From: Ming Lei <ming.lei@redhat.com>
To: Sagi Grimberg <sagi@grimberg.me>
Cc: Jens Axboe <axboe@kernel.dk>,
linux-nvme@lists.infradead.org, linux-block@vger.kernel.org,
Chao Leng <lengchao@huawei.com>, Keith Busch <kbusch@kernel.org>,
Christoph Hellwig <hch@lst.de>
Subject: Re: [PATCH v3 1/2] blk-mq: add async quiesce interface
Date: Tue, 28 Jul 2020 09:09:28 +0800 [thread overview]
Message-ID: <20200728010928.GA1303645@T590> (raw)
In-Reply-To: <bcb8f89b-8477-c48b-1e0f-947cbe741818@grimberg.me>
On Mon, Jul 27, 2020 at 02:00:15PM -0700, Sagi Grimberg wrote:
>
> > > > > > > +void blk_mq_quiesce_queue_async(struct request_queue *q)
> > > > > > > +{
> > > > > > > + struct blk_mq_hw_ctx *hctx;
> > > > > > > + unsigned int i;
> > > > > > > +
> > > > > > > + blk_mq_quiesce_queue_nowait(q);
> > > > > > > +
> > > > > > > + queue_for_each_hw_ctx(q, hctx, i) {
> > > > > > > + init_completion(&hctx->rcu_sync.completion);
> > > > > > > + init_rcu_head(&hctx->rcu_sync.head);
> > > > > > > + if (hctx->flags & BLK_MQ_F_BLOCKING)
> > > > > > > + call_srcu(hctx->srcu, &hctx->rcu_sync.head,
> > > > > > > + wakeme_after_rcu);
> > > > > > > + else
> > > > > > > + call_rcu(&hctx->rcu_sync.head,
> > > > > > > + wakeme_after_rcu);
> > > > > > > + }
> > > > > >
> > > > > > Looks not necessary to do anything in case of !BLK_MQ_F_BLOCKING, and single
> > > > > > synchronize_rcu() is OK for all hctx during waiting.
> > > > >
> > > > > That's true, but I want a single interface for both. v2 had exactly
> > > > > that, but I decided that this approach is better.
> > > >
> > > > Not sure one new interface is needed, and one simple way is to:
> > > >
> > > > 1) call blk_mq_quiesce_queue_nowait() for each request queue
> > > >
> > > > 2) wait in driver specific way
> > > >
> > > > Or just wondering why nvme doesn't use set->tag_list to retrieve NS,
> > > > then you may add per-tagset APIs for the waiting.
> > >
> > > Because it puts assumptions on how quiesce works, which is something
> > > I'd like to avoid because I think its cleaner, what do others think?
> > > Jens? Christoph?
> >
> > I'd prefer to have it in a helper, and just have blk_mq_quiesce_queue()
> > call that.
>
> I agree with this approach as well.
>
> Jens, this mean that we use the call_rcu mechanism also for non-blocking
> hctxs, because the caller will call it for multiple request queues (see
> patch 2) and we don't want to call synchronize_rcu for every request
> queue serially, we want it to happen in parallel.
>
> Which leaves us with the patchset as it is, just to convert the
> rcu_synchronize structure to be dynamically allocated on the heap
> rather than keeping it statically allocated in the hctx.
