block: make queue flags non-atomic

We can save some atomic ops in the IO path, if we clearly define
the rules of how to modify the queue flags.

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

5 files changed, 44 insertions, 24 deletions

diff --git a/block/blk-core.c b/block/blk-core.c
index e447799256d6..d2f23ec5ebfa 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -198,7 +198,8 @@ void blk_plug_device(struct request_queue *q)
 	if (blk_queue_stopped(q))
 		return;
 
-	if (!test_and_set_bit(QUEUE_FLAG_PLUGGED, &q->queue_flags)) {
+	if (!test_bit(QUEUE_FLAG_PLUGGED, &q->queue_flags)) {
+		__set_bit(QUEUE_FLAG_PLUGGED, &q->queue_flags);
 		mod_timer(&q->unplug_timer, jiffies + q->unplug_delay);
 		blk_add_trace_generic(q, NULL, 0, BLK_TA_PLUG);
 	}
@@ -213,9 +214,10 @@ int blk_remove_plug(struct request_queue *q)
 {
 	WARN_ON(!irqs_disabled());
 
-	if (!test_and_clear_bit(QUEUE_FLAG_PLUGGED, &q->queue_flags))
+	if (!test_bit(QUEUE_FLAG_PLUGGED, &q->queue_flags))
 		return 0;
 
+	queue_flag_clear(QUEUE_FLAG_PLUGGED, q);
 	del_timer(&q->unplug_timer);
 	return 1;
 }
@@ -311,15 +313,16 @@ void blk_start_queue(struct request_queue *q)
 {
 	WARN_ON(!irqs_disabled());
 
-	clear_bit(QUEUE_FLAG_STOPPED, &q->queue_flags);
+	queue_flag_clear(QUEUE_FLAG_STOPPED, q);
 
 	/*
 	 * one level of recursion is ok and is much faster than kicking
 	 * the unplug handling
 	 */
-	if (!test_and_set_bit(QUEUE_FLAG_REENTER, &q->queue_flags)) {
+	if (!test_bit(QUEUE_FLAG_REENTER, &q->queue_flags)) {
+		queue_flag_set(QUEUE_FLAG_REENTER, q);
 		q->request_fn(q);
-		clear_bit(QUEUE_FLAG_REENTER, &q->queue_flags);
+		queue_flag_clear(QUEUE_FLAG_REENTER, q);
 	} else {
 		blk_plug_device(q);
 		kblockd_schedule_work(&q->unplug_work);
@@ -344,7 +347,7 @@ EXPORT_SYMBOL(blk_start_queue);
 void blk_stop_queue(struct request_queue *q)
 {
 	blk_remove_plug(q);
-	set_bit(QUEUE_FLAG_STOPPED, &q->queue_flags);
+	queue_flag_set(QUEUE_FLAG_STOPPED, q);
 }
 EXPORT_SYMBOL(blk_stop_queue);
 
@@ -373,11 +376,8 @@ EXPORT_SYMBOL(blk_sync_queue);
  * blk_run_queue - run a single device queue
  * @q:	The queue to run
  */
-void blk_run_queue(struct request_queue *q)
+void __blk_run_queue(struct request_queue *q)
 {
-	unsigned long flags;
-
-	spin_lock_irqsave(q->queue_lock, flags);
 	blk_remove_plug(q);
 
 	/*
@@ -385,15 +385,28 @@ void blk_run_queue(struct request_queue *q)
 	 * handling reinvoke the handler shortly if we already got there.
 	 */
 	if (!elv_queue_empty(q)) {
-		if (!test_and_set_bit(QUEUE_FLAG_REENTER, &q->queue_flags)) {
+		if (!test_bit(QUEUE_FLAG_REENTER, &q->queue_flags)) {
+			queue_flag_set(QUEUE_FLAG_REENTER, q);
 			q->request_fn(q);
-			clear_bit(QUEUE_FLAG_REENTER, &q->queue_flags);
+			queue_flag_clear(QUEUE_FLAG_REENTER, q);
 		} else {
 			blk_plug_device(q);
 			kblockd_schedule_work(&q->unplug_work);
 		}
 	}
+}
+EXPORT_SYMBOL(__blk_run_queue);
 
+/**
+ * blk_run_queue - run a single device queue
+ * @q: The queue to run
+ */
+void blk_run_queue(struct request_queue *q)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(q->queue_lock, flags);
+	__blk_run_queue(q);
 	spin_unlock_irqrestore(q->queue_lock, flags);
 }
 EXPORT_SYMBOL(blk_run_queue);
@@ -406,7 +419,7 @@ void blk_put_queue(struct request_queue *q)
 void blk_cleanup_queue(struct request_queue *q)
 {
 	mutex_lock(&q->sysfs_lock);
-	set_bit(QUEUE_FLAG_DEAD, &q->queue_flags);
+	queue_flag_set_unlocked(QUEUE_FLAG_DEAD, q);
 	mutex_unlock(&q->sysfs_lock);
 
 	if (q->elevator)
diff --git a/block/blk-merge.c b/block/blk-merge.c
index b5c5c4a9e3f0..73b23562af20 100644
--- a/block/blk-merge.c
+++ b/block/blk-merge.c
@@ -55,7 +55,7 @@ void blk_recalc_rq_segments(struct request *rq)
 	if (!rq->bio)
 		return;
 
