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

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

Whatever reason this was done for no longer applies, and it causes
lots of make warnings.
Signed-off-by: Mans Rullgard <mans@mansr.com>

Mainly ported from 8-bit H.264 qpel.

Some code ported from x264. LGPL ok by author.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>

This is the version available in Debian stable, so it should be a reasonable
baseline that can be expected to be present on all developer machines.

Moreover, this is the version that is used by the nightly cronjob that
generates the online html version.

This was missed in 5d20f19b since CONFIG_SMALL was always broken
for big endian.
Signed-off-by: Mans Rullgard <mans@mansr.com>

Testing a million random keys takes annoying long time.
1000 iterations should be enough.
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>

This removes a number of compiler warnings.
Signed-off-by: Mans Rullgard <mans@mansr.com>

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

This makes the code work with clang/x86_32 and removes several warnings.
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>

This functionality is only implemented for RGB24/32 and causes crashes
otherwise.

The previously suggested replacement - av_get_bits_per_sample_fmt() -
was also deprecated.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>

av_get_bits_per_sample_fmt() was deprecated.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>

The code for this variant does not compile.
Signed-off-by: Mans Rullgard <mans@mansr.com>

This fixes invalid unaligned stores in some ARM configurations.
Signed-off-by: Mans Rullgard <mans@mansr.com>

Also include it into the doxygen documentation

Before this, almost all module groups have been used for grouping functions
and fields in structures semantically. This causes them to not appear
properly in the file documentation and needlessly clutters up the "Modules"
index.

Additionally, this commit streamlines some spelling and appearances.

convert the comment that documents the metadata API to use
the doxygen markup

av_open_input_stream used to allow this, even though it makes no sense.
Make it just print a warning instead of failing, thus restoring
compatibility.

Note that avformat_open_input() will still reject this combination.
Signed-off-by: Anton Khirnov <anton@khirnov.net>

x86 does not use this variant so having inline asm there
is pointless.
Signed-off-by: Mans Rullgard <mans@mansr.com>

We have now switched to http://bugzilla.libav.org.

Add al05_44, al06_44, al17_44.

This ensures that any processing between the MDCT and exponent extraction will
be using clipped coefficients.

For 9/10bit, it means we don't have to upscale to 16bit before
actual scaling or pixel format conversion, and thus a performance
gain.

EBML_STOP leaves matroska->current_id set. Then matroska_read_seek changes
the stream position without resetting current_id.  The next
matroska_parse_cluster  fails due to calculation of incorrect pos.  So clear
current_id when avio_seek happens in matroska_read_seek.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>