[PATCH v1 1/2] mm/swap: refault on swap-in, not in the swap cache allocator

[Alexandre Ghiti]

__swap_cache_alloc() calls workingset_refault() on every folio it
allocates into the swap cache. That is correct for a swap-in, but wrong
for a zswap writeback buffer, which is allocated through the same path
(swap_cache_alloc_folio()): the buffer is not a refault of anything, yet
it gets counted as one, inflating anon refault/activation.

Refaulting is a swap-in event, not an allocation event: move it out of
__swap_cache_alloc() into the swap-in callers.

The allocator has already put the folio on the LRU by the time the callers
refault, so workingset_refault() now has to activate an already-LRU folio:
folio_set_active() only sets the flag before LRU insertion and would be a
no-op here, so use folio_activate() when the folio is already on the LRU.
Existing callers pass not-yet-LRU folios and are unaffected.

Fixes: aae466b0052e ("mm/swap: implement workingset detection for anonymous LRU")
Signed-off-by: Nhat Pham <nphamcs@xxxxxxxxx>
Signed-off-by: Alexandre Ghiti <alex@xxxxxxxx>
---
 mm/swap_state.c | 20 ++++++++++++++++++--
 mm/workingset.c |  5 ++++-
 2 files changed, 22 insertions(+), 3 deletions(-)

diff --git a/mm/swap_state.c b/mm/swap_state.c
index 9c3a5cf99778..0bb26d2ba2ba 100644
--- a/mm/swap_state.c
+++ b/mm/swap_state.c
@@ -477,8 +477,6 @@ static struct folio *__swap_cache_alloc(struct swap_cluster_info *ci,
 
 	/* memsw uncharges swap when folio is added to swap cache */
 	memcg1_swapin(folio);
-	if (shadow)
-		workingset_refault(folio, shadow);
 
 	node_stat_mod_folio(folio, NR_FILE_PAGES, nr_pages);
 	lruvec_stat_mod_folio(folio, NR_SWAPCACHE, nr_pages);
@@ -638,17 +636,26 @@ static struct folio *swap_cache_read_folio(swp_entry_t entry, gfp_t gfp,
 					   struct swap_iocb **plug, bool readahead)
 {
 	struct folio *folio;
+	void *shadow = NULL;
 
 	do {
 		folio = swap_cache_get_folio(entry);
 		if (folio)
 			return folio;
+		/*
+		 * Capture the slot's shadow before the allocation overwrites it,
+		 * so a fresh swap-in can be evaluated as a refault below.
+		 */
+		shadow = swap_cache_get_shadow(entry);
 		folio = swap_cache_alloc_folio(entry, gfp, BIT(0), NULL, mpol, ilx);
 	} while (PTR_ERR(folio) == -EEXIST);
 
 	if (IS_ERR_OR_NULL(folio))
 		return NULL;
 
+	if (shadow)
+		workingset_refault(folio, shadow);
+
 	swap_read_folio(folio, plug);
 	if (readahead) {
 		folio_set_readahead(folio);
@@ -678,17 +685,26 @@ struct folio *swapin_sync(swp_entry_t entry, gfp_t gfp, unsigned long orders,
 			   struct vm_fault *vmf, struct mempolicy *mpol, pgoff_t ilx)
 {
 	struct folio *folio;
+	void *shadow = NULL;
 
 	do {
 		folio = swap_cache_get_folio(entry);
 		if (folio)
 			return folio;
+		/*
+		 * Capture the slot's shadow before the allocation overwrites it,
+		 * so a fresh swap-in can be evaluated as a refault below.
+		 */
+		shadow = swap_cache_get_shadow(entry);
 		folio = swap_cache_alloc_folio(entry, gfp, orders, vmf, mpol, ilx);
 	} while (PTR_ERR(folio) == -EEXIST);
 
 	if (IS_ERR(folio))
 		return folio;
 
+	if (shadow)
+		workingset_refault(folio, shadow);
+
 	swap_read_folio(folio, NULL);
 	return folio;
 }
diff --git a/mm/workingset.c b/mm/workingset.c
index f351798e723a..7c30cf091cbf 100644
--- a/mm/workingset.c
+++ b/mm/workingset.c
@@ -577,7 +577,10 @@ void workingset_refault(struct folio *folio, void *shadow)
 	if (!workingset_test_recent(shadow, file, &workingset, true))
 		goto out;
 
-	folio_set_active(folio);
+	if (unlikely(folio_test_lru(folio)))
+		folio_activate(folio);
+	else
+		folio_set_active(folio);
 	workingset_age_nonresident(lruvec, nr);
 	mod_lruvec_state(lruvec, WORKINGSET_ACTIVATE_BASE + file, nr);
 
-- 
2.53.0-Meta