]> git.sesse.net Git - ffmpeg/log
ffmpeg
10 years agompeg12decdata: Remove unused #define
Diego Biurrun [Wed, 14 Aug 2013 16:57:08 +0000 (18:57 +0200)]
mpeg12decdata: Remove unused #define

10 years agoavcodec: Replace local extern declarations for tables with header #includes
Diego Biurrun [Thu, 15 Aug 2013 10:58:41 +0000 (12:58 +0200)]
avcodec: Replace local extern declarations for tables with header #includes

10 years agoswscale: Move extern declarations for tables to swscale_internal.h
Diego Biurrun [Thu, 15 Aug 2013 11:38:12 +0000 (13:38 +0200)]
swscale: Move extern declarations for tables to swscale_internal.h

Also add missing ff_ prefixes where necessary.

10 years agoswscale: Mark a bunch of tables only used within one file static
Diego Biurrun [Thu, 15 Aug 2013 11:29:01 +0000 (13:29 +0200)]
swscale: Mark a bunch of tables only used within one file static

10 years agoivi_common: Make some tables only used within the file static
Diego Biurrun [Thu, 15 Aug 2013 11:13:01 +0000 (13:13 +0200)]
ivi_common: Make some tables only used within the file static

10 years agortpproto: Check the right feature detection macro
Martin Storsjö [Thu, 15 Aug 2013 07:36:20 +0000 (10:36 +0300)]
rtpproto: Check the right feature detection macro

IPPROTO_IPV6 is unrelated here (it's only used in udp.c for
multicast sockopts), check for support for the sockaddr_in6
struct itself.

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agoelectronicarts: Let functions always returning the same value return void
Diego Biurrun [Wed, 14 Aug 2013 16:04:36 +0000 (18:04 +0200)]
electronicarts: Let functions always returning the same value return void

10 years agoelectronicarts: Improve some function/variable names
Diego Biurrun [Wed, 14 Aug 2013 15:06:41 +0000 (17:06 +0200)]
electronicarts: Improve some function/variable names

10 years agoelectronicarts: comment wording fixes
Diego Biurrun [Wed, 14 Aug 2013 14:53:08 +0000 (16:53 +0200)]
electronicarts: comment wording fixes

10 years agoelectronicarts: Remove bogus function documentation
Diego Biurrun [Wed, 14 Aug 2013 15:54:40 +0000 (17:54 +0200)]
electronicarts: Remove bogus function documentation

10 years agoelectronicarts: K&R formatting cosmetics
Diego Biurrun [Wed, 14 Aug 2013 14:45:06 +0000 (16:45 +0200)]
electronicarts: K&R formatting cosmetics

10 years ago8bps: decode 24bit files correctly as rgb32 on bigendian
Janne Grunau [Fri, 19 Jul 2013 11:51:51 +0000 (13:51 +0200)]
8bps: decode 24bit files correctly as rgb32 on bigendian

10 years agortpproto: Check for the right feature when reading a sockaddr_in6
Dave Yeo [Wed, 14 Aug 2013 22:22:13 +0000 (15:22 -0700)]
rtpproto: Check for the right feature when reading a sockaddr_in6

Some systems, such as OS/2, define AF_INET6 without a full
implementation.

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agosdp: Add an option for sending RTCP packets to the source of the last packets
Martin Storsjö [Tue, 13 Aug 2013 10:21:35 +0000 (13:21 +0300)]
sdp: Add an option for sending RTCP packets to the source of the last packets

An SDP description normally only contains the target IP address
and port for the packets. This means that we don't really have
any clue where to send the RTCP RR packets - previously they're
sent to the destination IP written in the SDP (at the same port),
which rarely is the actual peer. And if the source for the packets
is on a different port than the destination, it's never correct.

With a new option, we can choose to send the packets to the
address that the latest packet on each socket arrived from.
---
Some may even argue that this should be the default - perhaps,
but I'd rather keep it optional at first. Additionally, I'm not
sure if sending RTCP RR directly back to the source is
desireable for e.g. multicast.

