[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <1663731123-81536-1-git-send-email-liusong@linux.alibaba.com>
Date: Wed, 21 Sep 2022 11:32:03 +0800
From: Liu Song <liusong@...ux.alibaba.com>
To: axboe@...nel.dk
Cc: hch@....de, kbusch@...nel.org, linux-kernel@...r.kernel.org,
linux-nvme@...ts.infradead.org, liusong@...ux.alibaba.com,
sagi@...mberg.me
Subject: [RFC PATCH] blk-mq: hctx has only one ctx mapping is no need to redirect the completion
From: Liu Song <liusong@...ux.alibaba.com>
High-performance NVMe devices usually support a large hw queue, which
ensures a 1:1 mapping of hctx and ctx. In this case there will be no
remote request, so we don't need to care about it.
Signed-off-by: Liu Song <liusong@...ux.alibaba.com>
---
block/blk-mq.c | 8 +++++---
1 file changed, 5 insertions(+), 3 deletions(-)
diff --git a/block/blk-mq.c b/block/blk-mq.c
index c11949d..9626ea1 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1093,10 +1093,12 @@ bool blk_mq_complete_request_remote(struct request *rq)
WRITE_ONCE(rq->state, MQ_RQ_COMPLETE);
/*
- * For a polled request, always complete locally, it's pointless
- * to redirect the completion.
+ * For request which hctx has only one ctx mapping,
+ * or a polled request, always complete locally,
+ * it's pointless to redirect the completion.
*/
- if (rq->cmd_flags & REQ_POLLED)
+ if (rq->mq_hctx->nr_ctx == 1 ||
+ rq->cmd_flags & REQ_POLLED)
return false;
if (blk_mq_complete_need_ipi(rq)) {
--
1.8.3.1
Powered by blists - more mailing lists