From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1030369AbWGaW41 (ORCPT ); Mon, 31 Jul 2006 18:56:27 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1030370AbWGaW41 (ORCPT ); Mon, 31 Jul 2006 18:56:27 -0400 Received: from gprs189-60.eurotel.cz ([160.218.189.60]:57825 "EHLO amd.ucw.cz") by vger.kernel.org with ESMTP id S1030369AbWGaW41 (ORCPT ); Mon, 31 Jul 2006 18:56:27 -0400 Date: Tue, 1 Aug 2006 00:53:08 +0200 From: Pavel Machek To: Nathan Scott Cc: kernel list , "Rafael J. Wysocki" , xfs@oss.sgi.com Subject: Re: XFS vs. swsusp Message-ID: <20060731225308.GA4000@elf.ucw.cz> References: <20060731215933.GB3612@elf.ucw.cz> <20060801084143.D2286470@wobbly.melbourne.sgi.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20060801084143.D2286470@wobbly.melbourne.sgi.com> X-Warning: Reading this can be dangerous to your mental health. User-Agent: Mutt/1.5.11+cvs20060126 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Hi! > > Rafael has patches to add bdev freezing to swsusp. I'd like to know if > > they are neccessary (and why). > > > > 1) Is sync() enough to guarantee that all the data written before sync > > actually reach the platters? > > > > (Or is it that data only reach the journal? OTOH that would be okay, too). > > It ensures file data reaches its final resting place, and that > metadata changes have been logged. It does not necessarily .... Okay, good, being safely in the journal is okay. > > 2) If we stop all the user proceses and all the kernel threads, is > > that enough to prevent XFS from writing to disk? > > Yes, I believe so (if all user processes and kernel threads are > stopped, who else would be left to initiate I/O?). There is a Well, we were afraid that you'd do it from timer interrupt or something like that. > timer driven wakeup done on the per-fs xfssyncd kernel threads, > which do background metadata writeout and will write to the log > periodically... but if those processes are all stopped too, you > should be OK. Timer only wakes up xfssyncd thread, right? That's okay, as that thread will be stopped. Rafael, I do not think we need bdev freezing changes for XFS. Pavel -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html