platform/upstream/libav.git
12 years agoReport an error if pitch_lag is zero in AMR-NB decoder.
Oana Stratulat [Wed, 4 Jan 2012 23:08:05 +0000 (01:08 +0200)]
Report an error if pitch_lag is zero in AMR-NB decoder.

This fixes an infinite loop in the decoder on specially
crafted files, and fixes bug 151.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
12 years agoRevert "4xm: Prevent buffer overreads."
Ronald S. Bultje [Thu, 5 Jan 2012 05:27:31 +0000 (21:27 -0800)]
Revert "4xm: Prevent buffer overreads."

This reverts commit 295a7c0238e84b0ffa8f21ed938d45f51f54a4cd. The
patch breaks decoding of regular files (e.g. fate-4xm-2).

12 years ago4xm: Prevent buffer overreads.
Aneesh Dogra [Wed, 4 Jan 2012 19:58:21 +0000 (01:28 +0530)]
4xm: Prevent buffer overreads.

4xm decoder while decoding i2 frames can overread the buffer if proper checks
are not made.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
12 years ago4xm: pass the correct remaining buffer size to decode_i2_frame().
Aneesh Dogra [Wed, 4 Jan 2012 18:40:01 +0000 (00:10 +0530)]
4xm: pass the correct remaining buffer size to decode_i2_frame().

frame_size is the number of bytes left in the packet, so if we are passing
buf-4 we can safely read frame_size+4 bytes.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
12 years ago4xm: fix calculation of the next output line position in decode_i2_frame().
Aneesh Dogra [Wed, 4 Jan 2012 19:12:26 +0000 (00:42 +0530)]
4xm: fix calculation of the next output line position in decode_i2_frame().

The current code doesn't work unless width is an exact multiple of 16.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
12 years agomtv: Make sure audio_subsegments is not 0
Shitiz Garg [Thu, 15 Dec 2011 19:06:27 +0000 (00:36 +0530)]
mtv: Make sure audio_subsegments is not 0

audio_subsegments would be 0 and cause floating point exceptions
Fixes bugzilla #144

Signed-off-by: Justin Ruggles <justin.ruggles@gmail.com>
12 years agov4l2: use V4L2_FMT_FLAG_EMULATED only if it is defined
Janne Grunau [Wed, 4 Jan 2012 19:29:57 +0000 (20:29 +0100)]
v4l2: use V4L2_FMT_FLAG_EMULATED only if it is defined

V4L2_FMT_FLAG_EMULATED was added in 2.6.32.

12 years agoavconv: add symbolic names for -vsync parameters
Anton Khirnov [Sat, 26 Nov 2011 10:01:33 +0000 (11:01 +0100)]
avconv: add symbolic names for -vsync parameters

12 years agoflvdec: Fix compiler warning for uninitialized variables
Jean First [Fri, 30 Dec 2011 20:08:02 +0000 (21:08 +0100)]
flvdec: Fix compiler warning for uninitialized variables

These can't be used uninitialized in practice, but the
compiler doesn't realize it.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agortsp: Fix compiler warning for uninitialized variable
Jean First [Tue, 3 Jan 2012 22:01:28 +0000 (23:01 +0100)]
rtsp: Fix compiler warning for uninitialized variable

This one won't ever be used uninitialized in practice, but
the compiler doesn't realize it.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agoulti: convert to new bytestream API.
Ronald S. Bultje [Wed, 4 Jan 2012 04:41:18 +0000 (20:41 -0800)]
ulti: convert to new bytestream API.

12 years agoswscale: Use standard multiple inclusion guards in ppc/ header files.
Diego Biurrun [Tue, 3 Jan 2012 09:29:19 +0000 (10:29 +0100)]
swscale: Use standard multiple inclusion guards in ppc/ header files.

12 years agoPlace some START_TIMER invocations in separate blocks.
Diego Biurrun [Tue, 3 Jan 2012 12:25:25 +0000 (13:25 +0100)]
Place some START_TIMER invocations in separate blocks.

