On 07/12/2018 01:10 PM, David Woodhouse wrote:
> On Thu, 2018-07-12 at 13:00 +0200, Christian Borntraeger wrote:
>>
>> On s390 this seems to add about 10ns (~3%) for a guest exit/rentry
>> microbenchmark mostly due to rcu_eqs_enter and rcu_eqs_exit now being
>> visible in perf samples. The older interface was cheaper.
>
> Well, the older interface wasn't actually working, which made it
> moderately suboptimal :)
>
> But that is fixed by the first patch of the two, so perhaps the second
> ought to be conditional on CONFIG_NO_HZ_FULL since it's only really
> fixing a real bug there?
>
> Unless we can contrive some way to do the rcu_eqs_enter/exit only on
> the paths to/from userspace, and *not* when we loop in the kernel to
> handle interrupts (etc.) and immediately re-enter the guest without
> returning? That's somewhat non-trivial though...
Assuming that patch1 really fixes the issue (and all KVM hot loops are
supposed to have something like
if (need_resched())
schedule();
this could work out given that patch1 does what it is supposed to do.
So you are thinking of something like
diff --git a/include/linux/rcutree.h b/include/linux/rcutree.h
index d8ee9f949ce3..cac5de8a887c 100644
--- a/include/linux/rcutree.h
+++ b/include/linux/rcutree.h
@@ -72,8 +72,17 @@ void cond_synchronize_sched(unsigned long oldstate);
void rcu_idle_enter(void);
void rcu_idle_exit(void);
+#ifdef CONFIG_NO_HZ_FULL
void rcu_kvm_enter(void);
void rcu_kvm_exit(void);
+#else
+static inline void rcu_kvm_enter(void)
+{
+}
+static inline void rcu_kvm_exit(void)
+{
+}
+#endif
void rcu_irq_enter(void);
void rcu_irq_exit(void);
void rcu_irq_enter_irqson(void);
?
