From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S264912AbUELJln (ORCPT ); Wed, 12 May 2004 05:41:43 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S264921AbUELJln (ORCPT ); Wed, 12 May 2004 05:41:43 -0400 Received: from grendel.digitalservice.pl ([217.67.200.140]:31139 "HELO mail.digitalservice.pl") by vger.kernel.org with SMTP id S264912AbUELJll (ORCPT ); Wed, 12 May 2004 05:41:41 -0400 From: "R. J. Wysocki" Organization: SiSK To: Fernando Paredes , Linux Kernel Mailing List Subject: Re: Toshiba keyboard lockups Date: Wed, 12 May 2004 11:49:37 +0200 User-Agent: KMail/1.5 References: <40A162BA.90407@sun.com> In-Reply-To: <40A162BA.90407@sun.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-2" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200405121149.37334.rjwysocki@sisk.pl> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Wednesday 12 of May 2004 01:33, Fernando Paredes wrote: > There was a previous thread on this, last month. > > I updated to 2.6.6 and I still get these random lockups. Nothing in > dmesg or /var/log/messages. Too annoying as I have to reboot the machine > constantly. Does anyone know the status on this? Is t a toshiba hardware > bug (is that possible?) or a bug in serio.c or keybd.c? It's most probably Toshiba-related, because it does not happen on other hardware, it seems. I've got a simple patch from Grzegorz Kulewski to help trace the problem, but I haven't got a lockup since. The patch is as follows: --- /usr/src/linux-2.6.5/drivers/input/serio/serio.c.orig 2004-04-04 05:36:15.000000000 +0200 +++ /usr/src/linux-2.6.5/drivers/input/serio/serio.c 2004-04-09 18:28:50.268521936 +0200 @@ -166,6 +166,11 @@ static int serio_thread(void *nothing) static void serio_queue_event(struct serio *serio, int event_type) { struct serio_event *event; + + if (event_type == SERIO_RESCAN || event_type == SERIO_RECONNECT) { + printk(KERN_WARNING "serio: RESCAN || RECONNECT requested: %d!\n", event_type); + dump_stack(); + } if ((event = kmalloc(sizeof(struct serio_event), GFP_ATOMIC))) { event->type = event_type; Please try to apply it and you should get something in the logs when a lockup occurs (ie. kernel warning + call trace). RJW