linux-kernel - Re: general protection fault in lo

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] [day] [month] [year] [list]

Date:   Wed, 9 May 2018 06:06:50 +0900
From:   Tetsuo Handa <penguin-kernel@...ove.SAKURA.ne.jp>
To:     tytso@....edu
Cc:     syzbot+bf89c128e05dd6c62523@...kaller.appspotmail.com,
        axboe@...nel.dk, syzkaller-bugs@...glegroups.com, jack@...e.cz,
        gmazyland@...il.com, dvyukov@...gle.com,
        linux-block@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: general protection fault in lo_ioctl (2)

Theodore Y. Ts'o wrote:
> On Tue, May 08, 2018 at 08:05:12PM +0900, Tetsuo Handa wrote:
> > 
> > So, it is time to think how to solve this race condition, as well as how to solve
> > lockdep's deadlock warning (and I guess that syzbot is actually hitting deadlocks).
> > An approach which serializes loop operations using global lock was proposed at
> > https://groups.google.com/d/msg/syzkaller-bugs/2Rw8-OM6IbM/PzdobV8kAgAJ .
> > Please respond...
> 
> I'm looking at your patch which you proposed on this, and the locking
> architecture still looks way too complex.  Things like
> loop_mutex_owner, and all of the infrastructure around
> lo->ioctl_in_progress should be removed, if at all possible.

The patch in the above link no longer uses "lo->ioctl_in_progress".
You looked at previous version rather than current version.

> 
> I believe it should be possible to do things with a single global
> mutex, some code refactoring, and some unlocked versions of some of
> the functions.

The patch in the above link uses single global mutex "loop_mutex".

> 
> Again, this requires root, and it requires someone deliberately trying
> to induce a race.  So "it's time" is not necessarily the priority I
> would set for this item.  But if we are going to fix it, let's fix it
> right, and not make the code more complex and less maintainable, all
> in the name of trying to make a rare, not-likely-to-happen-in-real-life
> syzbot reported problem to go away.

While NULL pointer dereference would be rare, deadlocks might not be rare
enough to postpone the patch. Deadlocks can cause pile of false-positive
hung task reports and can prevent syzbot from finding other bugs. That's
why I say "it is time to think".