From: Jens Axboe <axboe@kernel.dk>
Date: Sun, 18 Nov 2018 23:15:35 +0000 (-0700)
Subject: blk-mq: never redirect polled IO completions
X-Git-Url: http://git.maquefel.me/?a=commitdiff_plain;h=4ab32bf3305eedb4d31f85cac68a67becab10494;p=linux.git

blk-mq: never redirect polled IO completions

It's pointless to do so, we are by definition on the CPU we want/need
to be, as that's the one waiting for a completion event.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
---

diff --git a/block/blk-mq.c b/block/blk-mq.c
index ba3c7b6476b77..37674c1766a75 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -585,7 +585,12 @@ static void __blk_mq_complete_request(struct request *rq)
 		return;
 	}
 
-	if (!test_bit(QUEUE_FLAG_SAME_COMP, &q->queue_flags)) {
+	/*
+	 * For a polled request, always complete locallly, it's pointless
+	 * to redirect the completion.
+	 */
+	if ((rq->cmd_flags & REQ_HIPRI) ||
+	    !test_bit(QUEUE_FLAG_SAME_COMP, &q->queue_flags)) {
 		q->mq_ops->complete(rq);
 		return;
 	}