>
> This is how it looks:
> --
> diff --git a/block/blk-mq.c b/block/blk-mq.c
> index abcf590f6238..d913924117d2 100644
> --- a/block/blk-mq.c
> +++ b/block/blk-mq.c
> @@ -209,6 +209,52 @@ void blk_mq_quiesce_queue_nowait(struct request_queue
> *q)
> }
> EXPORT_SYMBOL_GPL(blk_mq_quiesce_queue_nowait);
>
> +void blk_mq_quiesce_queue_async(struct request_queue *q)
> +{
> + struct blk_mq_hw_ctx *hctx;
> + unsigned int i;
> + int rcu = false;
> +
> + blk_mq_quiesce_queue_nowait(q);
> +
> + queue_for_each_hw_ctx(q, hctx, i) {
> + hctx->rcu_sync = kmalloc(sizeof(*hctx->rcu_sync),
> GFP_KERNEL);
> + if (!hctx->rcu_sync) {
> + /* fallback to serial rcu sync */
> + if (hctx->flags & BLK_MQ_F_BLOCKING)
> + synchronize_srcu(hctx->srcu);
> + else
> + rcu = true;
> + } else {
> + init_completion(&hctx->rcu_sync->completion);
> + init_rcu_head(&hctx->rcu_sync->head);
> + if (hctx->flags & BLK_MQ_F_BLOCKING)
> + call_srcu(hctx->srcu, &hctx->rcu_sync->head,
> + wakeme_after_rcu);
> + else
> + call_rcu(&hctx->rcu_sync->head,
> + wakeme_after_rcu);
> + }
> + }
> + if (rcu)
> + synchronize_rcu();
> +}
> +EXPORT_SYMBOL_GPL(blk_mq_quiesce_queue_async);
> +
> +void blk_mq_quiesce_queue_async_wait(struct request_queue *q)
> +{
> + struct blk_mq_hw_ctx *hctx;
> + unsigned int i;
> +
> + queue_for_each_hw_ctx(q, hctx, i) {
> + if (!hctx->rcu_sync)
> + continue;
> + wait_for_completion(&hctx->rcu_sync->completion);
> + destroy_rcu_head(&hctx->rcu_sync->head);
> + }
> +}
> +EXPORT_SYMBOL_GPL(blk_mq_quiesce_queue_async_wait);
> +
> /**
> * blk_mq_quiesce_queue() - wait until all ongoing dispatches have finished
> * @q: request queue.
> diff --git a/include/linux/blk-mq.h b/include/linux/blk-mq.h
> index 23230c1d031e..7213ce56bb31 100644
> --- a/include/linux/blk-mq.h
> +++ b/include/linux/blk-mq.h
> @@ -5,6 +5,7 @@
> #include <linux/blkdev.h>
> #include <linux/sbitmap.h>
> #include <linux/srcu.h>
> +#include <linux/rcupdate_wait.h>
>
> struct blk_mq_tags;
> struct blk_flush_queue;
> @@ -170,6 +171,7 @@ struct blk_mq_hw_ctx {
> */
> struct list_head hctx_list;
>
> + struct rcu_synchronize *rcu_sync;
The above pointer needn't to be added to blk_mq_hw_ctx, and it can be
allocated on heap and passed to the waiting helper.
Thanks,
Ming
_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme
next prev parent reply other threads:[~2020-07-28 1:09 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-07-26 0:22 [PATCH v3 0/2] improve quiesce time for large amount of namespaces Sagi Grimberg
2020-07-26 0:23 ` [PATCH v3 1/2] blk-mq: add async quiesce interface Sagi Grimberg
2020-07-26 9:31 ` Ming Lei
2020-07-26 16:27 ` Sagi Grimberg
2020-07-27 2:08 ` Ming Lei
2020-07-27 3:33 ` Chao Leng
2020-07-27 3:50 ` Ming Lei
2020-07-27 5:55 ` Chao Leng
2020-07-27 6:32 ` Ming Lei
2020-07-27 18:40 ` Sagi Grimberg
2020-07-27 18:38 ` Sagi Grimberg
2020-07-27 18:36 ` Sagi Grimberg
2020-07-27 20:37 ` Jens Axboe
2020-07-27 21:00 ` Sagi Grimberg
2020-07-27 21:05 ` Jens Axboe
2020-07-27 21:21 ` Keith Busch
2020-07-27 21:30 ` Jens Axboe
2020-07-28 1:09 ` Ming Lei [this message]
2020-07-26 0:23 ` [PATCH v3 2/2] nvme: improve quiesce time for large amount of namespaces Sagi Grimberg
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=20200728010928.GA1303645@T590 \
--to=ming.lei@redhat.com \
--cc=axboe@kernel.dk \
--cc=hch@lst.de \
--cc=kbusch@kernel.org \
--cc=lengchao@huawei.com \
--cc=linux-block@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).