Anton Khirnov <[email protected]> writes:

> Fixes avformat_find_stream_info() on streams with number of frames <
> thread count.
> ---
>  libavformat/utils.c |    5 +++++
>  1 files changed, 5 insertions(+), 0 deletions(-)
>
> diff --git a/libavformat/utils.c b/libavformat/utils.c
> index ea075e7..a4985da 100644
> --- a/libavformat/utils.c
> +++ b/libavformat/utils.c
> @@ -2292,6 +2292,11 @@ int avformat_find_stream_info(AVFormatContext *ic, 
> AVDictionary **options)
>          assert(!st->codec->codec);
>          codec = avcodec_find_decoder(st->codec->codec_id);
>  
> +        /* this function doesn't flush the decoders, so force thread count
> +         * to 1 to fix behavior when thread count > number of frames in the 
> file */

Why is it not fixed to flush the decoders?  Although single-threaded is
probably preferable here (a single frame is usually all that's
required), won't this still fail if the stream is shorter than the
nominal decoder delay (e.g. single-frame MPEG2)?

-- 
Måns Rullgård
[email protected]
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to