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agortpproto: Add an option for writing return packets to the address of the last receive...
Martin Storsjö [Tue, 13 Aug 2013 10:20:42 +0000 (13:20 +0300)]
rtpproto: Add an option for writing return packets to the address of the last received packets

If we've received packets on the same socket before, the return
packets are sent to that address. If we've only received packets
on the other socket, try to guess the source port for the other
one assuming the basic +1/-1 logic.

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agortpproto: Update the parameter documentation
Martin Storsjö [Tue, 13 Aug 2013 09:19:57 +0000 (12:19 +0300)]
rtpproto: Update the parameter documentation

Move the sources documentation up below the marker for deprecated
otpions. Also mention the new block parameter, that was added
in 749722209.

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agosegafilm: Error out on impossible packet size
Luca Barbato [Tue, 13 Aug 2013 05:40:38 +0000 (07:40 +0200)]
segafilm: Error out on impossible packet size

Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
10 years agoogg: Always alloc the private context in vorbis_header
Luca Barbato [Tue, 13 Aug 2013 05:28:41 +0000 (07:28 +0200)]
ogg: Always alloc the private context in vorbis_header

It is possible to have an initial broken header and then valid packets.

Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
10 years agortjpeg: Use init_get_bits8
Luca Barbato [Tue, 13 Aug 2013 05:05:28 +0000 (07:05 +0200)]
rtjpeg: Use init_get_bits8

CC:libav-stable@libav.org

10 years agortjpeg: return meaningful error codes
Luca Barbato [Sun, 11 Aug 2013 18:32:40 +0000 (20:32 +0200)]
rtjpeg: return meaningful error codes

10 years agonuv: Use av_fast_realloc
Luca Barbato [Tue, 13 Aug 2013 05:01:40 +0000 (07:01 +0200)]
nuv: Use av_fast_realloc

The decompressed buffer can be used after codec_reinit, so it must be
preserved.

Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
10 years agonuv: Reset the frame on resize
Luca Barbato [Tue, 13 Aug 2013 04:01:48 +0000 (06:01 +0200)]
nuv: Reset the frame on resize

Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
10 years agonuv: Pad the lzo outbuf
Luca Barbato [Mon, 12 Aug 2013 09:34:06 +0000 (11:34 +0200)]
nuv: Pad the lzo outbuf

And properly update the buf_size with the correct size.

Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
10 years agonuv: Do not ignore lzo decompression failures
Luca Barbato [Sun, 11 Aug 2013 22:16:12 +0000 (00:16 +0200)]
nuv: Do not ignore lzo decompression failures

Update the fate reference since the last broken frame is not decoded
anymore.

Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
10 years agopng: allow encoding 16-bit grayscale
Ian Taylor [Tue, 13 Aug 2013 02:02:55 +0000 (22:02 -0400)]
png: allow encoding 16-bit grayscale

Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
10 years agowmall: use AVFrame API properly
Hendrik Leppkes [Thu, 8 Aug 2013 16:48:43 +0000 (18:48 +0200)]
wmall: use AVFrame API properly

This fixes a bug with non-refcounted callers resulting in invalid memory access.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
10 years agortmp: Do not misuse memcmp
Luca Barbato [Thu, 8 Aug 2013 17:44:19 +0000 (19:44 +0200)]
rtmp: Do not misuse memcmp

CC: libav-stable@libav.org
10 years agodoc: Clarify the avconv section about -re
Martin Storsjö [Sat, 10 Aug 2013 09:13:08 +0000 (12:13 +0300)]
doc: Clarify the avconv section about -re

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agodoc: Add an example on publishing over RTMP
Martin Storsjö [Sat, 10 Aug 2013 09:08:43 +0000 (12:08 +0300)]
doc: Add an example on publishing over RTMP

