It never makes sense for this function to be called by users.

It never makes sense for this function to be called by users.

Cf. #10857
Signed-off-by: Anton Khirnov <anton@khirnov.net>

If the time code side data is overridden by the packet level, we also
make sure not to update `p->metadata` to a mismatched timecode.

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

We only need to consider side data types that may possibly come from the
packet.
Signed-off-by: Anton Khirnov <anton@khirnov.net>

Also fixes a memory leak where the side data was previously not properly
cleaned up on OOM.
Signed-off-by: Anton Khirnov <anton@khirnov.net>

If the time code side data is overridden by the packet level, we also
make sure not to update `out->metadata` to a mismatched timecode.
Signed-off-by: Anton Khirnov <anton@khirnov.net>

Needed by following commit.

If the time code side data is overridden by the packet level, we also
make sure not to update `out->metadata` to a mismatched timecode.

This function was already ignoring OOM errors.

The signature of these wrappers is more complicated due to a need to
distinguish between "failed allocating side data" and "side data was
already present".
Signed-off-by: Anton Khirnov <anton@khirnov.net>

This and the following commits fix #10857

Native access is from the code that declared the options, foreign access
is from code that is using the options. Forbid foreign access to
AVOption.offset/default_val, for which there is no good reason, and
which should allow us more freedom in extending their semantics in a
compatible way.

Will be useful in following commits.

Replace the opt_size() function, currently only called from
av_opt_copy(), with
* a constant array of element sizes
* a function that signals whether an option type is POD (i.e.
  memcpyable) or not

Will be useful in following commits.

Will be useful in following commits.

Will be useful in following commits.

Forgotten in 2a68d945.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Otherwise s->input_format is always yuv420p.

This fixes invalid output format for hwframe download in the command
below:
./ffmpeg -init_hw_device vulkan -f lavfi \
	-i testsrc=duration=1,format=nv12 \
	-vf 'hwupload,format=vulkan,scale_vulkan=1024:768,hwdownload,format=nv12' \
	-f null -
Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>

Reviewed-by: epirat07@gmail.com
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Fixes the hevc_deblock checkasm test with Valgrind.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

The longest string here takes four bytes, so using an array
of pointers is wasteful even when ignoring the cost of relocations;
the lack of relocations also implies that this array
will now be put into .rodata and not into .data.rel.ro.

Static asserts are used to ensure that all strings are always
properly zero-terminated.
Tested-by: Marth64 <marth64@proxyid.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

The allocators have been superseded by av_vdpau_bind_context().
The latter have only been added "to allow multiple forks to add
fields to the structure without breaking ABI" [1], but libav
is no more, so this is not needed any longer.

[1]: https://ffmpeg.org/pipermail/ffmpeg-devel/2013-August/146954.htmlSigned-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Forgotten in b773a8d8.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

For testing purposes.

Forgotten in a12cd3be.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Signed-off-by: James Almer <jamrial@gmail.com>

These fields are documented to be non-public and are only used in
decode.c

Move related fields closer together.
Signed-off-by: James Almer <jamrial@gmail.com>