Re: [PATCH v5 0/3] mm/vmscan: reduce lru_lock contention via vmstat-derived scan-balance cost

On Mon, 27 Jul 2026 09:23:22 -0700 Usama Arif <[email protected]> wrote:

> The actual code between v4 and v5 is the same. The last patch is
> now split into 2 commits and the commit messages are shorter and
> cleaner.

Yup.  A comment was altered.

Thanks, I updated mm.git's mm-unstable branch to this version.

> v4 -> v5: https://lore.kernel.org/all/[email protected]/
> - Cleaner and shorter commit message (Johannes and Shakeel)
> - Split PGROTATE_{ANON,FILE} definitions and producer accounting into
>   a standalone patch, and document their public diagnostic value (Shakeel)

Here's how 5 altered mm.git:

 mm/memcontrol.c |    6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

--- a/mm/memcontrol.c~b
+++ a/mm/memcontrol.c
@@ -526,9 +526,9 @@ unsigned long lruvec_page_state(struct l
  * Do NOT use for non-monotonic page-count reads where a transient negative
  * reading from per-CPU delta skew must present as zero.
  *
- * XXX: This helper (and its node/global peers) exists because we place
- * monotonically-incremented event counters (NR_VMSCAN_WRITE and PGROTATE_*)
- * into enum node_stat_item.
+ * XXX: This helper (and its node/global peers) exists because some
+ * monotonically-incremented event counters are stored in
+ * enum node_stat_item.
  */
 unsigned long lruvec_page_state_monotonic(struct lruvec *lruvec,
 					  enum node_stat_item idx)
_