This fixes compilation failures related to START_TIMER/STOP_TIMER macros and
-Werror=declaration-after-statement.  START_TIMER declares variables and thus
may not be placed after statements outside of a new block.

12 years agov4l2: list available formats
Luca Barbato [Tue, 27 Dec 2011 05:31:41 +0000 (06:31 +0100)]
v4l2: list available formats

Make use of the experimental framesize enumeration ioctl if available.

12 years agov4l2: set the proper codec_tag
Luca Barbato [Tue, 27 Dec 2011 06:41:01 +0000 (07:41 +0100)]
v4l2: set the proper codec_tag

Unbreak direct streamcopy.

12 years agov4l2: refactor device_open
Luca Barbato [Tue, 27 Dec 2011 05:27:41 +0000 (06:27 +0100)]
v4l2: refactor device_open

Check capabilities directly in the function, further simplify the code.

12 years agov4l2: simplify away io_method
Luca Barbato [Tue, 27 Dec 2011 04:52:02 +0000 (05:52 +0100)]
v4l2: simplify away io_method

Only mmap is supported.

12 years agov4l2: cosmetics
Luca Barbato [Tue, 27 Dec 2011 04:37:58 +0000 (05:37 +0100)]
v4l2: cosmetics

12 years agov4l2: uniform and format options
Luca Barbato [Tue, 27 Dec 2011 02:38:13 +0000 (03:38 +0100)]
v4l2: uniform and format options

12 years agov4l2: do not force interlaced mode
Luca Barbato [Mon, 21 Nov 2011 16:03:46 +0000 (17:03 +0100)]
v4l2: do not force interlaced mode

Video4linux2 supports both interlaced and non-interlaced mode, do not
ask for interlaced if not necessary.

12 years agoavio: exit early in fill_buffer without read_packet
Janne Grunau [Tue, 3 Jan 2012 17:11:36 +0000 (18:11 +0100)]
avio: exit early in fill_buffer without read_packet

Fixes an invalid free() with ass in avi. The sample in bug 98 passes
parts of AVPacket.data as buffer for the AVIOContext. Since the packet
is quite large fill_buffer tries to reallocate the buffer before doing
nothing. Fixes bug 98.

12 years agovc1dec: fix invalid memory access for small video dimensions
John Brooks [Sun, 11 Dec 2011 09:37:21 +0000 (02:37 -0700)]
vc1dec: fix invalid memory access for small video dimensions

For small video dimensions, these calculations of the upper bound
for pixel access may have a negative result. Using an unsigned
comparison to bound a potentially negative value only works if
the greater operand is non-negative. Fixed by doing edge emulation
when the upper bound is probably negative, everywhere that this
pattern appears.

Signed-off-by: Kostya Shishkov <kostya.shishkov@gmail.com>
12 years agorv34: fix invalid memory access for small video dimensions
John Brooks [Wed, 4 Jan 2012 06:27:35 +0000 (23:27 -0700)]
rv34: fix invalid memory access for small video dimensions

For small video dimensions calculations of the upper bound for pixel
access may result in negative value. Using an unsigned comparison
works only if the greater operand is non-negative. This is fixed by
doing edge emulation explicitly for such conditions.

Signed-off-by: Kostya Shishkov <kostya.shishkov@gmail.com>
12 years agorv34: joint coefficient decoding and dequantization
Christophe GISQUET [Sun, 1 Jan 2012 14:28:47 +0000 (15:28 +0100)]
rv34: joint coefficient decoding and dequantization

Perform dequantization while decoding coefficients instead of performing it
on the entire coefficients buffer.

Since quantized coefficients are very sparse, this usually causes a small
speedup. Speedup of around 1% on Panda board compared to the removed here
neon code. Global speedup is probably around 3%.

Signed-off-by: Kostya Shishkov <kostya.shishkov@gmail.com>
12 years agoavplay: Don't call avio_set_interrupt_cb(NULL)
Martin Storsjö [Fri, 30 Dec 2011 10:39:03 +0000 (12:39 +0200)]
avplay: Don't call avio_set_interrupt_cb(NULL)

Since we don't use avio_set_interrupt_cb for interrupt callbacks,
we don't need to call it to reset the interrupt cb either.