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agodoc: Add librtmp to the section header for the librtmp specific details
Martin Storsjö [Sat, 10 Aug 2013 09:03:54 +0000 (12:03 +0300)]
doc: Add librtmp to the section header for the librtmp specific details

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agodoc: Explain that the default RTMP user agent is different when publishing
Martin Storsjö [Sat, 10 Aug 2013 08:51:49 +0000 (11:51 +0300)]
doc: Explain that the default RTMP user agent is different when publishing

The fact that a different user agent is used is cruicial for getting
publishing authentication working. (When using librtmp, this other
user agent has to be specified manually, but that's not needed
with the libavformat internal RTMP support.)

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agodoc: Extend the rtmp example to include how to pass username/password
Martin Storsjö [Sat, 10 Aug 2013 08:51:31 +0000 (11:51 +0300)]
doc: Extend the rtmp example to include how to pass username/password

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agortmp: Detect and warn if the user tries to pass librtmp style parameters
Martin Storsjö [Sat, 10 Aug 2013 08:53:31 +0000 (11:53 +0300)]
rtmp: Detect and warn if the user tries to pass librtmp style parameters

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agortmp: rename data_size to size
Luca Barbato [Thu, 8 Aug 2013 16:52:11 +0000 (18:52 +0200)]
rtmp: rename data_size to size

10 years agortmp: Use PRId64 when needed.
Luca Barbato [Thu, 8 Aug 2013 15:43:15 +0000 (17:43 +0200)]
rtmp: Use PRId64 when needed.

10 years agolavc: Add refcounted api to AVPacket
Luca Barbato [Wed, 10 Jul 2013 08:52:56 +0000 (10:52 +0200)]
lavc: Add refcounted api to AVPacket

Provide a clean way to manipulate packets.

10 years agoh264: check one context_init() allocation
Vittorio Giovara [Fri, 2 Aug 2013 16:36:28 +0000 (18:36 +0200)]
h264: check one context_init() allocation

Signed-off-by: Anton Khirnov <anton@khirnov.net>
10 years agoh264: use explicit variable names for *_field_flag
Vittorio Giovara [Fri, 2 Aug 2013 16:36:29 +0000 (18:36 +0200)]
h264: use explicit variable names for *_field_flag

Signed-off-by: Anton Khirnov <anton@khirnov.net>
10 years agoh264: return meaningful values
Vittorio Giovara [Fri, 2 Aug 2013 16:36:27 +0000 (18:36 +0200)]
h264: return meaningful values

Signed-off-by: Anton Khirnov <anton@khirnov.net>
10 years agolibavutil: Make avpriv_open a library-internal function on msvcrt
Martin Storsjö [Fri, 9 Aug 2013 08:06:46 +0000 (11:06 +0300)]
libavutil: Make avpriv_open a library-internal function on msvcrt

Add one copy of the function into each of the libraries, similarly
to what we do for log2_tab. When using static libs, only one
copy of the file_open.o object file gets included, while when
using shared libraries, each of them get a copy of its own.

This fixes DLL builds with a statically linked C runtime, where
each DLL effectively has got its own instance of the C runtime,
where file descriptors can't be shared across runtimes.

On systems not using msvcrt, the function is not duplicated.

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agolibavutil: Move avpriv_open to a new file, file_open.c
Martin Storsjö [Fri, 9 Aug 2013 07:58:15 +0000 (10:58 +0300)]
libavutil: Move avpriv_open to a new file, file_open.c

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agobktr: Changed a missed occurrance of open into avpriv_open
Martin Storsjö [Fri, 9 Aug 2013 07:16:57 +0000 (10:16 +0300)]
bktr: Changed a missed occurrance of open into avpriv_open

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agolibavutil: Enable the MSVC DLL symbol loading workaround in shared builds as well
Martin Storsjö [Thu, 8 Aug 2013 14:16:26 +0000 (17:16 +0300)]
libavutil: Enable the MSVC DLL symbol loading workaround in shared builds as well

This used to only be necessary in static builds (when using the
dynamically linked C runtime), since the _imp prefixed symbols do
exist when linking to the actual DLL. When building testprogs,
however, the current library (e.g. libavutil for some of the testprogs)
is linked statically.

