[FFmpeg-devel] [PATCH] avfillter/buffersrc: activate and EOF fix

[Tristan Matthews]

On Thu, Nov 2, 2023 at 6:10 AM Paul B Mahol <onemda at gmail.com> wrote:
>
> On Thu, Nov 2, 2023 at 11:03 AM Nicolas George <george at nsup.org> wrote:
>
> > Paul B Mahol (12023-11-02):
> > > I analyzed it already, It is very sorry state of libavfilter that
> > buffersrc
> > > keeps sending frames to EOF filtergraph.
> >
> > So, Paul, I will explain this to you one last time.
> >
> > Remember high school, when you had math test, and if you just gave the
> > result you got almost no points even if the result was correct? That was
> > because solving a math exercise is not just about finding the result, it
> > is about showing how you reach the result and proving that the result is
> > correct.
> >
> > The same goes here. On your own projects, you can change randomly the
> > code until a bug is no longer triggered and call it fixed.
> >
> > But in FFmpeg, you are not alone, and when a change is not trivial you
> > have to prove to your fellow developers that it is correct and
> > necessary.
> >
> > So get to work, produce that proof, re-submit the patch with the proof
> > in the commit message, and then we can talk.
> >
> > As it is, the need to switch buffersrc to activate is not established,
> > and therefore it should not be done.
> >
>
> I do not understand. What proof you need?
> Have you even tested this JEEB script without patches applied?
> It should straight forward cause OOM bomb.
>
> Have you noticed that buffersrc filter never checks outlink status of its
> output link?
>

Just to confirm that I can reproduce JEEB's test, before the patches:

    Maximum resident set size (kbytes): 629568

with *both* patches applied:

    Maximum resident set size (kbytes): 80920

Best,
Tristan

>
> >
> > --
> >   Nicolas George
> > _______________________________________________
> > 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".
> >
> _______________________________________________
> 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".