ffmpeg.git
6 years agoffplay: use av_malloc_array()
Michael Niedermayer [Mon, 25 Aug 2014 16:32:53 +0000 (18:32 +0200)]
ffplay: use av_malloc_array()

Reviewed-by: Marton Balint <cus@passwd.hu>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agolibavformat/ftp: Do not leak memory in routine ftp_features
Alexander Strasser [Tue, 19 Aug 2014 22:41:55 +0000 (00:41 +0200)]
libavformat/ftp: Do not leak memory in routine ftp_features

Setting the pointer to NULL inside both ftp_send_command
and ftp_features is redundant. Generally always setting to
NULL in ftp_send_command seems safer, but throughout the file
that parameter was always passed initialized. So I do it here
too for consistency.

Should fix CID1231988 (RESOURCE_LEAK)

OKed-by: Lukasz Marek <lukasz.m.luki2@gmail.com>
Signed-off-by: Alexander Strasser <eclipse7@gmx.net>
6 years agohuffyuvdec: decode the last odd sample
Christophe Gisquet [Mon, 25 Aug 2014 20:24:30 +0000 (20:24 +0000)]
huffyuvdec: decode the last odd sample

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agohuffyuvenc: write last odd sample
Christophe Gisquet [Mon, 25 Aug 2014 20:24:29 +0000 (20:24 +0000)]
huffyuvenc: write last odd sample

If width is odd, last sample wouldn't be written.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoavcodec/vp9: Use av_malloc_array()
Michael Niedermayer [Mon, 25 Aug 2014 16:33:11 +0000 (18:33 +0200)]
avcodec/vp9: Use av_malloc_array()

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agolavf/matroskadec: Ensure cues_end is initialized
Vignesh Venkatasubramanian [Mon, 25 Aug 2014 16:15:13 +0000 (09:15 -0700)]
lavf/matroskadec: Ensure cues_end is initialized

Ensure that cues_start and cues_end are always initialized and
the webm_dash_manifest_cues function returns appropriate error if
they are not computed correctly.

This fixes coverity defect CID1231991.

Signed-off-by: Vignesh Venkatasubramanian <vigneshv@google.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoavutil/opencl: use av_malloc(z)_array()
Michael Niedermayer [Mon, 25 Aug 2014 14:06:10 +0000 (16:06 +0200)]
avutil/opencl: use av_malloc(z)_array()

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoavcodec/vorbisenc: use av_malloc(z)_array()
Michael Niedermayer [Mon, 25 Aug 2014 14:03:54 +0000 (16:03 +0200)]
avcodec/vorbisenc: use av_malloc(z)_array()

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agodoc/APIchanges: fill in dates and hashes
Michael Niedermayer [Mon, 25 Aug 2014 13:57:37 +0000 (15:57 +0200)]
doc/APIchanges: fill in dates and hashes

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoavcodec/mpegvideo_parser: consider vbv_delay in bitrate heuristic also for mpeg2
Michael Niedermayer [Wed, 20 Aug 2014 18:01:40 +0000 (20:01 +0200)]
avcodec/mpegvideo_parser: consider vbv_delay in bitrate heuristic also for mpeg2

Fixes bitrate detection in CBR mpeg2
Fixes ticket3678

6 years agoMerge commit '6ee1cb5740e7490151db7dcec7e20ceaf8a2fe1f'
Michael Niedermayer [Mon, 25 Aug 2014 11:26:42 +0000 (13:26 +0200)]
Merge commit '6ee1cb5740e7490151db7dcec7e20ceaf8a2fe1f'

* commit '6ee1cb5740e7490151db7dcec7e20ceaf8a2fe1f':
  libavformat: use MSG_NOSIGNAL when applicable

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit 'b263f8ffe7599d9cd27ec477a12700da8eb2790d'
Michael Niedermayer [Mon, 25 Aug 2014 10:27:19 +0000 (12:27 +0200)]
Merge commit 'b263f8ffe7599d9cd27ec477a12700da8eb2790d'

* commit 'b263f8ffe7599d9cd27ec477a12700da8eb2790d':
  lavf: add AVFormatContext.max_ts_probe

Conflicts:
doc/APIchanges
libavformat/avformat.h
libavformat/utils.c
libavformat/version.h

lavf-fate/mp3 changes as the estimated input bitrate changes and that is
copied to the output

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit '6ca11f7157d0ffd11ea9a4211b04981b46dc75d6'
Michael Niedermayer [Mon, 25 Aug 2014 10:09:29 +0000 (12:09 +0200)]
Merge commit '6ca11f7157d0ffd11ea9a4211b04981b46dc75d6'

* commit '6ca11f7157d0ffd11ea9a4211b04981b46dc75d6':
  doc/APIchanges: fill in missing hashes and dates

Conflicts:
doc/APIchanges

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agolibavformat: use MSG_NOSIGNAL when applicable
Rémi Denis-Courmont [Wed, 20 Aug 2014 20:06:07 +0000 (23:06 +0300)]
libavformat: use MSG_NOSIGNAL when applicable

If the remote end of a connection oriented socket hangs up, generating
an EPIPE error is preferable over an unhandled SIGPIPE signal.

