On 11/18/2014 11:43 AM, Eric Auger wrote:

On 11/14/2014 11:07 AM, Andre Przywara wrote:

quoted

Currently we unconditionally register the GICv2 emulation device
during the host's KVM initialization. Since with GICv3 support we
may end up with only v2 or only v3 or both supported, we move the
registration into the GIC probing function, where we will later know
which combination is valid.

Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Acked-by: Christoffer Dall <redacted>
---
Changelog v3...v4:
- add Acked-by

 include/linux/kvm_host.h |    1 +
 virt/kvm/arm/vgic-v2.c   |    2 ++
 virt/kvm/arm/vgic-v3.c   |    1 +
 virt/kvm/arm/vgic.c      |    5 ++---
 4 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index ea53b04..326ba7a 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h

@@ -1084,6 +1084,7 @@ void kvm_unregister_device_ops(u32 type);
 
 extern struct kvm_device_ops kvm_mpic_ops;
 extern struct kvm_device_ops kvm_xics_ops;
+extern struct kvm_device_ops kvm_arm_vgic_v2_ops;
 
 #ifdef CONFIG_HAVE_KVM_CPU_RELAX_INTERCEPT
 

diff --git a/virt/kvm/arm/vgic-v2.c b/virt/kvm/arm/vgic-v2.c
index 2935405..e1cd3cb 100644
--- a/virt/kvm/arm/vgic-v2.c
+++ b/virt/kvm/arm/vgic-v2.c

@@ -229,6 +229,8 @@ int vgic_v2_probe(struct device_node *vgic_node,
 		goto out_unmap;
 	}
 
+	kvm_register_device_ops(&kvm_arm_vgic_v2_ops, KVM_DEV_TYPE_ARM_VGIC_V2);
+
 	vgic->vcpu_base = vcpu_res.start;
 
 	kvm_info("%s@%llx IRQ%d\n", vgic_node->name,

diff --git a/virt/kvm/arm/vgic-v3.c b/virt/kvm/arm/vgic-v3.c
index 1c2c8ee..d14c75f 100644
--- a/virt/kvm/arm/vgic-v3.c
+++ b/virt/kvm/arm/vgic-v3.c

@@ -230,6 +230,7 @@ int vgic_v3_probe(struct device_node *vgic_node,
 		ret = -ENXIO;
 		goto out;
 	}
+	kvm_register_device_ops(&kvm_arm_vgic_v2_ops, KVM_DEV_TYPE_ARM_VGIC_V2);

I did not find any unregistration? I am wrong or isn't it relevant? If
confirmed might be added in kvm_arch_exit. Saw KVM-VFIO device ops
unregistration is done in kvm_exit/kvm_vfio_ops_exit.

well forget that one. I guess it is not relevant for that device which
is not going to be released.

BR

Eric

Eric

quoted

 
 	vgic->vcpu_base = vcpu_res.start;
 	vgic->vctrl_base = NULL;

diff --git a/virt/kvm/arm/vgic.c b/virt/kvm/arm/vgic.c
index 963b84e..e8003ca 100644
--- a/virt/kvm/arm/vgic.c
+++ b/virt/kvm/arm/vgic.c

@@ -2540,7 +2540,7 @@ static int vgic_create(struct kvm_device *dev, u32 type)
 	return kvm_vgic_create(dev->kvm, type);
 }
 
-static struct kvm_device_ops kvm_arm_vgic_v2_ops = {
+struct kvm_device_ops kvm_arm_vgic_v2_ops = {
 	.name = "kvm-arm-vgic",
 	.create = vgic_create,
 	.destroy = vgic_destroy,

@@ -2619,8 +2619,7 @@ int kvm_vgic_hyp_init(void)
 
 	on_each_cpu(vgic_init_maintenance_interrupt, NULL, 1);
 
-	return kvm_register_device_ops(&kvm_arm_vgic_v2_ops,
-				       KVM_DEV_TYPE_ARM_VGIC_V2);
+	return 0;
 
 out_free_irq:
 	free_percpu_irq(vgic->maint_irq, kvm_get_running_vcpus());