From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753972Ab0CKCW2 (ORCPT ); Wed, 10 Mar 2010 21:22:28 -0500 Received: from mail-pv0-f174.google.com ([74.125.83.174]:56062 "EHLO mail-pv0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753178Ab0CKCW1 (ORCPT ); Wed, 10 Mar 2010 21:22:27 -0500 DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:sender:date:x-google-sender-auth:message-id:subject :from:to:content-type; b=tOE/SqHtYhaebdJwwT/EaLkJ6EDMD2ReMpczeR+4a9XBMJmESBOWtQa3BVPbzOE8EJ 01UmsqRb6r43f0TpvdvHUafJBcFt4MvSMAZ0jQLtmKReNW8jnIGJcTEORZ9lOHi2Z5b6 RgcQdU27bB9NddcU4f1MnWH+AaX4ZO2Jzydbk= MIME-Version: 1.0 Date: Wed, 10 Mar 2010 21:22:23 -0500 X-Google-Sender-Auth: 0762c0370a8ac41b Message-ID: <40ec3ea41003101822q1f44b393y198d693c59792969@mail.gmail.com> Subject: Re: [regression]: soft lockup in dmesg after suspend/resume From: Chase Douglas To: linux-kernel@vger.kernel.org Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org I wanted to ask about potential work arounds for the issue noted at [1]. I wasn't subscribed to this list when that thread was started, so unfortunately I can't reply directly. I also haven't found any follow up to the thread, but please point me in the right direction if there has been any. We're seeing a number of reports against Ubuntu Lucid caused by timestamps being highly warped coming out of suspend/resume. When a warp occurs, the TSC register is throwing off timestamps with the upper 32 bits set to 0xFFFFFFFF. I'm not terribly familiar with the TSC or x86 clocking in general, but such time stamps should be many many years into the future before we see them (unless I've done my math wrong :). So with that in mind, it seems that it should be possible to test for such outlandish values coming out of suspend/resume. If the values are warped, would it be safe to switch over to using jiffies in native_sched_clock by setting tsc_disable=1 during runtime? Thanks, Chase [1] http://lkml.org/lkml/2010/1/4/7