[FFmpeg-devel] [PATCH 1/3 v2] avcodec: add an AVCodecContext field to signal types of packet, frame, and coded stream side data to export

[James Almer]

On 2/11/2020 10:06 AM, Michael Niedermayer wrote:
> On Mon, Feb 10, 2020 at 07:40:39PM -0300, James Almer wrote:
>> On 2/10/2020 7:14 PM, Michael Niedermayer wrote:
>>> On Mon, Feb 10, 2020 at 03:26:44PM -0300, James Almer wrote:
>>>> Add an initial mvs flag to is, analog to the export_mvs flags2 one.
>>>>
>>>> Signed-off-by: James Almer <jamrial at gmail.com>
>>>> ---
>>>>  libavcodec/avcodec.h                 | 18 ++++++++++++++++++
>>>>  libavcodec/mpegpicture.c             |  2 +-
>>>>  libavcodec/mpegutils.c               |  2 +-
>>>>  libavcodec/options_table.h           |  2 ++
>>>>  libavcodec/snowdec.c                 |  2 +-
>>>>  libavcodec/utils.c                   |  3 +++
>>>>  tests/ref/fate/api-mjpeg-codec-param |  2 ++
>>>>  tests/ref/fate/api-png-codec-param   |  2 ++
>>>>  8 files changed, 30 insertions(+), 3 deletions(-)
>>>
>>> Breaks 
>>> ./ffplay  -debug 16384  -flags2 +export_mvs -i panda.mp4 -vf codecview=mv=pf+bf+bb
>>
>> Broken how?
> 
> The filter no longer shows any motion vectors after this patch
> 
> Thanks

What codec does that file use? I see some modules copying flags2 from
one avctx to another (pthread_frame, tdsc, tiff, imm5), so i'm inclined
to think this may be an issue with keeping the avctx user values synced
between threads.

Does the following change fix it?

> diff --git a/libavcodec/pthread_frame.c b/libavcodec/pthread_frame.c
> index 36ac0ac1e5..b5bd494474 100644
> --- a/libavcodec/pthread_frame.c
> +++ b/libavcodec/pthread_frame.c
> @@ -334,6 +334,7 @@ static int update_context_from_user(AVCodecContext *dst, AVCodecContext *src)
> 
>      dst->slice_flags = src->slice_flags;
>      dst->flags2      = src->flags2;
> +    dst->export_side_data = src->export_side_data;
> 
>      copy_fields(skip_loop_filter, subtitle_header);

If not, i could use that sample.