mm: fix slab->page _count corruption when using slub

On arches that do not support this_cpu_cmpxchg_double() slab_lock is used
to do atomic cmpxchg() on double word which contains page->_count.  The
page count can be changed from get_page() or put_page() without taking
slab_lock.  That corrupts page counter.

Fix it by moving page->_count out of cmpxchg_double data.  So that slub
does no change it while updating slub meta-data in struct page.

[[email protected]: use standard comment layout, tweak comment text]
Reported-by: Amey Bhide <[email protected]>
Signed-off-by: Pravin B Shelar <[email protected]>
Acked-by: Christoph Lameter <[email protected]>
Cc: Pekka Enberg <[email protected]>
Cc: Andrea Arcangeli <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

include/linux/mm_types.h

@@ -57,8 +57,18 @@ struct page {
 		};
 
 		union {
+#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
+	defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
 			/* Used for cmpxchg_double in slub */
 			unsigned long counters;
+#else
+			/*
+			 * Keep _count separate from slub cmpxchg_double data.
+			 * As the rest of the double word is protected by
+			 * slab_lock but _count is not.
+			 */
+			unsigned counters;
+#endif
 
 			struct {