[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <Z5JzA2OMokMz4Mic@google.com>
Date: Thu, 23 Jan 2025 16:49:07 +0000
From: Carlos Llamas <cmllamas@...gle.com>
To: Aleksandr Nogikh <nogikh@...gle.com>
Cc: syzbot <syzbot+7015dcf45953112c8b45@...kaller.appspotmail.com>,
Li Li <dualli@...gle.com>, arve@...roid.com, brauner@...nel.org,
gregkh@...uxfoundation.org, joel@...lfernandes.org,
linux-kernel@...r.kernel.org, linux-next@...r.kernel.org,
maco@...roid.com, sfr@...b.auug.org.au, surenb@...gle.com,
syzkaller-bugs@...glegroups.com, tkjos@...roid.com
Subject: Re: [syzbot] [kernel?] linux-next test error: KASAN:
slab-use-after-free Write in binder_add_device
On Thu, Jan 23, 2025 at 01:32:29PM +0100, Aleksandr Nogikh wrote:
> The problem began to appear after:
>
> commit 12d909cac1e1c4147cc3417fee804ee12fc6b984
> Author: Li Li <dualli@...gle.com>
> Date: Wed Dec 18 13:29:34 2024 -0800
>
> binderfs: add new binder devices to binder_devices
>
Correct. I tried to mark this commit with a #syz blame or something but
I couldn't find anything. The problem here is we add binderfs devices to
the binder_devices list but we don't remove them when these are kfreed
e.g. during umount.
This is then fairly easy to reproduce, something like:
$ mount -t binder binder /dev/binderfs
$ umount /dev/binderfs
$ mount -t binder binder /dev/binderfs
It should be a simply fix. I'll send a patch later today.
Thanks,
--
Carlos Llamas
Powered by blists - more mailing lists