Merge branch 'for-linus' into for-4.12/block

We've added a considerable amount of fixes for stalls and issues
with the blk-mq scheduling in the 4.11 series since forking
off the for-4.12/block branch. We need to do improvements on
top of that for 4.12, so pull in the previous fixes to make
our lives easier going forward.

Signed-off-by: Jens Axboe <axboe@fb.com>

6 files changed, 262 insertions, 146 deletions

diff --git a/block/blk-mq-sched.c b/block/blk-mq-sched.c
index 09af8ff18719a..c974a1bbf4cba 100644
--- a/block/blk-mq-sched.c
+++ b/block/blk-mq-sched.c
@@ -171,7 +171,8 @@ void blk_mq_sched_put_request(struct request *rq)
 
 void blk_mq_sched_dispatch_requests(struct blk_mq_hw_ctx *hctx)
 {
-	struct elevator_queue *e = hctx->queue->elevator;
+	struct request_queue *q = hctx->queue;
+	struct elevator_queue *e = q->elevator;
 	const bool has_sched_dispatch = e && e->type->ops.mq.dispatch_request;
 	bool did_work = false;
 	LIST_HEAD(rq_list);
@@ -203,10 +204,10 @@ void blk_mq_sched_dispatch_requests(struct blk_mq_hw_ctx *hctx)
 	 */
 	if (!list_empty(&rq_list)) {
 		blk_mq_sched_mark_restart_hctx(hctx);
-		did_work = blk_mq_dispatch_rq_list(hctx, &rq_list);
+		did_work = blk_mq_dispatch_rq_list(q, &rq_list);
 	} else if (!has_sched_dispatch) {
 		blk_mq_flush_busy_ctxs(hctx, &rq_list);
-		blk_mq_dispatch_rq_list(hctx, &rq_list);
+		blk_mq_dispatch_rq_list(q, &rq_list);
 	}
 
 	/*
@@ -222,7 +223,7 @@ void blk_mq_sched_dispatch_requests(struct blk_mq_hw_ctx *hctx)
 			if (!rq)
 				break;
 			list_add(&rq->queuelist, &rq_list);
-		} while (blk_mq_dispatch_rq_list(hctx, &rq_list));
+		} while (blk_mq_dispatch_rq_list(q, &rq_list));
 	}
 }
 
@@ -317,25 +318,68 @@ static bool blk_mq_sched_bypass_insert(struct blk_mq_hw_ctx *hctx,
 	return true;
 }
 
-static void blk_mq_sched_restart_hctx(struct blk_mq_hw_ctx *hctx)
+static bool blk_mq_sched_restart_hctx(struct blk_mq_hw_ctx *hctx)
 {
 	if (test_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state)) {
 		clear_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state);
-		if (blk_mq_hctx_has_pending(hctx))
+		if (blk_mq_hctx_has_pending(hctx)) {
 			blk_mq_run_hw_queue(hctx, true);
+			return true;
+		}
 	}
+	return false;
 }
 
-void blk_mq_sched_restart_queues(struct blk_mq_hw_ctx *hctx)
-{
-	struct request_queue *q = hctx->queue;
-	unsigned int i;
+/**
+ * list_for_each_entry_rcu_rr - iterate in a round-robin fashion over rcu list
+ * @pos:    loop cursor.
+ * @skip:   the list element that will not be examined. Iteration starts at
+ *          @skip->next.
+ * @head:   head of the list to examine. This list must have at least one
+ *          element, namely @skip.
+ * @member: name of the list_head structure within typeof(*pos).
+ */
+#define list_for_each_entry_rcu_rr(pos, skip, head, member)		\
+	for ((pos) = (skip);						\
+	     (pos = (pos)->member.next != (head) ? list_entry_rcu(	\
+			(pos)->member.next, typeof(*pos), member) :	\
+	      list_entry_rcu((pos)->member.next->next, typeof(*pos), member)), \
+	     (pos) != (skip); )
 
