diff --git a/block/blk-mq-sched.c b/block/blk-mq-sched.c
index 623c258d0b34e086d4b5564617d849ef2b7fe6b8..56368f1934aaa1ee32c62ff9e36d7cbb27b0ed57 100644
--- a/block/blk-mq-sched.c
+++ b/block/blk-mq-sched.c
@@ -81,6 +81,8 @@ void blk_mq_sched_restart(struct blk_mq_hw_ctx *hctx)
 	blk_mq_run_hw_queue(hctx, true);
 }
 
+#define BLK_MQ_BUDGET_DELAY	3		/* ms units */
+
 /*
  * Only SCSI implements .get_budget and .put_budget, and SCSI restarts
  * its queue by itself in its completion handler, so we don't need to
@@ -105,6 +107,14 @@ static void blk_mq_do_dispatch_sched(struct blk_mq_hw_ctx *hctx)
 		rq = e->type->ops.mq.dispatch_request(hctx);
 		if (!rq) {
 			blk_mq_put_dispatch_budget(hctx);
+			/*
+			 * We're releasing without dispatching. Holding the
+			 * budget could have blocked any "hctx"s with the
+			 * same queue and if we didn't dispatch then there's
+			 * no guarantee anyone will kick the queue.  Kick it
+			 * ourselves.
+			 */
+			blk_mq_delay_run_hw_queues(q, BLK_MQ_BUDGET_DELAY);
 			break;
 		}
 
@@ -151,6 +161,14 @@ static void blk_mq_do_dispatch_ctx(struct blk_mq_hw_ctx *hctx)
 		rq = blk_mq_dequeue_from_ctx(hctx, ctx);
 		if (!rq) {
 			blk_mq_put_dispatch_budget(hctx);
+			/*
+			 * We're releasing without dispatching. Holding the
+			 * budget could have blocked any "hctx"s with the
+			 * same queue and if we didn't dispatch then there's
+			 * no guarantee anyone will kick the queue.  Kick it
+			 * ourselves.
+			 */
+			blk_mq_delay_run_hw_queues(q, BLK_MQ_BUDGET_DELAY);
 			break;
 		}