8 messages in com.xensource.lists.xen-develRE: [Xen-devel] Timer going backwards...
FromSent OnAttachments
Robert Hulme16 May 2007 01:21 
Ian Pratt28 May 2007 13:03 
Jan Beulich29 May 2007 00:00 
Jan Beulich30 May 2007 08:20 
Keir Fraser30 May 2007 08:39 
Jan Beulich30 May 2007 09:02 
Keir Fraser30 May 2007 09:24 
Keir Fraser30 May 2007 10:30 
Subject:RE: [Xen-devel] Timer going backwards and Unable to handle kernel NULLpointer
From:Ian Pratt (Ian.@cl.cam.ac.uk)
Date:05/28/2007 01:03:20 PM
List:com.xensource.lists.xen-devel

I've got a Thunder K8SRE with two dual core Opteron processors and eight 1 GB SDRAM sticks in it. Booting in to SMP mode resulted in this:

May 16 00:43:37 weebl kernel: Timer ISR/3: Time went backwards: delta=-12967620 delta_cpu=217015619 shadow=46263756493 off=423276508 processed=46700000000 cpu_processed=46470016761 May 16 00:43:37 weebl kernel: 0: 46680000000 May 16 00:43:37 weebl kernel: 1: 46700016761 May 16 00:43:37 weebl kernel: 2: 45670016761 May 16 00:43:37 weebl kernel: 3: 46470016761

The only time I've seen this is on a system that was overheating and bouncing in and out of thermal throttling. Any chance that could be happening here?

May 16 06:28:35 weebl kernel: Call Trace: <ffffffff8011a275>{mm_unpin+427} <ffffffff8011a4a4>{_arch_exit_mmap+476} May 16 06:28:35 weebl kernel: <ffffffff8014fedd>{generic_file_aio_read+52} <ffffffff801724a6>{do_sync_read+199} May 16 06:28:35 weebl kernel: <ffffffff801612d2>{exit_mmap+34} <ffffffff80128e05>{mmput+40} May 16 06:28:35 weebl kernel:

That's a nasty crash. There were some cleanups in this area in Xen 3.1, but I don't think it was to address a specific crash like this. Might be worth trying 3.1 on the machine.

Best, Ian