linux-kernel - Re: KASAN: use-after-free Read in usbvision_v4l2

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]

Date:   Tue, 10 Dec 2019 15:13:28 -0500 (EST)
From:   Alan Stern <stern@...land.harvard.edu>
To:     Andrey Konovalov <andreyknvl@...gle.com>
cc:     syzbot <syzbot+c7b0ec009a216143df30@...kaller.appspotmail.com>,
        Hans Verkuil <hverkuil@...all.nl>,
        Souptick Joarder <jrdr.linux@...il.com>,
        LKML <linux-kernel@...r.kernel.org>,
        <linux-media@...r.kernel.org>,
        USB list <linux-usb@...r.kernel.org>,
        Mauro Carvalho Chehab <mchehab@...nel.org>,
        Richard Fontana <rfontana@...hat.com>,
        syzkaller-bugs <syzkaller-bugs@...glegroups.com>,
        Thomas Gleixner <tglx@...utronix.de>
Subject: Re: KASAN: use-after-free Read in usbvision_v4l2_open

On Tue, 10 Dec 2019, Andrey Konovalov wrote:

> On Tue, Dec 10, 2019 at 8:48 PM Alan Stern <stern@...land.harvard.edu> wrote:

> > This looks like a race in v4l2_open(): The function drops the
> > videodev_lock mutex before calling the video driver's open routine, and
> > the device can be unregistered during the short time between.
> >
> > This patch tries to make the race much more likely to happen, for
> > testing and verification.
> >
> > Andrey, will syzbot run the same test with this patch, even though it
> > says it doesn't have a reproducer?
> 
> Hi Alan,
> 
> No, unfortunately there's nothing to run if there's no reproducer.
> It's technically possible to run the same program log that triggered
> the bug initially, but since the bug wasn't reproduced with this log
> even without the patch, there isn't much sense in running it with the
> patch applied.

Actually it does make sense.  That bug was caused by a race, and the 
patch tries to make the race much more likely to happen, so the same 
test should fail again.

But never mind; I'll try a different approach.  There's another syzbot
bug report, one with a reproducer, and with this patch in place it
should trigger the same race.  I'll try submitting it that way.

By the way, do you know why syzbot sent _two_ reply messages?  One with 
message ID <00000000000031a0af05995eca0b@...gle.com> and the other with 
message ID <000000000000441a4205995eca11@...gle.com>?  It seems like 
overkill.

Alan Stern