blk-cgroup: fix list corruption from reorder of WRITE ->lqueued
authorMing Lei <ming.lei@redhat.com>
Wed, 15 May 2024 01:31:57 +0000 (09:31 +0800)
committerJens Axboe <axboe@kernel.dk>
Thu, 16 May 2024 02:14:20 +0000 (20:14 -0600)
commitd0aac2363549e12cc79b8e285f13d5a9f42fd08e
tree454260b00912e2dbb2c92e7214affd3d48975d28
parent6da6680632792709cecf2b006f2fe3ca7857e791
blk-cgroup: fix list corruption from reorder of WRITE ->lqueued

__blkcg_rstat_flush() can be run anytime, especially when blk_cgroup_bio_start
is being executed.

If WRITE of `->lqueued` is re-ordered with READ of 'bisc->lnode.next' in
the loop of __blkcg_rstat_flush(), `next_bisc` can be assigned with one
stat instance being added in blk_cgroup_bio_start(), then the local
list in __blkcg_rstat_flush() could be corrupted.

Fix the issue by adding one barrier.

Cc: Tejun Heo <tj@kernel.org>
Cc: Waiman Long <longman@redhat.com>
Fixes: 3b8cc6298724 ("blk-cgroup: Optimize blkcg_rstat_flush()")
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20240515013157.443672-3-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
block/blk-cgroup.c