page migration: fix NR_FILE_PAGES accounting
authorChristoph Lameter <clameter@sgi.com>
Mon, 23 Apr 2007 21:41:09 +0000 (14:41 -0700)
committerLinus Torvalds <torvalds@woody.linux-foundation.org>
Tue, 24 Apr 2007 15:23:08 +0000 (08:23 -0700)
NR_FILE_PAGES must be accounted for depending on the zone that the page
belongs to.  If we replace the page in the radix tree then we may have to
shift the count to another zone.

Suggested-by: Ethan Solomita <solo@google.com>
Eventually-typed-in-by: Christoph Lameter <clameter@sgi.com>
Cc: Martin Bligh <mbligh@mbligh.org>
Cc: <stable@kernel.org>
Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/migrate.c

index 7a66ca25dc8aa99b77f2a4f041fed0cd8194b6aa..a91ca00abebea3c7b0a8f0317cabfbc02f403d74 100644 (file)
@@ -297,7 +297,7 @@ static int migrate_page_move_mapping(struct address_space *mapping,
        void **pslot;
 
        if (!mapping) {
-               /* Anonymous page */
+               /* Anonymous page without mapping */
                if (page_count(page) != 1)
                        return -EAGAIN;
                return 0;
@@ -333,6 +333,19 @@ static int migrate_page_move_mapping(struct address_space *mapping,
         */
        __put_page(page);
 
+       /*
+        * If moved to a different zone then also account
+        * the page for that zone. Other VM counters will be
+        * taken care of when we establish references to the
+        * new page and drop references to the old page.
+        *
+        * Note that anonymous pages are accounted for
+        * via NR_FILE_PAGES and NR_ANON_PAGES if they
+        * are mapped to swap space.
+        */
+       __dec_zone_page_state(page, NR_FILE_PAGES);
+       __inc_zone_page_state(newpage, NR_FILE_PAGES);
+
        write_unlock_irq(&mapping->tree_lock);
 
        return 0;