[FFmpeg-devel] [RFC] ignore invalid user-supplied width/height

[Michael Niedermayer]

On Sun, Sep 12, 2010 at 04:39:20PM +0200, Reimar D?ffinger wrote:
> On Sun, Sep 12, 2010 at 02:45:28PM +0200, Michael Niedermayer wrote:
> > > Index: libavcodec/utils.c
> > > ===================================================================
> > > --- libavcodec/utils.c  (revision 25017)
> > > +++ libavcodec/utils.c  (working copy)
> > > @@ -485,10 +485,15 @@ int attribute_align_arg avcodec_open(AVCodecContex
> > >      else if(avctx->width && avctx->height)
> > >          avcodec_set_dimensions(avctx, avctx->width, avctx->height);
> > >  
> > > +    if ((avctx->coded_width || avctx->coded_height || avctx->width || avctx->height)
> > > +        && (  av_check_image_size(avctx->coded_width, avctx->coded_height, 0, avctx) < 0
> > > +           || av_check_image_size(avctx->width,       avctx->height,       0, avctx) < 0) {
> > > +        av_log(avctx, AV_LOG_WARNING, "ignoring invalid width/height values\n");
> > > +        avcodec_set_dimensions(avctx, 0, 0);
> > > +    }
> > 
> > this function is also used for encoders i think and i dont think this is
> > correct for them
> 
> Well,
> 1) 0,0 has always been allowed through, so at least it does not make anything worse

encoding video with 0,0 is surely not ok


> 2) an AVCodec can have both encode and decode, so we can't really do a 100% correct
>    check here. We do already have a check in avcodec_encode_video anyway (which admittedly
>    is too late for any init code).
> 
> I am fairly confident that this proposed patch fixes two issues (failures due to corrupted
> values from the container and bogus width/height values being allow through as long
> as one of them and the coded_width/coded_height values are 0) and does not introduce
> any new ones.

ja ja, commit it if you in awake state think its ok


> And sorry for munging this in this thread, but why is lowres only checked against
> max_lowres if avctx->codec->init is non-NULL? Doesn't make sense to me.

neither to me

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

The worst form of inequality is to try to make unequal things equal.
-- Aristotle
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20100913/9ae1eb4f/attachment.pgp>