From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1761849AbXGJB46 (ORCPT ); Mon, 9 Jul 2007 21:56:58 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1757270AbXGJB4v (ORCPT ); Mon, 9 Jul 2007 21:56:51 -0400 Received: from ozlabs.org ([203.10.76.45]:57197 "EHLO ozlabs.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756010AbXGJB4u (ORCPT ); Mon, 9 Jul 2007 21:56:50 -0400 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <18066.59225.533802.387340@cargo.ozlabs.ibm.com> Date: Tue, 10 Jul 2007 11:56:41 +1000 From: Paul Mackerras To: nigel@suspend2.net Cc: "Rafael J. Wysocki" , Matthew Garrett , Miklos Szeredi , linux-kernel@vger.kernel.org, Kyle Moffett , Pavel Machek , linux-pm@lists.linux-foundation.org Subject: Re: [linux-pm] Re: hibernation/snapshot design [was Re: [PATCH] Remove process freezer from suspend to RAM pathway] In-Reply-To: <200707101133.01566.nigel@nigel.suspend2.net> References: <20070708221315.GF5401@elf.ucw.cz> <200707090101.28832.rjw@sisk.pl> <200707101133.01566.nigel@nigel.suspend2.net> X-Mailer: VM 7.19 under Emacs 21.4.1 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Nigel Cunningham writes: > (Finally catching up on emails since Saturday!) Yes, I know what you mean; I was occupied with other work stuff on Friday and not well on Saturday, and by that stage there was a backlog of about 500 emails for me to wade through... :/ > The freezer is also essential for doing a full image of memory - you > have to be able to guarantee that some of the image won't change in > order to be able to write it out before the atomic copy. Of course the > freezer alone isn't enough, but it solves 95% of the issue. Sure. As I pointed out in another email, it's unrealistic to expect the freezer to be able to stop everything in a state where any arbitrary mutex is free. But as long as you don't need to do anything that depends on getting a mutex while the system is frozen, there should be no problem. Paul.