page allocator: do not check for compound pages during the page allocator sanity...
authorNick Piggin <nickpiggin@yahoo.com.au>
Tue, 16 Jun 2009 22:32:10 +0000 (15:32 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 17 Jun 2009 02:47:34 +0000 (19:47 -0700)
A number of sanity checks are made on each page allocation and free
including that the page count is zero.  page_count() checks for compound
pages and checks the count of the head page if true.  However, in these
paths, we do not care if the page is compound or not as the count of each
tail page should also be zero.

This patch makes two changes to the use of page_count() in the free path.
It converts one check of page_count() to a VM_BUG_ON() as the count should
have been unconditionally checked earlier in the free path.  It also
avoids checking for compound pages.

[mel@csn.ul.ie: Wrote changelog]
Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Nick Piggin <nickpiggin@yahoo.com.au>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/page_alloc.c

index fd8e3ca0cf3b9aeee96bb70dd9f5c33bffbbc335..8485735fc6906dcccb8117382e4170a15f601507 100644 (file)
@@ -421,7 +421,7 @@ static inline int page_is_buddy(struct page *page, struct page *buddy,
                return 0;
 
        if (PageBuddy(buddy) && page_order(buddy) == order) {
-               BUG_ON(page_count(buddy) != 0);
+               VM_BUG_ON(page_count(buddy) != 0);
                return 1;
        }
        return 0;
@@ -497,7 +497,7 @@ static inline int free_pages_check(struct page *page)
 {
        if (unlikely(page_mapcount(page) |
                (page->mapping != NULL)  |
-               (page_count(page) != 0)  |
+               (atomic_read(&page->_count) != 0) |
                (page->flags & PAGE_FLAGS_CHECK_AT_FREE))) {
                bad_page(page);
                return 1;
@@ -642,7 +642,7 @@ static int prep_new_page(struct page *page, int order, gfp_t gfp_flags)
 {
        if (unlikely(page_mapcount(page) |
                (page->mapping != NULL)  |
-               (page_count(page) != 0)  |
+               (atomic_read(&page->_count) != 0)  |
                (page->flags & PAGE_FLAGS_CHECK_AT_PREP))) {
                bad_page(page);
                return 1;