[PATCH] mm: pgtable: fix incorrect reclaim of non-empty PTE pages

[Qi Zheng <zhengqi.arch@bytedance.com>]

In zap_pte_range(), if the pte lock was released midway, the pte entries
may be refilled with physical pages by another thread, which may cause a
non-empty PTE page to be reclaimed and eventually cause the system to
crash.

To fix it, fall back to the slow path in this case to recheck if all pte
entries are still none.

Fixes: 6375e95f381e ("mm: pgtable: reclaim empty PTE page in madvise(MADV_DONTNEED)")
Reported-by: Christian Brauner <brauner@kernel.org>
Closes: https://lore.kernel.org/all/20250207-anbot-bankfilialen-acce9d79a2c7@brauner/
Reported-by: Qu Wenruo <quwenruo.btrfs@gmx.com>
Closes: https://lore.kernel.org/all/152296f3-5c81-4a94-97f3-004108fba7be@gmx.com/
Tested-by: Zi Yan <ziy@nvidia.com>
Cc: stable@vger.kernel.org
Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
---
 mm/memory.c | 17 ++++++++++++++---
 1 file changed, 14 insertions(+), 3 deletions(-)

diff --git a/mm/memory.c b/mm/memory.c
index a8196ae72e9ae..7c7193cb21248 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -1721,7 +1721,7 @@ static unsigned long zap_pte_range(struct mmu_gather *tlb,
 	pmd_t pmdval;
 	unsigned long start = addr;
 	bool can_reclaim_pt = reclaim_pt_is_enabled(start, end, details);
-	bool direct_reclaim = false;
+	bool direct_reclaim = true;
 	int nr;
 
 retry:
@@ -1736,8 +1736,10 @@ static unsigned long zap_pte_range(struct mmu_gather *tlb,
 	do {
 		bool any_skipped = false;
 
-		if (need_resched())
+		if (need_resched()) {
+			direct_reclaim = false;
 			break;
+		}
 
 		nr = do_zap_pte_range(tlb, vma, pte, addr, end, details, rss,
 				      &force_flush, &force_break, &any_skipped);
@@ -1745,11 +1747,20 @@ static unsigned long zap_pte_range(struct mmu_gather *tlb,
 			can_reclaim_pt = false;
 		if (unlikely(force_break)) {
 			addr += nr * PAGE_SIZE;
+			direct_reclaim = false;
 			break;
 		}
 	} while (pte += nr, addr += PAGE_SIZE * nr, addr != end);
 
-	if (can_reclaim_pt && addr == end)
+	/*
+	 * Fast path: try to hold the pmd lock and unmap the PTE page.
+	 *
+	 * If the pte lock was released midway (retry case), or if the attempt
+	 * to hold the pmd lock failed, then we need to recheck all pte entries
+	 * to ensure they are still none, thereby preventing the pte entries
+	 * from being repopulated by another thread.
+	 */
+	if (can_reclaim_pt && direct_reclaim && addr == end)
 		direct_reclaim = try_get_and_clear_pmd(mm, pmd, &pmdval);
 
 	add_mm_rss_vec(mm, rss);
-- 
2.20.1