Message-ID: <alpine.DEB.1.00.0907162315450.17021@mail.selltech.ca>
Date:	Thu, 16 Jul 2009 23:32:33 -0700 (PDT)
From:	"Li, Ming Chun" <macli@....ubc.ca>
To:	KOSAKI Motohiro <kosaki.motohiro@...fujitsu.com>
Cc:	LKML <linux-kernel@...r.kernel.org>,
	"linux-mm@...ck.org" <linux-mm@...ck.org>
Subject: Re: [PATCH] mm: count only reclaimable lru pages

On Fri, 17 Jul 2009, KOSAKI Motohiro wrote:

> > On Thu, 16 Jul 2009, David Howells wrote:
> > 
> > > Rik van Riel <riel@...hat.com> wrote:
> > > 
> > > > It's part of a series of patches, including the three posted by Kosaki-san
> > > > last night (to track the number of isolated pages) and the patch I posted
> > > > last night (to throttle reclaim when too many pages are isolated).
> > > 
> > > Okay; Rik gave me a tarball of those patches, which I applied and re-ran the
> > > test.  The first run of msgctl11 produced lots of:
> > > 
> > > 	[root@...romeda ltp]# while ./testcases/bin/msgctl11; do :; done
> > 
> > I applied the series of patches on 2.6.31-rc3 and run 
> > 
> > while ./testcases/bin/msgctl11; do :; done 
> > 
> > four times, only got one OOM kill in the first round and the system is 
> > quite responsive all the time.
> > 
> > # while ./testcases/bin/msgctl11; do :; done
> > 
> > ---
> >  kernel: [  735.507878] msgctl11 invoked oom-killer: gfp_mask=0x84d0, order=0, oom_adj=0
> 
> GFP_KERNEL | __GFP_REPEAT __GFP_ZERO

ah, ./scripts/gfp-translate 0x84d0 __GFP_WAIT | __GFP_IO | __GFP_FS | 
__GFP_REPEAT | __GFP_ZERO
 
> 
> >  kernel: [  735.507884] msgctl11 cpuset=/ mems_allowed=0
> >  kernel: [  735.507888] Pid: 20631, comm: msgctl11 Not tainted 2.6.31-rc3-custom #1
> >  kernel: [  735.507891] Call Trace:
> >  kernel: [  735.507900]  [<c01ad781>] oom_kill_process+0x161/0x280
> >  kernel: [  735.507905]  [<c01adcd3>] ? select_bad_process+0x63/0xd0
> >  kernel: [  735.507909]  [<c01add8e>] __out_of_memory+0x4e/0xb0
> >  kernel: [  735.507913]  [<c01ade42>] out_of_memory+0x52/0xa0
> >  kernel: [  735.507917]  [<c01b0b07>] __alloc_pages_nodemask+0x4d7/0x4f0
> >  kernel: [  735.507922]  [<c01b0b77>] __get_free_pages+0x17/0x30
> >  kernel: [  735.507927]  [<c012baa6>] pgd_alloc+0x36/0x250
> >  kernel: [  735.507932]  [<c01f4ad3>] ? dup_fd+0x23/0x340
> >  kernel: [  735.507936]  [<c01422f7>] ? dup_mm+0x47/0x350
> >  kernel: [  735.507939]  [<c0141dd9>] mm_init+0xa9/0xe0
> >  kernel: [  735.507943]  [<c0142329>] dup_mm+0x79/0x350
> >  kernel: [  735.507947]  [<c01ffe22>] ? copy_fs_struct+0x22/0x90
> >  kernel: [  735.507951]  [<c01432d5>] ? copy_process+0xc75/0x1070
> >  kernel: [  735.507955]  [<c0143090>] copy_process+0xa30/0x1070
> >  kernel: [  735.507959]  [<c054b204>] ? schedule+0x494/0xa80
> >  kernel: [  735.507963]  [<c014373f>] do_fork+0x6f/0x330
> >  kernel: [  735.507968]  [<c014fdce>] ? recalc_sigpending+0xe/0x40
> >  kernel: [  735.507972]  [<c0107716>] sys_clone+0x36/0x40
> >  kernel: [  735.507976]  [<c0108dd4>] sysenter_do_call+0x12/0x28
> >  kernel: [  735.507979] Mem-Info:
> >  kernel: [  735.507981] DMA per-cpu:
> >  kernel: [  735.507983] CPU    0: hi:    0, btch:   1 usd:   0
> >  kernel: [  735.507986] CPU    1: hi:    0, btch:   1 usd:   0
> >  kernel: [  735.507988] Normal per-cpu:
> >  kernel: [  735.507990] CPU    0: hi:  186, btch:  31 usd:  17
> >  kernel: [  735.507993] CPU    1: hi:  186, btch:  31 usd: 180
> >  kernel: [  735.507994] HighMem per-cpu:
> >  kernel: [  735.507997] CPU    0: hi:   42, btch:   7 usd:  22
> >  kernel: [  735.507999] CPU    1: hi:   42, btch:   7 usd:   0
> >  kernel: [  735.508008] active_anon:82389 inactive_anon:2043 isolated_anon:32
> >  kernel: [  735.508009]  active_file:2201 inactive_file:5773 isolated_file:31
> >  kernel: [  735.508010]  unevictable:0 dirty:4 writeback:0 unstable:0 buffer:19
> >  kernel: [  735.508011]  free:1825 slab_reclaimable:655 slab_unreclaimable:19679
> >  kernel: [  735.508012]  mapped:1309 shmem:113 pagetables:66757 bounce:0
> 
> a lot free pages. but...
> 
> >  kernel: [  735.508020] DMA free:3520kB min:64kB low:80kB high:96kB active_anon:2240kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15832kB mlocked:0kB dirty:0kB writeback:0kB mapped:0kB shmem:0kB slab_reclaimable:0kB slab_unreclaimable:132kB kernel_stack:120kB pagetables:2436kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
> >  kernel: [  735.508026] lowmem_reserve[]: 0 867 998 998
> >  kernel: [  735.508035] Normal free:3632kB min:3732kB low:4664kB high:5596kB active_anon:269136kB inactive_anon:0kB active_file:56kB inactive_file:20kB unevictable:0kB isolated(anon):128kB isolated(file):124kB present:887976kB mlocked:0kB dirty:0kB writeback:0kB mapped:4kB shmem:0kB slab_reclaimable:2620kB slab_unreclaimable:78584kB kernel_stack:77328kB pagetables:227972kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:222 all_unreclaimable? no
> >  kernel: [  735.508042] lowmem_reserve[]: 0 0 1052 1052
> 
> DMA and Normal zone doesn't have enough free pages.
 
Caculate this way?  
DMA: 3520K < 64K + 998 * 4K
Normal: 3632k < 3732k + 1052 * 4k 

>
> >  kernel: [  735.508051] HighMem free:148kB min:128kB low:268kB high:408kB active_anon:58180kB inactive_anon:8172kB active_file:8748kB inactive_file:23072kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:134688kB mlocked:0kB dirty:16kB writeback:0kB mapped:5232kB shmem:452kB slab_reclaimable:0kB slab_unreclaimable:0kB kernel_stack:0kB pagetables:36620kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
> >  kernel: [  735.508057] lowmem_reserve[]: 0 0 0 0
> 
> HighMem zone only have enough free pages and reclaimable file cache pages.
> 
GFP_KERNEL | GFP_REPEAT | GFP_ZERO could not access HighMem free pages?

Vincent Li
Biomedical Research Center
University of British Columbia
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists