From: Ming Lei <tom.leiming@gmail.com>
To: Keith Busch <keith.busch@linux.intel.com>
Cc: Jens Axboe <axboe@kernel.dk>,
linux-block <linux-block@vger.kernel.org>,
linux-nvme <linux-nvme@lists.infradead.org>,
Ming Lei <ming.lei@redhat.com>,
Keith Busch <keith.busch@intel.com>,
Bart Van Assche <bart.vanassche@wdc.com>,
Christoph Hellwig <hch@lst.de>
Subject: Re: [RFC PATCH 3/3] blk-mq: Remove generation seqeunce
Date: Tue, 22 May 2018 23:23:08 +0800 [thread overview]
Message-ID: <CACVXFVP3a3-TW15Z9x5HAGaJaxmv2k0XcwtJCoQAR38e_NShMQ@mail.gmail.com> (raw)
In-Reply-To: <20180522151757.GB8295@localhost.localdomain>
On Tue, May 22, 2018 at 11:17 PM, Keith Busch
<keith.busch@linux.intel.com> wrote:
> On Tue, May 22, 2018 at 11:07:07PM +0800, Ming Lei wrote:
>> > At approximately the same time, you're saying the driver that returned
>> > EH_HANDLED may then call blk_mq_complete_request() in reference to the
>> > *old* request that it returned EH_HANDLED for, incorrectly completing
>>
>> Because this request has been completed above by blk-mq timeout,
>> then this old request won't be completed any more via blk_mq_complete_request()
>> either from normal path or what ever, such as cancel.
>
>> > the new request before it is done. That will inevitably lead to data
>> > corruption. Is that happening today in any driver?
>>
>> No such issue since current blk-mq makes sure one req is only completed
>> once, but your patch changes to depend on driver to make sure that.
>
> The blk-mq timeout complete makes the request available for allocation
> as a new command, at which point blk_mq_complete_request can be called
> again. If a driver is somehow relying on blk-mq to prevent a double
> completion for a previously completed request context, they're already
> in a lot of trouble.
Yes, previously there is the atomic flag of REQ_ATOM_COMPLETE for
covering the atomic completion, and recently Tejun changes to aborted
state with generation counter, but both provides sort of atomic completion.
So even though it is much simplified by using request refcount, the atomic
completion should be provided by blk-mq, or drivers have to be audited
to avoid double completion.
Thanks,
Ming Lei
WARNING: multiple messages have this Message-ID (diff)
From: tom.leiming@gmail.com (Ming Lei)
Subject: [RFC PATCH 3/3] blk-mq: Remove generation seqeunce
Date: Tue, 22 May 2018 23:23:08 +0800 [thread overview]
Message-ID: <CACVXFVP3a3-TW15Z9x5HAGaJaxmv2k0XcwtJCoQAR38e_NShMQ@mail.gmail.com> (raw)
In-Reply-To: <20180522151757.GB8295@localhost.localdomain>
On Tue, May 22, 2018 at 11:17 PM, Keith Busch
<keith.busch@linux.intel.com> wrote:
> On Tue, May 22, 2018@11:07:07PM +0800, Ming Lei wrote:
>> > At approximately the same time, you're saying the driver that returned
>> > EH_HANDLED may then call blk_mq_complete_request() in reference to the
>> > *old* request that it returned EH_HANDLED for, incorrectly completing
>>
>> Because this request has been completed above by blk-mq timeout,
>> then this old request won't be completed any more via blk_mq_complete_request()
>> either from normal path or what ever, such as cancel.
>
>> > the new request before it is done. That will inevitably lead to data
>> > corruption. Is that happening today in any driver?
>>
>> No such issue since current blk-mq makes sure one req is only completed
>> once, but your patch changes to depend on driver to make sure that.
>
> The blk-mq timeout complete makes the request available for allocation
> as a new command, at which point blk_mq_complete_request can be called
> again. If a driver is somehow relying on blk-mq to prevent a double
> completion for a previously completed request context, they're already
> in a lot of trouble.
