diff --git a/block/blk-mq.c b/block/blk-mq.c
index 96debbe63ad205793bbf78bbe7129cc65e9b25a0..2b0f0a9f40d9d28d1b1ef7ee7865549068026538 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1521,17 +1521,62 @@ bool blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx, bool async)
 	return false;
 }
 EXPORT_SYMBOL(blk_mq_run_hw_queue);
+/*
+ * Is the request queue handled by an IO scheduler that does not respect
+ * hardware queues when dispatching?
+ */
+static bool blk_mq_has_sqsched(struct request_queue *q)
+{
+	struct elevator_queue *e = q->elevator;
 
-void blk_mq_run_hw_queues(struct request_queue *q, bool async)
+	if (e && e->type->ops.mq.dispatch_request &&
+	    !(e->type->elevator_features & ELEVATOR_F_MQ_AWARE))
+		return true;
+	return false;
+}
+
+/*
+ * Return prefered queue to dispatch from (if any) for non-mq aware IO
+ * scheduler.
+ */
+static struct blk_mq_hw_ctx *blk_mq_get_sq_hctx(struct request_queue *q)
 {
 	struct blk_mq_hw_ctx *hctx;
+
+	/*
+	 * If the IO scheduler does not respect hardware queues when
+	 * dispatching, we just don't bother with multiple HW queues and
+	 * dispatch from hctx for the current CPU since running multiple queues
+	 * just causes lock contention inside the scheduler and pointless cache
+	 * bouncing.
+	 */
+	hctx = blk_mq_map_queue(q, raw_smp_processor_id());
+	if (!blk_mq_hctx_stopped(hctx))
+		return hctx;
+	return NULL;
+}
+
+
+void blk_mq_run_hw_queues(struct request_queue *q, bool async)
+{
+	struct blk_mq_hw_ctx *hctx, *sq_hctx;
 	int i;
 
+	sq_hctx = NULL;
+	if (blk_mq_has_sqsched(q))
+		sq_hctx = blk_mq_get_sq_hctx(q);
 	queue_for_each_hw_ctx(q, hctx, i) {
 		if (blk_mq_hctx_stopped(hctx))
 			continue;
 
-		blk_mq_run_hw_queue(hctx, async);
+		/*
+		 * Dispatch from this hctx either if there's no hctx preferred
+		 * by IO scheduler or if it has requests that bypass the
+		 * scheduler.
+		 */
+		if (!sq_hctx || sq_hctx == hctx ||
+		    !list_empty_careful(&hctx->dispatch))
+			blk_mq_run_hw_queue(hctx, async);
 	}
 }
 EXPORT_SYMBOL(blk_mq_run_hw_queues);
@@ -1543,9 +1588,12 @@ EXPORT_SYMBOL(blk_mq_run_hw_queues);
  */
 void blk_mq_delay_run_hw_queues(struct request_queue *q, unsigned long msecs)
 {
-	struct blk_mq_hw_ctx *hctx;
+	struct blk_mq_hw_ctx *hctx, *sq_hctx;
 	int i;
 
+	sq_hctx = NULL;
+	if (blk_mq_has_sqsched(q))
+		sq_hctx = blk_mq_get_sq_hctx(q);
 	queue_for_each_hw_ctx(q, hctx, i) {
 		if (blk_mq_hctx_stopped(hctx))
 			continue;
@@ -1559,7 +1607,14 @@ void blk_mq_delay_run_hw_queues(struct request_queue *q, unsigned long msecs)
 		if (delayed_work_pending(&hctx->run_work))
 			continue;
 
-		blk_mq_delay_run_hw_queue(hctx, msecs);
+		/*
+		 * Dispatch from this hctx either if there's no hctx preferred
+		 * by IO scheduler or if it has requests that bypass the
+		 * scheduler.
+		 */
+		if (!sq_hctx || sq_hctx == hctx ||
+		    !list_empty_careful(&hctx->dispatch))
+			blk_mq_delay_run_hw_queue(hctx, msecs);
 	}
 }
 EXPORT_SYMBOL(blk_mq_delay_run_hw_queues);
diff --git a/block/kyber-iosched.c b/block/kyber-iosched.c
index d8f3cb1bffa618f54a63e5b66b741b6577ab597d..bc0376c338ffba965b48607321fc5fa61e5bafc3 100644
--- a/block/kyber-iosched.c
+++ b/block/kyber-iosched.c
@@ -972,6 +972,7 @@ static struct elevator_type kyber_sched = {
 #endif
 	.elevator_attrs = kyber_sched_attrs,
 	.elevator_name = "kyber",
+	.elevator_features = ELEVATOR_F_MQ_AWARE,
 	.elevator_owner = THIS_MODULE,
 };
 
diff --git a/include/linux/elevator.h b/include/linux/elevator.h
index 663ce1780c5d4a5bbeaf002926c9f02365333bcd..70c3c1fd5cf08bbaf96fd720457e7259fa754223 100644
--- a/include/linux/elevator.h
+++ b/include/linux/elevator.h
@@ -155,6 +155,7 @@ struct elevator_type
 	size_t icq_align;	/* ditto */
 	struct elv_fs_entry *elevator_attrs;
 	char elevator_name[ELV_NAME_MAX];
+	unsigned int elevator_features;
 	const char *elevator_alias;
 	struct module *elevator_owner;
 	bool uses_mq;
@@ -276,6 +277,9 @@ enum {
 #define rq_entry_fifo(ptr)	list_entry((ptr), struct request, queuelist)
 #define rq_fifo_clear(rq)	list_del_init(&(rq)->queuelist)
 
+/* Supports scheduling on multiple hardware queues */
+#define ELEVATOR_F_MQ_AWARE		(1U << 1)
+
 #else /* CONFIG_BLOCK */
 
 static inline void load_default_elevator_module(void) { }