From mboxrd@z Thu Jan 1 00:00:00 1970 From: Wu Fengguang Subject: Re: OOM killer in 2.6.31-rc2 Date: Fri, 10 Jul 2009 21:18:01 +0800 Message-ID: <20090710131801.GA17773@localhost> References: <200907061056.00229.gene.heskett@verizon.net> <200907091042.38022.gene.heskett@verizon.net> <19030.22024.132029.196682@stoffel.org> <200907091703.06691.gene.heskett@verizon.net> <19031.15772.404288.544946@stoffel.org> Mime-Version: 1.0 Return-path: DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:received:received:date:from:to:cc :subject:message-id:references:mime-version:content-type :content-disposition:in-reply-to:user-agent; bh=scV2d+HOCJmGa58qnCVCH2hmeUJTZjFJcY2Kp4d607I=; b=bKyoj/S8hyF6xPhxmYT/LcDNmaZzb20NArIbIFhC5jY3YmHRAbZxIogoSn2wyQnSDK W0qYQwsipxiZKm7+R/IrBu+nvzpA6Z8xW8wiiZTWf5ufqBL/0vaGrUTNPeqkw6yKPcRo L55+GmHWGA7q+JNEVbgLZIMKt+DGvRaiLC0sU= Content-Disposition: inline In-Reply-To: <19031.15772.404288.544946-HgN6juyGXH5AfugRpC6u6w@public.gmane.org> Sender: kernel-testers-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-ID: Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: John Stoffel Cc: Gene Heskett , Linux Kernel list , "linux-mm-Bw31MaZKKs3YtjvyW6yDsg@public.gmane.org" , Minchan Kim , "Rafael J. Wysocki" , Kernel Testers List , David Howells , KOSAKI Motohiro On Fri, Jul 10, 2009 at 09:09:48AM -0400, John Stoffel wrote: > >>>>> "Gene" == Gene Heskett writes: > > Gene> On Thursday 09 July 2009, John Stoffel wrote: > >>>>>>> "Gene" == Gene Heskett writes: > >> > Gene> On Wednesday 08 July 2009, Wu Fengguang wrote: > >>>> On Wed, Jul 08, 2009 at 01:15:15PM +0800, Wu Fengguang wrote: > >>>>> On Tue, Jul 07, 2009 at 11:42:07PM -0400, Gene Heskett wrote: > >> > Gene> [...] > >> > >>>>> I guess your near 800MB slab cache is somehow under scanned. > >>>> > >>>> Gene, can you run .31 with this patch? When OOM happens, it will tell > >>>> us whether the majority slab pages are reclaimable. Another way to > >>>> find things out is to run `slabtop` when your system is moderately > >>>> loaded. > >> > Gene> Its been running continuously, and after 24 hours is now showing: > >> > >> Just wondering, is this your M2N-SLI Deluxe board? > Gene> Yes. > >> I've got the same > >> board, with 4Gb of RAM and I haven't noticed any loss of RAM from my > >> looking (quickly) at top output. > > Gene> I am short approximately 500 megs according to top: > Gene> Mem: 3634228k total, 3522984k used, 111244k free, 308096k buffers > Gene> Swap: 8385912k total, 568k used, 8385344k free, 2544716k cached > > Gene> From dmesg: > Gene> [ 0.000000] TOM2: 0000000120000000 aka 4608M Gene> [...] > Gene> [ 0.000000] 2694MB HIGHMEM available. > Gene> [ 0.000000] 887MB LOWMEM available. > > Gene> The bios signon does say 4092M IIRC. > > >> But I also haven't bothered to upgrade the BIOS on this board at all > >> since I got it back in March of 2008. No need in my book so far. > > Gene> I had been running the original bios, #1502, because 1604 and > Gene> 1701 had very poor uptimes. 1502 caused an oops about 15 lines > Gene> into the boot but that triggered a remap and it was bulletproof > Gene> after that running a 32 bit 64G+PAE kernel. (I haven't quite > Gene> made the jump to a 64 bit install, yet...) > > Why haven't you made the laep to 64bit yet? To me, that seems to be > the real solution here, not hacks like the HIGHMEM4G and HIGHMEM64G, > esp when your hardware is 64Bit by default. Sure 64bit kernel would be the best option for Gene :) > I've made the leap and I've never looked back. Haven't missed any > 32bit only apps, and if I really needed them, I'd just load the 32bit > libraries if need be. But for now I'd appreciate a lot if Gene can run a HIGHMEM64G kernel with the provided patch, so as to collect one full OOM message for us to analyze :) The previous OOM message missed the most important data from zone Normal.. Thanks, Fengguang