From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S271270AbTG2GKd (ORCPT ); Tue, 29 Jul 2003 02:10:33 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S271278AbTG2GKd (ORCPT ); Tue, 29 Jul 2003 02:10:33 -0400 Received: from main.gmane.org ([80.91.224.249]:26084 "EHLO main.gmane.org") by vger.kernel.org with ESMTP id S271270AbTG2GKa (ORCPT ); Tue, 29 Jul 2003 02:10:30 -0400 X-Injected-Via-Gmane: http://gmane.org/ To: linux-kernel@vger.kernel.org From: Warren Turkal Subject: RE: PROBLEM: ACPI hangs when invoked from keyboard Date: Tue, 29 Jul 2003 01:06:08 -0500 Message-ID: References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7Bit X-Complaints-To: usenet@main.gmane.org User-Agent: KNode/0.7.2 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Grover, Andrew wrote: > This isn't ACPI, it's because 2.5.74+ force the APIC enabled. You are correct, and here is my addition to this bug report. Some acpi events appear to lock the system. svn revisions for pre-bug and post-bug 2.5.73 -> 11245 2.5.74 -> 11487 I will try to compile kernel revisions (highest_good+(lowest_bad-highest_good)/2) until I enocunter the bug. divide and conquer it 11245 good <- 2.5.73 11366 good 11426 good 11457 good 11472 good 11479 good 11483 good 11487 good <- 2.5.74 should have been bad; reason below After further investigation, the remaining revisions from 11483 to 11487 (2.5.74) did not look relavant to my situation. There was mainly an SH arch merge and two very minimal updates to md and the bumping of the version in the top Makefile. At this point, I relized that I had been compiling these kerenels without SMP, which is not what I did when I noticed the bug. Therefor, it is an SMP-kernel-only bug. I have a 1.9GHz Pentium 4M. My understanding is that it does not support hyperthreading. Why would I see an SMP bug at this point? divide and conquer with SMP kernel on 11245 good <- 2.5.73 11306 good 11313 good 11314 good 11315 bad 11317 bad 11321 bad 11336 bad 11366 bad 11487 bad <-- 2.5.74 Therefore, the breaking patch is the changeset from svn revion 11314 to 11315, which is reproduced below, including the patch comment. patch comment: wt@braindead:/usr/src/linux-trunk$ svn log -r 11315 ------------------------------------------------------------------------ rev 11315: mikpe | 2003-06-24 14:50:23 -0500 (Tue, 24 Jun 2003) | 17 lines [PATCH] enable local APIC on P4 The current local APIC code refuses to enable the local APIC on a P4 if the BIOS booted us with the local APIC disabled. This patch removes this unnecessary restriction. Please apply. Most P4 machines do boot with the local APIC enabled, but Keith Owens reported that the P4 based Compaq Evo N800v disables the local APIC, even though the machine actually works if Linux enables it. It is possible that some P4 machines with broken BIOSen were saved by our refusal to enable the local APIC. We can handle them via the DMI blacklist rules instead. BKrev: 3ef8ab7fg7Brw4y1sF1EaSr5fMHG7g ------------------------------------------------------------------------ wt@braindead:/usr/src/linux-trunk$ patch itself: wt@braindead:/usr/src/linux-trunk$ svn diff -r 11314:11315 Index: arch/i386/kernel/apic.c =================================================================== --- arch/i386/kernel/apic.c (revision 11314) +++ arch/i386/kernel/apic.c (revision 11315) @@ -616,7 +616,7 @@ goto no_apic; case X86_VENDOR_INTEL: if (boot_cpu_data.x86 == 6 || - (boot_cpu_data.x86 == 15 && cpu_has_apic) || + boot_cpu_data.x86 == 15 || (boot_cpu_data.x86 == 5 && cpu_has_apic)) break; goto no_apic; wt@braindead:/usr/src/linux-trunk$ Conclusion: Either there is a bug with supporting my type of APIC or my APIC must have a bug and is disabled by the BIOS for a reason. Either my hardware should be blacklisted or the old behavior should be restored. AFAIK, I have an Intel 845 chipset. I don't know what other info is needed for blacklisting. Thanks, Warren Turkal