Re: [RFC] Make need_resched() return true when rcu_urgent_qs requested

["Paul E. McKenney" <paulmck@linux.vnet.ibm.com>]

On Thu, Jul 19, 2018 at 12:23:34PM +0200, Christian Borntraeger wrote:
> 
> 
> On 07/19/2018 09:20 AM, David Woodhouse wrote:
> > On Thu, 2018-07-19 at 08:45 +0200, Christian Borntraeger wrote:
> >>
> >>> My thought would be something like this:
> >>>  
> >>>        if (context_tracking_cpu_is_enabled())
> >>>                rcu_kvm_enter();
> >>>        else
> >>>                rcu_virt_note_context_switch(smp_processor_id());
> >>
> >> In the past we needed that (when we introduced that). At least with every
> >> host interrupt we called this making an rcu event at least every HZ.
> >> Will the changes in need_resched make this part unnecessary?
> > 
> > Yes, the change in need_resched() should make this part unnecessary.
> > Unless your architecture's version of the vcpu_run() loop just loops
> > forever even when TIF_NEED_RESCHED is set? :)
> 
> Very early versions did that. The SIE instruction is interruptible
> so you can continue to run the guest by simply returning from an host
> interrupt. All sane versions of KVM on s390 now make sure to make a
> short trip into C after a host interrupt. There we check for
> need_resched signals and machine checks so we are good.

OK, thank you all!  I will drop the two patches that add the
rcu_kvm_enter() and rcu_kvm_exit() calls.  Two less things to
worry about!  ;-)

							Thanx, Paul

> > I'm not sure about the context tracking condition in the code snippet
> > cited above, though. I think that's what caused my problem in the first
> > place — I have CONTEXT_TRACKING_FORCE && !NO_HZ_FULL. So in 4.15, that
> > means rcu_user_enter() did nothing and rcu_virt_note_context_switch()
> > wasn't called. Hence the observed stalls.
> > 
> > Should rcu_user_enter() itself be conditional on CONTEXT_TRACKING not
> > NO_HZ_FULL? 
> >