blk-mq: plug request for shared sbitmap

In case of shared sbitmap, request won't be held in plug list any more
sine commit 32bc15a ("blk-mq: Facilitate a shared sbitmap per
tagset"), this way makes request merge from flush plug list & batching
submission not possible, so cause performance regression.

Yanhui reports performance regression when running sequential IO
test(libaio, 16 jobs, 8 depth for each job) in VM, and the VM disk
is emulated with image stored on xfs/megaraid_sas.

Fix the issue by recovering original behavior to allow to hold request
in plug list.

Cc: Yanhui Ma <[email protected]>
Cc: John Garry <[email protected]>
Cc: Bart Van Assche <[email protected]>
Cc: [email protected]
Fixes: 32bc15a ("blk-mq: Facilitate a shared sbitmap per tagset")
Signed-off-by: Ming Lei <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jens Axboe <[email protected]>

Author: Ming Lei

block/blk-mq.c

@@ -2232,8 +2232,9 @@ blk_qc_t blk_mq_submit_bio(struct bio *bio)
 		/* Bypass scheduler for flush requests */
 		blk_insert_flush(rq);
 		blk_mq_run_hw_queue(data.hctx, true);
-	} else if (plug && (q->nr_hw_queues == 1 || q->mq_ops->commit_rqs ||
-				!blk_queue_nonrot(q))) {
+	} else if (plug && (q->nr_hw_queues == 1 ||
+		   blk_mq_is_sbitmap_shared(rq->mq_hctx->flags) ||
+		   q->mq_ops->commit_rqs || !blk_queue_nonrot(q))) {
 		/*
 		 * Use plugging if we have a ->commit_rqs() hook as well, as
 		 * we know the driver uses bd->last in a smart fashion.