[PATCH v2] block/blk-cgroup-rwstat: use data_race() for online test

Tao Cui <[email protected]> Mon, 3 Aug 2026 21:34:00 +0800
Newsgroups org.kernel.vger.cgroups,org.kernel.vger.linux-block,org.kernel.vger.linux-kernel
Message-ID <[email protected]>
From: Tao Cui <[email protected]>

blkg_rwstat_recursive_sum() reads pos_blkg->online without the queue
lock that its doc comment requires, since blkcg_print_blkgs() stopped
holding it in 56cc24f59c14.  Concurrent blkg_create/destroy flips
->online, tripping KCSAN.  The race is harmless (RCU-protected, stale
online only causes minor stat noise).  Use data_race() to annotate
the intentional lockless read.

Also update the stale doc comment that still requires the queue lock.

Reviewed-by: Yu Kuai <[email protected]>
Signed-off-by: Tao Cui <[email protected]>

---

Changes in v2:
- Update the stale kernel-doc comment that still mandates the queue lock.

Link: https://lore.kernel.org/all/[email protected]/
---
 block/blk-cgroup-rwstat.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/block/blk-cgroup-rwstat.c b/block/blk-cgroup-rwstat.c
index aae910713814..2e9bf081cce2 100644
--- a/block/blk-cgroup-rwstat.c
+++ b/block/blk-cgroup-rwstat.c
@@ -88,8 +88,7 @@ EXPORT_SYMBOL_GPL(blkg_prfill_rwstat);
  * @sum: blkg_rwstat_sample structure containing the results
  *
  * Collect the blkg_rwstat specified by @blkg, @pol and @off and all its
- * online descendants and their aux counts.  The caller must be holding the
- * queue lock for online tests.
+ * online descendants and their aux counts.
  *
  * If @pol is NULL, blkg_rwstat is at @off bytes into @blkg; otherwise, it
  * is at @off bytes into @blkg's blkg_policy_data of the policy.
@@ -107,7 +106,7 @@ void blkg_rwstat_recursive_sum(struct blkcg_gq *blkg, struct blkcg_policy *pol,
 	blkg_for_each_descendant_pre(pos_blkg, pos_css, blkg) {
 		struct blkg_rwstat *rwstat;
 
-		if (!pos_blkg->online)
+		if (!data_race(pos_blkg->online))
 			continue;
 
 		if (pol) {
-- 
2.43.0