This avoids a warning about use of deprecated functions.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agoswscale: remove obsolete comment.
Ronald S. Bultje [Mon, 2 Jan 2012 20:46:40 +0000 (12:46 -0800)]
swscale: remove obsolete comment.

12 years agoswscale: split scale.asm.
Ronald S. Bultje [Mon, 2 Jan 2012 20:31:38 +0000 (12:31 -0800)]
swscale: split scale.asm.

scale.asm keeps horizontal scaling functions, whereas output.asm gets
the vertical scaling/output functions.

12 years agoswscale: don't show full-chroma-int warning for non-RGB output.
Ronald S. Bultje [Mon, 2 Jan 2012 20:03:02 +0000 (12:03 -0800)]
swscale: don't show full-chroma-int warning for non-RGB output.

Non-RGB output always uses full chroma interpolation.

12 years agoswscale: add MMX files to MMX-OBJS instead of OBJS.
Ronald S. Bultje [Tue, 3 Jan 2012 21:05:37 +0000 (13:05 -0800)]
swscale: add MMX files to MMX-OBJS instead of OBJS.

12 years agoh264: return index in buffer on end-of-sequence.
Ronald S. Bultje [Sat, 31 Dec 2011 17:12:53 +0000 (09:12 -0800)]
h264: return index in buffer on end-of-sequence.

Fixes hangs if the last packet contains an end-of-sequence NAL unit,
bug 158.

12 years agoMerge some declarations and initializations.
Diego Biurrun [Tue, 3 Jan 2012 12:18:49 +0000 (13:18 +0100)]
Merge some declarations and initializations.

This fixes compilation failures related to START_TIMER/STOP_TIMER macros and
-Werror=declaration-after-statement.  START_TIMER declares variables and thus
may not be placed after statements outside of a new block.

12 years agocabac: drop unused and disabled get_cabac_u() / get_cabac_ueg() functions
Diego Biurrun [Tue, 27 Dec 2011 10:51:28 +0000 (11:51 +0100)]
cabac: drop unused and disabled get_cabac_u() / get_cabac_ueg() functions

12 years agocabac: drop unused STRICT_LIMITS code branch
Diego Biurrun [Tue, 27 Dec 2011 11:25:18 +0000 (12:25 +0100)]
cabac: drop unused STRICT_LIMITS code branch

12 years agolibspeexenc: fix pts calculations for more than 1 frame per packet
Justin Ruggles [Fri, 30 Dec 2011 22:56:04 +0000 (17:56 -0500)]
libspeexenc: fix pts calculations for more than 1 frame per packet

12 years agoadxdec: clear eof flag and channel states when seeking
Justin Ruggles [Tue, 3 Jan 2012 22:32:15 +0000 (17:32 -0500)]
adxdec: clear eof flag and channel states when seeking

12 years agofate: add ADX encoding/decoding test
Justin Ruggles [Mon, 19 Dec 2011 16:31:14 +0000 (11:31 -0500)]
fate: add ADX encoding/decoding test

12 years agoadx: add an adx muxer
Justin Ruggles [Mon, 19 Dec 2011 16:10:34 +0000 (11:10 -0500)]
adx: add an adx muxer

12 years agoadxenc: check output buffer size before writing
Justin Ruggles [Mon, 19 Dec 2011 15:56:18 +0000 (10:56 -0500)]
adxenc: check output buffer size before writing

12 years agoadxenc: use bytestream functions for header writing.
Justin Ruggles [Mon, 19 Dec 2011 15:51:03 +0000 (10:51 -0500)]
adxenc: use bytestream functions for header writing.

also add more documentation about the header structure

12 years agoadxenc: use BLOCK_SIZE and BLOCK_SAMPLES macros
Justin Ruggles [Mon, 19 Dec 2011 15:23:21 +0000 (10:23 -0500)]
adxenc: use BLOCK_SIZE and BLOCK_SAMPLES macros

12 years agoadxenc: use a loop to encode each channel
Justin Ruggles [Mon, 19 Dec 2011 15:18:09 +0000 (10:18 -0500)]
adxenc: use a loop to encode each channel

12 years agoadxenc: remove unneeded loops
Justin Ruggles [Mon, 19 Dec 2011 15:14:35 +0000 (10:14 -0500)]
adxenc: remove unneeded loops

avctx->frame_size is 32, so that is how many samples we process per call.

