blk-mq: unify hctx delayed_run_work and run_work

They serve the exact same purpose. Get rid of the non-delayed
work variant, and just run it without delay for the normal case.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <Bart.VanAssche@sandisk.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
diff --git a/block/blk-core.c b/block/blk-core.c
index 6bd4d17..3793967 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -269,7 +269,7 @@
 		int i;
 
 		queue_for_each_hw_ctx(q, hctx, i) {
-			cancel_work_sync(&hctx->run_work);
+			cancel_delayed_work_sync(&hctx->run_work);
 			cancel_delayed_work_sync(&hctx->delay_work);
 		}
 	} else {
diff --git a/block/blk-mq.c b/block/blk-mq.c
index e6aad49..5c68fce 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1166,13 +1166,9 @@
 		put_cpu();
 	}
 
-	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));
+	kblockd_schedule_delayed_work_on(blk_mq_hctx_next_cpu(hctx),
+					 &hctx->run_work,
+					 msecs_to_jiffies(msecs));
 }
 
 void blk_mq_delay_run_hw_queue(struct blk_mq_hw_ctx *hctx, unsigned long msecs)
@@ -1224,7 +1220,7 @@
 
 void blk_mq_stop_hw_queue(struct blk_mq_hw_ctx *hctx)
 {
-	cancel_work(&hctx->run_work);
+	cancel_delayed_work_sync(&hctx->run_work);
 	cancel_delayed_work(&hctx->delay_work);
 	set_bit(BLK_MQ_S_STOPPED, &hctx->state);
 }
@@ -1282,17 +1278,7 @@
 {
 	struct blk_mq_hw_ctx *hctx;
 
-	hctx = container_of(work, struct blk_mq_hw_ctx, run_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);
-
+	hctx = container_of(work, struct blk_mq_hw_ctx, run_work.work);
 	__blk_mq_run_hw_queue(hctx);
 }
 
@@ -1898,8 +1884,7 @@
 	if (node == NUMA_NO_NODE)
 		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->run_work, blk_mq_run_work_fn);
 	INIT_DELAYED_WORK(&hctx->delay_work, blk_mq_delay_work_fn);
 	spin_lock_init(&hctx->lock);
 	INIT_LIST_HEAD(&hctx->dispatch);
diff --git a/include/linux/blk-mq.h b/include/linux/blk-mq.h
index 32bd8eb..c7cc903 100644
--- a/include/linux/blk-mq.h
+++ b/include/linux/blk-mq.h
@@ -15,7 +15,7 @@
 		unsigned long		state;		/* BLK_MQ_S_* flags */
 	} ____cacheline_aligned_in_smp;
 
-	struct work_struct	run_work;
+	struct delayed_work	run_work;
 	cpumask_var_t		cpumask;
 	int			next_cpu;
 	int			next_cpu_batch;
@@ -51,7 +51,6 @@
 
 	atomic_t		nr_active;
 
-	struct delayed_work	delayed_run_work;
 	struct delayed_work	delay_work;
 
 	struct hlist_node	cpuhp_dead;