This fixes make fate on DLL builds when using the dynamically
linked C runtime.

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agoff_socket: put out-of-line and fallback to fcntl() for close-on-exec
Rémi Denis-Courmont [Thu, 8 Aug 2013 19:26:42 +0000 (22:26 +0300)]
ff_socket: put out-of-line and fallback to fcntl() for close-on-exec

This supports non-Linux systems (SOCK_CLOEXEC is non-standard) and
older Linux kernels to the extent possible.

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agoFATE: add a TAK test
Anton Khirnov [Tue, 6 Aug 2013 07:40:54 +0000 (09:40 +0200)]
FATE: add a TAK test

10 years agofile: Move win32 utf8->wchar open wrapper to libavutil
Martin Storsjö [Thu, 8 Aug 2013 08:29:57 +0000 (11:29 +0300)]
file: Move win32 utf8->wchar open wrapper to libavutil

When libavformat was changed to use the new avpriv_open function
in 51eb213d001, this silently bypassed the existing wrapper for
win32. Move the win32 wrapper into libavutil/file.c to make sure
it gets called everywhere (not just in the libavformat case).

This makes sure that non-ascii file names gets opened properly
(where file names internally are stored as utf8, but they get
converted to wchar_t and opened with _wsopen).

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agoarm: Add assembly version of h264_find_start_code_candidate
Ben Avison [Mon, 5 Aug 2013 12:12:48 +0000 (13:12 +0100)]
arm: Add assembly version of h264_find_start_code_candidate

               Before          After
               Mean   StdDev   Mean   StdDev  Change
This function   508.8 23.4      185.4  9.0    +174.4%
Overall        3068.5 31.7     2752.1 29.4     +11.5%

In combination with the preceding patch:
                Before          After
                Mean   StdDev   Mean   StdDev  Change
Overall         2925.6 26.2     2752.1 29.4     +6.3%

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agoh264dsp: Factorize code into a new function, h264_find_start_code_candidate
Ben Avison [Mon, 5 Aug 2013 12:12:47 +0000 (13:12 +0100)]
h264dsp: Factorize code into a new function, h264_find_start_code_candidate

This performs the start code search which was previously part of
h264_find_frame_end() - the most CPU intensive part of the function.

By itself, this results in a performance regression:
              Before          After
              Mean   StdDev   Mean   StdDev  Change
Overall time  2925.6 26.2     3068.5 31.7    -4.7%

but this can more than be made up for by platform-optimised
implementations of the function.

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agoh264_parser: Initialize the h264dsp context in the parser as well
Ben Avison [Mon, 5 Aug 2013 12:12:46 +0000 (13:12 +0100)]
h264_parser: Initialize the h264dsp context in the parser as well

Each AVStream struct for an H.264 elementary stream actually has two
copies of the H264DSPContext struct (and in fact all the other members
of H264Context as well):

((H264Context *) ((AVStream *)st)->codec->priv_data)->h264dsp
((H264Context *) ((AVStream *)st)->parser->priv_data)->h264dsp

but only the first of these was actually being initialised. This
prevented the addition of platform-specific implementations of
parser-related functions.

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agoVoxware MetaSound decoder
Kostya Shishkov [Mon, 5 Aug 2013 10:55:30 +0000 (12:55 +0200)]
Voxware MetaSound decoder

10 years agoavio: Add const qualifiers to ffio_read_indirect
Ben Avison [Wed, 7 Aug 2013 22:22:38 +0000 (23:22 +0100)]
avio: Add const qualifiers to ffio_read_indirect

This provides at least some protection against potential accidental
corruption of AVIO buffer workspace.