12 years agoadx: add AVFMT_GENERIC_INDEX to demuxer flags.
Justin Ruggles [Mon, 19 Dec 2011 15:12:29 +0000 (10:12 -0500)]
adx: add AVFMT_GENERIC_INDEX to demuxer flags.

fixes seeking

12 years agoadxenc: avoid stereo deinterleaving
Justin Ruggles [Mon, 19 Dec 2011 14:52:32 +0000 (09:52 -0500)]
adxenc: avoid stereo deinterleaving

12 years agoadxenc: remove unnecessary setting of coded_frame->key_frame.
Justin Ruggles [Mon, 19 Dec 2011 14:45:50 +0000 (09:45 -0500)]
adxenc: remove unnecessary setting of coded_frame->key_frame.

It is already set by avcodec_alloc_frame().

12 years agoadxenc: log an error message and return AVERROR(EINVAL) for invalid channels
Justin Ruggles [Mon, 19 Dec 2011 14:45:17 +0000 (09:45 -0500)]
adxenc: log an error message and return AVERROR(EINVAL) for invalid channels

12 years agoadxenc: cosmetics: pretty-printing
Justin Ruggles [Mon, 19 Dec 2011 14:39:05 +0000 (09:39 -0500)]
adxenc: cosmetics: pretty-printing

12 years agoadxenc: change some data types
Justin Ruggles [Mon, 19 Dec 2011 14:29:11 +0000 (09:29 -0500)]
adxenc: change some data types

12 years agoadxenc: remove unneeded log message
Justin Ruggles [Mon, 19 Dec 2011 14:28:15 +0000 (09:28 -0500)]
adxenc: remove unneeded log message

12 years agoadxenc: remove unneeded comments
Justin Ruggles [Mon, 19 Dec 2011 14:28:02 +0000 (09:28 -0500)]
adxenc: remove unneeded comments

12 years agoAPIchanges: mention cleaned up header includes
Jonathan Nieder [Tue, 3 Jan 2012 06:40:08 +0000 (00:40 -0600)]
APIchanges: mention cleaned up header includes

Fixes bug 188.

Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Janne Grunau <janne-libav@jannau.net>
12 years agofate: add dxtory test
Oana Stratulat [Sun, 1 Jan 2012 18:55:21 +0000 (20:55 +0200)]
fate: add dxtory test

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
12 years agoadx_parser: rewrite.
Michael Niedermayer [Sun, 18 Dec 2011 17:57:56 +0000 (18:57 +0100)]
adx_parser: rewrite.

The previous code ended in multiple different infinite
loops. See stl_ten_1_big.sfd as example with and without zzuf

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: Justin Ruggles <justin.ruggles@gmail.com>
12 years agoadxdec: Validate channel count to fix a division by zero.
Justin Ruggles [Mon, 19 Dec 2011 13:43:10 +0000 (08:43 -0500)]
adxdec: Validate channel count to fix a division by zero.

12 years agoadxdec: Do not require extradata.
Michael Niedermayer [Sun, 18 Dec 2011 19:10:30 +0000 (20:10 +0100)]
adxdec: Do not require extradata.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: Justin Ruggles <justin.ruggles@gmail.com>
12 years agocmdutils: K&R reformatting cosmetics
Aneesh Dogra [Sat, 31 Dec 2011 12:31:54 +0000 (18:01 +0530)]
cmdutils: K&R reformatting cosmetics

Signed-off-by: Diego Biurrun <diego@biurrun.de>
12 years agoalacdec: implement the 2-pass prediction type.
Justin Ruggles [Wed, 7 Dec 2011 16:56:57 +0000 (11:56 -0500)]
alacdec: implement the 2-pass prediction type.

The reference encoder does not generate any streams using this, but the
reference decoder can handle it, so we should as well.

