From mboxrd@z Thu Jan 1 00:00:00 1970 From: Johannes Stezenbach Subject: Re: linux-3.2.1 + KVM: TSC unstable on Core i5-2400S / H67 Date: Thu, 26 Jan 2012 16:35:13 +0100 Message-ID: <20120126153513.GA2963@sig21.net> References: <20120124113608.GA16876@sig21.net> <20120125122702.GA30888@amt.cnet> <20120125132503.GA3399@sig21.net> <20120126094659.GA25282@sig21.net> <4F21668F.8080005@siemens.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Marcelo Tosatti , kvm@vger.kernel.org, linux-kernel@vger.kernel.org To: Jan Kiszka Return-path: Received: from bar.sig21.net ([80.81.252.164]:32878 "EHLO bar.sig21.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751483Ab2AZPfU (ORCPT ); Thu, 26 Jan 2012 10:35:20 -0500 Content-Disposition: inline In-Reply-To: <4F21668F.8080005@siemens.com> Sender: kvm-owner@vger.kernel.org List-ID: On Thu, Jan 26, 2012 at 03:43:27PM +0100, Jan Kiszka wrote: > On 2012-01-26 10:46, Johannes Stezenbach wrote: > > [16353.270435] Restarting tasks ... > > [16353.270612] BUG: spinlock bad magic on CPU#2, dhclient3/2590 > > [16353.274702] lock: ffff880127b9ce08, .magic: 00000000, .owner: dhclient3/2590, .owner_cpu: 2 > > [16353.276594] Pid: 2590, comm: dhclient3 Not tainted 3.2.1 #2 > > [16353.278294] done. > > [16353.278314] video LNXVIDEO:00: Restoring backlight state > > [16353.281627] Call Trace: > > [16353.281633] [] spin_dump+0x8a/0x8f > > [16353.281636] [] spin_bug+0x26/0x28 > > [16353.281639] [] do_raw_spin_unlock+0x1e/0x83 > > [16353.281643] [] _raw_spin_unlock_irq+0x28/0x5b > > [16353.281647] [] get_signal_to_deliver+0x546/0x57c > > [16353.281651] [] do_signal+0x3e/0x62f > > [16353.281654] [] ? retint_restore_args+0x13/0x13 > > [16353.281657] [] ? copy_user_generic_string+0x19/0x40 > > [16353.281661] [] ? poll_select_copy_remaining+0xc5/0xe9 > > [16353.281664] [] do_notify_resume+0x2c/0x53 > > [16353.281666] [] ? trace_hardirqs_on_thunk+0x3a/0x3f > > [16353.281669] [] int_signal+0x12/0x17 > > [16353.472289] tsc = fffffd60248a9f39 last_guest_tsc = 229a823d4b88 > > [ 0.006666] Marking TSC unstable due to KVM discovered backwards TSC > > [16362.155512] Switching to clocksource hpet > > [16363.617844] atl1c 0000:06:00.0: atl1c: eth0 NIC Link is Up<1000 Mbps Full Duplex> > > [16363.618145] br0: port 1(eth0) entering forwarding state > > [16363.618153] br0: port 1(eth0) entering forwarding state > > Are you suspending your host while a VM is running? There is mosto Yes, that's what I'm doing. > probably a long-pending bug in the KVM TSC check logic that declares > that clocksource unstable after resume. I'm seeing this here for ages > and complained about it from the beginning (but I never debugged it :-/). Hm, OK. I hope someone has an idea how to fix it. I'm ready to test patches. > Of course, this doesn't explain your corrupted spin lock. That might be > a different issue. Yes, most probably it is something different since I haven't seen this before. However, I didn't use 3.2.1 long. Now I updated to 3.2.2 and enabled a few more debug options. We'll see if it triggers again. Thanks Johannes