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agolibavutil: Don't use fcntl if the function does not exist
Derek Buitenhuis [Wed, 7 Aug 2013 23:44:37 +0000 (19:44 -0400)]
libavutil: Don't use fcntl if the function does not exist

Not all platforms have the function.

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agocmdutils: Only do the windows-specific commandline parsing on _WIN32
Diogo Franco [Wed, 7 Aug 2013 11:25:51 +0000 (08:25 -0300)]
cmdutils: Only do the windows-specific commandline parsing on _WIN32

Fixes commandline parsing on Cygwin (on 64 bit, and on very recent
32 bit), where the configure check does find the CommandLineToArgvW
function (since it exists in the link libraries and in the headers),
but whose GetCommandLineW() only returns the application's path.
(This is due to a cygwin internal optimization, see
http://cygwin.com/ml/cygwin/2013-07/msg00538.html for details.)
Arguments are only given through main's argc/argv, and they're already
UTF-8.

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agolibavformat: use avpriv_open()
Rémi Denis-Courmont [Tue, 6 Aug 2013 18:19:28 +0000 (21:19 +0300)]
libavformat: use avpriv_open()

Signed-off-by: Anton Khirnov <anton@khirnov.net>
10 years agolibavdevice: use avpriv_open()
Rémi Denis-Courmont [Tue, 6 Aug 2013 18:19:27 +0000 (21:19 +0300)]
libavdevice: use avpriv_open()

Signed-off-by: Anton Khirnov <anton@khirnov.net>
10 years agolibavcodec: use avpriv_open()
Rémi Denis-Courmont [Tue, 6 Aug 2013 18:19:26 +0000 (21:19 +0300)]
libavcodec: use avpriv_open()

Signed-off-by: Anton Khirnov <anton@khirnov.net>
10 years agolibavutil: use avpriv_open()
Rémi Denis-Courmont [Tue, 6 Aug 2013 18:19:25 +0000 (21:19 +0300)]
libavutil: use avpriv_open()

Signed-off-by: Anton Khirnov <anton@khirnov.net>
10 years agolibavutil: add avpriv_open() to open files with close-on-exec flag
Rémi Denis-Courmont [Tue, 6 Aug 2013 18:19:24 +0000 (21:19 +0300)]
libavutil: add avpriv_open() to open files with close-on-exec flag

Signed-off-by: Anton Khirnov <anton@khirnov.net>
10 years agompegvideo_enc: drop outdated copy_picture_attributes() in favour of a modern av_frame...
Vittorio Giovara [Fri, 2 Aug 2013 07:25:45 +0000 (09:25 +0200)]
mpegvideo_enc: drop outdated copy_picture_attributes() in favour of a modern av_frame_copy_props()

Signed-off-by: Anton Khirnov <anton@khirnov.net>
10 years agortsp: Don't include the listen flag in the SDP demuxer flags
Martin Storsjö [Wed, 7 Aug 2013 07:46:38 +0000 (10:46 +0300)]
rtsp: Don't include the listen flag in the SDP demuxer flags

It's only relevant for the RTSP demuxer. Similarly, the custom_io
flag is only present in the SDP demuxer options list.

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agodeprecate AV_CODEC_ID_VOXWARE and introduce AV_CODEC_ID_METASOUND instead
Kostya Shishkov [Tue, 6 Aug 2013 07:57:05 +0000 (09:57 +0200)]
deprecate AV_CODEC_ID_VOXWARE and introduce AV_CODEC_ID_METASOUND instead

Voxware is the name of company, it has produced several audio codecs e.g.
MetaVoice family and MetaSound.