12 years agoalacenc: implement the 2-pass prediction type.
Justin Ruggles [Wed, 7 Dec 2011 18:01:01 +0000 (13:01 -0500)]
alacenc: implement the 2-pass prediction type.

This isn't used by the reference encoder, but it is supported by the decoder.

12 years agoalacenc: do not generate invalid multi-channel ALAC files
Justin Ruggles [Wed, 7 Dec 2011 17:04:21 +0000 (12:04 -0500)]
alacenc: do not generate invalid multi-channel ALAC files

12 years agoalacdec: fill in missing or guessed info about the extradata format.
Justin Ruggles [Wed, 7 Dec 2011 16:55:24 +0000 (11:55 -0500)]
alacdec: fill in missing or guessed info about the extradata format.

Now that there is official documentation from Apple about this, we don't have
to guess anymore.

12 years agoutvideo: proper median prediction for interlaced videos
Kostya Shishkov [Sun, 1 Jan 2012 16:44:08 +0000 (17:44 +0100)]
utvideo: proper median prediction for interlaced videos

12 years agolavu: bump lavu minor for av_popcount64
Janne Grunau [Tue, 3 Jan 2012 14:30:30 +0000 (15:30 +0100)]
lavu: bump lavu minor for av_popcount64

Add it to doc/APIchanges and update and correct other entries.

12 years agodca: K&R formatting cosmetics
Shitiz Garg [Tue, 3 Jan 2012 10:49:28 +0000 (16:19 +0530)]
dca: K&R formatting cosmetics

Signed-off-by: Diego Biurrun <diego@biurrun.de>
12 years agodct: K&R formatting cosmetics
Aneesh Dogra [Sun, 1 Jan 2012 16:46:37 +0000 (22:16 +0530)]
dct: K&R formatting cosmetics

Signed-off-by: Diego Biurrun <diego@biurrun.de>
12 years agolavf: flush decoders in avformat_find_stream_info().
Anton Khirnov [Mon, 2 Jan 2012 08:01:02 +0000 (09:01 +0100)]
lavf: flush decoders in avformat_find_stream_info().

This means that avformat_find_stream_info() now works properly with
multiple threads, so there's no need to force thread count to 1.

12 years agowin32: detect number of CPUs using affinity
Daniel Verkamp [Sun, 1 Jan 2012 23:54:40 +0000 (23:54 +0000)]
win32: detect number of CPUs using affinity

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
12 years agoAdd av_popcount64
Daniel Verkamp [Tue, 3 Jan 2012 01:23:48 +0000 (01:23 +0000)]
Add av_popcount64

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
12 years agosnow: Restore three mistakenly removed casts.
Diego Biurrun [Mon, 2 Jan 2012 16:29:10 +0000 (17:29 +0100)]
snow: Restore three mistakenly removed casts.

12 years agompegenc: use avctx->slices as number of slices
Janne Grunau [Sun, 1 Jan 2012 19:24:24 +0000 (20:24 +0100)]
mpegenc: use avctx->slices as number of slices

Adds a new member to MpegEncContext to hold the number of used slice
contexts. Fixes segfaults with '-threads 17 -thread_type slice' and
fate-vsynth{1,2}-mpeg{2,4}thread{,_ilace} with --disable-pthreads.

12 years agov410enc: fix undefined signed left shift caused by integer promotion
Janne Grunau [Mon, 2 Jan 2012 14:47:26 +0000 (15:47 +0100)]
v410enc: fix undefined signed left shift caused by integer promotion

12 years agoRelease notes: mention cleaned up header includes
Reinhard Tartler [Mon, 2 Jan 2012 16:01:56 +0000 (17:01 +0100)]
Release notes: mention cleaned up header includes

A rebuild in debian revealed that this broke compilation of a number packages:
http://bugs.debian.org/cgi-bin/pkgreport.cgi?users=siretart@tauware.de;tag=ftbfs-libav0.8

12 years agofix Changelog file
Reinhard Tartler [Mon, 2 Jan 2012 15:58:16 +0000 (16:58 +0100)]
fix Changelog file

Both the Indeo 4 decoder and the SMJPEG demuxer were included in the
0.8_beta1 release.

