[Devel] [PATCH RHEL9 COMMIT] mm: Fix mem_cgroup_migrate's warn_on itself

[Konstantin Khorenko]

The commit is pushed to "branch-rh9-5.14.0-162.6.1.vz9.18.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git
after rh9-5.14.0-162.6.1.vz9.18.10
------>
commit 6c6835cdae7442cfbf7bfbac5ff157754eb18266
Author: Alexander Atanasov <alexander.atanasov at virtuozzo.com>
Date:   Wed Feb 15 10:57:56 2023 +0200

    mm: Fix mem_cgroup_migrate's warn_on itself
    
    The original check in vz7 commit only expected to have Anon,Swap and
    Cached pages. Which is not a valid assumption any more since folios have
    more folio types.
    
    To fix this update the condition to handle more folio types and change
    its logic to be in form of:
      flag && (any other flag)
    rather than
      flag != (!any other flag).
    
    Do this to avoid warn_on if new flags are added.
    
    Ressurect and rewrite a comment about what does the check do,
    also put an explanation why the check is in mem_cgroup_migrate()
    and not when updating the flags.
    
    Fixes: b8bc3dbf5e18 ("mm: per memory cgroup page cache limit")
    
    https://jira.sw.ru/browse/PSBM-144609
    Signed-off-by: Alexander Atanasov <alexander.atanasov at virtuozzo.com>
---
 mm/memcontrol.c | 21 +++++++++++++++++++--
 1 file changed, 19 insertions(+), 2 deletions(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 9f566fd438ef..8cd3a134c226 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -7687,8 +7687,25 @@ void mem_cgroup_migrate(struct folio *old, struct folio *new)
 			page_counter_charge(&memcg->memsw, nr_pages);
 	}
 
-	WARN_ON((!PageAnon(&new->page) && !PageSwapBacked(&new->page)) !=
-		folio_memcg_cache(new));
+	/*
+	 * finist explained the idea behind adding a WARN_ON() here:
+	 * - we do not want to check flags correctness on each flag change
+	 *   because of performance
+	 * - we do want to have a warning in case we somehow messed-up and
+	 *   have got a folio with wrong bits set
+	 * - we do not insist to catch every first buggy page with wrong
+	 *   bits
+	 *
+	 * But we still want to get a warning about the problem sooner or
+	 * later if the problem with flags exists.
+	 *
+	 * To achieve this check if a folio that is marked as cache does
+	 * not have any other incompatible flags set.
+	 */
+	WARN_ON(folio_memcg_cache(new) &&
+		(folio_test_slab(new)	      || folio_test_anon(new)	   ||
+		 folio_test_swapbacked(new)   || folio_test_swapcache(new) ||
+		 folio_test_mappedtodisk(new) || folio_test_ksm(new)));
 
 	if (folio_memcg_cache(new))
 		page_counter_charge(&memcg->cache, nr_pages);