10 years agortpenc_chain: Check for errors from ffio_fdopen and ffio_open_dyn_packet_buf
Martin Storsjö [Wed, 7 Aug 2013 07:20:07 +0000 (10:20 +0300)]
rtpenc_chain: Check for errors from ffio_fdopen and ffio_open_dyn_packet_buf

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agortspenc: Check the return value from ffio_open_dyn_packet_buf
Martin Storsjö [Wed, 7 Aug 2013 07:07:48 +0000 (10:07 +0300)]
rtspenc: Check the return value from ffio_open_dyn_packet_buf

Also clear the AVIOContext handle after freeing, to avoid
possible dangling pointers if the later call fails.

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agopcm_bluray: Return AVERROR_INVALIDDATA instead of -1 on header errors
Christian Schmidt [Tue, 6 Aug 2013 08:21:42 +0000 (10:21 +0200)]
pcm_bluray: Return AVERROR_INVALIDDATA instead of -1 on header errors

Signed-off-by: Diego Biurrun <diego@biurrun.de>
10 years agoconfigure: Properly split dv1394 indev dependencies
Diego Biurrun [Tue, 6 Aug 2013 00:36:17 +0000 (02:36 +0200)]
configure: Properly split dv1394 indev dependencies

10 years agoconfigure: Properly set zlib dependencies for all components
Diego Biurrun [Tue, 6 Aug 2013 00:31:33 +0000 (02:31 +0200)]
configure: Properly set zlib dependencies for all components

10 years agoconfigure: The W64 demuxer should select the WAV demuxer, not depend on it
Diego Biurrun [Mon, 5 Aug 2013 21:05:30 +0000 (23:05 +0200)]
configure: The W64 demuxer should select the WAV demuxer, not depend on it

10 years agoconfigure: Properly split avserver component and system dependencies
Diego Biurrun [Tue, 27 Nov 2012 20:48:34 +0000 (21:48 +0100)]
configure: Properly split avserver component and system dependencies

10 years agoriff: Move demuxing code to a separate file.
Diego Biurrun [Sun, 4 Aug 2013 12:33:36 +0000 (14:33 +0200)]
riff: Move demuxing code to a separate file.

10 years agoriff: Move muxing code to a separate file
Diego Biurrun [Sun, 4 Aug 2013 12:10:54 +0000 (14:10 +0200)]
riff: Move muxing code to a separate file

10 years agoavformat/dv: K&R formatting cosmetics
Diego Biurrun [Tue, 6 Aug 2013 00:55:02 +0000 (02:55 +0200)]
avformat/dv: K&R formatting cosmetics

10 years agotcp: Use a different log message and level if there's more addresses to try
Martin Storsjö [Mon, 5 Aug 2013 16:44:20 +0000 (19:44 +0300)]
tcp: Use a different log message and level if there's more addresses to try

This lowers the level of warnings printed if trying to connect
to a host name that provides both v6 and v4 addresses but the
service only is available on the v4 address (often occurring for
'localhost', with servers that aren't v6-aware).

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agounix: Convert from AVERROR to errno range before comparing error codes
Martin Storsjö [Mon, 5 Aug 2013 16:39:08 +0000 (19:39 +0300)]
unix: Convert from AVERROR to errno range before comparing error codes

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agoavcodec/options: Drop deprecation warning suppression macros
Diego Biurrun [Mon, 5 Aug 2013 20:28:26 +0000 (22:28 +0200)]
avcodec/options: Drop deprecation warning suppression macros

The options table is used in tools built by the host compiler and the
deprecation macros pull in bits that are not safe to use if host and
target compiler differ.

10 years agoriff.h: Remove stray extern declaration for non-existing symbol
Diego Biurrun [Sun, 4 Aug 2013 12:08:47 +0000 (14:08 +0200)]
riff.h: Remove stray extern declaration for non-existing symbol

10 years agotwinvq: Split VQF-specific part from common TwinVQ decoder core
Kostya Shishkov [Sun, 4 Aug 2013 15:18:34 +0000 (17:18 +0200)]
twinvq: Split VQF-specific part from common TwinVQ decoder core

Signed-off-by: Diego Biurrun <diego@biurrun.de>
10 years agotwinvq: Prefix enums and defines shared with VoxWare MetaSound
Diego Biurrun [Mon, 5 Aug 2013 10:37:01 +0000 (12:37 +0200)]
twinvq: Prefix enums and defines shared with VoxWare MetaSound

10 years agotwinvq: move all bitstream reading into single place
Kostya Shishkov [Sun, 4 Aug 2013 15:17:25 +0000 (17:17 +0200)]
twinvq: move all bitstream reading into single place

This is required for the future addition of VoxWare MetaSound decoder, for its
functions are mostly the same but bitstream reader is completely different
and bitstream format is slightly different too.