-	if (test_bit(QUEUE_FLAG_RESTART, &q->queue_flags)) {
-		if (test_and_clear_bit(QUEUE_FLAG_RESTART, &q->queue_flags)) {
-			queue_for_each_hw_ctx(q, hctx, i)
-				blk_mq_sched_restart_hctx(hctx);
+/*
+ * Called after a driver tag has been freed to check whether a hctx needs to
+ * be restarted. Restarts @hctx if its tag set is not shared. Restarts hardware
+ * queues in a round-robin fashion if the tag set of @hctx is shared with other
+ * hardware queues.
+ */
+void blk_mq_sched_restart(struct blk_mq_hw_ctx *const hctx)
+{
+	struct blk_mq_tags *const tags = hctx->tags;
+	struct blk_mq_tag_set *const set = hctx->queue->tag_set;
+	struct request_queue *const queue = hctx->queue, *q;
+	struct blk_mq_hw_ctx *hctx2;
+	unsigned int i, j;
+
+	if (set->flags & BLK_MQ_F_TAG_SHARED) {
+		rcu_read_lock();
+		list_for_each_entry_rcu_rr(q, queue, &set->tag_list,
+					   tag_set_list) {
+			queue_for_each_hw_ctx(q, hctx2, i)
+				if (hctx2->tags == tags &&
+				    blk_mq_sched_restart_hctx(hctx2))
+					goto done;
+		}
+		j = hctx->queue_num + 1;
+		for (i = 0; i < queue->nr_hw_queues; i++, j++) {
+			if (j == queue->nr_hw_queues)
+				j = 0;
+			hctx2 = queue->queue_hw_ctx[j];
+			if (hctx2->tags == tags &&
+			    blk_mq_sched_restart_hctx(hctx2))
+				break;
 		}
+done:
+		rcu_read_unlock();
 	} else {
 		blk_mq_sched_restart_hctx(hctx);
 	}
@@ -431,11 +475,67 @@ static void blk_mq_sched_free_tags(struct blk_mq_tag_set *set,
 	}
 }
 
-int blk_mq_sched_setup(struct request_queue *q)
+static int blk_mq_sched_alloc_tags(struct request_queue *q,
+				   struct blk_mq_hw_ctx *hctx,
+				   unsigned int hctx_idx)
+{
+	struct blk_mq_tag_set *set = q->tag_set;
+	int ret;
+
+	hctx->sched_tags = blk_mq_alloc_rq_map(set, hctx_idx, q->nr_requests,
+					       set->reserved_tags);
+	if (!hctx->sched_tags)
+		return -ENOMEM;
+
+	ret = blk_mq_alloc_rqs(set, hctx->sched_tags, hctx_idx, q->nr_requests);
+	if (ret)
+		blk_mq_sched_free_tags(set, hctx, hctx_idx);
+
+	return ret;
+}
+
+static void blk_mq_sched_tags_teardown(struct request_queue *q)
 {
 	struct blk_mq_tag_set *set = q->tag_set;
 	struct blk_mq_hw_ctx *hctx;
-	int ret, i;
+	int i;
+
+	queue_for_each_hw_ctx(q, hctx, i)
+		blk_mq_sched_free_tags(set, hctx, i);
+}
+
+int blk_mq_sched_init_hctx(struct request_queue *q, struct blk_mq_hw_ctx *hctx,
+			   unsigned int hctx_idx)
+{
+	struct elevator_queue *e = q->elevator;
+
+	if (!e)
+		return 0;
+
+	return blk_mq_sched_alloc_tags(q, hctx, hctx_idx);
+}
+
+void blk_mq_sched_exit_hctx(struct request_queue *q, struct blk_mq_hw_ctx *hctx,
+			    unsigned int hctx_idx)
+{
+	struct elevator_queue *e = q->elevator;
+
+	if (!e)
+		return;
+
+	blk_mq_sched_free_tags(q->tag_set, hctx, hctx_idx);
+}
+
+int blk_mq_init_sched(struct request_queue *q, struct elevator_type *e)
+{
+	struct blk_mq_hw_ctx *hctx;
+	unsigned int i;
+	int ret;
+
+	if (!e) {
+		q->elevator = NULL;
+		return 0;
+	}
 
 	/*
 	 * Default to 256, since we don't split into sync/async like the
@@ -443,49 +543,30 @@ int blk_mq_sched_setup(struct request_queue *q)
 	 */
 	q->nr_requests = 2 * BLKDEV_MAX_RQ;
 