Signed-off-by: Martin Storsjö <martin@martin.st>
6 years agolavf: add AVFormatContext.max_ts_probe
Anton Khirnov [Tue, 19 Aug 2014 17:49:58 +0000 (17:49 +0000)]
lavf: add AVFormatContext.max_ts_probe

It allows to configure how long will avformat_find_stream_info() wait
to get the first timestamp.

6 years agodoc/APIchanges: fill in missing hashes and dates
Anton Khirnov [Sun, 18 May 2014 08:52:41 +0000 (10:52 +0200)]
doc/APIchanges: fill in missing hashes and dates

6 years agoRevert "pnmenc: use bits_per_raw_sample"
Michael Niedermayer [Mon, 25 Aug 2014 00:59:17 +0000 (02:59 +0200)]
Revert "pnmenc: use bits_per_raw_sample"

This reverts commit 1506ea947d3c527f67a5caa93cc7fa0d064bbf1f.

Revert requested by carl as well as author in "Re: [FFmpeg-devel] [PATCH]Fix pnm encoding with bpc set"

6 years agoavcodec/vc1dec: fix null pointer dereference
Michael Niedermayer [Mon, 25 Aug 2014 00:17:32 +0000 (02:17 +0200)]
avcodec/vc1dec: fix null pointer dereference

Fixes Ticket3860

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoffserver: tests, use Metadata title in config
Reynaldo H. Verdejo Pinochet [Fri, 22 Aug 2014 22:29:18 +0000 (18:29 -0400)]
ffserver: tests, use Metadata title in config

'Title' was deprecated with 9985710a5a927d52c345c4bdcc88f5c014ee2fe7

Signed-off-by: Reynaldo H. Verdejo Pinochet <reynaldo@osg.samsung.com>
6 years agoffserver: tests, use new port/bindaddress config
Reynaldo H. Verdejo Pinochet [Wed, 20 Aug 2014 04:49:38 +0000 (00:49 -0400)]
ffserver: tests, use new port/bindaddress config

Comming from commit 1013d8dd6967f1e776c08dc133ed6ae44ca3dc64
the old Port and BindAddress config vars have been deprecated
in favor of HTTPPort and HTTPBindAddress respectively.

Signed-off-by: Reynaldo H. Verdejo Pinochet <reynaldo@osg.samsung.com>
6 years agoavcodec/utils: add GBRP16 to avcodec_align_dimensions2()
Michael Niedermayer [Sun, 24 Aug 2014 21:33:40 +0000 (23:33 +0200)]
avcodec/utils: add GBRP16 to avcodec_align_dimensions2()

Fixes Ticket3869

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit '424b929b5cb9ca4094099f25179829260d4b0fa3'
Michael Niedermayer [Sun, 24 Aug 2014 19:01:29 +0000 (21:01 +0200)]
Merge commit '424b929b5cb9ca4094099f25179829260d4b0fa3'

* commit '424b929b5cb9ca4094099f25179829260d4b0fa3':
  pulse: Add a wallclock option to be compatible with other other captures

Conflicts:
libavdevice/pulse_audio_dec.c

wallclock mode was supported and default in FFmpeg already before this commit.
its thus left the default

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoavformat/matroskaenc: Check alpha_mode
Michael Niedermayer [Wed, 20 Aug 2014 16:40:20 +0000 (18:40 +0200)]
avformat/matroskaenc: Check alpha_mode

Fixes CID1231992

Suggested-by: Timothy Gu <timothygu99@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agolavc/h264_slice: Add a missing newline to an error message.
Carl Eugen Hoyos [Sun, 24 Aug 2014 13:33:10 +0000 (15:33 +0200)]
lavc/h264_slice: Add a missing newline to an error message.

6 years agoMerge remote-tracking branch 'ubitux/codecview'
Michael Niedermayer [Sun, 24 Aug 2014 12:53:31 +0000 (14:53 +0200)]
Merge remote-tracking branch 'ubitux/codecview'

* ubitux/codecview:
  avfilter: add codecview filter

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoavfilter: add codecview filter
Clément Bœsch [Mon, 18 Aug 2014 14:32:26 +0000 (16:32 +0200)]
avfilter: add codecview filter

6 years agopulse: Add a wallclock option to be compatible with other other captures
Luca Barbato [Sat, 23 Aug 2014 17:03:21 +0000 (19:03 +0200)]
pulse: Add a wallclock option to be compatible with other other captures

alsa and x11grab use av_gettime() to report timestamps.

