[MPlayer-dev-eng] Bug in mplayers EVO demuxing code

[infernix]

Hi,

I've ran into an issue with mplayer SVN and .EVO demuxing. The files 
mentioned are available at http://dx.infernix.net/ffvc1/ and in the 
mplayer incoming dir called 'evodemuxbug'.

The file artifacts12.evo is a sample that was cut from an EVO file with 
a VC1 video stream and one (or two) DD+ audio streams.

The problem is that, when player back or demuxed with mplayer/mencoder 
(also with -demuxer lavf), there are some frames which give errors that 
lead to visual blocking (usually at the bottom of the frame):

[vc1 @ 0xe32140]Bits overconsumption: 242693 > 242560

This would be correct if the source has an error. However, this is not 
the case. When artifacts12.evo is demuxed to vc1 and this vc1 is playerd 
back/encoded with mplayer/mencoder, this error occurs.

When artifacts12.evo is demuxed with evodemux 
(http://pel.hu/down/EVOdemux.exe, works in wine) the resulting VC1 plays 
back perfectly.

So instead of this being an error in the source file, I think it is an 
error with demuxing.

I've uploaded artifacts12-mplayerdemuxed-errorframe.vc1 and 
artifacts12-evodemux-noerrorframe.vc1 to the same 'evodemuxbug' dir as 
artifacts12.evo.

Could someone have a look at this issue? I'd really appreciate it; 
Kostya put a lot of effort into getting ffvc1 working nicely (it's near 
perfect nowadays), and the most visually apparent bug with .evo playback 
is caused by this demuxing bug.

Thanks!

Regards,

infernix