-	/*
-	 * We're switching to using an IO scheduler, so setup the hctx
-	 * scheduler tags and switch the request map from the regular
-	 * tags to scheduler tags. First allocate what we need, so we
-	 * can safely fail and fallback, if needed.
-	 */
-	ret = 0;
 	queue_for_each_hw_ctx(q, hctx, i) {
-		hctx->sched_tags = blk_mq_alloc_rq_map(set, i,
-				q->nr_requests, set->reserved_tags);
-		if (!hctx->sched_tags) {
-			ret = -ENOMEM;
-			break;
-		}
-		ret = blk_mq_alloc_rqs(set, hctx->sched_tags, i, q->nr_requests);
+		ret = blk_mq_sched_alloc_tags(q, hctx, i);
 		if (ret)
-			break;
+			goto err;
 	}
 
-	/*
-	 * If we failed, free what we did allocate
-	 */
-	if (ret) {
-		queue_for_each_hw_ctx(q, hctx, i) {
-			if (!hctx->sched_tags)
-				continue;
-			blk_mq_sched_free_tags(set, hctx, i);
-		}
-
-		return ret;
-	}
+	ret = e->ops.mq.init_sched(q, e);
+	if (ret)
+		goto err;
 
 	return 0;
+
+err:
+	blk_mq_sched_tags_teardown(q);
+	q->elevator = NULL;
+	return ret;
 }
 
-void blk_mq_sched_teardown(struct request_queue *q)
+void blk_mq_exit_sched(struct request_queue *q, struct elevator_queue *e)
 {
-	struct blk_mq_tag_set *set = q->tag_set;
-	struct blk_mq_hw_ctx *hctx;
-	int i;
-
-	queue_for_each_hw_ctx(q, hctx, i)
-		blk_mq_sched_free_tags(set, hctx, i);
+	if (e->type->ops.mq.exit_sched)
+		e->type->ops.mq.exit_sched(e);
+	blk_mq_sched_tags_teardown(q);
+	q->elevator = NULL;
 }
 
 int blk_mq_sched_init(struct request_queue *q)
diff --git a/block/blk-mq-sched.h b/block/blk-mq-sched.h
index a75b16b123f7a..3a9e6e40558b5 100644
--- a/block/blk-mq-sched.h
+++ b/block/blk-mq-sched.h
@@ -19,7 +19,7 @@ bool blk_mq_sched_try_merge(struct request_queue *q, struct bio *bio,
 				struct request **merged_request);
 bool __blk_mq_sched_bio_merge(struct request_queue *q, struct bio *bio);
 bool blk_mq_sched_try_insert_merge(struct request_queue *q, struct request *rq);
-void blk_mq_sched_restart_queues(struct blk_mq_hw_ctx *hctx);
+void blk_mq_sched_restart(struct blk_mq_hw_ctx *hctx);
 
 void blk_mq_sched_insert_request(struct request *rq, bool at_head,
 				 bool run_queue, bool async, bool can_block);
@@ -32,8 +32,13 @@ void blk_mq_sched_move_to_dispatch(struct blk_mq_hw_ctx *hctx,
 			struct list_head *rq_list,
 			struct request *(*get_rq)(struct blk_mq_hw_ctx *));
 
-int blk_mq_sched_setup(struct request_queue *q);
-void blk_mq_sched_teardown(struct request_queue *q);
+int blk_mq_init_sched(struct request_queue *q, struct elevator_type *e);
+void blk_mq_exit_sched(struct request_queue *q, struct elevator_queue *e);
+
+int blk_mq_sched_init_hctx(struct request_queue *q, struct blk_mq_hw_ctx *hctx,
+			   unsigned int hctx_idx);
+void blk_mq_sched_exit_hctx(struct request_queue *q, struct blk_mq_hw_ctx *hctx,
+			    unsigned int hctx_idx);
 
 int blk_mq_sched_init(struct request_queue *q);
 
@@ -131,20 +136,6 @@ static inline void blk_mq_sched_mark_restart_hctx(struct blk_mq_hw_ctx *hctx)
 		set_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state);
 }
 
