[MPlayer-dev-eng] bug in demux_mkv

[Sean D'Epagnier]

On Sun, Jul 15, 2007 at 10:02:52AM +0200, Reimar D?ffinger wrote:

> Index: libmpdemux/demux_mkv.c
> ===================================================================
> --- libmpdemux/demux_mkv.c	(revision 23784)
> +++ libmpdemux/demux_mkv.c	(working copy)
> @@ -2520,7 +2520,7 @@
>    else if (demuxer->audio->id != -2)  /* -2 = no audio at all */
>      track = demux_mkv_find_track_by_num (mkv_d, demuxer->audio->id,
>                                           MATROSKA_TRACK_AUDIO);
> -  else
> +  if (!track)
>      {
>        mp_msg (MSGT_DEMUX, MSGL_INFO, MSGTR_MPDEMUX_MKV_NoAudioTrackFound);
>        demuxer->audio->id = -2;

This test is really not required since we know track is always NULL, but it
may make the code clearer.

> @@ -2534,7 +2534,7 @@
>            continue;
>        if(demuxer->a_streams[track->tnum])
>          {
> -          if(track && mkv_d->tracks[i] == track)
> +          if(mkv_d->tracks[i] == track)
>              {
>                demuxer->audio->id = track->tnum;
>                demuxer->audio->sh = demuxer->a_streams[track->tnum];

It's true it makes no sense to test if track is NULL here since it already
crashed at "if(demuxer->a_streams[track->tnum])" if it was NULL.

There is still nothing in this patch I can see that tests if track is NULL
before hitting "if(demuxer->a_streams[track->tnum])"


It was mentioned that mplayer does correctly autodetect the audio track.
I have both 64 and 32bit x86 systems.  Anyone else able to run:

mplayer http://people.freedesktop.org/~geckosenator/test.mkv

Tell us if it crashes or not.  It does for me.

Sean