] = -sign * x2[1]; @@ -716,6 +724,7 @@ static av_always_inline uint32_t quant_band_template(CeltFrame *f, OpusRangeCode float *next_lowband_out1 = NULL; int next_level = 0; int rebalance; + uint32_t cmt; /* Give more bits to low-energy MDCTs than they would * otherwise deserve */ @@ -746,55 +755,34 @@ static av_always_inline uint32_t quant_band_template(CeltFrame *f, OpusRangeCode if (mbits >= sbits) { /* In stereo mode, we do not apply a scaling to the mid * because we need the normalized mid for folding later */ - cm = quant ? ff_celt_encode_band(f, rc, band, X, NULL, N, mbits, blocks, - lowband, duration, next_lowband_out1, - next_level, stereo ? 1.0f : (gain * mid), - lowband_scratch, fill) : - ff_celt_decode_band(f, rc, band, X, NULL, N, mbits, blocks, - lowband, duration, next_lowband_out1, - next_level, stereo ? 1.0f : (gain * mid), - lowband_scratch, fill); - + cm = rec(f, rc, band, X, NULL, N, mbits, blocks, lowband, + duration, next_lowband_out1, next_level, + stereo ? 1.0f : (gain * mid), lowband_scratch, fill); rebalance = mbits - (rebalance - f->remaining2); if (rebalance > 3 << 3 && itheta != 0) sbits += rebalance - (3 << 3); /* For a stereo split, the high bits of fill are always zero, * so no folding will be done to the side. */ - cm |= quant ? ff_celt_encode_band(f, rc, band, Y, NULL, N, sbits, blocks, - next_lowband2, duration, NULL, - next_level, gain * side, NULL, - fill >> blocks) << ((B0 >> 1) & (stereo - 1)) : - ff_celt_decode_band(f, rc, band, Y, NULL, N, sbits, blocks, - next_lowband2, duration, NULL, - next_level, gain * side, NULL, - fill >> blocks) << ((B0 >> 1) & (stereo - 1)); + cmt = rec(f, rc, band, Y, NULL, N, sbits, blocks, next_lowband2, + duration, NULL, next_level, gain * side, NULL, + fill >> blocks); + cm |= cmt << ((B0 >> 1) & (stereo - 1)); } else { /* For a stereo split, the high bits of fill are always zero, * so no folding will be done to the side. */ - cm = quant ? ff_celt_encode_band(f, rc, band, Y, NULL, N, sbits, blocks, - next_lowband2, duration, NULL, - next_level, gain * side, NULL, - fill >> blocks) << ((B0 >> 1) & (stereo - 1)) : - ff_celt_decode_band(f, rc, band, Y, NULL, N, sbits, blocks, - next_lowband2, duration, NULL, - next_level, gain * side, NULL, - fill >> blocks) << ((B0 >> 1) & (stereo - 1)); - + cm = rec(f, rc, band, Y, NULL, N, sbits, blocks, next_lowband2, + duration, NULL, next_level, gain * side, NULL, fill >> blocks); + cm <<= ((B0 >> 1) & (stereo - 1)); rebalance = sbits - (rebalance - f->remaining2); if (rebalance > 3 << 3 && itheta != 16384) mbits += rebalance - (3 << 3); /* In stereo mode, we do not apply a scaling to the mid because * we need the normalized mid for folding later */ - cm |= quant ? ff_celt_encode_band(f, rc, band, X, NULL, N, mbits, blocks, - lowband, duration, next_lowband_out1, - next_level, stereo ? 1.0f : (gain * mid), - lowband_scratch, fill) : - ff_celt_decode_band(f, rc, band, X, NULL, N, mbits, blocks, - lowband, duration, next_lowband_out1, - next_level, stereo ? 1.0f : (gain * mid), - lowband_scratch, fill); + cm |= rec(f, rc, band, X, NULL, N, mbits, blocks, lowband, duration, + next_lowband_out1, next_level, stereo ? 1.0f : (gain * mid), + lowband_scratch, fill); } } } else {

1 0

avcodec/opus_pvq: fix recursive inlining compilation failures
by James Almer 27 Apr '17

