block: move queue enter logic into blk_mq_submit_bio()

Retain the old logic for the fops based submit, but for our internal
blk_mq_submit_bio(), move the queue entering logic into the core
function itself.

We need to be a bit careful if going into the scheduler, as a scheduler
or queue mappings can arbitrarily change before we have entered the queue.
Have the bio scheduler mapping do that separately, it's a very cheap
operation compared to actually doing merging locking and lookups.
Reviewed-by: Christoph Hellwig <hch@lst.de>
[axboe: update to check merge post submit_bio_checks() doing remap...]
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block/blk-core.c

@@ -744,7 +744,7 @@ static inline blk_status_t blk_check_zone_append(struct request_queue *q,
 	return BLK_STS_OK;
 }
 
-static noinline_for_stack bool submit_bio_checks(struct bio *bio)
+noinline_for_stack bool submit_bio_checks(struct bio *bio)
 {
 	struct block_device *bdev = bio->bi_bdev;
 	struct request_queue *q = bdev_get_queue(bdev);
@@ -862,24 +862,25 @@ static noinline_for_stack bool submit_bio_checks(struct bio *bio)
 	return false;
 }
 
-static void __submit_bio(struct bio *bio)
+static void __submit_bio_fops(struct gendisk *disk, struct bio *bio)
 {
-	struct gendisk *disk = bio->bi_bdev->bd_disk;
-
 	if (unlikely(bio_queue_enter(bio) != 0))
 		return;
-
-	if (!submit_bio_checks(bio) || !blk_crypto_bio_prep(&bio))
-		goto queue_exit;
-	if (!disk->fops->submit_bio) {
-		blk_mq_submit_bio(bio);
-		return;
-	}
+	if (submit_bio_checks(bio) && blk_crypto_bio_prep(&bio))
 		disk->fops->submit_bio(bio);
-queue_exit:
 	blk_queue_exit(disk->queue);
 }
 
+static void __submit_bio(struct bio *bio)
+{
+	struct gendisk *disk = bio->bi_bdev->bd_disk;
+
+	if (!disk->fops->submit_bio)
+		blk_mq_submit_bio(bio);
+	else
+		__submit_bio_fops(disk, bio);
+}
+
 /*
  * The loop in this function may be a bit non-obvious, and so deserves some
  * explanation:

block/blk-mq-sched.c

@@ -370,15 +370,20 @@ bool blk_mq_sched_bio_merge(struct request_queue *q, struct bio *bio,
 	bool ret = false;
 	enum hctx_type type;
 
-	if (e && e->type->ops.bio_merge)
-		return e->type->ops.bio_merge(q, bio, nr_segs);
+	if (bio_queue_enter(bio))
+		return false;
+
+	if (e && e->type->ops.bio_merge) {
+		ret = e->type->ops.bio_merge(q, bio, nr_segs);
+		goto out_put;
+	}
 
 	ctx = blk_mq_get_ctx(q);
 	hctx = blk_mq_map_queue(q, bio->bi_opf, ctx);
 	type = hctx->type;
 	if (!(hctx->flags & BLK_MQ_F_SHOULD_MERGE) ||
 	    list_empty_careful(&ctx->rq_lists[type]))
-		return false;
+		goto out_put;
 
 	/* default per sw-queue merge */
 	spin_lock(&ctx->lock);
@@ -391,6 +396,8 @@ bool blk_mq_sched_bio_merge(struct request_queue *q, struct bio *bio,
 		ret = true;
 
 	spin_unlock(&ctx->lock);
+out_put:
+	blk_queue_exit(q);
 	return ret;
 }
 

block/blk-mq.c

@@ -2478,9 +2478,23 @@ static inline unsigned short blk_plug_max_rq_count(struct blk_plug *plug)
 	return BLK_MAX_REQUEST_COUNT;
 }
 
