linux-kernel - Re: WARNING: bad unlock balance in ovl

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAOQ4uxhb422zv4wEXKnvMTZkbmPcp8ZL_EMjV=qnUsGj0m1WhA@mail.gmail.com>
Date:   Mon, 2 Mar 2020 14:19:32 +0200
From:   Amir Goldstein <amir73il@...il.com>
To:     Miklos Szeredi <miklos@...redi.hu>
Cc:     syzbot <syzbot+66a9752fa927f745385e@...kaller.appspotmail.com>,
        Dmitry Vyukov <dvyukov@...gle.com>,
        linux-kernel <linux-kernel@...r.kernel.org>,
        overlayfs <linux-unionfs@...r.kernel.org>,
        Miklos Szeredi <mszeredi@...hat.com>,
        syzkaller-bugs <syzkaller-bugs@...glegroups.com>
Subject: Re: WARNING: bad unlock balance in ovl_llseek

On Mon, Mar 2, 2020 at 1:17 PM Miklos Szeredi <miklos@...redi.hu> wrote:
>
> On Mon, Mar 2, 2020 at 12:10 PM Amir Goldstein <amir73il@...il.com> wrote:
>
> > > =====================================
> > > WARNING: bad unlock balance detected!
> > > 5.6.0-rc3-syzkaller #0 Not tainted
> > > -------------------------------------
> > > syz-executor194/8947 is trying to release lock (&ovl_i_lock_key[depth]) at:
> > > [<ffffffff828b7835>] ovl_inode_unlock fs/overlayfs/overlayfs.h:328 [inline]
> > > [<ffffffff828b7835>] ovl_llseek+0x215/0x2c0 fs/overlayfs/file.c:193
> > > but there are no more locks to release!
> > >
> >
> > This is strange. I don't see how that can happen nor how my change would
> > have caused this regression. If anything, the lock chance may have brought
> > a bug in stack file ops to light, but don't see the bug.
>
> The bug is that ovl_inode_lock() is interruptible and that the caller
> doesn't check for error.
>
> I think the fix is to make this lock uninterruptible (probably rename
> the current helper to _interruptible and use the current name as the
> uninterruptible version).
>

Ok.

Thanks,
Amir.