On 28/08/17 19:18, Christoffer Dall wrote:

On Mon, Jul 31, 2017 at 06:26:22PM +0100, Marc Zyngier wrote:

quoted

When the guest issues a MOVI, we need to tell the physical ITS
that we're now targetting a new vcpu. This is done by extracting
the current mapping, updating the target, and reapplying the
mapping. The core ITS code should do the right thing.

Signed-off-by: Marc Zyngier <redacted>
---
 virt/kvm/arm/vgic/vgic-its.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/virt/kvm/arm/vgic/vgic-its.c b/virt/kvm/arm/vgic/vgic-its.c
index 79bac93d3e7d..aaad577ce328 100644
--- a/virt/kvm/arm/vgic/vgic-its.c
+++ b/virt/kvm/arm/vgic/vgic-its.c

@@ -706,6 +706,19 @@ static int vgic_its_cmd_handle_movi(struct kvm *kvm, struct vgic_its *its,
 	ite->irq->target_vcpu = vcpu;
 	spin_unlock(&ite->irq->irq_lock);
 
+	if (ite->irq->hw) {
+		struct its_vlpi_map map;
+		int ret;
+
+		ret = its_get_vlpi(ite->irq->host_irq, &map);
+		if (ret)
+			return ret;
+
+		map.vpe_idx = vcpu->vcpu_id;
+
+		return its_map_vlpi(ite->irq->host_irq, &map);

Since you're not holding the irq_lock across these two calls, would it
be possible that the forwarding was removed through some other call path
here, and could you end up passing an invalid host_irq to its_map_vlpi?

I believe we should be OK here, as we hold the ITS mutex during any
command processing, and both the forward/unforward paths take that same
mutex.

On a slightly different note, it looks like the MOVI code could benefit
from using vgic_its_resolve_lpi(), which has been introduce earlier in
this series.

Thanks,

	M.
-- 
Jazz is not dead. It just smells funny...