Re: 2.6.20-mm1 [kernel BUG at mm/swap.c:442]

From: Christoph Lameter
Date: Fri Feb 16 2007 - 10:57:40 EST


On Fri, 16 Feb 2007, James Morris wrote:

> Then, I get this reliably as ntpd starts up:
> [ 92.905514] [<c014e7ff>] lru_add_drain+0x57/0x8d
> [ 92.905519] [<c015a03f>] free_pages_and_swap_cache+0x12/0x85
> [ 92.905526] [<c0155acb>] unmap_region+0xfd/0x129
> [ 92.905530] [<c0156404>] do_munmap+0x153/0x1b4
> [ 92.905534] [<c015648a>] sys_munmap+0x25/0x34
> [ 92.905538] [<c01028fc>] syscall_call+0x7/0xb
> [ 92.905542] =======================
> [ 92.905544] Code: 74 1a 85 d2 74 0b 8d 82 80 05 00 00 e8 6a 3a 1a 00 8d 86 80 05 00 00 e8 7c 36 1a 00 8b 03 a9 00 00 10 00 74
> 3f 8b 03 a8 20 74 04 <0f> 0b eb fe f0 0f ba 2b 05 f0 0f ba 33 14 f0 0f ba 2b 06 ba 03

Andrew already has this fix which cures it for me. PG_mlocked pages can
be freed in some situations and thus we need the correct handling in the
page allocator:


Index: linux-2.6.20/include/linux/page-flags.h
===================================================================
--- linux-2.6.20.orig/include/linux/page-flags.h 2007-02-15 20:42:42.000000000 -0800
+++ linux-2.6.20/include/linux/page-flags.h 2007-02-15 20:43:33.000000000 -0800
@@ -261,6 +261,7 @@ static inline void SetPageUptodate(struc
#define PageMlocked(page) test_bit(PG_mlocked, &(page)->flags)
#define SetPageMlocked(page) set_bit(PG_mlocked, &(page)->flags)
#define ClearPageMlocked(page) clear_bit(PG_mlocked, &(page)->flags)
+#define __ClearPageMlocked(page) __clear_bit(PG_mlocked, &(page)->flags)

struct page; /* forward declaration */

Index: linux-2.6.20/mm/page_alloc.c
===================================================================
--- linux-2.6.20.orig/mm/page_alloc.c 2007-02-15 20:42:42.000000000 -0800
+++ linux-2.6.20/mm/page_alloc.c 2007-02-15 20:55:23.000000000 -0800
@@ -203,6 +203,7 @@ static void bad_page(struct page *page)
1 << PG_slab |
1 << PG_swapcache |
1 << PG_writeback |
+ 1 << PG_mlocked |
1 << PG_buddy );
set_page_count(page, 0);
reset_page_mapcount(page);
@@ -442,6 +443,11 @@ static inline int free_pages_check(struc
bad_page(page);
if (PageDirty(page))
__ClearPageDirty(page);
+ if (PageMlocked(page)) {
+ /* Page is unused so no need to take the lru lock */
+ __ClearPageMlocked(page);
+ dec_zone_page_state(page, NR_MLOCK);
+ }
/*
* For now, we report if PG_reserved was found set, but do not
* clear it, and do not free the page. But we shall soon need
@@ -588,6 +594,7 @@ static int prep_new_page(struct page *pa
1 << PG_swapcache |
1 << PG_writeback |
1 << PG_reserved |
+ 1 << PG_mlocked |
1 << PG_buddy ))))
bad_page(page);

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/