12 years agoFix a bunch of typos.
Diego Biurrun [Tue, 20 Dec 2011 14:39:22 +0000 (15:39 +0100)]
Fix a bunch of typos.

12 years agoDrop some pointless void* return value casts from av_malloc() invocations.
Diego Biurrun [Sat, 31 Dec 2011 15:30:59 +0000 (16:30 +0100)]
Drop some pointless void* return value casts from av_malloc() invocations.

12 years agowavpack: fix typos in previous cosmetic clean-up commit
Justin Ruggles [Mon, 2 Jan 2012 13:54:30 +0000 (08:54 -0500)]
wavpack: fix typos in previous cosmetic clean-up commit

12 years agowavpack: cosmetics: K&R pretty-printing
Justin Ruggles [Mon, 2 Jan 2012 05:14:16 +0000 (00:14 -0500)]
wavpack: cosmetics: K&R pretty-printing

12 years agoavconv: remove the 'codec framerate is different from stream' warning
Anton Khirnov [Sat, 24 Dec 2011 14:08:07 +0000 (15:08 +0100)]
avconv: remove the 'codec framerate is different from stream' warning

Codec timebase is not framerate and the situation when it's different
from AVStream.r_frame_rate is quite common, nothing to warn about.

12 years agowavpack: determine sample_fmt before requesting a buffer
Hendrik Leppkes [Fri, 16 Dec 2011 21:41:29 +0000 (22:41 +0100)]
wavpack: determine sample_fmt before requesting a buffer

Signed-off-by: Justin Ruggles <justin.ruggles@gmail.com>
12 years agobmv audio: implement new audio decoding API
Justin Ruggles [Sat, 31 Dec 2011 20:47:30 +0000 (15:47 -0500)]
bmv audio: implement new audio decoding API

12 years agompegaudiodec: skip all channels when skipping granules
Justin Ruggles [Sun, 1 Jan 2012 00:24:30 +0000 (19:24 -0500)]
mpegaudiodec: skip all channels when skipping granules

Also fix calculation of new position when switching buffers.
This fixes "overread" error messages when seeking.

12 years agompegenc: simplify muxrate calculation
Janne Grunau [Sun, 1 Jan 2012 22:32:15 +0000 (23:32 +0100)]
mpegenc: simplify muxrate calculation

The fate-h264-bsf-mp4toannexb failures were caused by an integer
overflow of the unneeded multiplication.

Inspired by patch by: Michael Niedermayer <michaelni@gmx.at>

12 years agoavconv: make -frames work for all types of streams, not just video.
Anton Khirnov [Sun, 1 Jan 2012 20:51:26 +0000 (21:51 +0100)]
avconv: make -frames work for all types of streams, not just video.

12 years agobfi: K&R cosmetics
Aneesh Dogra [Sun, 1 Jan 2012 16:37:54 +0000 (22:07 +0530)]
bfi: K&R cosmetics

Signed-off-by: Anton Khirnov <anton@khirnov.net>
12 years agobgmc: K&R cleanup
Aneesh Dogra [Sat, 31 Dec 2011 12:30:22 +0000 (18:00 +0530)]
bgmc: K&R cleanup

Signed-off-by: Anton Khirnov <anton@khirnov.net>
12 years agorawdec: Set start_time to 0 for raw audio files.
Reimar Döffinger [Wed, 28 Dec 2011 05:50:32 +0000 (00:50 -0500)]
rawdec: Set start_time to 0 for raw audio files.

