From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755507AbZEEHkJ (ORCPT ); Tue, 5 May 2009 03:40:09 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752680AbZEEHjz (ORCPT ); Tue, 5 May 2009 03:39:55 -0400 Received: from casper.infradead.org ([85.118.1.10]:45047 "EHLO casper.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752159AbZEEHjy (ORCPT ); Tue, 5 May 2009 03:39:54 -0400 Subject: Re: [RFC][PATCH] vmscan: don't export nr_saved_scan in /proc/zoneinfo From: Peter Zijlstra To: Andrew Morton Cc: Wu Fengguang , torvalds@linux-foundation.org, kosaki.motohiro@jp.fujitsu.com, lee.schermerhorn@hp.com, riel@redhat.com, linux-mm@kvack.org, linux-kernel@vger.kernel.org, npiggin@suse.de, cl@linux-foundation.org In-Reply-To: <20090504144915.8d0716d7.akpm@linux-foundation.org> References: <200904302208.n3UM8t9R016687@imap1.linux-foundation.org> <20090501012212.GA5848@localhost> <20090430194907.82b31565.akpm@linux-foundation.org> <20090502023125.GA29674@localhost> <20090502024719.GA29730@localhost> <20090504144915.8d0716d7.akpm@linux-foundation.org> Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: Tue, 05 May 2009 09:38:10 +0200 Message-Id: <1241509090.11059.31.camel@twins> Mime-Version: 1.0 X-Mailer: Evolution 2.26.1 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Mon, 2009-05-04 at 14:49 -0700, Andrew Morton wrote: > On Sat, 2 May 2009 10:47:19 +0800 > Wu Fengguang wrote: > > > The lru->nr_saved_scan's are not meaningful counters for even kernel > > developers. They typically are smaller than 32 and are always 0 for > > large lists. So remove them from /proc/zoneinfo. > > > > Hopefully this interface change won't break too many scripts. > > /proc/zoneinfo is too unstructured to be script friendly, and I wonder > > the affected scripts - if there are any - are still bleeding since the > > not long ago commit "vmscan: split LRU lists into anon & file sets", > > which also touched the "scanned" line :) > > > > If we are to re-export accumulated vmscan counts in the future, they > > can go to new lines in /proc/zoneinfo instead of the current form, or > > to /sys/devices/system/node/node0/meminfo? > > > > /proc/zoneinfo is unsalvageable :( Shifting future work over to > /sys/devices/system/node/nodeN/meminfo and deprecating /proc/zoneinfo > sounds good to me. If only one could find things put in sysfs :-)