[vlc-devel] [PATCH] stream_filter: add ADF stream filter (Fixes #17501)

Filip Roséen filip at atch.se
Tue Oct 25 18:27:09 CEST 2016


Hi Francois,

On 2016-10-25 17:59, Francois Cartegnie wrote:

> Le 25/10/2016 à 17:31, Filip Roséen a écrit :
> > Hi Francois,
> > 
> > On 2016-10-25 17:18, Francois Cartegnie wrote:
> > 
> >> Le 25/10/2016 à 06:56, Tristan Matthews a écrit :
> >>> +    /* Probe for XOR'd ID3 tag. */
> >>> +    if( memcmp( peek, "\x6B\x66\x11", 3 ) )
> >>> +        return VLC_EGENERIC;
> >>
> >> Being before any demuxer and packetizer,
> >> expect high rate of false positives.
> > 
> >  - Why would this cause a *"high rate of false positives"?
> > 
> > It is equivalent to a *demuxer* probing for the same data, and as far
> > as I could see when I looked up the *magic-bytes* used, there is no
> > *file-format* that is publically documented that uses that marker.
> 
> Probing a single sequence is *weak*.
> and data has no alignment guarantee.
> 
> Hence the demuxers priorities and the packetizer based demuxers and
> multiple frames probing for most audio codecs. (and libav has even more
> intensive probing)
> 
> Putting this as stream filter gives it higher priority than any demuxer.
> 
> Will break any non aligned, sync code based, if the stream starts with
> "\x6B\x66\x11" (raw video, raw audio, ts, ps...), where demux with low
> score would not.

That is a very valid point, thank you for the elaboration!

> 
> Francois
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> https://mailman.videolan.org/listinfo/vlc-devel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/vlc-devel/attachments/20161025/148e3561/attachment.html>


More information about the vlc-devel mailing list