* KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> wrote:

quoted text
> > but it appears we dont do quicklist trimming anywhere else! So if a 
> > system has no idle time, the quicklist can grow unbounded, and 
> > that's a real memory leak IMO.
> 
> I test following method.
> 
> 1. $ hackbench 100 process 1000
> 2. $ cat /proc/meminfo
> 
> quicklist consume 1GB memory of 8GB total memory system.
> it seems too large cache ;)
> 
> IMHO we need shrink pgtable cache mecanism.

ouch! Could you try the patch below? How large is the quicklist cache 
with this applied?

	Ingo

---------------------->
Subject: x86: trim quicklist more agressively
From: Ingo Molnar <mingo@elte.hu>
Date: Sun Mar 09 12:59:41 CET 2008

trim the quicklists more agressively, up to 1024 pages at once.
(which pretty much means we keep this special-purpose cache as
small as possible)

Signed-off-by: Ingo Molnar <mingo@elte.hu>
---
 arch/x86/mm/pgtable_32.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Index: linux-x86.q/arch/x86/mm/pgtable_32.c
===================================================================
--- linux-x86.q.orig/arch/x86/mm/pgtable_32.c
+++ linux-x86.q/arch/x86/mm/pgtable_32.c
@@ -358,7 +358,7 @@ void pgd_free(struct mm_struct *mm, pgd_
 
 void check_pgt_cache(void)
 {
-	quicklist_trim(0, pgd_dtor, 25, 16);
+	quicklist_trim(0, pgd_dtor, 25, 1024);
 }
 
 void __pte_free_tlb(struct mmu_gather *tlb, struct page *pte)
--
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/