Re: [PATCH 1/4] block: add mq_ops->queue_rqs hook

[Christoph Hellwig <hch@infradead.org>]

> @@ -2208,6 +2208,19 @@ static void blk_mq_plug_issue_direct(struct blk_plug *plug, bool from_schedule)
>  	int queued = 0;
>  	int errors = 0;
>  
> +	/*
> +	 * Peek first request and see if we have a ->queue_rqs() hook. If we
> +	 * do, we can dispatch the whole plug list in one go. We already know
> +	 * at this point that all requests belong to the same queue, caller
> +	 * must ensure that's the case.
> +	 */
> +	rq = rq_list_peek(&plug->mq_list);
> +	if (rq->q->mq_ops->queue_rqs) {
> +		rq->q->mq_ops->queue_rqs(&plug->mq_list);
> +		if (rq_list_empty(plug->mq_list))
> +			return;
> +	}

I'd move this straight into the caller to simplify the follow, something
like the patch below.  Also I wonder if the slow path might want to
be moved from blk_mq_flush_plug_list into a separate helper to keep
the fast path as straight as possible?

diff --git a/block/blk-mq.c b/block/blk-mq.c
index 3ab34c4f20daf..370f4b2ab4511 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -2255,6 +2255,14 @@ void blk_mq_flush_plug_list(struct blk_plug *plug, bool from_schedule)
 	plug->rq_count = 0;
 
 	if (!plug->multiple_queues && !plug->has_elevator && !from_schedule) {
+		struct request_queue *q = rq_list_peek(&plug->mq_list)->q;
+
+		if (q->mq_ops->queue_rqs) {
+			q->mq_ops->queue_rqs(&plug->mq_list);
+			if (rq_list_empty(plug->mq_list))
+				return;
+		}
+
 		blk_mq_plug_issue_direct(plug, false);
 		if (rq_list_empty(plug->mq_list))
 			return;