The rounding used in the PTS calculations in filter_frame() does
not actually match the number of samples output by the resampler.
This leads to off-by-1 errors in the timestamps indicating gaps and
underruns, even when the input timestamps are all contiguous.
Bug-Id: 753
Signed-off-by: Anton Khirnov <anton@khirnov.net>
(cherry picked from commit 6cbbf0592f)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
(cherry picked from commit ca8c62d187)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Also convert a debug av_log() to av_dlog().
(cherry picked from commit a89dd9a72c6e9c3111d6f34d9b08cd624fe76358)
Signed-off-by: Diego Biurrun <diego@biurrun.de>
This fixes a segmentation fault because request_frame in fifo.c assumes
that the call to ff_request_frame will populate fifo->root.next.
Before, it was possible for request_frame in af_compand to not do this,
resulting in a null pointer access. Now, request_frame in af_compand
always will return at least one frame or an error, as per the API
specifications in avfilter.h for request_frame.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
(cherry picked from commit d3cfd7aff8)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Arch specific calls should have an emms_c following to keep the cpu
state consistent.
Reported-By: wm4
CC: libav-stable@libav.org
(cherry picked from commit e995cf1bcc)
Applications that have been linked against an older release of Libav and pick up
updated libraries experience segmentation faults because they pick up the new
libavfilter, which assumes AVFrames have been allocated by libavutil and thus
contain new reference-counting related fields. This will break for AVFrames that
have been allocated by old libavcodec.
All scheduled API changes are deferred to the next bump.
Before, it just returned width/height. Correct is width/height*sar.
That way it is consistent with DAR as in probe output and setdar.
Signed-off-by: Rudolf Polzer <divverent@xonotic.org>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Sometimes, if pthread_create() failed, then pthread_cond_wait() could
accidentally be called in the worker threads after the uninit function
had already called pthread_cond_broadcast(), leading to a deadlock.
Don't call pthread_cond_wait() if c->done is set.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
The pixel format tables are never modified, mark them as constant.
Signed-off-by: Diego Elio Pettenò <flameeyes@flameeyes.eu>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
When using AV_NOPTS_VALUE (which expands to INT64_C(0x8000000000000000))
as union initializer for a double field, the c99 converter needs to
interpret this constant when filling the union initializer, and it is
interpreted as a positive value.
When converting AV_NOPTS_VALUE to a double, MSVC 2010 ends up with
the same positive value as the c99 converter, while MSVC 2012 gets
a negative value.
This results in an infite loop in various FATE tests on MSVC 2012.
Signed-off-by: Martin Storsjö <martin@martin.st>
