From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S267294AbUJRSVW (ORCPT ); Mon, 18 Oct 2004 14:21:22 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S267335AbUJRSUl (ORCPT ); Mon, 18 Oct 2004 14:20:41 -0400 Received: from cantor.suse.de ([195.135.220.2]:52175 "EHLO Cantor.suse.de") by vger.kernel.org with ESMTP id S267341AbUJRSQ4 (ORCPT ); Mon, 18 Oct 2004 14:16:56 -0400 Date: Mon, 18 Oct 2004 20:16:54 +0200 From: Andi Kleen To: "Randy.Dunlap" Cc: bevand_m@epita.fr, linux-kernel@vger.kernel.org, discuss@x86-64.org Subject: Re: NMI watchdog detected lockup Message-Id: <20041018201654.58905384.ak@suse.de> In-Reply-To: <41740430.30604@osdl.org> References: <4172F91D.8090109@osdl.org> <4173F9A7.2090504@osdl.org> <20041018200017.0098710d.ak@suse.de> <41740430.30604@osdl.org> X-Mailer: Sylpheed version 0.9.11 (GTK+ 1.2.10; i686-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Mon, 18 Oct 2004 10:58:08 -0700 "Randy.Dunlap" wrote: > > Something on your system creates bogus NMI interrupts. What chipset > > are you using exactly? > > > > Sometimes chipsets can be programmed to raise NMIs when an PCI bus > > error occurs. > > > > 21 is the normal state (PIT timer running, but no errors logged) > > > > If you have an AMD 8131 it could be in theory erratum 54, but then > > normally one of the error bits in reason should be set. > > Yes, it's an AMD-8111 / 8131 / 8151 / K8-northbridge machine. It's probably one of your IO cards. I would remove them one by one or possibly switch them to different slots (PCI vs PCI-X) -Andi