From: Omar Sandoval <osandov@osandov.com>
To: Jens Axboe <axboe@fb.com>, linux-block@vger.kernel.org
Cc: Christoph Hellwig <hch@lst.de>,
Keith Busch <keith.busch@intel.com>, Josef Bacik <jbacik@fb.com>,
kernel-team@fb.com
Subject: [PATCH] blk-mq: remap queues when adding/removing hardware queues
Date: Fri, 31 Mar 2017 11:59:24 -0700 [thread overview]
Message-ID: <52fb0623a8496808622c718f0f6372d37574dbe1.1490986618.git.osandov@fb.com> (raw)
From: Omar Sandoval <osandov@fb.com>
blk_mq_update_nr_hw_queues() used to remap hardware queues, which is the
behavior that drivers expect. However, commit 4e68a011428a changed
blk_mq_queue_reinit() to not remap queues for the case of CPU
hotplugging, inadvertently making blk_mq_update_nr_hw_queues() not remap
queues as well. This breaks, for example, NBD's multi-connection mode,
leaving the added hardware queues unused. Fix it by making
blk_mq_queue_reinit() optionally remap queues, which we do when updating
the number of hardware queues but not when hotplugging.
Fixes: 4e68a011428a ("blk-mq: don't redistribute hardware queues on a CPU hotplug event")
Signed-off-by: Omar Sandoval <osandov@fb.com>
---
The only callers of blk_mq_update_nr_hw_queues() are nbd and nbd. I *think*
nbd_dev_add() also wants this remap behavior.
block/blk-mq.c | 25 ++++++++++++++++++-------
1 file changed, 18 insertions(+), 7 deletions(-)
diff --git a/block/blk-mq.c b/block/blk-mq.c
index 061fc2cc88d3..1abbf7c83193 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -2343,14 +2343,27 @@ void blk_mq_free_queue(struct request_queue *q)
blk_mq_exit_hw_queues(q, set, set->nr_hw_queues);
}
+static int blk_mq_update_queue_map(struct blk_mq_tag_set *set)
+{
+ if (set->ops->map_queues)
+ return set->ops->map_queues(set);
+ else
+ return blk_mq_map_queues(set);
+}
+
+
/* Basically redo blk_mq_init_queue with queue frozen */
static void blk_mq_queue_reinit(struct request_queue *q,
- const struct cpumask *online_mask)
+ const struct cpumask *online_mask,
+ bool remap_queues)
{
WARN_ON_ONCE(!atomic_read(&q->mq_freeze_depth));
blk_mq_sysfs_unregister(q);
+ if (remap_queues)
+ blk_mq_update_queue_map(q->tag_set);
+
/*
* redo blk_mq_init_cpu_queues and blk_mq_init_hw_queues. FIXME: maybe
* we should change hctx numa_node according to new topology (this
@@ -2387,7 +2400,7 @@ static void blk_mq_queue_reinit_work(void)
blk_mq_freeze_queue_wait(q);
list_for_each_entry(q, &all_q_list, all_q_node)
- blk_mq_queue_reinit(q, &cpuhp_online_new);
+ blk_mq_queue_reinit(q, &cpuhp_online_new, false);
list_for_each_entry(q, &all_q_list, all_q_node)
blk_mq_unfreeze_queue(q);
@@ -2532,10 +2545,7 @@ int blk_mq_alloc_tag_set(struct blk_mq_tag_set *set)
if (!set->mq_map)
goto out_free_tags;
- if (set->ops->map_queues)
- ret = set->ops->map_queues(set);
- else
- ret = blk_mq_map_queues(set);
+ ret = blk_mq_update_queue_map(set);
if (ret)
goto out_free_mq_map;
@@ -2629,11 +2639,12 @@ void blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, int nr_hw_queues)
set->nr_hw_queues = nr_hw_queues;
list_for_each_entry(q, &set->tag_list, tag_set_list) {
blk_mq_realloc_hw_ctxs(set, q);
- blk_mq_queue_reinit(q, cpu_online_mask);
+ blk_mq_queue_reinit(q, cpu_online_mask, true);
}
list_for_each_entry(q, &set->tag_list, tag_set_list)
blk_mq_unfreeze_queue(q);
+
}
EXPORT_SYMBOL_GPL(blk_mq_update_nr_hw_queues);
--
2.12.1
next reply other threads:[~2017-03-31 19:00 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-03-31 18:59 Omar Sandoval [this message]
2017-03-31 19:01 ` [PATCH] blk-mq: remap queues when adding/removing hardware queues Omar Sandoval
2017-03-31 20:30 ` Keith Busch
2017-03-31 20:30 ` Omar Sandoval
2017-03-31 20:44 ` Keith Busch
2017-03-31 20:43 ` [PATCH v2] " Omar Sandoval
2017-03-31 20:46 ` Omar Sandoval
2017-03-31 20:48 ` [PATCH v3] " Omar Sandoval
2017-03-31 21:09 ` Keith Busch
2017-04-04 6:24 ` Christoph Hellwig
2017-04-05 12:01 ` 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=52fb0623a8496808622c718f0f6372d37574dbe1.1490986618.git.osandov@fb.com \
--to=osandov@osandov.com \
--cc=axboe@fb.com \
--cc=hch@lst.de \
--cc=jbacik@fb.com \
--cc=keith.busch@intel.com \
--cc=kernel-team@fb.com \
--cc=linux-block@vger.kernel.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 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).