[PATCH] KVM: arm/arm64: Handle forward time correction gracefully

[Christoffer Dall]

On Wed, Apr 06, 2016 at 09:37:22AM +0100, Marc Zyngier wrote:
> On a host that runs NTP, corrections can have a direct impact on
> the background timer that we program on the behalf of a vcpu.
> 
> In particular, NTP performing a forward correction will result in
> a timer expiring sooner than expected from a guest point of view.
> Not a big deal, we kick the vcpu anyway.
> 
> But on wake-up, the vcpu thread is going to perform a check to
> find out whether or not it should block. And at that point, the
> timer check is going to say "timer has not expired yet, go back
> to sleep". This results in the timer event being lost forever.
> 
> There are multiple ways to handle this. One would be record that
> the timer has expired and let kvm_cpu_has_pending_timer return
> true in that case, but that would be fairly invasive. Another is
> to check for the "short sleep" condition in the hrtimer callback,
> and restart the timer for the remaining time when the condition
> is detected.
> 
> This patch implements the latter, with a bit of refactoring in
> order to avoid too much code duplication.
> 
> Reported-by: Alexander Graf <agraf at suse.de>
> Signed-off-by: Marc Zyngier <marc.zyngier at arm.com>
> ---
>  virt/kvm/arm/arch_timer.c | 47 +++++++++++++++++++++++++++++++++++++----------
>  1 file changed, 37 insertions(+), 10 deletions(-)
> 
> diff --git a/virt/kvm/arm/arch_timer.c b/virt/kvm/arm/arch_timer.c
> index a9ad4fe..4d0e77a 100644
> --- a/virt/kvm/arm/arch_timer.c
> +++ b/virt/kvm/arm/arch_timer.c
> @@ -98,10 +98,46 @@ static void kvm_timer_inject_irq_work(struct work_struct *work)
>  	kvm_vcpu_kick(vcpu);
>  }
>  
> +static u64 kvm_timer_compute_delta(struct kvm_vcpu *vcpu)
> +{
> +	cycle_t cval, now;
> +
> +	cval = vcpu->arch.timer_cpu.cntv_cval;
> +	now = kvm_phys_timer_read() - vcpu->kvm->arch.timer.cntvoff;
> +
> +	if (now < cval) {
> +		u64 ns;
> +
> +		ns = cyclecounter_cyc2ns(timecounter->cc,
> +					 cval - now,
> +					 timecounter->mask,
> +					 &timecounter->frac);
> +		return ns;
> +	}
> +
> +	return 0;
> +}
> +
>  static enum hrtimer_restart kvm_timer_expire(struct hrtimer *hrt)
>  {
>  	struct arch_timer_cpu *timer;
> +	struct kvm_vcpu *vcpu;
> +	u64 ns;
> +
>  	timer = container_of(hrt, struct arch_timer_cpu, timer);
> +	vcpu = container_of(timer, struct kvm_vcpu, arch.timer_cpu);
> +
> +	/*
> +	 * Check that the timer has really expired from the guest's
> +	 * PoV (NTP on the host may have forced it to expire
> +	 * early). If we should have slept longer, restart it.
> +	 */
> +	ns = kvm_timer_compute_delta(vcpu);
> +	if (unlikely(ns)) {
> +		hrtimer_forward_now(hrt, ns_to_ktime(ns));
> +		return HRTIMER_RESTART;
> +	}
> +
>  	queue_work(wqueue, &timer->expired);
>  	return HRTIMER_NORESTART;
>  }
> @@ -176,8 +212,6 @@ static int kvm_timer_update_state(struct kvm_vcpu *vcpu)
>  void kvm_timer_schedule(struct kvm_vcpu *vcpu)
>  {
>  	struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;
> -	u64 ns;
> -	cycle_t cval, now;
>  
>  	BUG_ON(timer_is_armed(timer));
>  
> @@ -197,14 +231,7 @@ void kvm_timer_schedule(struct kvm_vcpu *vcpu)
>  		return;
>  
>  	/*  The timer has not yet expired, schedule a background timer */
> -	cval = timer->cntv_cval;
> -	now = kvm_phys_timer_read() - vcpu->kvm->arch.timer.cntvoff;
> -
> -	ns = cyclecounter_cyc2ns(timecounter->cc,
> -				 cval - now,
> -				 timecounter->mask,
> -				 &timecounter->frac);
> -	timer_arm(timer, ns);
> +	timer_arm(timer, kvm_timer_compute_delta(vcpu));
>  }
>  
>  void kvm_timer_unschedule(struct kvm_vcpu *vcpu)
> -- 
> 2.1.4
> 

How do you guys feel about adding this to the patch for improved sleep
at night (pun intended):

diff --git a/virt/kvm/arm/arch_timer.c b/virt/kvm/arm/arch_timer.c
index a9ad4fe..230f720 100644
--- a/virt/kvm/arm/arch_timer.c
+++ b/virt/kvm/arm/arch_timer.c
@@ -91,6 +91,8 @@ static void kvm_timer_inject_irq_work(struct work_struct *work)
 	vcpu = container_of(work, struct kvm_vcpu, arch.timer_cpu.expired);
 	vcpu->arch.timer_cpu.armed = false;
 
+	BUG_ON(!kvm_timer_should_fire(vcpu));
+
 	/*
 	 * If the vcpu is blocked we want to wake it up so that it will see
 	 * the timer has expired when entering the guest.


Otherwise:

Reviewed-by: Christoffer Dall <christoffer.dall at linaro.org>

And I can queue this with CC to stable, but I would like Alex's
tested-by if possible.

Thanks,
-Christoffer