vmscan: add barrier to prevent evictable page in unevictable list

When a race between putback_lru_page() and shmem_lock with lock=0 happens,
progrom execution order is as follows, but clear_bit in processor #1 could
be reordered right before spin_unlock of processor #1.  Then, the page
would be stranded on the unevictable list.

spin_lock
SetPageLRU
spin_unlock
                                clear_bit(AS_UNEVICTABLE)
                                spin_lock
                                if PageLRU()
                                        if !test_bit(AS_UNEVICTABLE)
                                        	move evictable list
smp_mb
if !test_bit(AS_UNEVICTABLE)
        move evictable list
                                spin_unlock

But, pagevec_lookup() in scan_mapping_unevictable_pages() has
rcu_read_[un]lock() so it could protect reordering before reaching
test_bit(AS_UNEVICTABLE) on processor #1 so this problem never happens.
But it's a unexpected side effect and we should solve this problem
properly.

This patch adds a barrier after mapping_clear_unevictable.

I didn't meet this problem but just found during review.

Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik van Riel <riel@redhat.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Johannes Weiner <jweiner@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

2 files changed, 12 insertions, 5 deletions

diff --git a/mm/shmem.c b/mm/shmem.c
index 2d3577295298..fa4fa6ce13bc 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -1068,6 +1068,12 @@ int shmem_lock(struct file *file, int lock, struct user_struct *user)
                 user_shm_unlock(inode->i_size, user);
                 info->flags &= ~VM_LOCKED;
                 mapping_clear_unevictable(file->f_mapping);
+                /*
+                 * Ensure that a racing putback_lru_page() can see
+                 * the pages of this mapping are evictable when we
+                 * skip them due to !PageLRU during the scan.
+                 */
+                smp_mb__after_clear_bit();
                 scan_mapping_unevictable_pages(file->f_mapping);
         }
         retval = 0;
diff --git a/mm/vmscan.c b/mm/vmscan.c
index 3886b0bd7869..f51a33e8ed89 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -633,13 +633,14 @@ redo:
                 lru = LRU_UNEVICTABLE;
                 add_page_to_unevictable_list(page);
                 /*
-                 * When racing with an mlock clearing (page is
-                 * unlocked), make sure that if the other thread does
-                 * not observe our setting of PG_lru and fails
-                 * isolation, we see PG_mlocked cleared below and move
+                 * When racing with an mlock or AS_UNEVICTABLE clearing
+                 * (page is unlocked) make sure that if the other thread
+                 * does not observe our setting of PG_lru and fails
+                 * isolation/check_move_unevictable_page,
+                 * we see PG_mlocked/AS_UNEVICTABLE cleared below and move
                  * the page back to the evictable list.
                  *
-                 * The other side is TestClearPageMlocked().
+                 * The other side is TestClearPageMlocked() or shmem_lock().
                  */
                 smp_mb();
         }