-/*
- * Mark a hardware queue and the request queue it belongs to as needing a
- * restart.
- */
-static inline void blk_mq_sched_mark_restart_queue(struct blk_mq_hw_ctx *hctx)
-{
-	struct request_queue *q = hctx->queue;
-
-	if (!test_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state))
-		set_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state);
-	if (!test_bit(QUEUE_FLAG_RESTART, &q->queue_flags))
-		set_bit(QUEUE_FLAG_RESTART, &q->queue_flags);
-}
-
 static inline bool blk_mq_sched_needs_restart(struct blk_mq_hw_ctx *hctx)
 {
 	return test_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state);
diff --git a/block/blk-mq.c b/block/blk-mq.c
index f7cd3208bcdf7..724bcec0ca4f8 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -351,7 +351,7 @@ void __blk_mq_finish_request(struct blk_mq_hw_ctx *hctx, struct blk_mq_ctx *ctx,
 		blk_mq_put_tag(hctx, hctx->tags, ctx, rq->tag);
 	if (sched_tag != -1)
 		blk_mq_sched_completed_request(hctx, rq);
-	blk_mq_sched_restart_queues(hctx);
+	blk_mq_sched_restart(hctx);
 	blk_queue_exit(q);
 }
 
@@ -863,12 +863,8 @@ bool blk_mq_get_driver_tag(struct request *rq, struct blk_mq_hw_ctx **hctx,
 		.flags = wait ? 0 : BLK_MQ_REQ_NOWAIT,
 	};
 
-	if (rq->tag != -1) {
-done:
-		if (hctx)
-			*hctx = data.hctx;
-		return true;
-	}
+	if (rq->tag != -1)
+		goto done;
 
 	if (blk_mq_tag_is_reserved(data.hctx->sched_tags, rq->internal_tag))
 		data.flags |= BLK_MQ_REQ_RESERVED;
@@ -880,10 +876,12 @@ done:
 			atomic_inc(&data.hctx->nr_active);
 		}
 		data.hctx->tags->rqs[rq->tag] = rq;
-		goto done;
 	}
 
-	return false;
+done:
+	if (hctx)
+		*hctx = data.hctx;
+	return rq->tag != -1;
 }
 
 static void __blk_mq_put_driver_tag(struct blk_mq_hw_ctx *hctx,
@@ -980,17 +978,20 @@ static bool blk_mq_dispatch_wait_add(struct blk_mq_hw_ctx *hctx)
 	return true;
 }
 
-bool blk_mq_dispatch_rq_list(struct blk_mq_hw_ctx *hctx, struct list_head *list)
+bool blk_mq_dispatch_rq_list(struct request_queue *q, struct list_head *list)
 {
-	struct request_queue *q = hctx->queue;
+	struct blk_mq_hw_ctx *hctx;
 	struct request *rq;
 	int errors, queued, ret = BLK_MQ_RQ_QUEUE_OK;
 
+	if (list_empty(list))
+		return false;
+
 	/*
 	 * Now process all the entries, sending them to the driver.
 	 */
 	errors = queued = 0;
-	while (!list_empty(list)) {
+	do {
 		struct blk_mq_queue_data bd;
 
 		rq = list_first_entry(list, struct request, queuelist);
@@ -1053,7 +1054,7 @@ bool blk_mq_dispatch_rq_list(struct blk_mq_hw_ctx *hctx, struct list_head *list)
 
 		if (ret == BLK_MQ_RQ_QUEUE_BUSY)
 			break;
-	}
+	} while (!list_empty(list));
 
 	hctx->dispatched[queued_to_index(queued)]++;
 
@@ -1138,7 +1139,8 @@ static int blk_mq_hctx_next_cpu(struct blk_mq_hw_ctx *hctx)
 	return hctx->next_cpu;
 }
 
-void blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx, bool async)
+static void __blk_mq_delay_run_hw_queue(struct blk_mq_hw_ctx *hctx, bool async,
+					unsigned long msecs)
 {
 	if (unlikely(blk_mq_hctx_stopped(hctx) ||
 		     !blk_mq_hw_queue_mapped(hctx)))
@@ -1155,7 +1157,24 @@ void blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx, bool async)
 		put_cpu();
 	}
 
