From: Keith Busch <keith.busch@intel.com>
To: Christoph Hellwig <hch@infradead.org>
Cc: Jens Axboe <axboe@kernel.dk>,
linux-block@vger.kernel.org, linux-nvme@lists.infradead.org
Subject: Re: [PATCH 1/6] nvme: don't disable local ints for polled queue
Date: Wed, 14 Nov 2018 08:37:07 -0700 [thread overview]
Message-ID: <20181114153706.GA11416@localhost.localdomain> (raw)
In-Reply-To: <20181114084322.GA14954@infradead.org>
On Wed, Nov 14, 2018 at 12:43:22AM -0800, Christoph Hellwig wrote:
> static int nvme_poll_noirq(struct blk_mq_hw_ctx *hctx, unsigned int tag)
> {
> struct nvme_queue *nvmeq = hctx->driver_data;
> > u16 start, end;
> bool found;
> >
> > if (!nvme_cqe_pending(nvmeq))
> > return 0;
> >
> > + spin_lock(&nvmeq->cq_lock);
> > found = nvme_process_cq(nvmeq, &start, &end, tag);
> > + spin_unlock(&nvmeq->cq_lock);
> > +
> > nvme_complete_cqes(nvmeq, start, end);
> > return found;
>
> And while we are at it: I think for the irq-driven queues in a
> separate queue for poll setup we might not even need to take the
> CQ lock. Which might be an argument for only allowing polling
> if we have the separate queues just to keep everything simple.
That's a pretty cool observation. We still poll interrupt driven queues
in the timeout path as a sanity check (it really has helped in debugging
timeout issues), but we can temporarily disable the cq's irq and be
lockless.
WARNING: multiple messages have this Message-ID (diff)
From: keith.busch@intel.com (Keith Busch)
Subject: [PATCH 1/6] nvme: don't disable local ints for polled queue
Date: Wed, 14 Nov 2018 08:37:07 -0700 [thread overview]
Message-ID: <20181114153706.GA11416@localhost.localdomain> (raw)
In-Reply-To: <20181114084322.GA14954@infradead.org>
On Wed, Nov 14, 2018@12:43:22AM -0800, Christoph Hellwig wrote:
> static int nvme_poll_noirq(struct blk_mq_hw_ctx *hctx, unsigned int tag)
> {
> struct nvme_queue *nvmeq = hctx->driver_data;
> > u16 start, end;
> bool found;
> >
> > if (!nvme_cqe_pending(nvmeq))
> > return 0;
> >
> > + spin_lock(&nvmeq->cq_lock);
> > found = nvme_process_cq(nvmeq, &start, &end, tag);
> > + spin_unlock(&nvmeq->cq_lock);
> > +
> > nvme_complete_cqes(nvmeq, start, end);
> > return found;
>
> And while we are at it: I think for the irq-driven queues in a
> separate queue for poll setup we might not even need to take the
> CQ lock. Which might be an argument for only allowing polling
> if we have the separate queues just to keep everything simple.
That's a pretty cool observation. We still poll interrupt driven queues
in the timeout path as a sanity check (it really has helped in debugging
timeout issues), but we can temporarily disable the cq's irq and be
lockless.
next prev parent reply other threads:[~2018-11-14 15:40 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-11-10 15:13 [PATCHSET 0/6] Various block optimizations Jens Axboe
2018-11-10 15:13 ` [PATCH 1/6] nvme: don't disable local ints for polled queue Jens Axboe
2018-11-10 15:13 ` Jens Axboe
2018-11-14 8:43 ` Christoph Hellwig
2018-11-14 8:43 ` Christoph Hellwig
2018-11-14 15:31 ` Jens Axboe
2018-11-14 15:31 ` Jens Axboe
2018-11-14 15:37 ` Keith Busch [this message]
2018-11-14 15:37 ` Keith Busch
2018-11-14 10:18 ` Max Gurtovoy
2018-11-14 10:18 ` Max Gurtovoy
2018-11-10 15:13 ` [PATCH 2/6] block: add queue_is_mq() helper Jens Axboe
2018-11-10 15:13 ` [PATCH 3/6] blk-mq: embed blk_mq_ops directly in the request queue Jens Axboe
2018-11-14 15:20 ` Christoph Hellwig
2018-11-14 15:33 ` Jens Axboe
2018-11-14 15:39 ` Christoph Hellwig
2018-11-14 15:47 ` Jens Axboe
2018-11-14 15:58 ` Christoph Hellwig
2018-11-10 15:13 ` [PATCH 4/6] block: avoid ordered task state change for polled IO Jens Axboe
2018-11-12 9:35 ` jianchao.wang
2018-11-12 16:26 ` Jens Axboe
2018-11-13 2:02 ` jianchao.wang
2018-11-10 15:13 ` [PATCH 5/6] blk-rq-qos: inline check for q->rq_qos functions Jens Axboe
2018-11-14 15:21 ` Christoph Hellwig
2018-11-14 15:33 ` Jens Axboe
2018-11-14 15:40 ` Christoph Hellwig
2018-11-14 17:16 ` Jens Axboe
2018-11-10 15:13 ` [PATCH 6/6] block: add polled wakeup task helper Jens Axboe
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=20181114153706.GA11416@localhost.localdomain \
--to=keith.busch@intel.com \
--cc=axboe@kernel.dk \
--cc=hch@infradead.org \
--cc=linux-block@vger.kernel.org \
--cc=linux-nvme@lists.infradead.org \
/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.