mm: vmscan: fix typo in isolating lru pages

It is not the tag page but the cursor page that we should process, and it
looks a typo.

Signed-off-by: Hillf Danton <dhillf@gmail.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Hugh Dickins <hughd@google.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/vmscan.c

@@ -1173,7 +1173,7 @@ static unsigned long isolate_lru_pages(unsigned long nr_to_scan,
 			if (__isolate_lru_page(cursor_page, mode, file) == 0) {
 				list_move(&cursor_page->lru, dst);
 				mem_cgroup_del_lru(cursor_page);
-				nr_taken += hpage_nr_pages(page);
+				nr_taken += hpage_nr_pages(cursor_page);
 				nr_lumpy_taken++;
 				if (PageDirty(cursor_page))
 					nr_lumpy_dirty++;