From: Salman Qazi <sqazi@google.com>
To: Jens Axboe <axboe@kernel.dk>, Ming Lei <ming.lei@redhat.com>,
Bart Van Assche <bvanassche@acm.org>,
linux-block@vger.kernel.org, linux-kernel@vger.kernel.org
Cc: Jesse Barnes <jsbarnes@google.com>,
Gwendal Grignou <gwendal@google.com>,
Hannes Reinecke <hare@suse.com>, Christoph Hellwig <hch@lst.de>,
Salman Qazi <sqazi@google.com>
Subject: [PATCH] block: Limit number of items taken from the I/O scheduler in one go
Date: Wed, 5 Feb 2020 11:57:06 -0800 [thread overview]
Message-ID: <20200205195706.49438-1-sqazi@google.com> (raw)
In-Reply-To: <20200205045526.GA15286@ming.t460p>
Flushes bypass the I/O scheduler and get added to hctx->dispatch
in blk_mq_sched_bypass_insert. This can happen while a kworker is running
hctx->run_work work item and is past the point in
blk_mq_sched_dispatch_requests where hctx->dispatch is checked.
The blk_mq_do_dispatch_sched call is not guaranteed to end in bounded time,
because the I/O scheduler can feed an arbitrary number of commands.
Since we have only one hctx->run_work, the commands waiting in
hctx->dispatch will wait an arbitrary length of time for run_work to be
rerun.
A similar phenomenon exists with dispatches from the software queue.
The solution is to poll hctx->dispatch in blk_mq_do_dispatch_sched and
blk_mq_do_dispatch_ctx and return from the run_work handler and let it
rerun.
Signed-off-by: Salman Qazi <sqazi@google.com>
---
block/blk-mq-sched.c | 37 +++++++++++++++++++++++++++++++------
1 file changed, 31 insertions(+), 6 deletions(-)
diff --git a/block/blk-mq-sched.c b/block/blk-mq-sched.c
index ca22afd47b3d..52249fddeb66 100644
--- a/block/blk-mq-sched.c
+++ b/block/blk-mq-sched.c
@@ -84,12 +84,16 @@ void blk_mq_sched_restart(struct blk_mq_hw_ctx *hctx)
* Only SCSI implements .get_budget and .put_budget, and SCSI restarts
* its queue by itself in its completion handler, so we don't need to
* restart queue if .get_budget() returns BLK_STS_NO_RESOURCE.
+ *
+ * Returns true if hctx->dispatch was found non-empty and
+ * run_work has to be run again.
*/
-static void blk_mq_do_dispatch_sched(struct blk_mq_hw_ctx *hctx)
+static bool blk_mq_do_dispatch_sched(struct blk_mq_hw_ctx *hctx)
{
struct request_queue *q = hctx->queue;
struct elevator_queue *e = q->elevator;
LIST_HEAD(rq_list);
+ bool ret = false;
do {
struct request *rq;
@@ -97,6 +101,11 @@ static void blk_mq_do_dispatch_sched(struct blk_mq_hw_ctx *hctx)
if (e->type->ops.has_work && !e->type->ops.has_work(hctx))
break;
+ if (!list_empty_careful(&hctx->dispatch)) {
+ ret = true;
+ break;
+ }
+
if (!blk_mq_get_dispatch_budget(hctx))
break;
@@ -113,6 +122,8 @@ static void blk_mq_do_dispatch_sched(struct blk_mq_hw_ctx *hctx)
*/
list_add(&rq->queuelist, &rq_list);
} while (blk_mq_dispatch_rq_list(q, &rq_list, true));
+
+ return ret;
}
static struct blk_mq_ctx *blk_mq_next_ctx(struct blk_mq_hw_ctx *hctx,
@@ -130,16 +141,25 @@ static struct blk_mq_ctx *blk_mq_next_ctx(struct blk_mq_hw_ctx *hctx,
* Only SCSI implements .get_budget and .put_budget, and SCSI restarts
* its queue by itself in its completion handler, so we don't need to
* restart queue if .get_budget() returns BLK_STS_NO_RESOURCE.
+ *
+ * Returns true if hctx->dispatch was found non-empty and
+ * run_work has to be run again.
