Re: rcu: INFO: rcu_preempt detected stalls on CPUs/tasks on v3.6

[John Stultz <johnstul@us.ibm.com>]

On 08/06/2012 10:21 AM, John Stultz wrote:
> On 08/05/2012 09:55 AM, Sasha Levin wrote:
>> On 07/30/2012 03:17 PM, Avi Kivity wrote:
>>> Possible causes:
>>>   - the APIC calibration in the guest failed, so it is programming too
>>> low values into the timer
>>>   - it actually needs 1 us wakeups and then can't keep up (esp. as kvm
>>> interrupt injection is slowing it down)
>>>
>>> You can try to find out by changing
>>> arch/x86/kvm/lapic.c:start_lapic_timer() to impose a minimum wakeup of
>>> (say) 20 microseconds which will let the guest live long enough for you
>>> to ftrace it and see what kind of timers it is programming.
>> I've kept trying to narrow it down, and found out It's triggerable 
>> using adjtimex().
>>
>> At that point I've bisected it, and got the following commit (parties 
>> Cc'ed):
>>
>> commit 5baefd6d84163443215f4a99f6a20f054ef11236
>> Author: John Stultz <johnstul@us.ibm.com>
>> Date:   Tue Jul 10 18:43:25 2012 -0400
>>
>>      hrtimer: Update hrtimer base offsets each hrtimer_interrupt
>>
>>
>> I've also confirmed that reverting the commit above on top of 
>> linux-next indeed fixes the issue.
> Hey Sasha,
>     Thanks for the heads up.  I don't have a clear sense of what could 
> be wrong here yet, but if you see this with 3.6-rc but not 3.5,  could 
> you try the fix(1d17d17484d40f2d5b35c79518597a2b25296996) Ingo just 
> made on tip/timers/urgent?

Reading over the thread here, the large timeouts also made me think that 
it could also be related to this pending fix:
     http://lkml.org/lkml/2012/8/1/436

Its not a clear cut solution though, since the edge case that limits 
usually results in a hang since we stop expiring timers all together.

Still working to reproduce what you're seeing, and will let you know as 
soon as I have any more info.

thanks
-john