Signed-off-by: Diego Biurrun <diego@biurrun.de>
10 years agompegts: Remove one 64-bit integer modulus operation per packet
Ben Avison [Mon, 5 Aug 2013 12:12:49 +0000 (13:12 +0100)]
mpegts: Remove one 64-bit integer modulus operation per packet

The common case of the pointer having increased by one packet (which results
in no change to the modulus) can be detected with a 64-bit subtraction,
which is far cheaper than a division on many platforms.

           Before          After
           Mean   StdDev   Mean   StdDev  Change
Divisions   248.3  8.8      51.5   7.4    +381.7%
Overall    2773.2 25.6     2372.5 43.1     +16.9%

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agovc1: check mb_height validity.
Luca Barbato [Mon, 5 Aug 2013 04:30:24 +0000 (06:30 +0200)]
vc1: check mb_height validity.

Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
10 years agovc1: check the source buffer in vc1_mc functions
Luca Barbato [Mon, 5 Aug 2013 04:27:12 +0000 (06:27 +0200)]
vc1: check the source buffer in vc1_mc functions

Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
10 years agobink: Bound check the quantization matrix.
Luca Barbato [Sun, 4 Aug 2013 16:48:20 +0000 (18:48 +0200)]
bink: Bound check the quantization matrix.

Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
10 years agonetwork: Use SOCK_CLOEXEC when available
Luca Barbato [Sat, 3 Aug 2013 13:00:11 +0000 (15:00 +0200)]
network: Use SOCK_CLOEXEC when available

10 years agolavf: Support unix sockets
Luca Barbato [Mon, 29 Jul 2013 23:34:56 +0000 (01:34 +0200)]
lavf: Support unix sockets

10 years agolibx264: add shortcut for the bluray compatibility option
Luca Barbato [Mon, 5 Aug 2013 11:23:52 +0000 (13:23 +0200)]
libx264: add shortcut for the bluray compatibility option

As for intra-refresh it is just a commodity.

10 years agompegts: Make discard_pid() faster for single-program streams
Ben Avison [Mon, 5 Aug 2013 12:12:51 +0000 (13:12 +0100)]
mpegts: Make discard_pid() faster for single-program streams

When a stream contains a single program, there's no point in doing a
PID -> program lookup. Normally the one and only program isn't disabled,
so no packets should be discarded.

              Before          After
              Mean   StdDev   Mean   StdDev  Change
discard_pid()   73.8  9.4       20.2  1.5    +264.8%
Overall       2300.8 28.0     2253.1 20.6      +2.1%

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agompegts: Remove one memcpy per packet
Ben Avison [Wed, 31 Jul 2013 22:46:08 +0000 (23:46 +0100)]
mpegts: Remove one memcpy per packet

This was being performed to ensure that a complete packet was held in
contiguous memory, prior to parsing the packet. However, the source buffer
is typically large enough that the packet was already contiguous, so it is
beneficial to return the packet by reference in most cases.

         Before          After
         Mean   StdDev   Mean   StdDev  Change
memcpy    720.7  32.7     649.8  25.1   +10.9%
Overall  2372.7  46.1    2291.7  21.8    +3.5%

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agoavio: Add an internal function for reading without copying
Ben Avison [Wed, 31 Jul 2013 22:46:08 +0000 (23:46 +0100)]
avio: Add an internal function for reading without copying

As long as there is enough contiguous data in the avio buffer,
just return a pointer to it instead of copying it to the caller
provided buffer.

