From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261546AbVHBOsr (ORCPT ); Tue, 2 Aug 2005 10:48:47 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S261556AbVHBOqn (ORCPT ); Tue, 2 Aug 2005 10:46:43 -0400 Received: from atrey.karlin.mff.cuni.cz ([195.113.31.123]:63920 "EHLO atrey.karlin.mff.cuni.cz") by vger.kernel.org with ESMTP id S261546AbVHBOpX (ORCPT ); Tue, 2 Aug 2005 10:45:23 -0400 Date: Tue, 2 Aug 2005 16:45:16 +0200 From: Pavel Machek To: "Eric W. Biederman" Cc: Benjamin Herrenschmidt , Andrew Morton , Linux Kernel list Subject: Re: Calling suspend() in halt/restart/shutdown -> not a good idea Message-ID: <20050802144516.GC2465@atrey.karlin.mff.cuni.cz> References: <1122908972.18835.153.camel@gaston> <20050802095312.GA1442@elf.ucw.cz> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.6+20040907i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Hi! > >> Why are we calling driver suspend routines in these ? This is _not_ > > > > Well, reason is that if you remove device_suspend() you'll get > > emergency hard disk park during powerdown. As harddrives can survive > > only limited number of emergency stops, that is not a good idea. > > Then the practical question is: do we suspend the disk by > calling device_suspend() for every device. Or do we modify > the ->shutdown() method for the disk. The additional data in pm_message_t are usefull, and sharing code between suspend-to-ram and suspend-to-disk is usefull => option #1... Pavel -- Boycott Kodak -- for their patent abuse against Java.