2023-07-05 05:09:01

by Oleksij Rempel

[permalink] [raw]
Subject: Re: [PATCH] can: j1939: prevent deadlock by changing j1939_socks_lock to rwlock

On Tue, Jul 04, 2023 at 10:55:47AM -0700, Astra Joan wrote:
> Hi Oleksij,
>
> Thank you for providing help with the bug fix! The patch was created
> when I was working on another bug:
>
> https://syzkaller.appspot.com/bug?extid=881d65229ca4f9ae8c84
>
> But the patch was not a direct fix of the problem reported in the
> unregister_netdevice function call. Instead, it suppresses potential
> deadlock information to guarantee the real bug would show up. Since I
> have verified that the patch resolved a deadlock situation involving
> the exact same locks, I'm pretty confident it would be a proper fix for
> the current bug in this thread.
>
> I'm not sure, though, about how I could instruct Syzbot to create a
> reproducer to properly test this patch. Could you or anyone here help
> me find the next step? Thank you so much!

Sorry, I'm not syzbot expert. I hope someone else can help here.

Regards,
Oleksij
--
Pengutronix e.K. | |
Steuerwalder Str. 21 | http://www.pengutronix.de/ |
31137 Hildesheim, Germany | Phone: +49-5121-206917-0 |
Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |


2023-07-05 05:41:10

by Dmitry Vyukov

[permalink] [raw]
Subject: Re: [PATCH] can: j1939: prevent deadlock by changing j1939_socks_lock to rwlock

On Wed, 5 Jul 2023 at 06:40, Oleksij Rempel <[email protected]> wrote:
>
> On Tue, Jul 04, 2023 at 10:55:47AM -0700, Astra Joan wrote:
> > Hi Oleksij,
> >
> > Thank you for providing help with the bug fix! The patch was created
> > when I was working on another bug:
> >
> > https://syzkaller.appspot.com/bug?extid=881d65229ca4f9ae8c84
> >
> > But the patch was not a direct fix of the problem reported in the
> > unregister_netdevice function call. Instead, it suppresses potential
> > deadlock information to guarantee the real bug would show up. Since I
> > have verified that the patch resolved a deadlock situation involving
> > the exact same locks, I'm pretty confident it would be a proper fix for
> > the current bug in this thread.
> >
> > I'm not sure, though, about how I could instruct Syzbot to create a
> > reproducer to properly test this patch. Could you or anyone here help
> > me find the next step? Thank you so much!
>
> Sorry, I'm not syzbot expert. I hope someone else can help here.

+syzkaller mailing list

Hi Astra,

You mean you have a reproducer and you want syzbot to run it with your patch?
No such feature exists at the moment.

Presumably you already run it locally, so I am not sure syzbot can add
much value here.