On Tue, Mar 17, 2015 at 06:02:41PM +0000, Andre Przywara wrote:
> Hej,
>
> On 14/03/15 14:30, Christoffer Dall wrote:
> > On Fri, Mar 13, 2015 at 04:10:09PM +0000, Andre Przywara wrote:
> >> Using the framework provided by the recent vgic.c changes we register
> >> a kvm_io_bus device when initializing the virtual GICv2.
> >>
> >> Signed-off-by: Andre Przywara <andre.przywara@xxxxxxx>
> >> ---
> >> include/kvm/arm_vgic.h | 1 +
> >> virt/kvm/arm/vgic-v2-emul.c | 13 +++++++++----
> >> virt/kvm/arm/vgic.c | 16 ++++++++++++++++
> >> 3 files changed, 26 insertions(+), 4 deletions(-)
> >>
> >> diff --git a/include/kvm/arm_vgic.h b/include/kvm/arm_vgic.h
> >> index 4bfc6a3..74a4ac4 100644
> >> --- a/include/kvm/arm_vgic.h
> >> +++ b/include/kvm/arm_vgic.h
> >> @@ -245,6 +245,7 @@ struct vgic_dist {
> >> unsigned long *irq_pending_on_cpu;
> >>
> >> struct vgic_vm_ops vm_ops;
> >> + struct vgic_io_device dist_iodev;
> >> };
> >>
> >> struct vgic_v2_cpu_if {
> >> diff --git a/virt/kvm/arm/vgic-v2-emul.c b/virt/kvm/arm/vgic-v2-emul.c
> >> index 0defac6..6f685c9 100644
> >> --- a/virt/kvm/arm/vgic-v2-emul.c
> >> +++ b/virt/kvm/arm/vgic-v2-emul.c
> >> @@ -490,6 +490,7 @@ static bool vgic_v2_queue_sgi(struct kvm_vcpu *vcpu, int irq)
> >> static int vgic_v2_map_resources(struct kvm *kvm,
> >> const struct vgic_params *params)
> >> {
> >> + struct vgic_dist *dist = &kvm->arch.vgic;
> >> int ret = 0;
> >>
> >> if (!irqchip_in_kernel(kvm))
> >> @@ -500,13 +501,17 @@ static int vgic_v2_map_resources(struct kvm *kvm,
> >> if (vgic_ready(kvm))
> >> goto out;
> >>
> >> - if (IS_VGIC_ADDR_UNDEF(kvm->arch.vgic.vgic_dist_base) ||
> >> - IS_VGIC_ADDR_UNDEF(kvm->arch.vgic.vgic_cpu_base)) {
> >> + if (IS_VGIC_ADDR_UNDEF(dist->vgic_dist_base) ||
> >> + IS_VGIC_ADDR_UNDEF(dist->vgic_cpu_base)) {
> >> kvm_err("Need to set vgic cpu and dist addresses first\n");
> >> ret = -ENXIO;
> >> goto out;
> >> }
> >>
> >> + vgic_register_kvm_io_dev(kvm, dist->vgic_dist_base,
> >> + KVM_VGIC_V2_DIST_SIZE,
> >> + vgic_dist_ranges, -1, &dist->dist_iodev);
> >> +
> >> /*
> >> * Initialize the vgic if this hasn't already been done on demand by
> >> * accessing the vgic state from userspace.
> >> @@ -517,7 +522,7 @@ static int vgic_v2_map_resources(struct kvm *kvm,
> >> goto out;
> >> }
> >>
> >> - ret = kvm_phys_addr_ioremap(kvm, kvm->arch.vgic.vgic_cpu_base,
> >> + ret = kvm_phys_addr_ioremap(kvm, dist->vgic_cpu_base,
> >> params->vcpu_base, KVM_VGIC_V2_CPU_SIZE,
> >> true);
> >> if (ret) {
> >> @@ -525,7 +530,7 @@ static int vgic_v2_map_resources(struct kvm *kvm,
> >> goto out;
> >> }
> >>
> >> - kvm->arch.vgic.ready = true;
> >> + dist->ready = true;
> >> out:
> >> if (ret)
> >> kvm_vgic_destroy(kvm);
> >> diff --git a/virt/kvm/arm/vgic.c b/virt/kvm/arm/vgic.c
> >> index 71389b8..b1dd717 100644
> >> --- a/virt/kvm/arm/vgic.c
> >> +++ b/virt/kvm/arm/vgic.c
> >> @@ -908,6 +908,20 @@ int vgic_register_kvm_io_dev(struct kvm *kvm, gpa_t base, int len,
> >> return 0;
> >> }
> >>
> >> +static void vgic_unregister_kvm_io_dev(struct kvm *kvm)
> >> +{
> >> + struct vgic_dist *dist = &kvm->arch.vgic;
> >> +
> >> + if (!dist || !kvm->buses[KVM_MMIO_BUS])
> >> + return;
> >> +
> >> + mutex_lock(&kvm->slots_lock);
> >> + if (dist->dist_iodev.dev.ops)
> >> + kvm_io_bus_unregister_dev(kvm, KVM_MMIO_BUS,
> >> + &dist->dist_iodev.dev);
> >
> > why are we only unregisttering the iodev when we have ops?
>
> Because vgic_unregister_kvm_io_dev() is called by kvm_vgic_destroy(),
> which we call on some occasions during vgic_init() when we encounter an
> error. There is quite a window of cases where the kvm_io_bus devices
> haven't been registered yet, so we shouldn't try to unregister them at
> this place. Using the .ops parameter seemed like an elegant way to
> detect this case.
> Does that make sense? Or have I missed something?
>
That's ok, but it derserves a comment explaining this.
-Christoffer
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