Yes, previously there is the atomic flag of REQ_ATOM_COMPLETE for
covering the atomic completion, and recently Tejun changes to aborted
state with generation counter, but both provides sort of atomic completion.
So even though it is much simplified by using request refcount, the atomic
completion should be provided by blk-mq, or drivers have to be audited
to avoid double completion.
Thanks,
Ming Lei
next prev parent reply other threads:[~2018-05-22 15:23 UTC|newest]
Thread overview: 128+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-05-21 23:11 [RFC PATCH 0/3] blk-mq: Timeout rework Keith Busch
2018-05-21 23:11 ` Keith Busch
2018-05-21 23:11 ` [RFC PATCH 1/3] blk-mq: Reference count request usage Keith Busch
2018-05-21 23:11 ` Keith Busch
2018-05-22 2:27 ` Ming Lei
2018-05-22 2:27 ` Ming Lei
2018-05-22 15:19 ` Christoph Hellwig
2018-05-22 15:19 ` Christoph Hellwig
2018-05-21 23:11 ` [RFC PATCH 2/3] blk-mq: Fix timeout and state order Keith Busch
2018-05-21 23:11 ` Keith Busch
2018-05-22 2:28 ` Ming Lei
2018-05-22 2:28 ` Ming Lei
2018-05-22 15:24 ` Christoph Hellwig
2018-05-22 15:24 ` Christoph Hellwig
2018-05-22 16:27 ` Bart Van Assche
2018-05-22 16:27 ` Bart Van Assche
2018-05-21 23:11 ` [RFC PATCH 3/3] blk-mq: Remove generation seqeunce Keith Busch
2018-05-21 23:11 ` Keith Busch
2018-05-21 23:29 ` Bart Van Assche
2018-05-21 23:29 ` Bart Van Assche
2018-05-22 14:15 ` Keith Busch
2018-05-22 14:15 ` Keith Busch
2018-05-22 16:29 ` Bart Van Assche
2018-05-22 16:29 ` Bart Van Assche
2018-05-22 16:34 ` Keith Busch
2018-05-22 16:34 ` Keith Busch
2018-05-22 16:48 ` Bart Van Assche
2018-05-22 16:48 ` Bart Van Assche
2018-05-22 2:49 ` Ming Lei
2018-05-22 2:49 ` Ming Lei
2018-05-22 3:16 ` Jens Axboe
2018-05-22 3:16 ` Jens Axboe
2018-05-22 3:47 ` Ming Lei
2018-05-22 3:47 ` Ming Lei
2018-05-22 3:51 ` Jens Axboe
2018-05-22 3:51 ` Jens Axboe
2018-05-22 8:51 ` Ming Lei
2018-05-22 8:51 ` Ming Lei
2018-05-22 14:35 ` Jens Axboe
2018-05-22 14:35 ` Jens Axboe
2018-05-22 14:20 ` Keith Busch
2018-05-22 14:20 ` Keith Busch
2018-05-22 14:37 ` Ming Lei
2018-05-22 14:37 ` Ming Lei
2018-05-22 14:46 ` Keith Busch
2018-05-22 14:46 ` Keith Busch
2018-05-22 14:57 ` Ming Lei
2018-05-22 14:57 ` Ming Lei
2018-05-22 15:01 ` Keith Busch
2018-05-22 15:01 ` Keith Busch
2018-05-22 15:07 ` Ming Lei
2018-05-22 15:07 ` Ming Lei
2018-05-22 15:17 ` Keith Busch
2018-05-22 15:17 ` Keith Busch
2018-05-22 15:23 ` Ming Lei [this message]
2018-05-22 15:23 ` Ming Lei
2018-05-22 16:17 ` Christoph Hellwig
2018-05-22 16:17 ` Christoph Hellwig
2018-05-23 0:34 ` Ming Lei
2018-05-23 0:34 ` Ming Lei
