Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <0e1b3d10-ae79-f987-187e-58109441ccee@kernel.dk>
Date:   Tue, 10 May 2022 06:45:31 -0600
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux aarch64; rv:91.0) Gecko/20100101
 Thunderbird/91.8.1
Subject: Re: [syzbot] KASAN: use-after-free Read in bio_poll
Content-Language: en-US
To:     Christoph Hellwig <hch@lst.de>, Ming Lei <ming.lei@redhat.com>
Cc:     syzbot <syzbot+99938118dfd9e1b0741a@syzkaller.appspotmail.com>,
        andrii@kernel.org, ast@kernel.org, bpf@vger.kernel.org,
        daniel@iogearbox.net, john.fastabend@gmail.com, kafai@fb.com,
        kpsingh@kernel.org, linux-block@vger.kernel.org,
        linux-kernel@vger.kernel.org, netdev@vger.kernel.org,
        songliubraving@fb.com, syzkaller-bugs@googlegroups.com, yhs@fb.com
References: <00000000000029572505de968021@google.com>
 <a72282ef-650c-143b-4b88-5185009c3ec2@kernel.dk> <YnmuRuO4yplt8p/p@T590>
 <20220510055039.GA10576@lst.de>
From:   Jens Axboe <axboe@kernel.dk>
In-Reply-To: <20220510055039.GA10576@lst.de>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 5/9/22 11:50 PM, Christoph Hellwig wrote:
> On Tue, May 10, 2022 at 08:13:58AM +0800, Ming Lei wrote:
>>> Guys, should we just queue:
>>>
>>> ommit 9650b453a3d4b1b8ed4ea8bcb9b40109608d1faf
>>> Author: Ming Lei <ming.lei@redhat.com>
>>> Date:   Wed Apr 20 22:31:10 2022 +0800
>>>
>>>     block: ignore RWF_HIPRI hint for sync dio
>>>
>>> up for 5.18 and stable?
>>
>> I am fine with merging to 5.18 & stable.
> 
> I'm fine, too.  But are we sure this actually is one and the same
> issue?  Otherwise I'll try to find some time to feed it to syzbot
> first.

I re-wrote the reproducer a bit and can reproduce it, so I can certainly
test a backport. But yes, I was skeptical on this being the same issue
too. My initial reaction was that this is likely due to the bio being
"downgraded" from polled to IRQ driven, and hence completes without an
extra reference before the bio_poll() is done on it. Which is not the
issue described in the referenced commit.

-- 
Jens Axboe