This reverts commit 4284d791bc.
It causes crashes on various h264 and DNXHD/VC3 streams, where the
decoders write to arbitrary memory far after what we've allocated.
As a side effect, left/right green bars goes away when using
xvimagesink. I just think that xv cropping is broken, so this is
probably just hiding a bug.
While all this information is in the .la files, libtool seems to get
confused with ordering in presence of static system libraries. This could
cause missing symbol error at link time. Adding these depenencies explicitly
workaround the issue.
This should help libtool in getting the internal linking right.
Effectively, libtool can sometime get the link order wrong when
presented with a mix of .la and -l arguments. These .la file are
also required by the android build system and were previously
created by cerbero.
The install line was using -t parameter which is not supported on OSX.
Instead, use automake DATA installation mechanism, this way we rely on
automake to generate portable scripts.
In the last iteration, we kept the original method to link the shared
plugin and edited the .a and .la files so satisfy what cerbero needed.
Unfortunately, that required adding .a file into the archive which is
not allowed with iOS ar command for universal builds.
This patch uses standard method to link a static library. One of the
benefit is that it removes some libtool warning about portability.
For the static case, we implement an install hook that installs
FFMPEG internal .a files in the plugin directory (so it does not get
confused with a possible system FFMPEG. This makes the static plugin
usable without depending on cerbero recipe.
Libtool will produce libgstlibav.la and libgstlibav.lai (the installed
version). We need to edit at least the installed version for the final
linking of static application to work.
Some libtool will endup removing the shared build when running a static
build. That had unwanted side effect. Rather then fighting libtool to
get to build each static and shared seperatly, let libtool build with
the LIBAV_DEPS added to LIBADD (list of libav*.a) and finally remove the
extra .a from the archive and fix the .la to what cerbero will expect.
When building plugins with internal FFMPEG, we use different link
flags depending if it is static or shared. As we want to build both
static and dynamic plugins at once, rewrite the rules so we can
pass the right flags.
https://bugzilla.gnome.org/show_bug.cgi?id=779344
It's actually a parser but it a) can only work with the ffmpeg GIF
decoder that is deactivated anyway, and b) it currently causes infinite
linking of avdemux_gif elements with a multiqueue in between in
decodebin.
https://bugzilla.gnome.org/show_bug.cgi?id=775516
We expect it to be a int or uint, however it changed the type to a
int64_t in later versions of ffmpeg. As such it would be passed as a 64
bit value to varargs functions, while the consumer of the arguments
assumes only 32 bits. This causes crashes.
https://bugzilla.gnome.org/show_bug.cgi?id=771092
When switching playback modes, like from TRICKMODE or TRICKMODE_KEY_UNITS
back to regular playback, we need to make sure we set the skip mode
back to the default setting.
While this field would be properly reset when we *have* feedback from
downstream (i.e. diff != G_MAXINT64), it would not be reset during
the initial phase (i.e. when the decoder hasn't pushed a buffer yet,
and therefore the sink hasn't sent back QoS information).
This avoids dropping plenty of frames when going back to regular playback
Several decoders will only be able to report a real latency (has_b_frames)
once they're actually initialized (i.e. when they return their first frame).
Doing it earlier (in set_format) doesn't guarantee that the AVCodecContext
has_b_frames has been properly initialized.
https://bugzilla.gnome.org/show_bug.cgi?id=766362
Otherwise we will consider them as one frame of raw audio that is still
pending, and shift all timestamps by the amount of time spent with header
buffers.
https://bugzilla.gnome.org/show_bug.cgi?id=765797
Various functions in libavcodec need them, like the format, sample rate, etc.
and just having them in the context is not enough.
This fixes draining for codecs like MP2 that require a fixed frame size and
require libav to pad the last frame if required.
The GObject macros either for GstFFMpegVidDec and GstFFMpegVidEnc can
break the compilation because they are not GTypes, since each av video
elements are registered in runtime.
https://bugzilla.gnome.org/show_bug.cgi?id=764162
Remove calls to gst_pad_has_current_caps() which then go on to call
gst_pad_get_current_caps() as the caps can go to NULL in between. Instead just
use gst_pad_get_current_caps() and check for NULL.
https://bugzilla.gnome.org/show_bug.cgi?id=759539
It has its own allocator that is not necessarily doing the same as malloc and
will then usually crash. E.g. on Windows or when memalign() is available.
