Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
In-Reply-To: <be262e04-df8a-c248-26a8-d42bf081be1c@virtuozzo.com>
References: <00000000000059484105767ac88f@google.com> <be262e04-df8a-c248-26a8-d42bf081be1c@virtuozzo.com>
From:   Miklos Szeredi <miklos@szeredi.hu>
Date:   Mon, 24 Sep 2018 16:27:21 +0200
Message-ID: <CAJfpeguTiN22KqMCH9NMs0Hj+ZcshdzFUk7Mb38Jt0ewHEOHpg@mail.gmail.com>
Subject: Re: KASAN: use-after-free Read in fuse_dev_do_read
To:     Kirill Tkhai <ktkhai@virtuozzo.com>
Cc:     syzbot <syzbot+4e975615ca01f2277bdd@syzkaller.appspotmail.com>,
        Dmitry Vyukov <dvyukov@google.com>,
        linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
        syzkaller-bugs <syzkaller-bugs@googlegroups.com>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, Sep 24, 2018 at 2:06 PM, Kirill Tkhai <ktkhai@virtuozzo.com> wrote:

> Fix from my previous message makes the use-after-free does not reproduce
> with the reproducer in my setup.

Excellent.

>
> I can prepare the patch, but before this some comments from Miklos would
> be welcome.
>
> Miklos, what you think about this?

I like the patch.

We could optimize away the get/put by moving the set_bit/test_bit part
inside the fpq->lock-ed region and only get the refcount for the
(unlikely) interrupted case.  OTOH it's probably not worth the extra
complexity, so let's stay with this simpler fix.

Thanks,
Miklos