On Fri, Nov 23, 2012 at 1:43 PM, Janne Grunau <janne-li...@jannau.net> wrote:
> Sample count and delta are in ISO 14496-12 clearly specified as
> unsigned int.
> Fixes an assert caused by negative avg_frame_rate deduced from a
> negative duraton computed in mov_read_stts().
> ---
> libavformat/isom.h | 4 ++--
> libavformat/mov.c | 4 ++--
> 2 files changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/libavformat/isom.h b/libavformat/isom.h
> index 932b1d0..f4a178b 100644
> --- a/libavformat/isom.h
> +++ b/libavformat/isom.h
> @@ -43,8 +43,8 @@ int ff_mov_lang_to_iso639(unsigned code, char to[4]);
> */
>
> typedef struct MOVStts {
> - int count;
> - int duration;
> + unsigned count;
> + unsigned duration;
> } MOVStts;
>
> typedef struct MOVStsc {
> diff --git a/libavformat/mov.c b/libavformat/mov.c
> index 35c07bd..a121aa5 100644
> --- a/libavformat/mov.c
> +++ b/libavformat/mov.c
> @@ -1664,8 +1664,8 @@ static int mov_read_stts(MOVContext *c, AVIOContext
> *pb, MOVAtom atom)
> return AVERROR(ENOMEM);
>
> for (i = 0; i < entries && !pb->eof_reached; i++) {
> - int sample_duration;
> - int sample_count;
> + unsigned sample_duration;
> + unsigned sample_count;
>
> sample_count=avio_rb32(pb);
> sample_duration = avio_rb32(pb);
I have a sample which only has a proper timeline when sample_duration
is negative, and thus is required to be signed.
I vote for fixing the logic that trips your assert, and not break this file.
In case you are interested:
http://files.1f0.de/samples/mp4-negative-stts-problem.mp4
Admitedly, this file is not spec compliant, but breaking it seems like
something that can be avoided.
- Hendrik
_______________________________________________
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
