Re: 2.6.26-rc5-mm2: OOM with 1G free swap

[Nick Piggin]

On Wednesday 11 June 2008 22:57, Rik van Riel wrote:
> On Tue, 10 Jun 2008 23:27:05 -0700
>
> Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
> > Well I assume that Rik ran LTP.  Perhaps a merge problem.
> >
> > Zero pages on active_anon and inactive_anon.  I suspect we lost those
> > pages.
>
> Known problem.  I fixed this one in the updates I sent you last night.

Oh good. Yeah I was just running some tests, and got as far as verifying
that the upstream kernel + lockless pagecache patches reclaims file pages
like a dream, but -mm2 sucks very badly at it.

During which, I also did find by inspection a little problem with my
speculative references patch. Andrew please apply this fix.

Fix the VM_BUG_ON assertion check to actually do what I want, noted by
Christoph.

Also, fix an error-path-leak type issue with frozen refcount not being
unfrozen. Found by review. In practice, this check is very rare to hit
because a page dirtier is likely to hold the refcount elevated for much
longer than it takes to check and non-racy-recheck. So it doesn't pose
a big problem for users of -mm, but of course needs fixing.

---
Index: linux-2.6/mm/vmscan.c
===================================================================
--- linux-2.6.orig/mm/vmscan.c	2008-06-11 23:36:07.000000000 +1000
+++ linux-2.6/mm/vmscan.c	2008-06-11 23:36:18.000000000 +1000
@@ -415,8 +415,10 @@ static int __remove_mapping(struct addre
 	if (!page_freeze_refs(page, 2))
 		goto cannot_free;
 	/* note: atomic_cmpxchg in page_freeze_refs provides the smp_rmb */
-	if (unlikely(PageDirty(page)))
+	if (unlikely(PageDirty(page))) {
+		page_unfreeze_refs(page, 2);
 		goto cannot_free;
+	}
 
 	if (PageSwapCache(page)) {
 		swp_entry_t swap = { .val = page_private(page) };
Index: linux-2.6/include/linux/pagemap.h
===================================================================
--- linux-2.6.orig/include/linux/pagemap.h	2008-06-11 23:36:07.000000000 +1000
+++ linux-2.6/include/linux/pagemap.h	2008-06-11 23:36:18.000000000 +1000
@@ -165,7 +165,7 @@ static inline int page_cache_get_specula
 		return 0;
 	}
 #endif
-	VM_BUG_ON(PageCompound(page) && (struct page *)page_private(page) != page);
+	VM_BUG_ON(PageTail(page));
 
 	return 1;
 }