[FFmpeg-devel] [PATCH 1/3] AviSynth demuxer rewrite.

[Stephen Hutchinson]

On Wed, Feb 27, 2013 at 9:03 AM, Michael Niedermayer <michaelni at gmx.at> wrote:
>> +    pkt->size = ((avs->vi->width * avs->vi->height) * bits) / 8;
>
> can the multiplication overflow before the /8 ?

My reason for using the parentheses was as a visual aid, and because
the last version of the patch from October wouldn't encode anything
because the parentheses that were there were wrapped around the bits /
8 section instead of making the division by 8 come absolutely last,
resulting in packet size errors.

If I'm understanding the concept correctly, it'd depend on what the
width and height of the input is.  Even 4320p in BGRA wouldn't result
in a number too large for 32-bit int to handle, again if I'm
understanding it right.

> also the avx/avisynth code in ffmpeg will need a maintainer
> are you or someone else from the avxsynth team interrested?
> If so please add yourself to the MAINTAINERS file

>From the list of tasks, I'd only feel comfortable with testing
patches, and a rough maybe on making sure it doesn't break with future
updates (although I also don't know if I'd necessarily be able to fix
it if that happened).  I don't feel confident enough to handle
refactoring or most sorts of review.

I've thrown the question over to the AvxSynth guys, maybe they will.