[FFmpeg-devel] [PATCH 4/6] Revert "avcodec/pngdec: fix possible race condition with APNG decoding"

[Paul B Mahol]

On Wed, Feb 17, 2021 at 12:31 PM Anton Khirnov <anton at khirnov.net> wrote:

> Quoting Paul B Mahol (2021-02-17 11:45:02)
> > On Wed, Feb 17, 2021 at 9:14 AM Anton Khirnov <anton at khirnov.net> wrote:
> >
> > > Quoting Paul B Mahol (2021-02-16 21:57:47)
> > > > Do you have actual proof for such claims?
> > >
> > > The burden of proof is on you here - you are supposed to prove that
> your
> > > commit fixes something. And when I asked you about details, you
> couldn't
> > > even tell me what the bug was, much less why would switching the order
> > > of allocations be the correct fix for it.
> > >
> >
> > It is not about order of allocations.
>
> What is it about then?
>
> >
> >
> > >
> > > >
> > > > There is no point in reverting commit if that commit fixed some
> behavior.
> > >
> > > It didn't fix anything though. Before your commit, frame 69 of the
> > > sample in #9017 is almost black with 1 thread and looks okay with 2
> > > threads. After your commit, it is almost black in both cases. So your
> > > commit made it consistently wrong. This shows up in my tests and is
> also
> > >
> >
> >
> > You are mistaken. It fixed race. Prove it does not.
>
> You have it backwards. You are making a claim - that your patch fixes a
> race. It is your responsibility to prove that your patch is correct. So
> far you did not even explain what this supposed race is.
>

I'm loosing my precious time discussing irrelevant matters here.

Is helgrind report clean with your patch set?


> --
> Anton Khirnov
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
> To unsubscribe, visit link above, or email
> ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".