-	kblockd_schedule_work_on(blk_mq_hctx_next_cpu(hctx), &hctx->run_work);
+	if (msecs == 0)
+		kblockd_schedule_work_on(blk_mq_hctx_next_cpu(hctx),
+					 &hctx->run_work);
+	else
+		kblockd_schedule_delayed_work_on(blk_mq_hctx_next_cpu(hctx),
+						 &hctx->delayed_run_work,
+						 msecs_to_jiffies(msecs));
+}
+
+void blk_mq_delay_run_hw_queue(struct blk_mq_hw_ctx *hctx, unsigned long msecs)
+{
+	__blk_mq_delay_run_hw_queue(hctx, true, msecs);
+}
+EXPORT_SYMBOL(blk_mq_delay_run_hw_queue);
+
+void blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx, bool async)
+{
+	__blk_mq_delay_run_hw_queue(hctx, async, 0);
 }
 
 void blk_mq_run_hw_queues(struct request_queue *q, bool async)
@@ -1258,6 +1277,15 @@ static void blk_mq_run_work_fn(struct work_struct *work)
 	__blk_mq_run_hw_queue(hctx);
 }
 
+static void blk_mq_delayed_run_work_fn(struct work_struct *work)
+{
+	struct blk_mq_hw_ctx *hctx;
+
+	hctx = container_of(work, struct blk_mq_hw_ctx, delayed_run_work.work);
+
+	__blk_mq_run_hw_queue(hctx);
+}
+
 static void blk_mq_delay_work_fn(struct work_struct *work)
 {
 	struct blk_mq_hw_ctx *hctx;
@@ -1822,6 +1850,8 @@ static void blk_mq_exit_hctx(struct request_queue *q,
 				       hctx->fq->flush_rq, hctx_idx,
 				       flush_start_tag + hctx_idx);
 
+	blk_mq_sched_exit_hctx(q, hctx, hctx_idx);
+
 	if (set->ops->exit_hctx)
 		set->ops->exit_hctx(hctx, hctx_idx);
 
@@ -1858,6 +1888,7 @@ static int blk_mq_init_hctx(struct request_queue *q,
 		node = hctx->numa_node = set->numa_node;
 
 	INIT_WORK(&hctx->run_work, blk_mq_run_work_fn);
+	INIT_DELAYED_WORK(&hctx->delayed_run_work, blk_mq_delayed_run_work_fn);
 	INIT_DELAYED_WORK(&hctx->delay_work, blk_mq_delay_work_fn);
 	spin_lock_init(&hctx->lock);
 	INIT_LIST_HEAD(&hctx->dispatch);
@@ -1888,9 +1919,12 @@ static int blk_mq_init_hctx(struct request_queue *q,
 	    set->ops->init_hctx(hctx, set->driver_data, hctx_idx))
 		goto free_bitmap;
 
+	if (blk_mq_sched_init_hctx(q, hctx, hctx_idx))
+		goto exit_hctx;
+
 	hctx->fq = blk_alloc_flush_queue(q, hctx->numa_node, set->cmd_size);
 	if (!hctx->fq)
-		goto exit_hctx;
+		goto sched_exit_hctx;
 
 	if (set->ops->init_request &&
 	    set->ops->init_request(set->driver_data,
@@ -1905,6 +1939,8 @@ static int blk_mq_init_hctx(struct request_queue *q,
 
  free_fq:
 	kfree(hctx->fq);
+ sched_exit_hctx:
+	blk_mq_sched_exit_hctx(q, hctx, hctx_idx);
  exit_hctx:
 	if (set->ops->exit_hctx)
 		set->ops->exit_hctx(hctx, hctx_idx);
@@ -2129,8 +2165,6 @@ void blk_mq_release(struct request_queue *q)
 	struct blk_mq_hw_ctx *hctx;
 	unsigned int i;
 
-	blk_mq_sched_teardown(q);
-
 	/* hctx kobj stays in hctx */
 	queue_for_each_hw_ctx(q, hctx, i) {
 		if (!hctx)
@@ -2461,6 +2495,14 @@ static int blk_mq_alloc_rq_maps(struct blk_mq_tag_set *set)
 	return 0;
 }
 
