[PATCH v2 12/14] mm: multi-gen LRU: walk_pte_range() use pte_offset_map_rw_nolock()

[Qi Zheng]

In walk_pte_range(), we may modify the pte entry after holding the ptl, so
convert it to using pte_offset_map_rw_nolock(). At this time, the write
lock of mmap_lock is not held, and the pte_same() check is not performed
after the ptl held, so we should get pmdval and do pmd_same() check to
ensure the stability of pmd entry.

Signed-off-by: Qi Zheng <zhengqi.arch at bytedance.com>
---
 mm/vmscan.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index 5dc96a843466a..89ef1ac8eb1a6 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -3396,8 +3396,10 @@ static bool walk_pte_range(pmd_t *pmd, unsigned long start, unsigned long end,
 	struct pglist_data *pgdat = lruvec_pgdat(walk->lruvec);
 	DEFINE_MAX_SEQ(walk->lruvec);
 	int old_gen, new_gen = lru_gen_from_seq(max_seq);
+	pmd_t pmdval;
 
-	pte = pte_offset_map_nolock(args->mm, pmd, start & PMD_MASK, &ptl);
+	pte = pte_offset_map_rw_nolock(args->mm, pmd, start & PMD_MASK, &pmdval,
+				       &ptl);
 	if (!pte)
 		return false;
 	if (!spin_trylock(ptl)) {
@@ -3405,6 +3407,11 @@ static bool walk_pte_range(pmd_t *pmd, unsigned long start, unsigned long end,
 		return false;
 	}
 
+	if (unlikely(!pmd_same(pmdval, pmdp_get_lockless(pmd)))) {
+		pte_unmap_unlock(pte, ptl);
+		return false;
+	}
+
 	arch_enter_lazy_mmu_mode();
 restart:
 	for (i = pte_index(start), addr = start; addr != end; i++, addr += PAGE_SIZE) {
-- 
2.20.1