Bug-Id: 647

6 years agoCorrect few "ffmpeg" typos
Paul B Mahol [Sat, 23 Aug 2014 19:39:22 +0000 (19:39 +0000)]
Correct few "ffmpeg" typos

Signed-off-by: Paul B Mahol <onemda@gmail.com>
6 years agoavcodec/xan: fix style issue
Paul B Mahol [Sat, 23 Aug 2014 19:22:13 +0000 (19:22 +0000)]
avcodec/xan: fix style issue

Signed-off-by: Paul B Mahol <onemda@gmail.com>
6 years agox86: hevc_mc: split differently calls
Christophe Gisquet [Sun, 24 Aug 2014 08:46:30 +0000 (08:46 +0000)]
x86: hevc_mc: split differently calls

In some cases, 2 or 3 calls are performed to functions for unusual
widths. Instead, perform 2 calls for different widths to split the
workload.

The 8+16 and 4+8 widths for respectively 8 and more than 8 bits can't
be processed that way without modifications: some calls use unaligned
buffers, and having branches to handle this was resulting in no
micro-benchmark benefit.

For block_w == 12 (around 1% of the pixels of the sequence):
Before:
12758 decicycles in epel_uni, 4093 runs, 3 skips
19389 decicycles in qpel_uni, 8187 runs, 5 skips
22699 decicycles in epel_bi, 32743 runs, 25 skips
34736 decicycles in qpel_bi, 32733 runs, 35 skips

After:
11929 decicycles in epel_uni, 4096 runs, 0 skips
18131 decicycles in qpel_uni, 8184 runs, 8 skips
20065 decicycles in epel_bi, 32750 runs, 18 skips
31458 decicycles in qpel_bi, 32753 runs, 15 skips

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agox86: hevc_mc: correct unneeded use of SSE4 code
Christophe Gisquet [Sun, 24 Aug 2014 08:46:31 +0000 (08:46 +0000)]
x86: hevc_mc: correct unneeded use of SSE4 code

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agopnmenc: use bits_per_raw_sample
Christophe Gisquet [Wed, 20 Aug 2014 08:10:45 +0000 (08:10 +0000)]
pnmenc: use bits_per_raw_sample

This allows writing actual bitdepth in RGB48 when it isn't actually 16.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoChangelog: add "H.261 RTP payload format (RFC 4587)"
Michael Niedermayer [Sun, 24 Aug 2014 02:11:37 +0000 (04:11 +0200)]
Changelog: add "H.261 RTP payload format (RFC 4587)"

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge remote-tracking branch 'cehoyos/master'
Michael Niedermayer [Sun, 24 Aug 2014 01:58:11 +0000 (03:58 +0200)]
Merge remote-tracking branch 'cehoyos/master'

* cehoyos/master:
  fate: Fix ffprobe tests with --target-path set.
  doc/ffmpeg: Try to clarify that the input option -r is not the same as -framerate.
  Do not print a useless error number if mov header reading fails.

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoAdd support for H.261 RTP payload format (RFC 4587)
ThomasVolkert [Sat, 23 Aug 2014 19:25:39 +0000 (21:25 +0200)]
Add support for H.261 RTP payload format (RFC 4587)

6 years agofate: Fix ffprobe tests with --target-path set.
Carl Eugen Hoyos [Sat, 16 Aug 2014 10:19:05 +0000 (12:19 +0200)]
fate: Fix ffprobe tests with --target-path set.

6 years agodoc/ffmpeg: Try to clarify that the input option -r is not the same as -framerate.
Carl Eugen Hoyos [Sun, 24 Aug 2014 00:49:06 +0000 (02:49 +0200)]
doc/ffmpeg: Try to clarify that the input option -r is not the same as -framerate.

6 years agoDo not print a useless error number if mov header reading fails.
Carl Eugen Hoyos [Sun, 24 Aug 2014 00:45:58 +0000 (02:45 +0200)]
Do not print a useless error number if mov header reading fails.

The error string is printed by ffmpeg as for other demuxers.

6 years agoavcodec/tiff: Make pixel format checks tighter
Michael Niedermayer [Sat, 23 Aug 2014 23:54:45 +0000 (01:54 +0200)]
avcodec/tiff: Make pixel format checks tighter

Fixes assertion failure with imgsrc001-lzma.tiff

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agolavc/tiff: add support for LZMA compression
James Almer [Sat, 23 Aug 2014 02:39:57 +0000 (23:39 -0300)]
lavc/tiff: add support for LZMA compression

Derived from deflate code.
Requires liblzma.

Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit '749b1f1359b5af0a08221923b016551b18ab6171'
Michael Niedermayer [Sat, 23 Aug 2014 20:46:06 +0000 (22:46 +0200)]
Merge commit '749b1f1359b5af0a08221923b016551b18ab6171'

* commit '749b1f1359b5af0a08221923b016551b18ab6171':
  configure: add --enable-rpath

Conflicts:
configure

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit '5caf039ba2b4be067568a30146f29008d8db28d0'
Michael Niedermayer [Sat, 23 Aug 2014 20:08:15 +0000 (22:08 +0200)]
Merge commit '5caf039ba2b4be067568a30146f29008d8db28d0'

* commit '5caf039ba2b4be067568a30146f29008d8db28d0':
  Prepare for 11_beta2 Release

Conflicts:
RELEASE

Not merged as theres no FFmpeg 11_beta2

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit 'f9f34cb9983ec6f4ef119c34b726d3b39c143110'
Michael Niedermayer [Sat, 23 Aug 2014 19:39:14 +0000 (21:39 +0200)]
Merge commit 'f9f34cb9983ec6f4ef119c34b726d3b39c143110'

* commit 'f9f34cb9983ec6f4ef119c34b726d3b39c143110':
  ogg: Use separate classes for the aliases

Conflicts:
libavformat/oggenc.c

See: 2ccc6ff03acc3ca31db1aeb828c747d05b5cb6aa
Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoavutil/pixelutils: faster pixelutils_sad_16x16
Clément Bœsch [Sat, 23 Aug 2014 18:03:10 +0000 (20:03 +0200)]
avutil/pixelutils: faster pixelutils_sad_16x16

501 to 439 decicycles.

See 45c7f3997ea11c3d1007b2126b1c0049a8c27105.

6 years agoconfigure: add --enable-rpath
Reinhard Tartler [Sat, 23 Aug 2014 14:04:26 +0000 (10:04 -0400)]
configure: add --enable-rpath

This option facilitates testing shared libarary builds: for instance
fate builders do no longer need to set LD_LIBRARY_PATH as the binaries will
get the right search paths hardcoded into their executable file.

This option is only meant to be used for testing purposes: The installed
libraries must not move around in the file system, and doing so will
cause a lot of subtle problems. For more information why using RPATH is
dangerous, please refer to

https://blog.flameeyes.eu/2010/06/the-why-and-how-of-rpath

6 years agoPrepare for 11_beta2 Release
Reinhard Tartler [Sat, 23 Aug 2014 14:12:16 +0000 (10:12 -0400)]
Prepare for 11_beta2 Release

6 years agodoc/filters: fix Shwo/Show typo
Clément Bœsch [Sat, 23 Aug 2014 14:07:55 +0000 (16:07 +0200)]
doc/filters: fix Shwo/Show typo

6 years agodict.c: Free non-strduped av_dict_set arguments on error.
Reimar Döffinger [Tue, 29 Jul 2014 19:16:35 +0000 (21:16 +0200)]
dict.c: Free non-strduped av_dict_set arguments on error.

Unfortunately this was not explicitly documented and thus
might be risky.
But all uses I could find in FFmpeg and one in VLC had a memleak
in these cases, and I could not find any that relied on the previous
behaviour.

Signed-off-by: Reimar Döffinger <Reimar.Doeffinger@gmx.de>
6 years agoavutil/pixelutils: faster pixelutils_sad_[au]_16x16
Clément Bœsch [Thu, 14 Aug 2014 20:30:55 +0000 (22:30 +0200)]
avutil/pixelutils: faster pixelutils_sad_[au]_16x16

~560 → ~500 decicycles

This is following the comments from Michael in
https://ffmpeg.org/pipermail/ffmpeg-devel/2014-August/160599.html

Using 2 registers for accumulator didn't help. On the other hand,
some re-ordering between the movs and psadbw allowed going ~538 to ~500.

6 years agoavfilter/showcqt: add fontcolor option
Muhammad Faiz [Sat, 23 Aug 2014 08:01:22 +0000 (10:01 +0200)]
avfilter/showcqt: add fontcolor option

6 years agoavcodec/h264_slice: More complete cleanup in h264_slice_header_init()
Michael Niedermayer [Thu, 21 Aug 2014 14:33:03 +0000 (16:33 +0200)]
avcodec/h264_slice: More complete cleanup in h264_slice_header_init()

Fixes null pointer dereference
Fixes Ticket3873

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoavcodec/h264: do proper cleanup in ff_h264_alloc_tables() in case DPB alloc fails
Michael Niedermayer [Thu, 21 Aug 2014 14:15:16 +0000 (16:15 +0200)]
avcodec/h264: do proper cleanup in ff_h264_alloc_tables() in case DPB alloc fails

