[patch] reproducible athlon mce fix

From: Geoffrey Lee <glee@gnupilgrims.org>
To: linux-kernel@vger.kernel.org, davej@redhat.com
Subject: [patch] reproducible athlon mce fix
Date: Sun, 2 Nov 2003 13:57:48 +0800	[thread overview]
Message-ID: <20031102055748.GA1218@anakin.wychk.org> (raw)

[-- Attachment #1: Type: text/plain, Size: 660 bytes --]

Hi all,


After switching from 2.4.22 to 2.6.0-test9, I have received reproducible
MCE non-fatal error check messages in my kernel log.  (For example, one
shows up right after my first scsi card init).

>From Dave Jones' patch here:

http://lists.insecure.org/lists/linux-kernel/2003/Sep/7362.html

and another message here:

http://lkml.org/lkml/2003/10/7/214

would seem to imply that Athlons don't like having their Bank 0 poked at,
though that's what non-fatal.c does.  Would it be correct to make sure
that that non-fatal.c starts at bank 1, if it is an Athlon?

Dave, is the following patch correct?  Booted and tested, no ill effects
so far ...


	- g.

[-- Attachment #2: mce-fix.patch --]
[-- Type: text/plain, Size: 428 bytes --]

--- linux-2.6.0-test9/arch/i386/kernel/cpu/mcheck/non-fatal.c.orig	2003-11-02 13:31:43.000000000 +0800
+++ linux-2.6.0-test9/arch/i386/kernel/cpu/mcheck/non-fatal.c	2003-11-02 13:34:37.000000000 +0800
@@ -30,7 +30,11 @@
 	int i;
 
 	preempt_disable(); 
+#if CONFIG_MK7
+	for (i=1; i<nr_mce_banks; i++) {
+#else
 	for (i=0; i<nr_mce_banks; i++) {
+#endif
 		rdmsr (MSR_IA32_MC0_STATUS+i*4, low, high);
 
 		if (high & (1<<31)) {
