[FFmpeg-cvslog] r19979 - trunk/libavcodec/h264.c

[Reimar Döffinger]

On Wed, Sep 23, 2009 at 11:58:45AM +0200, michael wrote:
> Author: michael
> Date: Wed Sep 23 11:58:44 2009
> New Revision: 19979
> 
> Log:
> Check num_units_in_tick/time_scale to be valid and within the range we support.
> based on a patch by chrome
> 
> Modified:
>    trunk/libavcodec/h264.c
> 
> Modified: trunk/libavcodec/h264.c
> ==============================================================================
> --- trunk/libavcodec/h264.c	Wed Sep 23 11:18:23 2009	(r19978)
> +++ trunk/libavcodec/h264.c	Wed Sep 23 11:58:44 2009	(r19979)
> @@ -7087,6 +7087,10 @@ static inline int decode_vui_parameters(
>      if(sps->timing_info_present_flag){
>          sps->num_units_in_tick = get_bits_long(&s->gb, 32);
>          sps->time_scale = get_bits_long(&s->gb, 32);
> +        if(sps->num_units_in_tick-1 > 0x7FFFFFFEU || sps->time_scale-1 > 0x7FFFFFFEU){
> +            av_log(h->s.avctx, AV_LOG_ERROR, "time_scale/num_units_in_tick inavlid or unsupported (%d/%d)\n", sps->time_scale, sps->num_units_in_tick);
> +            return -1;
> +        }
>          sps->fixed_frame_rate_flag = get_bits1(&s->gb);
>      }

No sense to make it fallback to some sane value instead of failing in
your opinion?
Or just not until it happens with some real-world file?