+static int blk_mq_update_queue_map(struct blk_mq_tag_set *set)
+{
+	if (set->ops->map_queues)
+		return set->ops->map_queues(set);
+	else
+		return blk_mq_map_queues(set);
+}
+
 /*
  * Alloc a tag set to be associated with one or more request queues.
  * May fail with EINVAL for various error conditions. May adjust the
@@ -2515,10 +2557,7 @@ int blk_mq_alloc_tag_set(struct blk_mq_tag_set *set)
 	if (!set->mq_map)
 		goto out_free_tags;
 
-	if (set->ops->map_queues)
-		ret = set->ops->map_queues(set);
-	else
-		ret = blk_mq_map_queues(set);
+	ret = blk_mq_update_queue_map(set);
 	if (ret)
 		goto out_free_mq_map;
 
@@ -2610,6 +2649,7 @@ void blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, int nr_hw_queues)
 		blk_mq_freeze_queue(q);
 
 	set->nr_hw_queues = nr_hw_queues;
+	blk_mq_update_queue_map(set);
 	list_for_each_entry(q, &set->tag_list, tag_set_list) {
 		blk_mq_realloc_hw_ctxs(set, q);
 		blk_mq_queue_reinit(q, cpu_online_mask);
diff --git a/block/blk-mq.h b/block/blk-mq.h
index 8d49c06fc5203..7e6f2e467696e 100644
--- a/block/blk-mq.h
+++ b/block/blk-mq.h
@@ -30,7 +30,7 @@ void blk_mq_freeze_queue(struct request_queue *q);
 void blk_mq_free_queue(struct request_queue *q);
 int blk_mq_update_nr_requests(struct request_queue *q, unsigned int nr);
 void blk_mq_wake_waiters(struct request_queue *q);
-bool blk_mq_dispatch_rq_list(struct blk_mq_hw_ctx *, struct list_head *);
+bool blk_mq_dispatch_rq_list(struct request_queue *, struct list_head *);
 void blk_mq_flush_busy_ctxs(struct blk_mq_hw_ctx *hctx, struct list_head *list);
 bool blk_mq_hctx_has_pending(struct blk_mq_hw_ctx *hctx);
 bool blk_mq_get_driver_tag(struct request *rq, struct blk_mq_hw_ctx **hctx,
diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index 45854266e398b..c47db43a40cc9 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -803,7 +803,7 @@ static void blk_release_queue(struct kobject *kobj)
 
 	if (q->elevator) {
 		ioc_clear_queue(q);
-		elevator_exit(q->elevator);
+		elevator_exit(q, q->elevator);
 	}
 
 	blk_free_queue_stats(q->stats);
diff --git a/block/elevator.c b/block/elevator.c
index 01139f549b5be..dbeecf7be719e 100644
--- a/block/elevator.c
+++ b/block/elevator.c
@@ -242,26 +242,21 @@ int elevator_init(struct request_queue *q, char *name)
 		}
 	}
 
-	if (e->uses_mq) {
-		err = blk_mq_sched_setup(q);
-		if (!err)
-			err = e->ops.mq.init_sched(q, e);
-	} else
+	if (e->uses_mq)
+		err = blk_mq_init_sched(q, e);
+	else
 		err = e->ops.sq.elevator_init_fn(q, e);
-	if (err) {
-		if (e->uses_mq)
-			blk_mq_sched_teardown(q);
+	if (err)
 		elevator_put(e);
-	}
 	return err;
 }
 EXPORT_SYMBOL(elevator_init);
 
-void elevator_exit(struct elevator_queue *e)
+void elevator_exit(struct request_queue *q, struct elevator_queue *e)
 {
 	mutex_lock(&e->sysfs_lock);
 	if (e->uses_mq && e->type->ops.mq.exit_sched)
-		e->type->ops.mq.exit_sched(e);
+		blk_mq_exit_sched(q, e);
 	else if (!e->uses_mq && e->type->ops.sq.elevator_exit_fn)
 		e->type->ops.sq.elevator_exit_fn(e);
 	mutex_unlock(&e->sysfs_lock);
@@ -946,6 +941,45 @@ void elv_unregister(struct elevator_type *e)
 }
 EXPORT_SYMBOL_GPL(elv_unregister);
 
