Skip to contentSkip to conversation

blk-mq: Allow blocking queue tag iter callbacks

PATCH In review Keith Busch 3 participants 3 messages Sep 25, 2018 – Sep 26, 2018 Block layer

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 blk-mq: Allow blocking queue tag iter callbacks

A recent commit runs tag iterator callbacks under the rcu read lock, but existing callbacks do not satisfy the non-blocking requirement. The commit intended to prevent an iterator from accessing a queue that's being modified. This patch fixes the original issue by taking a queue reference instead of reading it, which allows callbacks to make blocking calls.

Fixes f5bbbbe4d6357 ("blk-mq: sync the update nr_hw_queues with blk_mq_queue_tag_busy_iter")
Cc Jianchao Wang <[email protected]>

---
v1 -> v2:

Leave the iterator interface as-is, making this change transparent to the callers. This will increment the queue usage counter in the timeout path twice, which is harmless.

Changelog.

 block/blk-mq-tag.c | 13 ++++---------
 1 file changed, 4 insertions(+), 9 deletions(-)
block/blk-mq-tag.c +4 -9
index 94e1ed667b6e..41317c50a446 100644
@@ -322,16 +322,11 @@ void blk_mq_queue_tag_busy_iter(struct request_queue *q, busy_iter_fn *fn,
 
 	/*
 	 * __blk_mq_update_nr_hw_queues will update the nr_hw_queues and
-	 * queue_hw_ctx after freeze the queue. So we could use q_usage_counter
-	 * to avoid race with it. __blk_mq_update_nr_hw_queues will users
-	 * synchronize_rcu to ensure all of the users go out of the critical
-	 * section below and see zeroed q_usage_counter.
+	 * queue_hw_ctx after freeze the queue, so we use q_usage_counter
+	 * to avoid race with it.
 	 */
-	rcu_read_lock();
-	if (percpu_ref_is_zero(&q->q_usage_counter)) {
-		rcu_read_unlock();
+	if (!percpu_ref_tryget(&q->q_usage_counter))
 		return;
-	}
 
 	queue_for_each_hw_ctx(q, hctx, i) {
 		struct blk_mq_tags *tags = hctx->tags;
@@ -347,7 +342,7 @@ void blk_mq_queue_tag_busy_iter(struct request_queue *q, busy_iter_fn *fn,
 			bt_for_each(hctx, &tags->breserved_tags, fn, priv, true);
 		bt_for_each(hctx, &tags->bitmap_tags, fn, priv, false);
 	}
-	rcu_read_unlock();
+	blk_queue_exit(q);
 }
 
 static int bt_alloc(struct sbitmap_queue *bt, unsigned int depth,
Signature · 1 line
2.14.4
Acked-by on v2 1/1
#1 lore raw

Hi Keith

Quoted 10 lines · Keith Busch

A recent commit runs tag iterator callbacks under the rcu read lock, but existing callbacks do not satisfy the non-blocking requirement. The commit intended to prevent an iterator from accessing a queue that's being modified. This patch fixes the original issue by taking a queue reference instead of reading it, which allows callbacks to make blocking calls.

Fixes f5bbbbe4d6357 ("blk-mq: sync the update nr_hw_queues with blk_mq_queue_tag_busy_iter")
Cc Jianchao Wang <[email protected]>
Quoted 7 lines

---
v1 -> v2:

Leave the iterator interface as-is, making this change transparent to the callers. This will increment the queue usage counter in the timeout path twice, which is harmless.

This is OK to me.

Quoted 39 lines

Changelog.

 block/blk-mq-tag.c | 13 ++++---------
 1 file changed, 4 insertions(+), 9 deletions(-)
block/blk-mq-tag.c +4 -9
index 94e1ed667b6e..41317c50a446 100644
@@ -322,16 +322,11 @@ void blk_mq_queue_tag_busy_iter(struct request_queue *q, busy_iter_fn *fn,
 
 	/*
 	 * __blk_mq_update_nr_hw_queues will update the nr_hw_queues and
-	 * queue_hw_ctx after freeze the queue. So we could use q_usage_counter
-	 * to avoid race with it. __blk_mq_update_nr_hw_queues will users
-	 * synchronize_rcu to ensure all of the users go out of the critical
-	 * section below and see zeroed q_usage_counter.
+	 * queue_hw_ctx after freeze the queue, so we use q_usage_counter
+	 * to avoid race with it.
 	 */
-	rcu_read_lock();
-	if (percpu_ref_is_zero(&q->q_usage_counter)) {
-		rcu_read_unlock();
+	if (!percpu_ref_tryget(&q->q_usage_counter))
 		return;
-	}
 
 	queue_for_each_hw_ctx(q, hctx, i) {
 		struct blk_mq_tags *tags = hctx->tags;
@@ -347,7 +342,7 @@ void blk_mq_queue_tag_busy_iter(struct request_queue *q, busy_iter_fn *fn,
 			bt_for_each(hctx, &tags->breserved_tags, fn, priv, true);
 		bt_for_each(hctx, &tags->bitmap_tags, fn, priv, false);
 	}
-	rcu_read_unlock();
+	blk_queue_exit(q);
 }
 
 static int bt_alloc(struct sbitmap_queue *bt, unsigned int depth,
maintainer
#2 lore raw
In reply to Keith Busch
Quoted 6 lines · Keith Busch

A recent commit runs tag iterator callbacks under the rcu read lock, but existing callbacks do not satisfy the non-blocking requirement. The commit intended to prevent an iterator from accessing a queue that's being modified. This patch fixes the original issue by taking a queue reference instead of reading it, which allows callbacks to make blocking calls.

Thanks Keith, queued up for 4.19.

Signature · 1 line
Jens Axboe