*/
-static void blk_mq_do_dispatch_ctx(struct blk_mq_hw_ctx *hctx)
+static bool blk_mq_do_dispatch_ctx(struct blk_mq_hw_ctx *hctx)
{
struct request_queue *q = hctx->queue;
LIST_HEAD(rq_list);
struct blk_mq_ctx *ctx = READ_ONCE(hctx->dispatch_from);
+ bool ret = false;
do {
struct request *rq;
+ if (!list_empty_careful(&hctx->dispatch)) {
+ ret = true;
+ break;
+ }
+
if (!sbitmap_any_bit_set(&hctx->ctx_map))
break;
@@ -165,6 +185,7 @@ static void blk_mq_do_dispatch_ctx(struct blk_mq_hw_ctx *hctx)
} while (blk_mq_dispatch_rq_list(q, &rq_list, true));
WRITE_ONCE(hctx->dispatch_from, ctx);
+ return ret;
}
void blk_mq_sched_dispatch_requests(struct blk_mq_hw_ctx *hctx)
@@ -172,6 +193,7 @@ void blk_mq_sched_dispatch_requests(struct blk_mq_hw_ctx *hctx)
struct request_queue *q = hctx->queue;
struct elevator_queue *e = q->elevator;
const bool has_sched_dispatch = e && e->type->ops.dispatch_request;
+ bool run_again = false;
LIST_HEAD(rq_list);
/* RCU or SRCU read lock is needed before checking quiesced flag */
@@ -208,19 +230,22 @@ void blk_mq_sched_dispatch_requests(struct blk_mq_hw_ctx *hctx)
blk_mq_sched_mark_restart_hctx(hctx);
if (blk_mq_dispatch_rq_list(q, &rq_list, false)) {
if (has_sched_dispatch)
- blk_mq_do_dispatch_sched(hctx);
+ run_again = blk_mq_do_dispatch_sched(hctx);
else
- blk_mq_do_dispatch_ctx(hctx);
+ run_again = blk_mq_do_dispatch_ctx(hctx);
}
} else if (has_sched_dispatch) {
- blk_mq_do_dispatch_sched(hctx);
+ run_again = blk_mq_do_dispatch_sched(hctx);
} else if (hctx->dispatch_busy) {
/* dequeue request one by one from sw queue if queue is busy */
- blk_mq_do_dispatch_ctx(hctx);
+ run_again = blk_mq_do_dispatch_ctx(hctx);
} else {
blk_mq_flush_busy_ctxs(hctx, &rq_list);
blk_mq_dispatch_rq_list(q, &rq_list, false);
}
+
+ if (run_again)
+ blk_mq_run_hw_queue(hctx, true);
}
bool blk_mq_sched_try_merge(struct request_queue *q, struct bio *bio,
--
2.25.0.341.g760bfbb309-goog
next prev parent reply other threads:[~2020-02-05 19:57 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-01-30 19:34 Hung tasks with multiple partitions Salman Qazi
2020-01-30 20:49 ` Bart Van Assche
2020-01-30 21:02 ` Salman Qazi
[not found] ` <20200203204554.119849-1-sqazi@google.com>
2020-02-03 20:59 ` [PATCH] block: Limit number of items taken from the I/O scheduler in one go Salman Qazi
2020-02-04 3:47 ` Bart Van Assche
2020-02-04 9:20 ` Ming Lei
2020-02-04 18:26 ` Salman Qazi
2020-02-04 19:37 ` Salman Qazi
2020-02-05 4:55 ` Ming Lei
2020-02-05 19:57 ` Salman Qazi [this message]
2020-02-06 10:18 ` Ming Lei
2020-02-06 21:12 ` Salman Qazi
2020-02-07 2:07 ` Ming Lei
2020-02-07 15:26 ` Bart Van Assche
2020-02-07 18:45 ` Salman Qazi
2020-02-07 19:04 ` Salman Qazi
2020-02-07 20:19 ` Bart Van Assche
2020-02-07 20:37 ` Salman Qazi
2020-04-20 16:42 ` Doug Anderson
2020-04-23 20:13 ` Jesse Barnes
2020-04-23 20:34 ` Jens Axboe
2020-04-23 20:40 ` Salman Qazi
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=20200205195706.49438-1-sqazi@google.com \
--to=sqazi@google.com \
--cc=axboe@kernel.dk \
--cc=bvanassche@acm.org \
--cc=gwendal@google.com \
--cc=hare@suse.com \
--cc=hch@lst.de \
--cc=jsbarnes@google.com \
--cc=linux-block@vger.kernel.org \
--cc=linux-kernel@vger.kernel.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 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).