From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932908AbaGUOKd (ORCPT ); Mon, 21 Jul 2014 10:10:33 -0400 Received: from atrey.karlin.mff.cuni.cz ([195.113.26.193]:41837 "EHLO atrey.karlin.mff.cuni.cz" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932383AbaGUOK3 (ORCPT ); Mon, 21 Jul 2014 10:10:29 -0400 Date: Mon, 21 Jul 2014 16:10:27 +0200 From: Pavel Machek To: Joerg Roedel Cc: "Rafael J. Wysocki" , Len Brown , linux-pm@vger.kernel.org, linux-kernel@vger.kernel.org Subject: Re: [PATCH 0/6 v2] PM / Hibernate: Memory bitmap scalability improvements Message-ID: <20140721141027.GA16042@amd.pavel.ucw.cz> References: <1405938422-21900-1-git-send-email-joro@8bytes.org> <20140721120053.GA14069@amd.pavel.ucw.cz> <20140721123613.GL30979@8bytes.org> <20140721130629.GA15122@amd.pavel.ucw.cz> <20140721133800.GM30979@8bytes.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20140721133800.GM30979@8bytes.org> User-Agent: Mutt/1.5.20 (2009-06-14) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi! > > > That would just cover the problem that the bitmap data structure and the > > > algorithm in swsusp_free do not scale well on bigmem machines. > > > > And is it a problem? Hibernation of 12TB machine will take 6 hours if > > you back your swap with SSDs. > > > > Does not scale == burns additional 60 seconds of CPU time. I think we > > can live with that... > > Problem is that these 76s are burned every time, whether you just use > 500MB or the full 12TB of the machine. On 12TB machine, how much memory is used after boot finishes? I suspect it is not 500MB. > Next problem is that the bitmaps are allocated (and need to be freed) > without even being sure that a resume will happen. Could we fix that instead? Allocating bitmaps before we see hibernation signature is a bug. [How long does 12TB machine boot?] > > ...because noone sane will hibernate 12TB machine. > > And Linux is only made for sane people? Thats pretty new to me ;-) Yeah, but please don't optimize for insane people :-). > > Yes, that's why I propose to apply just patch 6 -- to avoid soft > > lockup warnings. > > Only patch 6 would wrap the problem with the soft lockups, but the other > patches actually improve the resume and boot situation on those > machines. Yeah. Provide timings for 12TB machine suspending/resuming and we may talk about it. #6 is bugfix. We should take it. Rest is performance improvement for insane config. I don't think we should take it. Pavel -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html