linux-block.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] block: Remove unnecessary elevator operation checks
@ 2021-06-18  1:59 Damien Le Moal
  2021-06-18  8:52 ` Johannes Thumshirn
                   ` (2 more replies)
  0 siblings, 3 replies; 4+ messages in thread
From: Damien Le Moal @ 2021-06-18  1:59 UTC (permalink / raw)
  To: linux-block, Jens Axboe

The insert_requests and dispatch_request elevator operations are
mandatory for the correct execution of an elevator, and all implemented
elevators (bfq, kyber and mq-deadline) implement them. As a result,
there is no need to check for these operations before calling them when
a queue has an elevator set. This simplifies the code in
__blk_mq_sched_dispatch_requests() and blk_mq_sched_insert_request().

To avoid out-of-tree elevators to crash the kernel in case of bad
implementation, add a check in elv_register() to verify that these
operations are implemented.

A small, probably not significant, IOPS improvement of 0.1% is observed
with this patch applied (4.117 MIOPS to 4.123 MIOPS, average of 20 fio
runs doing 4K random direct reads with psync and 32 jobs).

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>
---
 block/blk-mq-sched.c | 13 ++++++-------
 block/elevator.c     |  4 ++++
 2 files changed, 10 insertions(+), 7 deletions(-)

diff --git a/block/blk-mq-sched.c b/block/blk-mq-sched.c
index a9182d2f8ad3..bf2c105e81a2 100644
--- a/block/blk-mq-sched.c
+++ b/block/blk-mq-sched.c
@@ -294,8 +294,7 @@ static int blk_mq_do_dispatch_ctx(struct blk_mq_hw_ctx *hctx)
 static int __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;
+	const bool has_sched = q->elevator;
 	int ret = 0;
 	LIST_HEAD(rq_list);
 
@@ -326,12 +325,12 @@ static int __blk_mq_sched_dispatch_requests(struct blk_mq_hw_ctx *hctx)
 	if (!list_empty(&rq_list)) {
 		blk_mq_sched_mark_restart_hctx(hctx);
 		if (blk_mq_dispatch_rq_list(hctx, &rq_list, 0)) {
-			if (has_sched_dispatch)
+			if (has_sched)
 				ret = blk_mq_do_dispatch_sched(hctx);
 			else
 				ret = blk_mq_do_dispatch_ctx(hctx);
 		}
-	} else if (has_sched_dispatch) {
+	} else if (has_sched) {
 		ret = blk_mq_do_dispatch_sched(hctx);
 	} else if (hctx->dispatch_busy) {
 		/* dequeue request one by one from sw queue if queue is busy */
@@ -463,7 +462,7 @@ void blk_mq_sched_insert_request(struct request *rq, bool at_head,
 		goto run;
 	}
 
-	if (e && e->type->ops.insert_requests) {
+	if (e) {
 		LIST_HEAD(list);
 
 		list_add(&rq->queuelist, &list);
@@ -494,9 +493,9 @@ void blk_mq_sched_insert_requests(struct blk_mq_hw_ctx *hctx,
 	percpu_ref_get(&q->q_usage_counter);
 
 	e = hctx->queue->elevator;
-	if (e && e->type->ops.insert_requests)
+	if (e) {
 		e->type->ops.insert_requests(hctx, list, false);
-	else {
+	} else {
 		/*
 		 * try to issue requests directly if the hw queue isn't
 		 * busy in case of 'none' scheduler, and this way may save
diff --git a/block/elevator.c b/block/elevator.c
index 06e203426410..85d0d4adbb64 100644
--- a/block/elevator.c
+++ b/block/elevator.c
@@ -522,6 +522,10 @@ void elv_unregister_queue(struct request_queue *q)
 
 int elv_register(struct elevator_type *e)
 {
+	/* insert_requests and dispatch_request are mandatory */
+	if (WARN_ON_ONCE(!e->ops.insert_requests || !e->ops.dispatch_request))
+		return -EINVAL;
+
 	/* create icq_cache if requested */
 	if (e->icq_size) {
 		if (WARN_ON(e->icq_size < sizeof(struct io_cq)) ||
-- 
2.31.1


^ permalink raw reply related	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2021-06-18 16:28 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-06-18  1:59 [PATCH] block: Remove unnecessary elevator operation checks Damien Le Moal
2021-06-18  8:52 ` Johannes Thumshirn
2021-06-18 14:52 ` Jens Axboe
2021-06-18 16:28 ` Bart Van Assche

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).