From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751053AbWGMRgk (ORCPT ); Thu, 13 Jul 2006 13:36:40 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751417AbWGMRgk (ORCPT ); Thu, 13 Jul 2006 13:36:40 -0400 Received: from ns.suse.de ([195.135.220.2]:7079 "EHLO mx1.suse.de") by vger.kernel.org with ESMTP id S1751053AbWGMRgk (ORCPT ); Thu, 13 Jul 2006 13:36:40 -0400 From: Andi Kleen To: Michal Schmidt Subject: Re: [patch] IDE: Touch NMI watchdog during resume from STR Date: Thu, 13 Jul 2006 19:36:34 +0200 User-Agent: KMail/1.9.3 Cc: linux-kernel@vger.kernel.org, Ingo Molnar References: <44B61D0A.7010305@stud.feec.vutbr.cz> <44B683EB.20709@stud.feec.vutbr.cz> In-Reply-To: <44B683EB.20709@stud.feec.vutbr.cz> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200607131936.34832.ak@suse.de> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Thursday 13 July 2006 19:33, Michal Schmidt wrote: > Andi Kleen wrote: > > Michal Schmidt writes: > >> if (stat == 0xff) > >> return -ENODEV; > >> touch_softlockup_watchdog(); > >> + touch_nmi_watchdog(); > > You can remove the touch_softlock_watchdog then. It's implied in > > touch_nmi_watchdog > > I don't think that's always true. There are architectures where > touch_nmi_watchdog is a NOP. This is in include/linux/nmi.h: > > #ifdef ARCH_HAS_NMI_WATCHDOG > extern void touch_nmi_watchdog(void); > #else > # define touch_nmi_watchdog() do { } while(0) > #endif That's broken code then. It should be defined to touch_softlock_watchdog for the !ARCH_HAS_NMI_WATCHDOG then. -Andi