On 2024/8/3 13:33, Yosry Ahmed wrote:
Move the comments and spin_{lock/unlock}() calls around in
shrink_worker() to make it obvious the lock is protecting the loop
updating zswap_next_shrink.
Signed-off-by: Yosry Ahmed <yosryahmed@xxxxxxxxxx>
Yeah, it's clearer.
Reviewed-by: Chengming Zhou <chengming.zhou@xxxxxxxxx>
Thanks.
---
This is intended to be squashed into "mm: zswap: fix global shrinker
memcg iteration".
---
mm/zswap.c | 14 ++++++--------
1 file changed, 6 insertions(+), 8 deletions(-)
diff --git a/mm/zswap.c b/mm/zswap.c
index babf0abbcc765..df620eacd1d11 100644
--- a/mm/zswap.c
+++ b/mm/zswap.c
@@ -1364,24 +1364,22 @@ static void shrink_worker(struct work_struct *w)
* until the next run of shrink_worker().
*/
do {
- spin_lock(&zswap_shrink_lock);
-
/*
* Start shrinking from the next memcg after zswap_next_shrink.
* When the offline cleaner has already advanced the cursor,
* advancing the cursor here overlooks one memcg, but this
* should be negligibly rare.
+ *
+ * If we get an online memcg, keep the extra reference in case
+ * the original one obtained by mem_cgroup_iter() is dropped by
+ * zswap_memcg_offline_cleanup() while we are shrinking the
+ * memcg.
*/
+ spin_lock(&zswap_shrink_lock);
do {
memcg = mem_cgroup_iter(NULL, zswap_next_shrink, NULL);
zswap_next_shrink = memcg;
} while (memcg && !mem_cgroup_tryget_online(memcg));
- /*
- * Note that if we got an online memcg, we will keep the extra
- * reference in case the original reference obtained by mem_cgroup_iter
- * is dropped by the zswap memcg offlining callback, ensuring that the
- * memcg is not killed when we are reclaiming.
- */
spin_unlock(&zswap_shrink_lock);
if (!memcg) {