None of the raw audio files have timestamps, thus setting
start_time to 0 should be reasonable.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
12 years agoDetect 'yuv2' as rawvideo also in avi.
Carl Eugen Hoyos [Wed, 28 Dec 2011 05:50:31 +0000 (00:50 -0500)]
Detect 'yuv2' as rawvideo also in avi.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
12 years agorawdec: propagate pict_type information to the output frame
Stefano Sabatini [Wed, 28 Dec 2011 05:50:28 +0000 (00:50 -0500)]
rawdec: propagate pict_type information to the output frame

Signed-off-by: Anton Khirnov <anton@khirnov.net>
12 years agorawdec: Support more QT 1bpp rawvideo files.
Piotr Bandurski [Wed, 28 Dec 2011 05:50:25 +0000 (00:50 -0500)]
rawdec: Support more QT 1bpp rawvideo files.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
12 years agoavconv: free bitstream filters
Janne Grunau [Sun, 1 Jan 2012 16:56:28 +0000 (17:56 +0100)]
avconv: free bitstream filters

12 years agothreads: limit the number of automatic threads to MAX_AUTO_THREADS
Janne Grunau [Sun, 1 Jan 2012 16:35:16 +0000 (17:35 +0100)]
threads: limit the number of automatic threads to MAX_AUTO_THREADS

The extra thread added in {frame_}*thread_init was not taken into
account. Explicitly sets thread_count to 1 if only one CPU core was
detected. Also fixes two typos in comments.

12 years agoavplay: K&R cleanup
Aneesh Dogra [Sun, 1 Jan 2012 12:21:26 +0000 (17:51 +0530)]
avplay: K&R cleanup

Signed-off-by: Anton Khirnov <anton@khirnov.net>
12 years agofate: use rgb24 as output format for dfa tests
Janne Grunau [Sun, 1 Jan 2012 15:32:25 +0000 (16:32 +0100)]
fate: use rgb24 as output format for dfa tests

Palette is as supposed in native endianness. Converting the pal8 output
to rgb24 is thus necessary for identical CRCs on big and little endian
systems.

12 years agothreads: set thread_count to 1 when thread support is disabled
Janne Grunau [Sun, 1 Jan 2012 13:17:32 +0000 (14:17 +0100)]
threads: set thread_count to 1 when thread support is disabled

12 years agothreads: introduce CODEC_CAP_AUTO_THREADS and add it to libx264
Janne Grunau [Sun, 1 Jan 2012 13:21:50 +0000 (14:21 +0100)]
threads: introduce CODEC_CAP_AUTO_THREADS and add it to libx264

Some external codecs have their own code to determine the best number
of threads. This number is not necessary the number of cpu cores.
Thread_count will be only 0 if the codec has CODEC_CAP_AUTO_THREADS.

12 years agoFATE: add tests for dfa
Oana Stratulat [Thu, 29 Dec 2011 20:47:59 +0000 (22:47 +0200)]
FATE: add tests for dfa

Signed-off-by: Justin Ruggles <justin.ruggles@gmail.com>
12 years agompegaudiodec: fix seeking.
Ronald S. Bultje [Fri, 30 Dec 2011 21:08:21 +0000 (13:08 -0800)]
mpegaudiodec: fix seeking.

The safe bitstream reader does not allow using skip_bits_long() to seek to a
point before the start of the buffer, which was needed by the mp3 decoder.
This change instead calculates the start point of the first valid granule and
skips to that position.

Signed-off-by: Justin Ruggles <justin.ruggles@gmail.com>
12 years agompegaudiodec: fix compilation when testing the unchecked bitstream reader
Justin Ruggles [Sat, 31 Dec 2011 18:36:48 +0000 (13:36 -0500)]
mpegaudiodec: fix compilation when testing the unchecked bitstream reader

12 years agothreads: add sysconf based number of CPUs detection
Janne Grunau [Fri, 23 Dec 2011 23:27:12 +0000 (00:27 +0100)]
threads: add sysconf based number of CPUs detection

Can act as fallback and should work on a couple of Unix systems.

12 years agothreads: always include necessary headers for number of CPUs detection
Janne Grunau [Fri, 23 Dec 2011 23:04:16 +0000 (00:04 +0100)]
threads: always include necessary headers for number of CPUs detection

Since the conditions for the actual usage are more specific a less
preferred method can be used. This would cause compilation errors
because necessary headers are not included.

12 years agothreads: default to automatic thread count detection
Janne Grunau [Sun, 25 Dec 2011 10:45:05 +0000 (11:45 +0100)]
threads: default to automatic thread count detection