cachestat: do not flush stats in recency check

syzbot detects that cachestat() is flushing stats, which can sleep, in its
RCU read section (see [1]).  This is done in the workingset_test_recent()
step (which checks if the folio's eviction is recent).

Move the stat flushing step to before the RCU read section of cachestat,
and skip stat flushing during the recency check.

[1]: https://lore.kernel.org/cgroups/[email protected]/

Link: https://lkml.kernel.org/r/[email protected]
Fixes: b006847 ("mm: workingset: move the stats flush into workingset_test_recent()")
Signed-off-by: Nhat Pham <[email protected]>
Reported-by: [email protected]
Closes: https://lore.kernel.org/cgroups/[email protected]/
Debugged-by: Johannes Weiner <[email protected]>
Suggested-by: Johannes Weiner <[email protected]>
Acked-by: Johannes Weiner <[email protected]>
Acked-by: Shakeel Butt <[email protected]>
Cc: Al Viro <[email protected]>
Cc: David Hildenbrand <[email protected]>
Cc: "Huang, Ying" <[email protected]>
Cc: Kairui Song <[email protected]>
Cc: Matthew Wilcox (Oracle) <[email protected]>
Cc: Ryan Roberts <[email protected]>
Cc: Yosry Ahmed <[email protected]>
Cc: <[email protected]>	[6.8+]
Signed-off-by: Andrew Morton <[email protected]>

Author: nhatsmrt

include/linux/swap.h

@@ -354,7 +354,8 @@ static inline swp_entry_t page_swap_entry(struct page *page)
 }
 
 /* linux/mm/workingset.c */
-bool workingset_test_recent(void *shadow, bool file, bool *workingset);
+bool workingset_test_recent(void *shadow, bool file, bool *workingset,
+				bool flush);
 void workingset_age_nonresident(struct lruvec *lruvec, unsigned long nr_pages);
 void *workingset_eviction(struct folio *folio, struct mem_cgroup *target_memcg);
 void workingset_refault(struct folio *folio, void *shadow);

mm/filemap.c

@@ -4248,6 +4248,9 @@ static void filemap_cachestat(struct address_space *mapping,
 	XA_STATE(xas, &mapping->i_pages, first_index);
 	struct folio *folio;
 
+	/* Flush stats (and potentially sleep) outside the RCU read section. */
+	mem_cgroup_flush_stats_ratelimited(NULL);
+
 	rcu_read_lock();
 	xas_for_each(&xas, folio, last_index) {
 		int order;
@@ -4311,7 +4314,7 @@ static void filemap_cachestat(struct address_space *mapping,
 					goto resched;
 			}
 #endif
-			if (workingset_test_recent(shadow, true, &workingset))
+			if (workingset_test_recent(shadow, true, &workingset, false))
 				cs->nr_recently_evicted += nr_pages;
 
 			goto resched;

mm/workingset.c

@@ -412,10 +412,12 @@ void *workingset_eviction(struct folio *folio, struct mem_cgroup *target_memcg)
  * @file: whether the corresponding folio is from the file lru.
  * @workingset: where the workingset value unpacked from shadow should
  * be stored.
+ * @flush: whether to flush cgroup rstat.
  *
  * Return: true if the shadow is for a recently evicted folio; false otherwise.
  */
-bool workingset_test_recent(void *shadow, bool file, bool *workingset)
+bool workingset_test_recent(void *shadow, bool file, bool *workingset,
+				bool flush)
 {
 	struct mem_cgroup *eviction_memcg;
 	struct lruvec *eviction_lruvec;
@@ -467,10 +469,16 @@ bool workingset_test_recent(void *shadow, bool file, bool *workingset)
 
 	/*
 	 * Flush stats (and potentially sleep) outside the RCU read section.
+	 *
+	 * Note that workingset_test_recent() itself might be called in RCU read
+	 * section (for e.g, in cachestat) - these callers need to skip flushing
+	 * stats (via the flush argument).
+	 *
 	 * XXX: With per-memcg flushing and thresholding, is ratelimiting
 	 * still needed here?
 	 */
-	mem_cgroup_flush_stats_ratelimited(eviction_memcg);
+	if (flush)
+		mem_cgroup_flush_stats_ratelimited(eviction_memcg);
 
 	eviction_lruvec = mem_cgroup_lruvec(eviction_memcg, pgdat);
 	refault = atomic_long_read(&eviction_lruvec->nonresident_age);
@@ -558,7 +566,7 @@ void workingset_refault(struct folio *folio, void *shadow)
 
 	mod_lruvec_state(lruvec, WORKINGSET_REFAULT_BASE + file, nr);
 
-	if (!workingset_test_recent(shadow, file, &workingset))
+	if (!workingset_test_recent(shadow, file, &workingset, true))
 		return;
 
 	folio_set_active(folio);