This updated version does not deviate from previous behavior on default value of 'buffer_size'

I skipped porting 'sources', 'block' options for now as they're parsed seriously. So i added TODO remarks.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

It is all dead and rotting code.

Also, move yuvcmp.c to tools. Its license is non-existent, however,
and should be clarified.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>

All whitespace plus a curly brace.
Signed-off-by: Ben Jackson <ben@ben.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

The YUV channels of VP6 are encoded in a highly linear fashion which does
not have any slice-like concept to thread.  The alpha channel of VP6A is
fairly independent of the YUV and comprises 40% of the work.  This patch
uses the THREAD_SLICE capability to split the YUV and A decodes into
separate threads.

Two bugs are fixed by splitting YUV and alpha state:
- qscale_table from VP6A decode was for alpha channel instead of YUV
- alpha channel filtering settings were overwritten by YUV header parse
Signed-off-by: Ben Jackson <ben@ben.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Rather than cleverly managing frame pointers with swaps to avoid
re-using "golden" frames, just do brute-force management of the
4 AVFrames.  New strategy is probably no more costly and is easier
to adapt to threaded usage.
Signed-off-by: Ben Jackson <ben@ben.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Makes golden_frame more like other frame data, paves way for threading
alpha channel decode.
Signed-off-by: Ben Jackson <ben@ben.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Pave the way for per-thread context initialization.
Signed-off-by: Ben Jackson <ben@ben.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Pave the way for per-thread context initialization.
Signed-off-by: Ben Jackson <ben@ben.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

According to the sample for trac ticket #1722, PGS subtitles
are decoded from several packets at the same DTS and varying PTS.
The PTS from the presentation segment seem to be the valid one;
in particular, clear subtitles are too early with the other PTS.

If the packet has a PTS and the corresponding time base is known,
set the pts field of the decoded subtitle structure before the
call to the decoder. The decoder is still allowed to change the
PTS if necessary.

Without that change, all subtitles will stay until the next one.

* qatar/master:
  riff: Add SVQ3 fourcc
  fate: ac3: add 4.0 and downmix tests
  configure: x86: improve ebp availability check
  vorbisdec: ensure FASTDIV denominator is never 1
  avformat: refactor avformat_close_input
  avformat: simplify avformat_close_input

Conflicts:
	configure
	libavcodec/vorbisdec.c
	libavformat/utils.c
Merged-by: Michael Niedermayer <michaelni@gmx.at>

This fixes a warning about a unused function
Alternatively the function could be removed
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Japanese DTV uses some non standard extensions in AAC audio.
One example is 'dual mono', which combines two independent
audio into one stereo stream, storing them in left and right channels
respectively.  Historically, dual mono audio has been used for
multi-lingual audio, one for local/native language, and another for english,
and usually the "main" (local language) channel should be output without
any user interactions.

The frames of those dual mono audio are allowed to set
ADTS channel_config field to 0, and just contain two SCE's *WITHOUT* PCE,
which is a non standard extension by Japanese DTV standard.
(ref. ARIB STD-B32 PartII 5.2.3)

This patch adds an AVPacket side data, AV_PKT_DATA_JP_DUALMONO,
which indicates that the AVPacket is likely to contain an audio frame
with the above dual mono extension, and has the parameter to specify
the desired channel selection in that case.
It also makes aacdec to detect dual mono and output just the desired
channel when this side data is attached.
Signed-off-by: Akihiro Tsukada <atsukada@users.sourceforge.net>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Stefano Sabatini <stefasab@gmail.com>

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Mans Rullgard <mans@mansr.com>

Some compilers are extra strict about register usage in main(),
disallowing ebp in inline asm there while allowing it elsewhere.
This change makes the test better reflect actual usage.
Signed-off-by: Mans Rullgard <mans@mansr.com>

In both usages of FASTDIV the denominator might be 1.

Using a branch could make the function slower than using a normal
division.

Both denominator and numerator can be multiplied by 2 safely and
using shifts is faster than using a branch.

Do not crash if the input format is not allocated yet.

avio_close checks by itself for NULL condition.

* qatar/master:
  x86: dsputil: Move Xvid IDCT put/add functions to a more suitable place
  trasher: Include all the necessary headers
  x86: Remove some leftover declarations for non-existent functions
  ARM: libavresample: NEON optimised generic fltp to s16 conversion
  ARM: libavresample: NEON optimised stereo fltp to s16 conversion
  ARM: libavresample: NEON optimised flat float to s16 conversion

Conflicts:
	libavcodec/x86/dsputil_mmx.c
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Fix regression introduced in f054dbee.
+10.23l.

The c99-to-c89 converter (for MSVC support) doesn't currently handle
designated initializers or compound literals with nested unions or
structs.

This is apparently the only place where this construct is used in the
FFmpeg codebase.

Previously the cast long long int -> double was always performed (and
then the double was converted back to long long int) even when it was
avoidable.

pthread_wait_cond can wake up for no reason (Wikipedia: Spurious_wakeup).
The FF_THREAD_SLICE thread mechanism could spontaneously execute jobs or
allow the caller of avctx->execute to return before all jobs were complete.
This adds tests to both cases to ensure the wakeup is real.
Signed-off-by: Ben Jackson <ben@ben.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Fixes Ticket1718
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>