2018-05-23 14:35 ` Keith Busch
2018-05-23 14:35 ` Keith Busch
2018-05-24 1:52 ` Ming Lei
2018-05-24 1:52 ` Ming Lei
2018-05-23 5:48 ` Hannes Reinecke
2018-05-23 5:48 ` Hannes Reinecke
2018-07-12 18:16 ` Bart Van Assche
2018-07-12 18:16 ` Bart Van Assche
2018-07-12 19:24 ` Keith Busch
2018-07-12 19:24 ` Keith Busch
2018-07-12 22:24 ` Bart Van Assche
2018-07-12 22:24 ` Bart Van Assche
2018-07-13 1:12 ` jianchao.wang
2018-07-13 1:12 ` jianchao.wang
2018-07-13 2:40 ` jianchao.wang
2018-07-13 2:40 ` jianchao.wang
2018-07-13 15:43 ` Keith Busch
2018-07-13 15:43 ` Keith Busch
2018-07-13 15:52 ` Bart Van Assche
2018-07-13 15:52 ` Bart Van Assche
2018-07-13 18:47 ` Keith Busch
2018-07-13 18:47 ` Keith Busch
2018-07-13 23:03 ` Bart Van Assche
2018-07-13 23:03 ` Bart Van Assche
2018-07-13 23:58 ` Keith Busch
2018-07-13 23:58 ` Keith Busch
2018-07-18 19:56 ` hch
2018-07-18 19:56 ` hch
2018-07-18 20:39 ` hch
2018-07-18 20:39 ` hch
2018-07-18 21:05 ` Bart Van Assche
2018-07-18 21:05 ` Bart Van Assche
2018-07-18 22:53 ` Keith Busch
2018-07-18 22:53 ` Keith Busch
2018-07-18 20:53 ` Keith Busch
2018-07-18 20:53 ` Keith Busch
2018-07-18 20:58 ` Bart Van Assche
2018-07-18 20:58 ` Bart Van Assche
2018-07-18 21:17 ` Keith Busch
2018-07-18 21:17 ` Keith Busch
2018-07-18 21:30 ` Bart Van Assche
2018-07-18 21:30 ` Bart Van Assche
2018-07-18 21:33 ` Keith Busch
2018-07-18 21:33 ` Keith Busch
2018-07-19 13:19 ` hch
2018-07-19 13:19 ` hch
2018-07-19 14:59 ` Keith Busch
2018-07-19 14:59 ` Keith Busch
2018-07-19 15:56 ` Keith Busch
2018-07-19 15:56 ` Keith Busch
2018-07-19 16:04 ` Bart Van Assche
2018-07-19 16:04 ` Bart Van Assche
2018-07-19 16:22 ` Keith Busch
2018-07-19 16:22 ` Keith Busch
2018-07-19 16:29 ` hch
2018-07-19 16:29 ` hch
2018-07-19 20:18 ` Keith Busch
2018-07-19 20:18 ` Keith Busch
2018-07-19 13:22 ` hch
2018-07-19 13:22 ` hch
2018-05-21 23:29 ` [RFC PATCH 0/3] blk-mq: Timeout rework Bart Van Assche
2018-05-21 23:29 ` Bart Van Assche
2018-05-22 14:06 ` Keith Busch
2018-05-22 14:06 ` Keith Busch
2018-05-22 16:30 ` Bart Van Assche
2018-05-22 16:30 ` Bart Van Assche
2018-05-22 16:44 ` Keith Busch
2018-05-22 16:44 ` Keith Busch
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=CACVXFVP3a3-TW15Z9x5HAGaJaxmv2k0XcwtJCoQAR38e_NShMQ@mail.gmail.com \
--to=tom.leiming@gmail.com \
--cc=axboe@kernel.dk \
--cc=bart.vanassche@wdc.com \
--cc=hch@lst.de \
--cc=keith.busch@intel.com \
--cc=keith.busch@linux.intel.com \
--cc=linux-block@vger.kernel.org \
--cc=linux-nvme@lists.infradead.org \
--cc=ming.lei@redhat.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.