From: Yu Kuai <yukuai1@huaweicloud.com>
To: Jan Kara <jack@suse.cz>, Jens Axboe <axboe@kernel.dk>
Cc: Yi Zhang <yi.zhang@redhat.com>,
linux-block <linux-block@vger.kernel.org>,
Paolo Valente <paolo.valente@linaro.org>,
"yukuai (C)" <yukuai3@huawei.com>
Subject: Re: [bug report]BUG: KFENCE: use-after-free read in bfq_exit_icq_bfqq+0x132/0x270
Date: Mon, 26 Dec 2022 09:17:16 +0800 [thread overview]
Message-ID: <a33dd515-b771-f400-ce49-e01f4ab31ad7@huaweicloud.com> (raw)
In-Reply-To: <20221222144949.sqmrmycfg2hg5ymq@quack3>
Hi, Jan!
在 2022/12/22 22:49, Jan Kara 写道:
> Thanks for report and the CC!
>
> On Mon 19-12-22 10:52:57, Jens Axboe wrote:
>> On 12/19/22 12:16 AM, Yi Zhang wrote:
>>> Below issue was triggered during blktests nvme-tcp with for-next
>>> (6.1.0, block, 2280cbf6), pls help check it
>>>
>>> [ 782.395936] run blktests nvme/013 at 2022-12-18 07:32:09
>>> [ 782.425739] nvmet: adding nsid 1 to subsystem blktests-subsystem-1
>>> [ 782.435780] nvmet_tcp: enabling port 0 (127.0.0.1:4420)
>>> [ 782.446357] nvmet: creating nvm controller 1 for subsystem
>>> blktests-subsystem-1 for NQN
>>> nqn.2014-08.org.nvmexpress:uuid:4c4c4544-0042-3910-8039-c6c04f544833.
>>> [ 782.460744] nvme nvme0: creating 32 I/O queues.
>>> [ 782.466760] nvme nvme0: mapped 32/0/0 default/read/poll queues.
>>> [ 782.479615] nvme nvme0: new ctrl: NQN "blktests-subsystem-1", addr
>>> 127.0.0.1:4420
>>> [ 783.612793] XFS (nvme0n1): Mounting V5 Filesystem
>>> [ 783.650705] XFS (nvme0n1): Ending clean mount
>>> [ 799.653271] ==================================================================
>>> [ 799.660496] BUG: KFENCE: use-after-free read in bfq_exit_icq_bfqq+0x132/0x270
>>> [ 799.669117] Use-after-free read at 0x000000008c692c21 (in kfence-#11):
>>> [ 799.675647] bfq_exit_icq_bfqq+0x132/0x270
>>> [ 799.679753] bfq_exit_icq+0x5b/0x80
>>> [ 799.683255] exit_io_context+0x81/0xb0
>>> [ 799.687015] do_exit+0x74b/0xaf0
>>> [ 799.690256] kthread_exit+0x25/0x30
>>> [ 799.693758] kthread+0xc8/0x110
>>> [ 799.696904] ret_from_fork+0x1f/0x30
>>> [ 799.701991] kfence-#11: 0x00000000f1839eaa-0x0000000011c747a1,
>>> size=568, cache=bfq_queue
>>> [ 799.711549] allocated by task 19533 on cpu 9 at 499.180335s:
>>> [ 799.717218] bfq_get_queue+0xe0/0x530
>>> [ 799.720884] bfq_get_bfqq_handle_split+0x75/0x120
>>> [ 799.725592] bfq_insert_requests+0x1d15/0x2710
>>> [ 799.730045] blk_mq_sched_insert_requests+0x5c/0x170
>>> [ 799.735021] blk_mq_flush_plug_list+0x115/0x2e0
>>> [ 799.739551] __blk_flush_plug+0xf2/0x130
>>> [ 799.743479] blk_finish_plug+0x25/0x40
>>> [ 799.747231] __iomap_dio_rw+0x520/0x7b0
>>> [ 799.751070] btrfs_dio_write+0x42/0x50
>>> [ 799.754832] btrfs_do_write_iter+0x2f4/0x5d0
>>> [ 799.759112] nvmet_file_submit_bvec+0xa6/0xe0 [nvmet]
>>> [ 799.764193] nvmet_file_execute_io+0x1a4/0x250 [nvmet]
>>> [ 799.769349] process_one_work+0x1c4/0x380
>>> [ 799.773361] worker_thread+0x4d/0x380
>>> [ 799.777028] kthread+0xe6/0x110
>>> [ 799.780174] ret_from_fork+0x1f/0x30
>>> [ 799.785252] freed by task 19533 on cpu 9 at 799.653250s:
>>> [ 799.790584] bfq_put_queue+0x183/0x2c0
>>> [ 799.794344] bfq_exit_icq_bfqq+0x129/0x270
>>> [ 799.798442] bfq_exit_icq+0x5b/0x80
>>> [ 799.801934] exit_io_context+0x81/0xb0
>>> [ 799.805687] do_exit+0x74b/0xaf0
>>> [ 799.808920] kthread_exit+0x25/0x30
>>> [ 799.812413] kthread+0xc8/0x110
>>> [ 799.815561] ret_from_fork+0x1f/0x30
>>> [ 799.820648] CPU: 9 PID: 19533 Comm: kworker/dying Not tainted 6.1.0 #1
>>> [ 799.827181] Hardware name: Dell Inc. PowerEdge R640/0X45NX, BIOS
>>> 2.15.1 06/15/2022
>>> [ 799.834746] ==================================================================
>>> [ 823.081364] XFS (nvme0n1): Unmounting Filesystem
>>> [ 823.159994] nvme nvme0: Removing ctrl: NQN "blktests-subsystem-1"
>
> Can you use addr2line to find which dereference is exactly causing the
> problem? Hum, it seems to point to some strange issue because we've just
> freed bfqq in this exit_io_context() invocation and seeing you are testing
> linux-block tree I think the problem might be caused by 64dc8c732f5c
> ("block, bfq: fix possible uaf for 'bfqq->bic'"). Kuai, I think we've
> messed up bfq_exit_icq_bfqq() and now bic_set_bfqq() can access already
> freed 'old_bfqq'. So we need something like:
>
>
> spin_lock_irqsave(&bfqd->lock, flags);
> bfqq->bic = NULL;
> - bfq_exit_bfqq(bfqd, bfqq);
> bic_set_bfqq(bic, NULL, is_sync);
> + bfq_exit_bfqq(bfqd, bfqq);
> spin_unlock_irqrestore(&bfqd->lock, flags);
>
> so free bfqq only after it is removed from the bic...
Sorry for the delay, and you're right, that's definitely a problem. 😒
Thanks,
Kuai
>
> Honza
>
prev parent reply other threads:[~2022-12-26 1:17 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-12-19 7:16 [bug report]BUG: KFENCE: use-after-free read in bfq_exit_icq_bfqq+0x132/0x270 Yi Zhang
2022-12-19 17:52 ` Jens Axboe
2022-12-22 14:49 ` Jan Kara
2022-12-26 1:17 ` Yu Kuai [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=a33dd515-b771-f400-ce49-e01f4ab31ad7@huaweicloud.com \
--to=yukuai1@huaweicloud.com \
--cc=axboe@kernel.dk \
--cc=jack@suse.cz \
--cc=linux-block@vger.kernel.org \
--cc=paolo.valente@linaro.org \
--cc=yi.zhang@redhat.com \
--cc=yukuai3@huawei.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.