+static int elevator_switch_mq(struct request_queue *q,
+			      struct elevator_type *new_e)
+{
+	int ret;
+
+	blk_mq_freeze_queue(q);
+	blk_mq_quiesce_queue(q);
+
+	if (q->elevator) {
+		if (q->elevator->registered)
+			elv_unregister_queue(q);
+		ioc_clear_queue(q);
+		elevator_exit(q, q->elevator);
+	}
+
+	ret = blk_mq_init_sched(q, new_e);
+	if (ret)
+		goto out;
+
+	if (new_e) {
+		ret = elv_register_queue(q);
+		if (ret) {
+			elevator_exit(q, q->elevator);
+			goto out;
+		}
+	}
+
+	if (new_e)
+		blk_add_trace_msg(q, "elv switch: %s", new_e->elevator_name);
+	else
+		blk_add_trace_msg(q, "elv switch: none");
+
+out:
+	blk_mq_unfreeze_queue(q);
+	blk_mq_start_stopped_hw_queues(q, true);
+	return ret;
+
+}
+
 /*
  * switch to new_e io scheduler. be careful not to introduce deadlocks -
  * we don't free the old io scheduler, before we have allocated what we
@@ -958,10 +992,8 @@ static int elevator_switch(struct request_queue *q, struct elevator_type *new_e)
 	bool old_registered = false;
 	int err;
 
-	if (q->mq_ops) {
-		blk_mq_freeze_queue(q);
-		blk_mq_quiesce_queue(q);
-	}
+	if (q->mq_ops)
+		return elevator_switch_mq(q, new_e);
 
 	/*
 	 * Turn on BYPASS and drain all requests w/ elevator private data.
@@ -973,11 +1005,7 @@ static int elevator_switch(struct request_queue *q, struct elevator_type *new_e)
 	if (old) {
 		old_registered = old->registered;
 
-		if (old->uses_mq)
-			blk_mq_sched_teardown(q);
-
-		if (!q->mq_ops)
-			blk_queue_bypass_start(q);
+		blk_queue_bypass_start(q);
 
 		/* unregister and clear all auxiliary data of the old elevator */
 		if (old_registered)
@@ -987,56 +1015,32 @@ static int elevator_switch(struct request_queue *q, struct elevator_type *new_e)
 	}
 
 	/* allocate, init and register new elevator */
-	if (new_e) {
-		if (new_e->uses_mq) {
-			err = blk_mq_sched_setup(q);
-			if (!err)
-				err = new_e->ops.mq.init_sched(q, new_e);
-		} else
-			err = new_e->ops.sq.elevator_init_fn(q, new_e);
-		if (err)
-			goto fail_init;
+	err = new_e->ops.sq.elevator_init_fn(q, new_e);
+	if (err)
+		goto fail_init;
 
-		err = elv_register_queue(q);
-		if (err)
-			goto fail_register;
-	} else
-		q->elevator = NULL;
+	err = elv_register_queue(q);
+	if (err)
+		goto fail_register;
 
 	/* done, kill the old one and finish */
 	if (old) {
-		elevator_exit(old);
-		if (!q->mq_ops)
-			blk_queue_bypass_end(q);
+		elevator_exit(q, old);
+		blk_queue_bypass_end(q);
 	}
 
-	if (q->mq_ops) {
-		blk_mq_unfreeze_queue(q);
-		blk_mq_start_stopped_hw_queues(q, true);
-	}
-
-	if (new_e)
-		blk_add_trace_msg(q, "elv switch: %s", new_e->elevator_name);
-	else
-		blk_add_trace_msg(q, "elv switch: none");
+	blk_add_trace_msg(q, "elv switch: %s", new_e->elevator_name);
 
 	return 0;
 
 fail_register:
-	if (q->mq_ops)
-		blk_mq_sched_teardown(q);
-	elevator_exit(q->elevator);
+	elevator_exit(q, q->elevator);
 fail_init:
 	/* switch failed, restore and re-register old elevator */
 	if (old) {
 		q->elevator = old;
 		elv_register_queue(q);
-		if (!q->mq_ops)
-			blk_queue_bypass_end(q);
-	}
-	if (q->mq_ops) {
-		blk_mq_unfreeze_queue(q);
-		blk_mq_start_stopped_hw_queues(q, true);
+		blk_queue_bypass_end(q);
 	}
 
 	return err;