[FFmpeg-devel] [PATCH 15/15] fftools/sync_queue: make sure non-limiting streams are not used as queue head

[James Almer]

On 5/23/2023 10:58 AM, Anton Khirnov wrote:
> A non-limiting stream could mistakenly end up being the queue head,
> which would then produce incorrect synchronization, seen e.g. in
> fate-matroska-flac-extradata-update for certain number of frame threads
> (e.g. 5).
> 
> Found-By: James Almer

Strictly speaking, it was FATE.

> ---
>   fftools/sync_queue.c | 13 +++++++++++--
>   1 file changed, 11 insertions(+), 2 deletions(-)
> 
> diff --git a/fftools/sync_queue.c b/fftools/sync_queue.c
> index c0f33e9235..bc107ba4fe 100644
> --- a/fftools/sync_queue.c
> +++ b/fftools/sync_queue.c
> @@ -217,17 +217,26 @@ static void finish_stream(SyncQueue *sq, unsigned int stream_idx)
>   
>   static void queue_head_update(SyncQueue *sq)
>   {
> +    av_assert0(sq->have_limiting);
> +
>       if (sq->head_stream < 0) {
> +        unsigned first_limiting = UINT_MAX;
> +
>           /* wait for one timestamp in each stream before determining
>            * the queue head */
>           for (unsigned int i = 0; i < sq->nb_streams; i++) {
>               SyncQueueStream *st = &sq->streams[i];
> -            if (st->limiting && st->head_ts == AV_NOPTS_VALUE)
> +            if (!st->limiting)
> +                continue;
> +            if (st->head_ts == AV_NOPTS_VALUE)
>                   return;
> +            if (first_limiting == UINT_MAX)
> +                first_limiting = i;
>           }
>   
>           // placeholder value, correct one will be found below
> -        sq->head_stream = 0;
> +        av_assert0(first_limiting < UINT_MAX);
> +        sq->head_stream = first_limiting;
>       }
>   
>       for (unsigned int i = 0; i < sq->nb_streams; i++) {

Can confirm it fixes the issue.