From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S266069AbUHWUGl (ORCPT ); Mon, 23 Aug 2004 16:06:41 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S266073AbUHWUEj (ORCPT ); Mon, 23 Aug 2004 16:04:39 -0400 Received: from mxfep02.bredband.com ([195.54.107.73]:16291 "EHLO mxfep02.bredband.com") by vger.kernel.org with ESMTP id S266566AbUHWSpo (ORCPT ); Mon, 23 Aug 2004 14:45:44 -0400 Subject: Re: 2.6.8.1-mm IRQ routing problems From: Alexander Nyberg To: Bjorn Helgaas Cc: Andrew Morton , "Randy.Dunlap" , linux-kernel@vger.kernel.org In-Reply-To: <200408231016.36318.bjorn.helgaas@hp.com> References: <1093088008.777.13.camel@boxen> <20040822180911.22bbbc96.akpm@osdl.org> <1093264936.834.1.camel@boxen> <200408231016.36318.bjorn.helgaas@hp.com> Content-Type: text/plain Message-Id: <1093286742.868.3.camel@boxen> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Mon, 23 Aug 2004 20:45:42 +0200 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Mon, 2004-08-23 at 18:16, Bjorn Helgaas wrote: > On Monday 23 August 2004 6:42 am, Alexander Nyberg wrote: > > On Mon, 2004-08-23 at 03:09, Andrew Morton wrote: > > > Alexander Nyberg wrote: > > > > > > > > Using 2.6.8.1-mm3 I ran into some problems on x86_64. This > > > > only happens when fsck runs at bootup because in my case > > > > of the max-mount-count being reached (I use ext3). Booting > > > > with pci=routeirq makes problem go away. > > > > > > > > Do I win the weird problem prize? > > > > > > I think this was fixed in -mm4. Please retest. > > > > Still happens in -mm4. > > Can you double-check this, and perhaps post the dmesg for the 2.6.8.1-mm4 > attempt? It still looks very much like the problem Randy fixed here: > > http://marc.theaimsgroup.com/?l=linux-kernel&m=109313574928853&w=2 > > I just checked, and Randy's patch is indeed in 2.6.8.1-mm4. If the oops > still occurs there, it must be a different problem, and the dmesg might > help diagnose it. > Hi, This fixes things for me. Signed-off-by: Alexander Nyberg --- linux-2.6.7/arch/x86_64/kernel/io_apic.c_orig 2004-08-23 18:32:10.000000000 +0200 +++ linux-2.6.7/arch/x86_64/kernel/io_apic.c 2004-08-23 20:42:33.236703256 +0200 @@ -80,7 +80,7 @@ int vector_irq[NR_VECTORS] = { [0 ... NR * shared ISA-space IRQs, so we have to support them. We are super * fast in the common case, and fast for shared ISA-space IRQs. */ -static void __init add_pin_to_irq(unsigned int irq, int apic, int pin) +static void add_pin_to_irq(unsigned int irq, int apic, int pin) { static int first_free_entry = NR_IRQS; struct irq_pin_list *entry = irq_2_pin + irq; @@ -656,11 +656,7 @@ static inline int IO_APIC_irq_trigger(in /* irq_vectors is indexed by the sum of all RTEs in all I/O APICs. */ u8 irq_vector[NR_IRQ_VECTORS] = { FIRST_DEVICE_VECTOR , 0 }; -#ifdef CONFIG_PCI_MSI int assign_irq_vector(int irq) -#else -int __init assign_irq_vector(int irq) -#endif { static int current_vector = FIRST_DEVICE_VECTOR, offset = 0;