Re: 5.14.0-rc1 KASAN use after free

[jim . cromie]

On Mon, Aug 2, 2021 at 8:21 AM Paolo Valente <paolo.valente@xxxxxxxxxx> wrote:
>
>
>
> > Il giorno 23 lug 2021, alle ore 15:08, jim.cromie@xxxxxxxxx ha scritto:
> >
> > On Sun, Jul 18, 2021 at 5:58 PM Jens Axboe <axboe@xxxxxxxxx> wrote:
> >>
> >> On 7/18/21 3:08 PM, Oleksandr Natalenko wrote:
> >>> + Paolo, Jens et al.
> >>>
> >>> On čtvrtek 15. července 2021 16:32:29 CEST jim.cromie@xxxxxxxxx wrote:
> >>>> hi all,
> >>>>
> >>>> I noticed this report this morning, from 3 days ago,
> >>>> about 10 minutes after boot.
> >>>> Its easiest to ignore it, and I dont want to make a fuss,
> >>>> but it looks useful to someone
> >>>>
> >>>>
> >>>> [   33.663464] Bluetooth: RFCOMM ver 1.11
> >>>> [  646.343628]
> >>>> ================================================================== [
> >>>> 646.343649] BUG: KASAN: use-after-free in bfq_get_queue+0x47d/0x900 [
> >>>> 646.343680] Read of size 8 at addr ffff88810d864a00 by task
> >>>> journal-offline/1639
> >>
> >> There are only a few commits between 5.13 and master in this area, see
> >> attached. I'd just start reverting from the top, one by one, and see
> >> which one is causing the issue. Jim, would that be feasible?
> >>
> >
> > oops, didn't see this earlier.
> > It hasnt happened since, I can try to recreate mid-next-week
> >
>
> Still nothing?
>

Nada.

up to an hour ago, I was still running that installed kernel.
I just rebooted to it and ran a virtme session on it
(because of a possible 9p related trigger)
no sign of kasan err.

Im gonna boot rc4 built in the same build-dir,
I dont think Ive messed with the config,
but its a long-shot anyway to reproduce,
since same kernel image didnt do it 2nd time.


> Thanks,
> Paolo
>
> >
> >> --
> >> Jens Axboe
>