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

The coefficient test for i16x16 add_pixels4 assumed luma plane.

Please use -x264opts to force open gop
This fixes Ticket268
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

This fix introduces a small memleak
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

* qatar/master: (28 commits)
  Replace usages of av_get_bits_per_sample_fmt() with av_get_bytes_per_sample().
  x86: cabac: fix register constraints for 32-bit mode
  cabac: move x86 asm to libavcodec/x86/cabac.h
  x86: h264: cast pointers to intptr_t rather than int
  x86: h264: remove hardcoded edi in decode_significance_8x8_x86()
  x86: h264: remove hardcoded esi in decode_significance[_8x8]_x86()
  x86: h264: remove hardcoded edx in decode_significance[_8x8]_x86()
  x86: h264: remove hardcoded eax in decode_significance[_8x8]_x86()
  x86: cabac: change 'a' constraint to 'r' in get_cabac_inline()
  x86: cabac: remove hardcoded esi in get_cabac_inline()
  x86: cabac: remove hardcoded edx in get_cabac_inline()
  x86: cabac: remove unused macro parameter
  x86: cabac: remove hardcoded ebx in inline asm
  x86: cabac: remove hardcoded struct offsets from inline asm
  cabac: remove inline asm under #if 0
  cabac: remove BRANCHLESS_CABAC_DECODER switch
  cabac: remove #if 0 cascade under never-set #ifdef ARCH_X86_DISABLED
  document libswscale bump
  error_resilience: skip last-MV predictor step if MVs are not available.
  error_resilience: actually add counter when adding a MV predictor.
  ...

Conflicts:
	Changelog
	libavcodec/error_resilience.c
	libavfilter/defaults.c
	libavfilter/vf_drawtext.c
	libswscale/swscale.h
	tests/ref/vsynth1/error
	tests/ref/vsynth2/error
Merged-by: Michael Niedermayer <michaelni@gmx.at>

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

Fixes ticket272

av_get_bits_per_sample_fmt() is deprecated.

Some operands need to be accessed in byte mode, which restricts the
available registers in 32-bit mode.  Using the 'q' constraint selects
a suitable register.
Signed-off-by: Mans Rullgard <mans@mansr.com>

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

Only the low-order bits are used here so the type is not important,
but this avoids a compiler warning.
Signed-off-by: Mans Rullgard <mans@mansr.com>

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

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

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

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

Nothing requires this value in %eax.
Signed-off-by: Mans Rullgard <mans@mansr.com>

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

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

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

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

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

A comment says it's not faster than the C code.
Signed-off-by: Mans Rullgard <mans@mansr.com>

The code does not compile without this set.
Signed-off-by: Mans Rullgard <mans@mansr.com>

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

There are 2 known Bink audio codecs. Additionally they have
a different flavor if they are found inside Bink-b "BIKb" file.
In order to guess the correct flavor, the demuxer sets the audio
codec_tag to be the same as the file format tag.
This causes problem because same tag is used for both audio codecs.
The hack works in FFmpeg because audio codecs are identified by their
codec_id, but other players rely on standard behavior.

This fix removes the codec_tag hack and instead uses artificial
extradata format to signal the codec flavor. This would also
allow proper embedding of Bink audio in other containers.
Signed-off-by: Ivan Kalvachev <ikalvachev@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Fixes crashes when playing broken MPEG2-TS streams.

Without, the predictor isn't actually used.

Fixes ticket270
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: Anton Khirnov <anton@khirnov.net>

Fix compilatin after removal of FF_INTERNAL_MEM_TYPE_MAX_VALUE.
Signed-off-by: Stefano Sabatini <stefano.sabatini-lala@poste.it>
Signed-off-by: Anton Khirnov <anton@khirnov.net>

Fix warnings of the type:
vf_drawtext.c:NNN: warning: missing braces around initializer
vf_drawtext.c:NNN: warning: (near initialization for ‘drawtext_options[X].default_val’)
Signed-off-by: Anton Khirnov <anton@khirnov.net>

Signed-off-by: Anton Khirnov <anton@khirnov.net>

The feature was dropped after the filter was partially rewritten and
recommitted.
Signed-off-by: Anton Khirnov <anton@khirnov.net>

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

On architectures such as x86 (both 32 bit and 64bit), the stack element
size is fixed, which maintains alignment.  Here, this change does not
break anything. However, we also support also other architectures where
this property is not maintained and therefore, applications will crash
horribly.

This change effectively forces all applications to be recompiled against
libswscale.

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

* qatar/master:
  Drop git- prefix from version labels
  Use the contents of RELEASE as version fallback instead of 'UNKNOWN'
  path64/open64: filter out unsupported flags
  Add Release Notes
  sipr: Drop unused DSPContext
  sipr: include string.h for mem*()
  sipr: Use memmove() to copy overlapped buffers.
  h264: drop some ugly casts

Conflicts:
	version.sh
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Fixed ticket245
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>