[FFmpeg-cvslog] r19838 - trunk/libavformat/idcin.c

[Vitor Sessak]

Reimar D?ffinger wrote:
> On Mon, Sep 14, 2009 at 05:44:05PM +0200, reimar wrote:
>> Author: reimar
>> Date: Mon Sep 14 17:44:05 2009
>> New Revision: 19838
>>
>> Log:
>> Check buffer size for idcin check, otherwise false positives are too likely for
>> small probe buffer sizes due to 0-padding (see probetest results).
> 
> I just realize that once again the padding does not have to be 0-ed,
> that might mean quite a few more tests (e.g. sierravmd.c, mm.c) need to
> check the buffer size despite the padding.
> Is that intended/what should be done?

IMHO, ideally the probe code should not be triggered neither by random 
junk padding nor by 0-byte padding (anyway, I think that if buffer_size 
< 256 the client app is broken)...

-Vitor