Quoting Luca Barbato (2014-11-08 08:16:08) > On 08/11/14 08:11, Anton Khirnov wrote: > > Quoting Luca Barbato (2014-11-02 19:49:34) > >> On 02/11/14 19:34, Anton Khirnov wrote: > >>> Quoting Vittorio Giovara (2014-10-30 01:51:53) > >>>> CC: [email protected] > >>>> Bug-Id: CID 29387 > >>>> --- > >>>> libavcodec/motion_est.c | 2 ++ > >>>> 1 file changed, 2 insertions(+) > >>>> > >>>> diff --git a/libavcodec/motion_est.c b/libavcodec/motion_est.c > >>>> index a765b26..1aa12f3 100644 > >>>> --- a/libavcodec/motion_est.c > >>>> +++ b/libavcodec/motion_est.c > >>>> @@ -805,6 +805,8 @@ static int interlaced_search(MpegEncContext *s, int > >>>> ref_index, > >>>> best_field= field_select; > >>>> } > >>>> } > >>>> + if (best_field < 0) > >>>> + return AVERROR_INVALIDDATA; > >>>> { > >>>> int16_t (*mv_table)[2]= mv_tables[block][best_field]; > >>>> > >>>> -- > >>>> 1.9.3 (Apple Git-50) > >>>> > >>> > >>> I do not think this can happen. And even if it does, it would mean that > >>> some other code went horribly wrong, so a check here does not do > >>> anything useful. > >>> > >> > >> AVERROR_BUG then. > > > > See my reply to 01/15 > > > I do not have time this month to check how motion_est could be made less > error prone so I do not have best solutions ^^; >
IMO the best solution is to fix real bugs and stop wasting time on imaginary problems. -- Anton Khirnov _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