In ffmpeg this is the same as FRONT_CENTER, but we distinguish between
FRONT_CENTER and MONO in GStreamer. Add an explicit mapping for this special
case in the translations functions.
https://bugzilla.gnome.org/show_bug.cgi?id=759846
Handling slice_offset in avviddec is resulting in invalid memory read.
Since rv decoders anyways handle slice_offset, removing the same to fix
memory mishandlings
https://bugzilla.gnome.org/show_bug.cgi?id=758726
Error out if system's libav* libraries are not
provided by FFmpeg. Libav-incompatible changes
were introduced to support the latter so we
can no longer support both.
https://bugzilla.gnome.org/show_bug.cgi?id=758183
If downstream does not provide a (usable) pool, we would use our internal
pool. But the internal pool might be configured with a different width/height
because of padding, which then will cause problems if we push buffers from it
directly downstream.
Instead create a new pool if the width/height is different.
This prevents crashes with vaapisink and d3dvideosink for example.
Based on the debugging results and discussions with
Nicolas Dufresne <nicolas.dufresne@collabora.com>
https://bugzilla.gnome.org/show_bug.cgi?id=758344
... since they handle separate cases in video decoder with different requirements.
Consider e.g. x264enc ! rtph264pay ! identity drop-probability=0.1 ! rtph264depay
to illustrate a need for such separation.
Multithreaded encoders are going to free this dummy codec data twice, e.g.
with this pipeline
gst-launch-1.0 videotestsrc num-buffers=40 ! \
videoconvert ! avenc_mjpeg ! fakesink
Since gst_buffer_pool_set_config() takes ownership of the config structure,
it is only necessary to free the structure before using it when the true
branch of if (gst_buffer_pool_config_validate_params) hasn't run.
gst_buffer_pool_set_config() always takes ownership of the structure
regardless of success or failure. Which means the return, checked with
if (!working_pool), has no relation to the state of the structure.
Change default alignment from 16 to 32 bytes, which fixes crashes
when decoding H.265 using AVX2-based decoder code paths and when
using ximagesink/glimagesink.
https://bugzilla.gnome.org/show_bug.cgi?id=754120
Make sure the alignment requirement in GstAllocationParams
matches the GstVideoAlignment requirements. This fixes
issues with avdec_h265 crashing in the avx2 code path when
used with playbin and ximagesink/glimagesink as videosink.
The internal video pool would allocate buffers with an
alignment of 15 even though GstVideoAlignment specified
a stride_align requirement of 31 (which comes from ffmpeg).
https://bugzilla.gnome.org/show_bug.cgi?id=754120
Add the codec name and bitrate into the output for informational
purposes. Bitrate in particular is now used by flvmux to set
videodatarate and audiodatarate in the resulting stream
Some check where incorect and also unsafe. The only reliable information
in get_buffer2 is the picture width/height really. The side effect is
that the width/height of the internal pool endup padded, so when we
switch we also need to switch to the a new width/height, hence we save
the pool info.
https://bugzilla.gnome.org/show_bug.cgi?id=753869
The size in the AVFrame in get_buffer2 don't match the output size,
instead they match ffmpeg's memory requirements, so we can't compare
them from the values of the output AVFrame. Those are comparable to
the values in the passed AVCodecContext.
ffmpeg doesn't provide the final's image width & height in the get_buffer2()
callback, so it's not possible to create an output state for GstVideoDecoder
at this stage. So only try to do direct rendering if the buffer pool has already
been negotiated based on the final decoded size.
This partially reverts the effects of 2e621f8dbhttps://bugzilla.gnome.org/show_bug.cgi?id=752802
If it is created earlier and the stride is invalid, then the frame
will be freed and it won't be possible to use it in the fallback path.
Not doing this causes a segfault because it will try to use
already freed memory.
ffmpeg seems to be the one of the two forks, which is most widely used by
Linux distributions and in general. Also Google is using it for e.g. Chrome
and has engineers working on finding and fixing security issues in it.
https://bugzilla.gnome.org/show_bug.cgi?id=751607
gstavvidenc.c: In function 'gst_ffmpegvidenc_flush_buffers':
gstavvidenc.c:733:7: error: ISO C90 forbids mixed declarations and code [-Werror=declaration-after-statement]
GstFFMpegVidEncClass *oclass =
^
cc1: all warnings being treated as errors
libav always uses planar audio formats nowadays, not much use in
us trying to allocate anything here until we add support for planar
aka non-interleaved audio formats at least in audioconvert.