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoogg: Use separate classes for the aliases
Luca Barbato [Sat, 23 Aug 2014 00:42:18 +0000 (02:42 +0200)]
ogg: Use separate classes for the aliases

Unbreak 051aadeed104ecbe8ee4850ec2d7e5394f5e1ccd

6 years agoavcodec/imc: Fix bitstream buffer padding
Michael Niedermayer [Fri, 22 Aug 2014 23:34:28 +0000 (01:34 +0200)]
avcodec/imc: Fix bitstream buffer padding

Fixes buffer overread

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoh264: do not return on sidedata allocation failure
Christophe Gisquet [Fri, 22 Aug 2014 14:49:54 +0000 (16:49 +0200)]
h264: do not return on sidedata allocation failure

Not having allocated it is not a good reason to leave the object
in an undetermined state. Though a particular setting like the
AV_EF_* flags could be useful to control that behaviour.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoavcodec/hevc_ps: fix 1 vs. 0 typo
Michael Niedermayer [Fri, 22 Aug 2014 21:00:41 +0000 (23:00 +0200)]
avcodec/hevc_ps: fix 1 vs. 0 typo

Found-by: Timothy Gu <timothygu99@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agovidstabutils: improve documentation
Timothy Gu [Fri, 22 Aug 2014 02:12:10 +0000 (19:12 -0700)]
vidstabutils: improve documentation

Signed-off-by: Timothy Gu <timothygu99@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agovidstab*: Remove accidentally exported av_2_vs_pixel_format()
Timothy Gu [Fri, 22 Aug 2014 02:12:09 +0000 (19:12 -0700)]
vidstab*: Remove accidentally exported av_2_vs_pixel_format()

Also correctly namespace other functions in vidstabutils, and decrease
difference from Libav.

Initial-patch-by: Vittorio Giovara <vittorio.giovara@gmail.com>
Signed-off-by: Timothy Gu <timothygu99@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit '3526ab891c28396ada8b58bf7647309bab30de1d'
Michael Niedermayer [Fri, 22 Aug 2014 19:55:00 +0000 (21:55 +0200)]
Merge commit '3526ab891c28396ada8b58bf7647309bab30de1d'

* commit '3526ab891c28396ada8b58bf7647309bab30de1d':
  qt-faststart: Undefine fseeko/ftello before defining them

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit '1019b7c4edff537499c4a6cb0d65abae04ce58f6'
Michael Niedermayer [Fri, 22 Aug 2014 19:49:00 +0000 (21:49 +0200)]
Merge commit '1019b7c4edff537499c4a6cb0d65abae04ce58f6'

* commit '1019b7c4edff537499c4a6cb0d65abae04ce58f6':
  os_support: Undefine lseek/stat/fstat before defining them

Conflicts:
libavformat/os_support.h

See: ef122ff5072463366c020157f0a27aad7e6610db
See: ed3c0fe85dd5c228e98126b18e50ffb617e77070
Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit '051aadeed104ecbe8ee4850ec2d7e5394f5e1ccd'
Michael Niedermayer [Fri, 22 Aug 2014 19:14:43 +0000 (21:14 +0200)]
Merge commit '051aadeed104ecbe8ee4850ec2d7e5394f5e1ccd'

* commit '051aadeed104ecbe8ee4850ec2d7e5394f5e1ccd':
  ogg: Provide aliases for Speex, Opus and audio-only ogg

Conflicts:
Changelog
libavformat/oggenc.c
libavformat/version.h

See: 2ccc6ff03acc3ca31db1aeb828c747d05b5cb6aa
Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit 'cb7b1a2dfb547ab78342a7a9d5cd729d77d90421'
Michael Niedermayer [Fri, 22 Aug 2014 18:59:49 +0000 (20:59 +0200)]
Merge commit 'cb7b1a2dfb547ab78342a7a9d5cd729d77d90421'

* commit 'cb7b1a2dfb547ab78342a7a9d5cd729d77d90421':
  electronicarts: set the framerate for TGQ/TQI

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit '7b6aae23e12f41cdfac7f1069debfe44d9a3d136'
Michael Niedermayer [Fri, 22 Aug 2014 17:52:49 +0000 (19:52 +0200)]
Merge commit '7b6aae23e12f41cdfac7f1069debfe44d9a3d136'

* commit '7b6aae23e12f41cdfac7f1069debfe44d9a3d136':
  electronicarts: read the framerate for MAD

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agowavpackenc: reset trailer info on block encoding
Christophe Gisquet [Fri, 22 Aug 2014 09:31:45 +0000 (09:31 +0000)]
wavpackenc: reset trailer info on block encoding

In some cases, in particular if several blocks are needed because of
the channel layout (e.g. 2.1), the information used to write the
trailing bits terminating the sample data was not reset.

