[PATCH v2 28/32 fix] mm/memory: allow pte_offset_map[_lock]() to fail: fix

[Hugh Dickins]

__wp_page_copy_user() was liable to call update_mmu_tlb() with NULL
vmf->pte in two places: not a problem today, but could become a problem
later when pte_offset_map_lock() fails.

Signed-off-by: Hugh Dickins <hughd@xxxxxxxxxx>
---
Andrew, please add this as a fix patch for later merge into my
"mm/memory: allow" patch in mm-unstable: it's something noticed while
researching the bug Nathan reported, but not so serious - thanks.

 mm/memory.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/mm/memory.c b/mm/memory.c
index 4ec46eecefd3..cdadcff5ab26 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -2843,7 +2843,8 @@ static inline int __wp_page_copy_user(struct page *dst, struct page *src,
 			 * Other thread has already handled the fault
 			 * and update local tlb only
 			 */
-			update_mmu_tlb(vma, addr, vmf->pte);
+			if (vmf->pte)
+				update_mmu_tlb(vma, addr, vmf->pte);
 			ret = -EAGAIN;
 			goto pte_unlock;
 		}
@@ -2867,7 +2868,8 @@ static inline int __wp_page_copy_user(struct page *dst, struct page *src,
 		vmf->pte = pte_offset_map_lock(mm, vmf->pmd, addr, &vmf->ptl);
 		if (unlikely(!vmf->pte || !pte_same(*vmf->pte, vmf->orig_pte))) {
 			/* The PTE changed under us, update local tlb */
-			update_mmu_tlb(vma, addr, vmf->pte);
+			if (vmf->pte)
+				update_mmu_tlb(vma, addr, vmf->pte);
 			ret = -EAGAIN;
 			goto pte_unlock;
 		}
-- 
2.35.3