-	cluster = q->queue_flags & (1 << QUEUE_FLAG_CLUSTER);
+	cluster = test_bit(QUEUE_FLAG_CLUSTER, &q->queue_flags);
 	hw_seg_size = seg_size = 0;
 	phys_size = hw_size = nr_phys_segs = nr_hw_segs = 0;
 	rq_for_each_segment(bv, rq, iter) {
@@ -128,7 +128,7 @@ EXPORT_SYMBOL(blk_recount_segments);
 static int blk_phys_contig_segment(struct request_queue *q, struct bio *bio,
 				   struct bio *nxt)
 {
-	if (!(q->queue_flags & (1 << QUEUE_FLAG_CLUSTER)))
+	if (!test_bit(QUEUE_FLAG_CLUSTER, &q->queue_flags))
 		return 0;
 
 	if (!BIOVEC_PHYS_MERGEABLE(__BVEC_END(bio), __BVEC_START(nxt)))
@@ -175,7 +175,7 @@ int blk_rq_map_sg(struct request_queue *q, struct request *rq,
 	int nsegs, cluster;
 
 	nsegs = 0;
-	cluster = q->queue_flags & (1 << QUEUE_FLAG_CLUSTER);
+	cluster = test_bit(QUEUE_FLAG_CLUSTER, &q->queue_flags);
 
 	/*
 	 * for each bio in rq
diff --git a/block/blk-settings.c b/block/blk-settings.c
index 77b51dc37a3c..6089384ab064 100644
--- a/block/blk-settings.c
+++ b/block/blk-settings.c
@@ -287,7 +287,7 @@ void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b)
 	t->max_segment_size = min(t->max_segment_size, b->max_segment_size);
 	t->hardsect_size = max(t->hardsect_size, b->hardsect_size);
 	if (!test_bit(QUEUE_FLAG_CLUSTER, &b->queue_flags))
-		clear_bit(QUEUE_FLAG_CLUSTER, &t->queue_flags);
+		queue_flag_clear(QUEUE_FLAG_CLUSTER, t);
 }
 EXPORT_SYMBOL(blk_queue_stack_limits);
 
diff --git a/block/blk-tag.c b/block/blk-tag.c
index 4780a46ce234..e176ddbe599e 100644
--- a/block/blk-tag.c
+++ b/block/blk-tag.c
@@ -70,7 +70,7 @@ void __blk_queue_free_tags(struct request_queue *q)
 	__blk_free_tags(bqt);
 
 	q->queue_tags = NULL;
-	q->queue_flags &= ~(1 << QUEUE_FLAG_QUEUED);
+	queue_flag_clear(QUEUE_FLAG_QUEUED, q);
 }
 
 /**
@@ -98,7 +98,7 @@ EXPORT_SYMBOL(blk_free_tags);
  **/
 void blk_queue_free_tags(struct request_queue *q)
 {
-	clear_bit(QUEUE_FLAG_QUEUED, &q->queue_flags);
+	queue_flag_clear(QUEUE_FLAG_QUEUED, q);
 }
 EXPORT_SYMBOL(blk_queue_free_tags);
 
@@ -188,7 +188,7 @@ int blk_queue_init_tags(struct request_queue *q, int depth,
 		rc = blk_queue_resize_tags(q, depth);
 		if (rc)
 			return rc;
-		set_bit(QUEUE_FLAG_QUEUED, &q->queue_flags);
+		queue_flag_set(QUEUE_FLAG_QUEUED, q);
 		return 0;
 	} else
 		atomic_inc(&tags->refcnt);
@@ -197,7 +197,7 @@ int blk_queue_init_tags(struct request_queue *q, int depth,
 	 * assign it, all done
 	 */
 	q->queue_tags = tags;
-	q->queue_flags |= (1 << QUEUE_FLAG_QUEUED);
+	queue_flag_set(QUEUE_FLAG_QUEUED, q);
 	INIT_LIST_HEAD(&q->tag_busy_list);
 	return 0;
 fail:
diff --git a/block/elevator.c b/block/elevator.c
index 88318c383608..e8a90fe23424 100644
--- a/block/elevator.c
+++ b/block/elevator.c
@@ -1070,7 +1070,7 @@ static int elevator_switch(struct request_queue *q, struct elevator_type *new_e)
 	 */
 	spin_lock_irq(q->queue_lock);
 
-	set_bit(QUEUE_FLAG_ELVSWITCH, &q->queue_flags);
+	queue_flag_set(QUEUE_FLAG_ELVSWITCH, q);
 
 	elv_drain_elevator(q);
 
@@ -1104,7 +1104,10 @@ static int elevator_switch(struct request_queue *q, struct elevator_type *new_e)
 	 * finally exit old elevator and turn off BYPASS.
 	 */
 	elevator_exit(old_elevator);
-	clear_bit(QUEUE_FLAG_ELVSWITCH, &q->queue_flags);
+	spin_lock_irq(q->queue_lock);
+	queue_flag_clear(QUEUE_FLAG_ELVSWITCH, q);
+	spin_unlock_irq(q->queue_lock);
+
 	return 1;
 
 fail_register:
@@ -1115,7 +1118,11 @@ fail_register:
 	elevator_exit(e);
 	q->elevator = old_elevator;
 	elv_register_queue(q);
-	clear_bit(QUEUE_FLAG_ELVSWITCH, &q->queue_flags);
+
+	spin_lock_irq(q->queue_lock);
+	queue_flag_clear(QUEUE_FLAG_ELVSWITCH, q);
+	spin_unlock_irq(q->queue_lock);
+
 	return 0;
 }