From: duoming@zju.edu.cn
To: "James Smart" <jsmart2021@gmail.com>
Cc: linux-nvme@lists.infradead.org, linux-kernel@vger.kernel.org,
james.smart@broadcom.com, kbusch@kernel.org, axboe@fb.com,
hch@lst.de, sagi@grimberg.me
Subject: Re: [PATCH] nvme-fc: fix sleep-in-atomic-context bug caused by nvme_fc_rcv_ls_req
Date: Tue, 4 Oct 2022 19:10:19 +0800 (GMT+08:00) [thread overview]
Message-ID: <118cdb2e.1067d5.183a2b07fa7.Coremail.duoming@zju.edu.cn> (raw)
In-Reply-To: <8aeea62b-c947-6414-bca1-3bd3f427cd56@gmail.com>
Hello,
On Mon, 3 Oct 2022 10:48:31 -0700 James Smart wrote:
> On 10/2/2022 7:56 PM, James Smart wrote:
> > On 10/2/2022 6:50 PM, duoming@zju.edu.cn wrote:
> >> Hello,
> >>
> >> On Sun, 2 Oct 2022 10:12:15 -0700 James Smart wrote:
> >>
> >>> On 10/1/2022 5:19 PM, Duoming Zhou wrote:
> >>>> The function lpfc_poll_timeout() is a timer handler that runs in an
> >>>> atomic context, but it calls "kzalloc(.., GFP_KERNEL)" that may sleep.
> >>>> As a result, the sleep-in-atomic-context bug will happen. The processes
> >>>> is shown below:
> >>>>
> >>>> lpfc_poll_timeout()
> >>>> lpfc_sli_handle_fast_ring_event()
> >>>> lpfc_sli_process_unsol_iocb()
> >>>> lpfc_complete_unsol_iocb()
> >>>> lpfc_nvme_unsol_ls_handler()
> >>>> lpfc_nvme_handle_lsreq()
> >>>> nvme_fc_rcv_ls_req()
> >>>> kzalloc(sizeof(.., GFP_KERNEL) //may sleep
> >>>>
> >>>> This patch changes the gfp_t parameter of kzalloc() from GFP_KERNEL to
> >>>> GFP_ATOMIC in order to mitigate the bug.
> >>>>
> >>>> Fixes: 14fd1e98afaf ("nvme-fc: Add Disconnect Association Rcv support")
> >>>> Signed-off-by: Duoming Zhou <duoming@zju.edu.cn>
> >>>> ---
> >>>> drivers/nvme/host/fc.c | 2 +-
> >>>> 1 file changed, 1 insertion(+), 1 deletion(-)
> >>>>
> >>>> diff --git a/drivers/nvme/host/fc.c b/drivers/nvme/host/fc.c
> >>>> index 127abaf9ba5..36698dfc8b3 100644
> >>>> --- a/drivers/nvme/host/fc.c
> >>>> +++ b/drivers/nvme/host/fc.c
> >>>> @@ -1754,7 +1754,7 @@ nvme_fc_rcv_ls_req(struct nvme_fc_remote_port
> >>>> *portptr,
> >>>> lsop = kzalloc(sizeof(*lsop) +
> >>>> sizeof(union nvmefc_ls_requests) +
> >>>> sizeof(union nvmefc_ls_responses),
> >>>> - GFP_KERNEL);
> >>>> + GFP_ATOMIC);
> >>>> if (!lsop) {
> >>>> dev_info(lport->dev,
> >>>> "RCV %s LS failed: No memory\n",
> >>>
> >>> I would prefer this was fixed within lpfc rather than introducing atomic
> >>> allocations (1st in either host or target transport). It was introduced
> >>> by lpfc change in irq handling style.
> >>
> >> Thank your for your reply and suggestions!
> >>
> >> Do you think change the lpfc_poll_timeout() to a delayed_work is better?
> >>
> >> Best regards,
> >> Duoming Zhou
> >
> > as a minimum: the lpfc_complete_unsol_iocb handler should be passing off
> > the iocb to a work queue routine - so that the context changes so that
> > either nvme host or nvmet ls callback routines can be called. If
> > possible, it should do the axchg alloc - to avoid a GFP_ATOMIC there as
> > well...
> >
> > It's usually best for these nvme LS's and ELS's to be done in a slow
> > path thread/work queue element. That may mean segmenting a little
> > earlier in the path.
> >
> > -- james
> >
>
> looking further... lpfc_poll_timeout() should only be used on an SLI-3
> adapter. The existing SLI-3 adapters don't support NVMe. So I'm a
> little confused by this stack trace.
I found this problem through a static analysis tool wroten by myself.
I think the hacker may simulate the hardware to trigger this stack trace.
So, I send the patch to correct the problem.
> Can you describe what the system config/software setup is and
> specifically what lpfc adapter is being used (dmesg attachment logs are
> sufficient, or lspci output).
Best regards,
Duoming Zhou
next prev parent reply other threads:[~2022-10-04 11:11 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-10-02 0:19 [PATCH] nvme-fc: fix sleep-in-atomic-context bug caused by nvme_fc_rcv_ls_req Duoming Zhou
2022-10-02 17:12 ` James Smart
2022-10-03 1:50 ` duoming
2022-10-03 2:56 ` James Smart
2022-10-03 17:48 ` James Smart
2022-10-04 11:10 ` duoming [this message]
2022-10-04 13:51 ` James Smart
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=118cdb2e.1067d5.183a2b07fa7.Coremail.duoming@zju.edu.cn \
--to=duoming@zju.edu.cn \
--cc=axboe@fb.com \
--cc=hch@lst.de \
--cc=james.smart@broadcom.com \
--cc=jsmart2021@gmail.com \
--cc=kbusch@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-nvme@lists.infradead.org \
--cc=sagi@grimberg.me \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).