+static bool blk_attempt_bio_merge(struct request_queue *q, struct bio *bio,
+				  unsigned int nr_segs, bool *same_queue_rq)
+{
+	if (!blk_queue_nomerges(q) && bio_mergeable(bio)) {
+		if (blk_attempt_plug_merge(q, bio, nr_segs, same_queue_rq))
+			return true;
+		if (blk_mq_sched_bio_merge(q, bio, nr_segs))
+			return true;
+	}
+	return false;
+}
+
 static struct request *blk_mq_get_new_requests(struct request_queue *q,
 					       struct blk_plug *plug,
-					       struct bio *bio)
+					       struct bio *bio,
+					       unsigned int nsegs,
+					       bool *same_queue_rq)
 {
 	struct blk_mq_alloc_data data = {
 		.q		= q,
@@ -2489,6 +2503,15 @@ static struct request *blk_mq_get_new_requests(struct request_queue *q,
 	};
 	struct request *rq;
 
+	if (unlikely(bio_queue_enter(bio)))
+		return NULL;
+	if (unlikely(!submit_bio_checks(bio)))
+		goto put_exit;
+	if (blk_attempt_bio_merge(q, bio, nsegs, same_queue_rq))
+		goto put_exit;
+
+	rq_qos_throttle(q, bio);
+
 	if (plug) {
 		data.nr_tags = plug->nr_ios;
 		plug->nr_ios = 1;
@@ -2502,25 +2525,34 @@ static struct request *blk_mq_get_new_requests(struct request_queue *q,
 	rq_qos_cleanup(q, bio);
 	if (bio->bi_opf & REQ_NOWAIT)
 		bio_wouldblock_error(bio);
+put_exit:
+	blk_queue_exit(q);
 	return NULL;
 }
 
 static inline struct request *blk_mq_get_request(struct request_queue *q,
 						 struct blk_plug *plug,
-						 struct bio *bio)
+						 struct bio *bio,
+						 unsigned int nsegs,
+						 bool *same_queue_rq)
 {
 	if (plug) {
 		struct request *rq;
 
 		rq = rq_list_peek(&plug->cached_rq);
 		if (rq) {
+			if (unlikely(!submit_bio_checks(bio)))
+				return NULL;
+			if (blk_attempt_bio_merge(q, bio, nsegs, same_queue_rq))
+				return NULL;
 			plug->cached_rq = rq_list_next(rq);
 			INIT_LIST_HEAD(&rq->queuelist);
+			rq_qos_throttle(q, bio);
 			return rq;
 		}
 	}
 
-	return blk_mq_get_new_requests(q, plug, bio);
+	return blk_mq_get_new_requests(q, plug, bio, nsegs, same_queue_rq);
 }
 
 /**
@@ -2546,26 +2578,20 @@ void blk_mq_submit_bio(struct bio *bio)
 	unsigned int nr_segs = 1;
 	blk_status_t ret;
 
+	if (unlikely(!blk_crypto_bio_prep(&bio)))
+		return;
+
 	blk_queue_bounce(q, &bio);
 	if (blk_may_split(q, bio))
 		__blk_queue_split(q, &bio, &nr_segs);
 
 	if (!bio_integrity_prep(bio))
-		goto queue_exit;
-
-	if (!blk_queue_nomerges(q) && bio_mergeable(bio)) {
-		if (blk_attempt_plug_merge(q, bio, nr_segs, &same_queue_rq))
-			goto queue_exit;
-		if (blk_mq_sched_bio_merge(q, bio, nr_segs))
-			goto queue_exit;
-	}
-
-	rq_qos_throttle(q, bio);
+		return;
 
 	plug = blk_mq_plug(q, bio);
-	rq = blk_mq_get_request(q, plug, bio);
+	rq = blk_mq_get_request(q, plug, bio, nr_segs, &same_queue_rq);
 	if (unlikely(!rq))
-		goto queue_exit;
+		return;
 
 	trace_block_getrq(bio);
 
@@ -2646,10 +2672,6 @@ void blk_mq_submit_bio(struct bio *bio)
 		/* Default case. */
 		blk_mq_sched_insert_request(rq, false, true, true);
 	}
-
-	return;
-queue_exit:
-	blk_queue_exit(q);
 }
 
 static size_t order_to_size(unsigned int order)

block/blk.h

@@ -56,6 +56,7 @@ void blk_freeze_queue(struct request_queue *q);
 void __blk_mq_unfreeze_queue(struct request_queue *q, bool force_atomic);
 void blk_queue_start_drain(struct request_queue *q);
 int __bio_queue_enter(struct request_queue *q, struct bio *bio);
+bool submit_bio_checks(struct bio *bio);
 
 static inline bool blk_try_enter_queue(struct request_queue *q, bool pm)
 {