On Wed, Nov 08, 2017 at 03:08:36PM +0000, Marc Zyngier wrote:
> On 07/11/17 21:28, Auger Eric wrote:
> > Hi Marc,
> >
> > On 27/10/2017 16:28, Marc Zyngier wrote:
> >> Upon updating a property, we propagate it all the way to the physical
> >> ITS, and ask for an INV command to be executed there.
> >>
> >> Acked-by: Christoffer Dall <cdall@xxxxxxxxxx>
> >> Signed-off-by: Marc Zyngier <marc.zyngier@xxxxxxx>
> >> ---
> >> virt/kvm/arm/vgic/vgic-its.c | 3 +++
> >> 1 file changed, 3 insertions(+)
> >>
> >> diff --git a/virt/kvm/arm/vgic/vgic-its.c b/virt/kvm/arm/vgic/vgic-its.c
> >> index 0b7e648e7a0c..2e77c7c83942 100644
> >> --- a/virt/kvm/arm/vgic/vgic-its.c
> >> +++ b/virt/kvm/arm/vgic/vgic-its.c
> >> @@ -296,6 +296,9 @@ static int update_lpi_config(struct kvm *kvm, struct vgic_irq *irq,
> >> spin_unlock(&irq->irq_lock);
> >> }
> >>
> >> + if (irq->hw)
> >> + return its_prop_update_vlpi(irq->host_irq, prop, true);
> >> +
> >> return 0;
> >> }
> > I am confused by the vgic_queue_irq_unlock() on the "hw" path. Why is it
> > needed in hw mode?
>
> It's not. I guess we could bypass this altogether and take a short cut
> after having updated the priority and enabled fields.
>
I can apply this on top of the series as well if you're happy with it:
commit b54fba93b1330803a59ca75f3a5102e22cadc871 (HEAD -> next-gicv4)
Author: Christoffer Dall <christoffer.dall@xxxxxxxxxx>
Date: Fri Nov 10 09:34:54 2017 +0100
KVM: arm/arm64: Don't queue VLPIs on INV/INVALL
Since VLPIs are injected directly by the hardware there's no need to
mark these as pending in software and queue them on the AP list.
Signed-off-by: Christoffer Dall <christoffer.dall@xxxxxxxxxx>
diff --git a/virt/kvm/arm/vgic/vgic-its.c b/virt/kvm/arm/vgic/vgic-its.c
index c93ecd4a903b..a3754ec719c4 100644
--- a/virt/kvm/arm/vgic/vgic-its.c
+++ b/virt/kvm/arm/vgic/vgic-its.c
@@ -292,11 +292,14 @@ static int update_lpi_config(struct kvm *kvm, struct vgic_irq *irq,
irq->priority = LPI_PROP_PRIORITY(prop);
irq->enabled = LPI_PROP_ENABLE_BIT(prop);
- vgic_queue_irq_unlock(kvm, irq, flags);
- } else {
- spin_unlock_irqrestore(&irq->irq_lock, flags);
+ if (!irq->hw) {
+ vgic_queue_irq_unlock(kvm, irq, flags);
+ return 0;
+ }
}
+ spin_unlock_irqrestore(&irq->irq_lock, flags);
+
if (irq->hw)
return its_prop_update_vlpi(irq->host_irq, prop, needs_inv);
Thanks,
-Christoffer
