From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262409AbVHAHTp (ORCPT ); Mon, 1 Aug 2005 03:19:45 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262419AbVHAHTo (ORCPT ); Mon, 1 Aug 2005 03:19:44 -0400 Received: from gprs189-60.eurotel.cz ([160.218.189.60]:22246 "EHLO amd.ucw.cz") by vger.kernel.org with ESMTP id S262409AbVHAHTo (ORCPT ); Mon, 1 Aug 2005 03:19:44 -0400 Date: Mon, 1 Aug 2005 09:19:32 +0200 From: Pavel Machek To: Andrew Morton Cc: Shaohua Li , linux-kernel@vger.kernel.org, ambx1@neo.rr.com, torvalds@osdl.org, hugh@veritas.com, linux@dominikbrodowski.net, daniel.ritz@gmx.ch, len.brown@intel.com Subject: Re: revert yenta free_irq on suspend Message-ID: <20050801071932.GK27580@elf.ucw.cz> References: <2e00842e116e.2e116e2e0084@columbus.rr.com> <1122861542.2953.8.camel@linux-hp.sh.intel.com> <20050731190645.748f57e9.akpm@osdl.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20050731190645.748f57e9.akpm@osdl.org> X-Warning: Reading this can be dangerous to your mental health. User-Agent: Mutt/1.5.9i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Hi! > > > In general, I think that calling free_irq is the right behavior. > > > Although irqs changing after suspend is rare, there are also some > > > more serious issues. This has been discussed in the past, and a > > > summary is as follows: > > > > irqs actually isn't changed after suspend currently, it's a considering > > for future usage like hotplug. > > Calling free_irq actually isn't a complete ACPI issue, but ACPI requires > > it to solve nasty 'sleep in atomic' warning. > > Is that the only problem? If so, then surely we can make free_irq() run > happily with interrupts disabled: unlink the IRQ handler synchronously, > defer the /proc teardown or something like that. No, the problem is that a) restoring interrupt links needs interrupts enabled [or rewriting half of ACPI interpretter] b) to solve a) [and to solve other stuff, too], we need free_irq/request_irq all over the tree. > > You will find such break > > with swsusp without ACPI. Could we revert the ACPI change in Linus's > > tree but keep it in -mm tree? So we get a chance to fix drivers. > > That depends on the amount of brokenness involved: if it's significant then > I'll get a ton of bug reports concerning something which we already know is > broken and we'll drive away our long-suffering testers. The amount of brokenness is not that bad, and it fixes some machines, too. Pavel -- if you have sharp zaurus hardware you don't need... you know my address