From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail190.messagelabs.com (mail190.messagelabs.com [216.82.249.51]) by kanga.kvack.org (Postfix) with SMTP id 96A348D000E for ; Wed, 1 Dec 2010 20:28:53 -0500 (EST) Received: from m1.gw.fujitsu.co.jp ([10.0.50.71]) by fgwmail6.fujitsu.co.jp (Fujitsu Gateway) with ESMTP id oB21SoM6025763 for (envelope-from kosaki.motohiro@jp.fujitsu.com); Thu, 2 Dec 2010 10:28:51 +0900 Received: from smail (m1 [127.0.0.1]) by outgoing.m1.gw.fujitsu.co.jp (Postfix) with ESMTP id 662E945DE55 for ; Thu, 2 Dec 2010 10:28:50 +0900 (JST) Received: from s1.gw.fujitsu.co.jp (s1.gw.fujitsu.co.jp [10.0.50.91]) by m1.gw.fujitsu.co.jp (Postfix) with ESMTP id 49DBB45DE5C for ; Thu, 2 Dec 2010 10:28:50 +0900 (JST) Received: from s1.gw.fujitsu.co.jp (localhost.localdomain [127.0.0.1]) by s1.gw.fujitsu.co.jp (Postfix) with ESMTP id 3C8F7E38001 for ; Thu, 2 Dec 2010 10:28:50 +0900 (JST) Received: from m107.s.css.fujitsu.com (m107.s.css.fujitsu.com [10.249.87.107]) by s1.gw.fujitsu.co.jp (Postfix) with ESMTP id D7C6BE38005 for ; Thu, 2 Dec 2010 10:28:49 +0900 (JST) From: KOSAKI Motohiro Subject: Re: [patch]vmscan: make kswapd use a correct order In-Reply-To: <1291249749.12777.86.camel@sli10-conroe> References: <20101201155854.GA3372@barrios-desktop> <1291249749.12777.86.camel@sli10-conroe> Message-Id: <20101202101555.157C.A69D9226@jp.fujitsu.com> MIME-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 8bit Date: Thu, 2 Dec 2010 10:28:49 +0900 (JST) Sender: owner-linux-mm@kvack.org To: Shaohua Li Cc: kosaki.motohiro@jp.fujitsu.com, Minchan Kim , linux-mm , Andrew Morton , Mel Gorman List-ID: > > In addtion, new order is always less than old order in that context. > > so big order page reclaim makes much safe for low order pages. > big order page reclaim makes we have more chances to reclaim useful > pages by lumpy, why it's safe? Because some crappy driver try high order GFP_ATOMIC allocation and they often don't have enough failure handling. Even though they have good allocation failure handling, network packet loss (wireless drivers are one of most big high order GFP_ATOMIC user) is usually big impact than page cache drop/re-readings. Thanks. -- 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/ . Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/ Don't email: email@kvack.org