This would cause potential desync on the decoder, although decoded
samples were actually mostly fine.

Fixes ticket #3879.

6 years agowavpack: check number of channels
Christophe Gisquet [Fri, 22 Aug 2014 09:31:44 +0000 (09:31 +0000)]
wavpack: check number of channels

This means container and codec disagree. The codec is supposed to
know better so this could be an error instead.

6 years agoMerge commit '4d6c5152849e23a4cc0f6a6ac2880c01ebcd301b'
Michael Niedermayer [Fri, 22 Aug 2014 17:01:26 +0000 (19:01 +0200)]
Merge commit '4d6c5152849e23a4cc0f6a6ac2880c01ebcd301b'

* commit '4d6c5152849e23a4cc0f6a6ac2880c01ebcd301b':
  electronicarts: do not fail on zero-sized chunks

Conflicts:
libavformat/electronicarts.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit 'dc4b2e7d33903a6b9380e8a84b22b3a20facbb08'
Michael Niedermayer [Fri, 22 Aug 2014 16:49:22 +0000 (18:49 +0200)]
Merge commit 'dc4b2e7d33903a6b9380e8a84b22b3a20facbb08'

* commit 'dc4b2e7d33903a6b9380e8a84b22b3a20facbb08':
  rv34: use ff_mpeg_update_thread_context only when decoder is fully initialized

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit 'b0bfd09f88da8b7c7666faf0ac7d5e74559dba9f'
Michael Niedermayer [Fri, 22 Aug 2014 16:43:26 +0000 (18:43 +0200)]
Merge commit 'b0bfd09f88da8b7c7666faf0ac7d5e74559dba9f'

* commit 'b0bfd09f88da8b7c7666faf0ac7d5e74559dba9f':
  configure: Suppress "potentially uninitialized variable" warnings from MSVC

Conflicts:
configure

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoqt-faststart: Undefine fseeko/ftello before defining them
Diego Biurrun [Fri, 15 Aug 2014 19:43:48 +0000 (21:43 +0200)]
qt-faststart: Undefine fseeko/ftello before defining them

This avoids a number of redefinition warnings on MinGW64.

6 years agoos_support: Undefine lseek/stat/fstat before defining them
Diego Biurrun [Fri, 15 Aug 2014 19:42:56 +0000 (21:42 +0200)]
os_support: Undefine lseek/stat/fstat before defining them

This avoids a number of redefinition warnings on MinGW64.

6 years agox86: hevcdsp: use compilation-time-fixed constant
Christophe Gisquet [Mon, 28 Jul 2014 17:17:28 +0000 (17:17 +0000)]
x86: hevcdsp: use compilation-time-fixed constant

The stride for some buffers is known.

Reviewed-by: Mickaël Raulet <mraulet@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agohevcdsp: remove more instances of compile-time-fixed parameters
Christophe Gisquet [Mon, 28 Jul 2014 10:13:06 +0000 (12:13 +0200)]
hevcdsp: remove more instances of compile-time-fixed parameters

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agohevcdsp: remove compilation-time-fixed parameter
Christophe Gisquet [Mon, 28 Jul 2014 17:17:26 +0000 (17:17 +0000)]
hevcdsp: remove compilation-time-fixed parameter

The dststride parameter is always MAX_PB_SIZE.

Reviewed-by: Mickaël Raulet <mraulet@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agohevc: move MAX_PB_SIZE declaration
Christophe Gisquet [Mon, 28 Jul 2014 17:17:25 +0000 (17:17 +0000)]
hevc: move MAX_PB_SIZE declaration

Reviewed-by: Mickaël Raulet <mraulet@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoogg: Provide aliases for Speex, Opus and audio-only ogg
Luca Barbato [Sun, 17 Aug 2014 18:55:37 +0000 (20:55 +0200)]
ogg: Provide aliases for Speex, Opus and audio-only ogg

Since they are aliases for ogg enabling any of them enables ogg as well.

6 years agox86: hevc_mc: assume 2nd source stride is 64
Christophe Gisquet [Mon, 28 Jul 2014 06:55:26 +0000 (08:55 +0200)]
x86: hevc_mc: assume 2nd source stride is 64

Reviewed-by: Mickaël Raulet <mraulet@gmail.com
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoelectronicarts: set the framerate for TGQ/TQI
Anton Khirnov [Wed, 20 Aug 2014 06:06:41 +0000 (06:06 +0000)]
electronicarts: set the framerate for TGQ/TQI

It is hardcoded to 15fps.

6 years agoelectronicarts: read the framerate for MAD
Anton Khirnov [Wed, 20 Aug 2014 05:52:44 +0000 (05:52 +0000)]
electronicarts: read the framerate for MAD

6 years agoelectronicarts: do not fail on zero-sized chunks
Anton Khirnov [Wed, 20 Aug 2014 05:40:53 +0000 (05:40 +0000)]
electronicarts: do not fail on zero-sized chunks

At least one FATE sample contains such chunks and happens to work simply
by accident (due to find_stream_info() swallowing the error).

CC: libav-stable@libav.org
6 years agorv34: use ff_mpeg_update_thread_context only when decoder is fully initialized
Janne Grunau [Thu, 21 Aug 2014 11:26:33 +0000 (13:26 +0200)]
rv34: use ff_mpeg_update_thread_context only when decoder is fully initialized

MpegEncContext based decoders are only fully initialized after the first
ff_thread_get_buffer() call. The RV30/40 decoders may fail before a frame
buffer was requested. ff_mpeg_update_thread_context() fails on half
initialized MpegEncContexts. Since this can only happen before a the
first frame was decoded there is no need to call
ff_mpeg_update_thread_context().

Based on patches by John Stebbins and tested by John Stebbins.

CC: libav-stable@libav.org
6 years agowavpackenc: make assert more thorough
Christophe Gisquet [Fri, 22 Aug 2014 09:31:43 +0000 (09:31 +0000)]
wavpackenc: make assert more thorough

It was only validating that normal data wasn't filling the buffer.
However, extra data may be written afterwards.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agowavpackenc: fix number of samples per block
Christophe Gisquet [Fri, 22 Aug 2014 09:31:42 +0000 (09:31 +0000)]
wavpackenc: fix number of samples per block

Currently, the encoder will try to reduce it down to 150000, but the
decoder will complain starting at 131072 (WV_MAX_SAMPLES). Therefore,
change the loop limit.

Fixes ticket #3881.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoavcodec/fic: Check if a frame is available before using it
Michael Niedermayer [Thu, 14 Aug 2014 03:32:44 +0000 (05:32 +0200)]
avcodec/fic: Check if a frame is available before using it

Fixes null pointer dereference
Fixes: ficvf.avi
Found-by: Piotr Bandurski <ami_stuff@o2.pl>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoavcodec: fix aac/ac3 parser bitstream buffer size
Michael Niedermayer [Thu, 21 Aug 2014 23:15:57 +0000 (01:15 +0200)]
avcodec: fix aac/ac3 parser bitstream buffer size

Buffers containing copies of the AAC and AC3 header bits were not padded
before parsing, violating init_get_bits() buffer padding requirement,
leading to potential buffer read overflows.
This change adds FF_INPUT_BUFFER_PADDING_SIZE bytes to the bit buffer
for parsing the header in each of aac_parser.c and ac3_parser.c.

Based on patch by: Matt Wolenetz <wolenetz@chromium.org>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agodpxenc: fix padding in encode_gbrp12
Christophe Gisquet [Wed, 20 Aug 2014 08:10:44 +0000 (08:10 +0000)]
dpxenc: fix padding in encode_gbrp12

It was added per pixel instead of per line.

Reviewed-by: James Darnley <james.darnley@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agohevc_ps: check overflow and test alternate syntax
Christophe Gisquet [Thu, 21 Aug 2014 16:57:18 +0000 (18:57 +0200)]
hevc_ps: check overflow and test alternate syntax

Some streams were found to have what appeared to be truncated SPS.
Their syntax seem to be valid at least until the end of the VUI, so
try that syntax if the parsing would overflow the SPS in the
conforming syntax.

Fixes ticket #3872.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit '13c90bc9a359e969cc2b7f7e8199b02a0e4c6ec9'
Michael Niedermayer [Thu, 21 Aug 2014 19:56:33 +0000 (21:56 +0200)]
Merge commit '13c90bc9a359e969cc2b7f7e8199b02a0e4c6ec9'

* commit '13c90bc9a359e969cc2b7f7e8199b02a0e4c6ec9':
  adts: Return more meaningful error codes

Conflicts:
libavformat/adtsenc.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit '593aaee953f8b07c141ff115e67bae85ef0350c7'
Michael Niedermayer [Thu, 21 Aug 2014 19:47:43 +0000 (21:47 +0200)]
Merge commit '593aaee953f8b07c141ff115e67bae85ef0350c7'

* commit '593aaee953f8b07c141ff115e67bae85ef0350c7':
  setpts: Add missing inttypes.h #include for PRId64

Conflicts:
libavfilter/setpts.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit '8fc6a70c2167b645b7a37d0cbc0e276e7b787cc9'
Michael Niedermayer [Thu, 21 Aug 2014 19:37:39 +0000 (21:37 +0200)]
Merge commit '8fc6a70c2167b645b7a37d0cbc0e276e7b787cc9'

* commit '8fc6a70c2167b645b7a37d0cbc0e276e7b787cc9':
  mpeg12enc: Add missing #include for PICT_FRAME

