lavc: replace references to deprecated AVCodecContext.error_recognition to use AVCodecContext.err_recognition
Signed-off-by: Anton Khirnov <anton@khirnov.net>

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

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

Right now it's not actually initialized on 32-bit, leading to crashes
on win32.

Signed-off-by: Janne Grunau <janne-libav@jannau.net>

The blue channel went missing when toggling fullscreen on Mac OS X.

Signed-off-by: Martin Storsjö <martin@martin.st>

This allows for knowing the output size before decoding even when there is no
header (e.g. FLV). Otherwise we would have to do a preliminary full frame
decode to determine the number of frames-per-packet.

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

It's only used in the JACK device.

Fixes linking shared lavd with JACK enabled.

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

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

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

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

The correct decoder private context needs to be used.
This fixes mp3on4 playback and seeking in avplay.

count.

This also allows for checking output data size based on the actual
number of channel instead of the maximum number of channels.

Avoids allocating unnecessary memory and ensures proper alignment.

Fixes segfault when decoding multi-channel MP3onMP4 files.

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

Enable multithreding when decoding with libopenjpeg
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>

This actually causes them to be inlined, leading to a significant
speedup (1-1.5% in my measurements).

Neon parts by Mans Rullgard <mans@mansr.com>.

Signed-off-by: Diego Biurrun <diego@biurrun.de>
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>

Associate segmentation_map[] with reference frame, rather than
decoding instance. This fixes cases where the map would be free()'ed
on e.g. a size change in one thread, whereas the other thread was
still accessing it. Also, it fixes cases where threads overwrite data
that is still being referenced by the previous thread, who thinks that
it's part of the frame previously decoded by the next thread.