GitLab

Most of the cases it's enought just to set pts so codec resyncs itself.
Also flushing on discontinue isn't good idea.

%p is only for pointers to void, not just any type of pointers. Due to
variable arguments, there is no implicit conversion here.

Both branches of the conditionals had the same results.

Signed-off-by: Boris Egorov <egorov@linux.com>
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

This reverts commit 5e36cb2c.

This reverts commit 48a867c7.

Revert for futher checking

If blocks are marked BLOCK_FLAG_DISCONTINUITY then do not drop them,
unless BLOCK_FLAGS_CORRUPTED is set. Instead flush old blocks from decoder
queue and restart at this new stream access point (new timeline).
Signed-off-by: Ilkka Ollakka <ileoo@videolan.org>

Signed-off-by: Ilkka Ollakka <ileoo@videolan.org>

Signed-off-by: Jean-Paul Saman <jpsaman@videolan.org>

Signed-off-by: Ilkka Ollakka <ileoo@videolan.org>

Error 8973 is triggered if the number of open HW decoding sessions is > than the number of available HW decoders

Further, a failing session creation should not invalidate the entire video decoding chain but just trigger the fallback

The FCC defines a "safe area" for caption rendering to be 80% of the
window height.  This patch changes it from 75% to 80%, and fixes a
bug in the original math due to integer arithmetic (the result of
(100/16 got truncated):

Original math:
100 / 16 * 8 / 10 = 4.8

Fixed math:
100 * 8 / 10 / 16 = 5
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

This solves linking issues with FFmpeg HEAD

This reduces the semaphore contention scope.

The coded video size is already fixed by the decoder when creating the
vidoe acceleration plugin, while the chroma is fixed by
vlc_va_GetChroma(). This callback is useless.

This call was originally meant to adjust the hardware acceleration to
video size changes, but that cannot happen there anymore. Also errors
cannot be handled properly there either.

This should never happen, and it would not work. We need the real
dimensions to initialize the video output thread.