Re: WARNING in mark_lock (3)

[Dmitry Vyukov]

On Thu, Nov 28, 2019 at 9:46 AM Sven Eckelmann <sven@xxxxxxxxxxxxx> wrote:
>
> On Thursday, 28 November 2019 09:40:32 CET Dmitry Vyukov wrote:
> > On Thu, Nov 28, 2019 at 8:25 AM Sven Eckelmann <sven@xxxxxxxxxxxxx> wrote:
> > >
> > > On Thursday, 28 November 2019 03:00:01 CET syzbot wrote:
> > > [...]
> > > > bisection log:  https://syzkaller.appspot.com/x/bisect.txt?x=132ee536e00000
> > > > start commit:   89d57ddd Merge tag 'media/v5.5-1' of git://git.kernel.org/..
> > > > git tree:       upstream
> > > > final crash:    https://syzkaller.appspot.com/x/report.txt?x=10aee536e00000
> > >
> > > Can the syzbot infrastructure be told to ignore this crash in the bisect run?
> > > Because this should be an unrelated crash which is (hopefully) fixed in
> > > 40e220b4218b ("batman-adv: Avoid free/alloc race when handling OGM buffer").
> >
> > +syzkaller mailing list for syzbot discussion
> >
> > Hi Sven,
> >
> > There is no such functionality at the moment.
> > What exactly do you mean? Somehow telling it interactively? Or
> > hardcode some set of crashes for linux? I don't see how any of these
> > options can really work...
>
> I was thinking more about rerunning the same bisect but tell it to assume
> "crashed: general protection fault in batadv_iv_ogm_queue_add" as OK instead
> of assuming that it is a crashed like the previous "crashed: WARNING in
> mark_lock". Just to get a non-bogus bisect result. Or try to rerun the
> bisect between 40e220b4218b and 89d57dddd7d319ded00415790a0bb3c954b7e386

But... but this done by a program. What do you mean by "tell it"?