mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: "Yu Kuai" <yukuai@fnnas.com>
To: "Li Nan" <linan666@huaweicloud.com>, <song@kernel.org>,
	 <linux-raid@vger.kernel.org>
Cc: <linux-kernel@vger.kernel.org>,
	 "yangerkun@huawei.com" <yangerkun@huawei.com>,
	<yukuai@fnnas.com>
Subject: Re: [PATCH] md: wake raid456 reshape waiters before suspend
Date: Tue, 7 Apr 2026 13:33:29 +0800	[thread overview]
Message-ID: <ba76b132-4bf9-445f-9a77-6194c0fc437b@fnnas.com> (raw)
In-Reply-To: <3f930776-20e7-f74b-067b-bfb348291a50@huaweicloud.com>

Hi,

在 2026/4/2 10:53, Li Nan 写道:
>
>
> 在 2026/3/27 22:07, Yu Kuai 写道:
>> During raid456 reshape, direct IO across the reshape position can sleep
>> in raid5_make_request() waiting for reshape progress while still
>> holding an active_io reference. If userspace then freezes reshape and
>> writes md/suspend_lo or md/suspend_hi, mddev_suspend() kills active_io
>> and waits for all in-flight IO to drain.
>>
>> This can deadlock: the IO needs reshape progress to continue, but the
>> reshape thread is already frozen, so the active_io reference is never
>> dropped and suspend never completes.
>>
>> raid5_prepare_suspend() already wakes wait_for_reshape for dm-raid. Do
>> the same for normal md suspend when reshape is already interrupted, so
>> waiting raid456 IO can abort, drop its reference, and let suspend
>> finish.
>>
>> The mdadm test tests/25raid456-reshape-deadlock reproduces the hang.
>>
>> Fixes: 714d20150ed8 ("md: add new helpers to suspend/resume array")
>> Signed-off-by: Yu Kuai <yukuai@fnnas.com>
>> ---
>>   drivers/md/md.c | 11 +++++++++++
>>   1 file changed, 11 insertions(+)
>>
>> diff --git a/drivers/md/md.c b/drivers/md/md.c
>> index 65c85973092a..3ae5f36b5baf 100644
>> --- a/drivers/md/md.c
>> +++ b/drivers/md/md.c
>> @@ -489,6 +489,17 @@ int mddev_suspend(struct mddev *mddev, bool 
>> interruptible)
>>       }
>>         percpu_ref_kill(&mddev->active_io);
>> +
>> +    /*
>> +     * RAID456 IO can sleep in wait_for_reshape while still holding an
>> +     * active_io reference. If reshape is already interrupted or 
>> frozen,
>> +     * wake those waiters so they can abort and drop the reference 
>> instead
>> +     * of deadlocking suspend.
>> +     */
>> +    if (mddev->pers && mddev->pers->prepare_suspend &&
>> +        reshape_interrupted(mddev))
>> +        mddev->pers->prepare_suspend(mddev);
>> +
>>       if (interruptible)
>>           err = wait_event_interruptible(mddev->sb_wait,
>> percpu_ref_is_zero(&mddev->active_io));
>
> If I only freeze raid5 without suspend, the IO still hangs and cannot
> complete. Should we fix this scenario?
Yes, for other path can freeze raid5 directly, we should fix it as well.
>
> -- 
> Thanks,
> Nan
>
-- 
Thansk,
Kuai

  reply	other threads:[~2026-04-07  5:33 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-03-27 14:07 Yu Kuai
2026-04-02  2:53 ` Li Nan
2026-04-07  5:33   ` Yu Kuai [this message]
2026-04-07  5:32 ` Yu Kuai

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=ba76b132-4bf9-445f-9a77-6194c0fc437b@fnnas.com \
    --to=yukuai@fnnas.com \
    --cc=linan666@huaweicloud.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-raid@vger.kernel.org \
    --cc=song@kernel.org \
    --cc=yangerkun@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®