From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail143.messagelabs.com (mail143.messagelabs.com [216.82.254.35]) by kanga.kvack.org (Postfix) with ESMTP id 1FA956B0044 for ; Sun, 20 Dec 2009 07:00:50 -0500 (EST) Date: Sun, 20 Dec 2009 13:00:43 +0100 From: Attila Kinali Subject: Re: page allocation failure - still unfixed in 2.6.32.1 Message-Id: <20091220130043.dac8aa88.attila@kinali.ch> In-Reply-To: References: <20091220124721.006da86a.attila@kinali.ch> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux-mm@kvack.org To: Mikael Abrahamsson Cc: linux-mm@kvack.org List-ID: On Sun, 20 Dec 2009 12:52:23 +0100 (CET) Mikael Abrahamsson wrote: > > The page allocation failure that was introduced in 2.6.31 and > > which has been discussed here a few times, is still present in > > 2.6.32.1. I can still (more or less) reproduce it on my home-fileserver: > > > > Dec 20 11:43:14 koyomi kernel: swapper: page allocation failure. order:3, mode:0x20 > > Dec 20 11:43:14 koyomi kernel: Pid: 0, comm: swapper Not tainted 2.6.32.1 #1 > > Dec 20 11:43:14 koyomi kernel: Call Trace: > > Are you sure these are new? I've been seeing them sporadically for years, > this latest one is on 2.6.28. They seem to happen whenever there are lots > of TCP sessions going, and I have raised the default TCP parameters for > long latency performance. Yes. I didn't see any such problems until i updated to 2.6.30.x. And as the machine is mostly used for NFSv3, 99% of the traffic is UDP and not TCP. A quick count shows currently 17 TCP connections running. So i dont think i'd hit any limit there. Attila Kinali -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org