Signed-off-by: Martin Storsjö <martin@martin.st>
10 years agobuild: Add missing img2.o dependency to apetag.o
Diego Biurrun [Sun, 4 Aug 2013 23:43:13 +0000 (01:43 +0200)]
build: Add missing img2.o dependency to apetag.o

10 years agoismindex: Replace mkdir ifdeffery by os_support.h #include
Diego Biurrun [Sun, 4 Aug 2013 22:06:06 +0000 (00:06 +0200)]
ismindex: Replace mkdir ifdeffery by os_support.h #include

os_support.h contains more precise workarounds for non-POSIX mkdir().

10 years agovdpau: deprecate bitstream buffers within the hardware context
Rémi Denis-Courmont [Thu, 25 Jul 2013 19:30:21 +0000 (22:30 +0300)]
vdpau: deprecate bitstream buffers within the hardware context

The bitstream buffers are now private and freed by libavcodec. For
backward compatibility, the hold bitstream buffer pointer is left NULL
(applications were supposed to av_freep() it).

Signed-off-by: Anton Khirnov <anton@khirnov.net>
10 years agovdpau: store picture data in picture's rather than codec's context
Rémi Denis-Courmont [Thu, 25 Jul 2013 19:30:20 +0000 (22:30 +0300)]
vdpau: store picture data in picture's rather than codec's context

Signed-off-by: Anton Khirnov <anton@khirnov.net>
10 years agovdpau: deprecate VDPAU codec capability
Rémi Denis-Courmont [Thu, 25 Jul 2013 19:30:19 +0000 (22:30 +0300)]
vdpau: deprecate VDPAU codec capability

Signed-off-by: Anton Khirnov <anton@khirnov.net>
10 years agovdpau: deprecate old codec-specific pixel formats
Rémi Denis-Courmont [Thu, 25 Jul 2013 19:30:18 +0000 (22:30 +0300)]
vdpau: deprecate old codec-specific pixel formats

Signed-off-by: Anton Khirnov <anton@khirnov.net>
10 years agovdpau: remove old-style decoders
Rémi Denis-Courmont [Thu, 25 Jul 2013 19:30:17 +0000 (22:30 +0300)]
vdpau: remove old-style decoders

Signed-off-by: Anton Khirnov <anton@khirnov.net>
10 years agoavconv: support -t as an input option.
Anton Khirnov [Tue, 18 Jun 2013 09:12:09 +0000 (11:12 +0200)]
avconv: support -t as an input option.

It limits the duration of the data read from a given input.

10 years agoavconv: make input -ss accurate when transcoding
Anton Khirnov [Sat, 15 Jun 2013 07:59:40 +0000 (09:59 +0200)]
avconv: make input -ss accurate when transcoding

Insert (a)trim filters on the corresponding inputs, so the extra frames
are decoded and discarded.

10 years agoavconv: distinguish between -ss 0 and -ss not being used
Anton Khirnov [Sat, 15 Jun 2013 07:35:10 +0000 (09:35 +0200)]
avconv: distinguish between -ss 0 and -ss not being used

Using -ss 0 to drop frames with negative timestamps is a perfectly valid
use case.

10 years agoavconv: insert extra filters in the same way for both graph inputs and outputs
Anton Khirnov [Fri, 14 Jun 2013 16:25:30 +0000 (18:25 +0200)]
avconv: insert extra filters in the same way for both graph inputs and outputs

This will allow using the same code when one filter can be inserted both
on input and output.

10 years agolavfi/fifo: fix flushing when using request_samples
Anton Khirnov [Sun, 4 Aug 2013 10:10:23 +0000 (12:10 +0200)]
lavfi/fifo: fix flushing when using request_samples

If any samples are still buffered when request_frame returns EOF, they
won't be returned currently.