From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932503AbWFHFLd (ORCPT ); Thu, 8 Jun 2006 01:11:33 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751287AbWFHFLd (ORCPT ); Thu, 8 Jun 2006 01:11:33 -0400 Received: from omx2-ext.sgi.com ([192.48.171.19]:17293 "EHLO omx2.sgi.com") by vger.kernel.org with ESMTP id S1751228AbWFHFLd (ORCPT ); Thu, 8 Jun 2006 01:11:33 -0400 X-Mailer: exmh version 2.7.0 06/18/2004 with nmh-1.1-RC1 From: Keith Owens To: "Brendan Trotter" cc: "Andi Kleen" , "Ashok Raj" , linux-kernel@vger.kernel.org Subject: Re: NMI problems with Dell SMP Xeons In-reply-to: Your message of "Wed, 07 Jun 2006 15:18:57 GMT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 08 Jun 2006 15:11:15 +1000 Message-ID: <6888.1149743475@kao2.melbourne.sgi.com> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org "Brendan Trotter" (on Wed, 7 Jun 2006 15:18:57 +0000) wrote: >Hi, > >On 6/7/06, Keith Owens wrote: >> This problem is not KDB specific, although that is where it was first >> noticed. Any code that sends a broadcast IPI 2 or an NMI IPI will >> crash these Dell boxes when there is a mismatch between the cpus known >> to the BIOS and the cpus known to the OS. > >I missed this (broadcast IPI 2 causing problems)... It turns out that there is an uncommented hack in x86_64's version of __prepare_ICR(). case NMI_VECTOR: icr |= APIC_DM_NMI; break; That hack is not in i386, which means that IPI 2 does different things in i386 vs. x86_64 mode on the same hardware. I am still tracking down all the differences before doing a clean up patch that makes i386 and x86_64 behave the same.