Re: [BUG] cpu hotplug vs scheduler

[Heiko Carstens <heiko.carstens@de.ibm.com>]

On Tue, May 13, 2008 at 06:33:00PM +0300, Avi Kivity wrote:
> Avi Kivity wrote:
>> I'm testing host cpu hotplug with kvm.  Basically running 7 guests on  
>> a 4 core machine, offlining and onlining host cpus at random.   
>> Eventually I hit this:
>>
>
> [snip oops]
>
>> This seems to be the assignment to cfs_rq after pick_next_entity().
>>
>> I'm running kvm.git, which is currently 2.6.26-rc2 plus a few kvm  
>> patches.  It could be kvm's fault, but it doesn't appear so from the  
>> traces.
>>
>
> I get the same oops running with the kvm modules unloaded.  None of  
> kvm.git's changes are in core code, so this is definitely a mainline  
> problem.
>
> It took about 40 minutes to reproduce; 1200 cpus onlined and 1200 cpus  
> offlined.

FWIW, this happens on s390 as well (without kvm support). But until now
I didn't have time to look into it. Didn't report it here yet, since we
had plenty of cpu hotplug changes in arch code, so I suspected a bug there.
This doesn't seem to be the case.

It takes only a few minutes to trigger the bug here.