On 6/22/23 04:42, Ming Lei wrote:
When __blkcg_rstat_flush() is called from cgroup_rstat_flush*() code
path, interrupt is always disabled.
When we start to flush blkcg per-cpu stats list in __blkg_release()
for avoiding to leak blkcg_gq's reference in commit 20cb1c2fb756
("blk-cgroup: Flush stats before releasing blkcg_gq"), local irq
isn't disabled yet, then lockdep warning may be triggered because
the dependent cgroup locks may be acquired from irq(soft irq) handler.
Fix the issue by disabling local irq always.
Fixes: 20cb1c2fb756 ("blk-cgroup: Flush stats before releasing blkcg_gq")
Reported-by: Shinichiro Kawasaki <shinichiro.kawasaki@xxxxxxx>
Closes: https://lore.kernel.org/linux-block/pz2wzwnmn5tk3pwpskmjhli6g3qly7eoknilb26of376c7kwxy@qydzpvt6zpis/T/#u
Cc: stable@xxxxxxxxxxxxxxx
Cc: Jay Shin <jaeshin@xxxxxxxxxx>
Cc: Tejun Heo <tj@xxxxxxxxxx>
Cc: Waiman Long <longman@xxxxxxxxxx>
Signed-off-by: Ming Lei <ming.lei@xxxxxxxxxx>
---
block/blk-cgroup.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index f0b5c9c41cde..dce1548a7a0c 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -970,6 +970,7 @@ static void __blkcg_rstat_flush(struct blkcg *blkcg, int cpu)
struct llist_head *lhead = per_cpu_ptr(blkcg->lhead, cpu);
struct llist_node *lnode;
struct blkg_iostat_set *bisc, *next_bisc;
+ unsigned long flags;
rcu_read_lock();
@@ -983,7 +984,7 @@ static void __blkcg_rstat_flush(struct blkcg *blkcg, int cpu)
* When flushing from cgroup, cgroup_rstat_lock is always held, so
* this lock won't cause contention most of time.
*/
- raw_spin_lock(&blkg_stat_lock);
+ raw_spin_lock_irqsave(&blkg_stat_lock, flags);
/*
* Iterate only the iostat_cpu's queued in the lockless list.
@@ -1009,7 +1010,7 @@ static void __blkcg_rstat_flush(struct blkcg *blkcg, int cpu)
blkcg_iostat_update(parent, &blkg->iostat.cur,
&blkg->iostat.last);
}
- raw_spin_unlock(&blkg_stat_lock);
+ raw_spin_unlock_irqrestore(&blkg_stat_lock, flags);
out:
rcu_read_unlock();
}
Reviewed-by: Waiman Long <longman@xxxxxxxxxx>
