| Message ID | 20211019024132.432458-2-yukuai3@huawei.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | blk-cgroup: synchoronize blkg creation against policy deactivation | expand |
Hello, On Tue, Oct 19, 2021 at 10:41:31AM +0800, Yu Kuai wrote: > blkcg_deactivate_policy() can be called from either > blk_cleanup_queue() for all policies or elevator_switch() for bfq > policy. Thus there is no need to freeze queue in > blkcg_deactivate_policy() since the caller freeze the queue aready. > > Signed-off-by: Yu Kuai <yukuai3@huawei.com> I would just keep the calls. The operation requires the q to be frozen and there isn't a good way to annotate that the caller must be holding a usage count, so it's useful even just as documentation. Thanks.
diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c index cec86a705c89..ca60233c8392 100644 --- a/block/blk-cgroup.c +++ b/block/blk-cgroup.c @@ -1398,9 +1398,6 @@ void blkcg_deactivate_policy(struct request_queue *q, if (!blkcg_policy_enabled(q, pol)) return; - if (queue_is_mq(q)) - blk_mq_freeze_queue(q); - spin_lock_irq(&q->queue_lock); __clear_bit(pol->plid, q->blkcg_pols); @@ -1419,9 +1416,6 @@ void blkcg_deactivate_policy(struct request_queue *q, } spin_unlock_irq(&q->queue_lock); - - if (queue_is_mq(q)) - blk_mq_unfreeze_queue(q); } EXPORT_SYMBOL_GPL(blkcg_deactivate_policy);
blkcg_deactivate_policy() can be called from either blk_cleanup_queue() for all policies or elevator_switch() for bfq policy. Thus there is no need to freeze queue in blkcg_deactivate_policy() since the caller freeze the queue aready. Signed-off-by: Yu Kuai <yukuai3@huawei.com> --- block/blk-cgroup.c | 6 ------ 1 file changed, 6 deletions(-)