27 Apr '17

ffmpeg | branch: master | James Almer <jamrial(a)gmail.com> | Thu Apr 27 00:33:24 2017 -0300| [453f8ea2423617ed6d027f1442500084ae11b432] | committer: James Almer avcodec/opus_pvq: fix recursive inlining compilation failures They were introduced by f16180f44859587e0211170bd82b1bfd26f24b5a > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=453f8ea2423617ed6… --- libavcodec/opus_pvq.c | 48 ++++++++++++++++++++++++++++++++---------------- 1 file changed, 32 insertions(+), 16 deletions(-) diff --git a/libavcodec/opus_pvq.c b/libavcodec/opus_pvq.c index 3109e70f17..d4c2bcbdd6 100644 --- a/libavcodec/opus_pvq.c +++ b/libavcodec/opus_pvq.c @@ -746,10 +746,14 @@ static av_always_inline uint32_t quant_band_template(CeltFrame *f, OpusRangeCode if (mbits >= sbits) { /* In stereo mode, we do not apply a scaling to the mid * because we need the normalized mid for folding later */ - cm = quant_band_template(f, rc, band, X, NULL, N, mbits, blocks, - lowband, duration, next_lowband_out1, - next_level, stereo ? 1.0f : (gain * mid), - lowband_scratch, fill, quant); + cm = quant ? ff_celt_encode_band(f, rc, band, X, NULL, N, mbits, blocks, + lowband, duration, next_lowband_out1, + next_level, stereo ? 1.0f : (gain * mid), + lowband_scratch, fill) : + ff_celt_decode_band(f, rc, band, X, NULL, N, mbits, blocks, + lowband, duration, next_lowband_out1, + next_level, stereo ? 1.0f : (gain * mid), + lowband_scratch, fill); rebalance = mbits - (rebalance - f->remaining2); if (rebalance > 3 << 3 && itheta != 0) @@ -757,17 +761,25 @@ static av_always_inline uint32_t quant_band_template(CeltFrame *f, OpusRangeCode /* For a stereo split, the high bits of fill are always zero, * so no folding will be done to the side. */ - cm |= quant_band_template(f, rc, band, Y, NULL, N, sbits, blocks, - next_lowband2, duration, NULL, - next_level, gain * side, NULL, - fill >> blocks, quant) << ((B0 >> 1) & (stereo - 1)); + cm |= quant ? ff_celt_encode_band(f, rc, band, Y, NULL, N, sbits, blocks, + next_lowband2, duration, NULL, + next_level, gain * side, NULL, + fill >> blocks) << ((B0 >> 1) & (stereo - 1)) : + ff_celt_decode_band(f, rc, band, Y, NULL, N, sbits, blocks, + next_lowband2, duration, NULL, + next_level, gain * side, NULL, + fill >> blocks) << ((B0 >> 1) & (stereo - 1)); } else { /* For a stereo split, the high bits of fill are always zero, * so no folding will be done to the side. */ - cm = quant_band_template(f, rc, band, Y, NULL, N, sbits, blocks, - next_lowband2, duration, NULL, - next_level, gain * side, NULL, - fill >> blocks, quant) << ((B0 >> 1) & (stereo - 1)); + cm = quant ? ff_celt_encode_band(f, rc, band, Y, NULL, N, sbits, blocks, + next_lowband2, duration, NULL, + next_level, gain * side, NULL, + fill >> blocks) << ((B0 >> 1) & (stereo - 1)) : + ff_celt_decode_band(f, rc, band, Y, NULL, N, sbits, blocks, + next_lowband2, duration, NULL, + next_level, gain * side, NULL, + fill >> blocks) << ((B0 >> 1) & (stereo - 1)); rebalance = sbits - (rebalance - f->remaining2); if (rebalance > 3 << 3 && itheta != 16384) @@ -775,10 +787,14 @@ static av_always_inline uint32_t quant_band_template(CeltFrame *f, OpusRangeCode /* In stereo mode, we do not apply a scaling to the mid because * we need the normalized mid for folding later */ - cm |= quant_band_template(f, rc, band, X, NULL, N, mbits, blocks, - lowband, duration, next_lowband_out1, - next_level, stereo ? 1.0f : (gain * mid), - lowband_scratch, fill, quant); + cm |= quant ? ff_celt_encode_band(f, rc, band, X, NULL, N, mbits, blocks, + lowband, duration, next_lowband_out1, + next_level, stereo ? 1.0f : (gain * mid), + lowband_scratch, fill) : + ff_celt_decode_band(f, rc, band, X, NULL, N, mbits, blocks, + lowband, duration, next_lowband_out1, + next_level, stereo ? 1.0f : (gain * mid), + lowband_scratch, fill); } } } else {

1 0

avcodec/mdec: Fix runtime error: left shift of negative value -127
by Michael Niedermayer 27 Apr '17

27 Apr '17

ffmpeg | branch: master | Michael Niedermayer <michael(a)niedermayer.cc> | Thu Apr 27 02:27:16 2017 +0200| [6ca82975b7a8eaf676a52738ec8e7e36732327cc] | committer: Michael Niedermayer avcodec/mdec: Fix runtime error: left shift of negative value -127 Fixes undefined behavior Fixes: 1275/clusterfuzz-testcase-minimized-6718162017976320 Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg Signed-off-by: Michael Niedermayer <michael(a)niedermayer.cc> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=6ca82975b7a8eaf67… --- libavcodec/mdec.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libavcodec/mdec.c b/libavcodec/mdec.c index 1cc4ca4742..42bd561cd7 100644 --- a/libavcodec/mdec.c +++ b/libavcodec/mdec.c @@ -73,7 +73,7 @@ static inline int mdec_decode_block_intra(MDECContext *a, int16_t *block, int n) if (diff >= 0xffff) return AVERROR_INVALIDDATA; a->last_dc[component] += diff; - block[0] = a->last_dc[component] << 3; + block[0] = a->last_dc[component] * (1 << 3); } i = 0;

1 0