See: 9517900bef528527d8507da5a1992f99513d71b4
Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit '7cb66ebc0be48489785f7166c9d15eac594b0763'
Michael Niedermayer [Thu, 21 Aug 2014 19:30:31 +0000 (21:30 +0200)]
Merge commit '7cb66ebc0be48489785f7166c9d15eac594b0763'

* commit '7cb66ebc0be48489785f7166c9d15eac594b0763':
  error_resilience: Drop asserts from guess_mv()

Conflicts:
libavcodec/error_resilience.c

Not merged, the asserts in FFmpeg build and work fine

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoMerge commit '11cd727fbd603197cb1e49654fce3352d56f8fd8'
Michael Niedermayer [Thu, 21 Aug 2014 19:23:14 +0000 (21:23 +0200)]
Merge commit '11cd727fbd603197cb1e49654fce3352d56f8fd8'

* commit '11cd727fbd603197cb1e49654fce3352d56f8fd8':
  vsrc_movie: Adjust a silly typo from b977b287f61fea48ecd6251d54a26334213b7ec6

Conflicts:
libavfilter/src_movie.c

No change, the typo is not part of FFmpeg

Merged-by: Michael Niedermayer <michaelni@gmx.at>
6 years agox86/hevc_res_add: refactor ff_hevc_transform_add{16,32}_8
James Almer [Wed, 20 Aug 2014 22:36:29 +0000 (19:36 -0300)]
x86/hevc_res_add: refactor ff_hevc_transform_add{16,32}_8

* Reduced xmm register count to 7 (As such they are now enabled for x86_32).
* Removed four movdqa (affects the sse2 version only).
* pxor is now used to clear m0 only once.

~5% faster.

Reviewed-by: Christophe Gisquet <christophe.gisquet@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
6 years agowavpackenc: assert on too small buffer
Christophe Gisquet [Tue, 19 Aug 2014 12:05:56 +0000 (14:05 +0200)]
wavpackenc: assert on too small buffer

bytestream2_* will not cause buffer overflow, but in that case, this means
the allocation would be incorrect and the encoded result invalid. Therefore,
assert no overflow occurred.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoconfigure: Suppress "potentially uninitialized variable" warnings from MSVC
Diego Biurrun [Fri, 15 Aug 2014 18:03:27 +0000 (20:03 +0200)]
configure: Suppress "potentially uninitialized variable" warnings from MSVC

The same is done for GCC and clang already.

6 years agoavformat/mov: use 64bit for size in mov_skip_multiple_stsd()
Michael Niedermayer [Thu, 21 Aug 2014 15:28:11 +0000 (17:28 +0200)]
avformat/mov: use 64bit for size in mov_skip_multiple_stsd()

Fixes integer overflow
Fixes Ticket 3866

Reviewed-by: Christophe Gisquet <christophe.gisquet@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agodoc/filters/apad: extend documentation
Stefano Sabatini [Mon, 18 Aug 2014 12:52:08 +0000 (14:52 +0200)]
doc/filters/apad: extend documentation

6 years agolavfi/apad: fix logic when whole_len or pad_len options are specified
Stefano Sabatini [Mon, 18 Aug 2014 12:51:25 +0000 (14:51 +0200)]
lavfi/apad: fix logic when whole_len or pad_len options are specified

In particular, allow pad_len and whole_len to have value set to 0, which
means that no padding will be added. Previously a value set to 0 meant
that that the filter had to pad forever.

The new semantics is clearer, also simplifies scripting since the option
value might be automatically computed, so that no checks need to be done
in case it is 0.

The old semantics was never documented and the logic was broken (the
filter was always adding samples indefinitely), so this should not break
backward compatibility.

6 years agoimc: reject files with unfathomable sampling rates
Christophe Gisquet [Thu, 21 Aug 2014 10:21:07 +0000 (12:21 +0200)]
imc: reject files with unfathomable sampling rates

With huge sampling rates, the table derivation method does not converge fast
enough. While fixing it using e.g. Newton-Rhapson-like methods (the curve is
nicely convex) is possible, it is much simpler to reject these cases.

The value of 96000 was arbitrarily chosen as a realistic value, though
1000000 would still work and converge.

Fixes ticket #3868.

Suggested-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoavfilter/vf_lenscorrection: fix memleak
Michael Niedermayer [Thu, 21 Aug 2014 10:16:07 +0000 (12:16 +0200)]
avfilter/vf_lenscorrection: fix memleak

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
6 years agoadts: Return more meaningful error codes
Nidhi Makhijani [Fri, 15 Aug 2014 11:37:24 +0000 (17:07 +0530)]
adts: Return more meaningful error codes

Signed-off-by: Diego Biurrun <diego@biurrun.de>