When rebasing to a more recent RH upstream, the split nature of kvm_arch_irq_routing_update / kvm_arch_post_irq_routing_update was lost.
As a result, in current VZ kernels the ACK notifiers aren't attached to Hyper-V synthetic interrupts, which causes the userspace that observed contention on the SynIC message page and went to sleep, to never be woken up again. Signed-off-by: Roman Kagan <rka...@virtuozzo.com> --- arch/x86/kvm/irq_comm.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/arch/x86/kvm/irq_comm.c b/arch/x86/kvm/irq_comm.c index c2d61d8..2bfb9a6 100644 --- a/arch/x86/kvm/irq_comm.c +++ b/arch/x86/kvm/irq_comm.c @@ -399,11 +399,11 @@ int kvm_arch_set_irq(struct kvm_kernel_irq_routing_entry *irq, struct kvm *kvm, } } -void kvm_arch_irq_routing_update(struct kvm *kvm) +void kvm_arch_post_irq_routing_update(struct kvm *kvm) { - if (ioapic_in_kernel(kvm) || !irqchip_in_kernel(kvm)) + if (!irqchip_split(kvm)) return; - kvm_hv_irq_routing_update(kvm); + kvm_make_scan_ioapic_request(kvm); } void kvm_scan_ioapic_routes(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap) @@ -435,3 +435,8 @@ void kvm_scan_ioapic_routes(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap) } srcu_read_unlock(&kvm->irq_srcu, idx); } + +void kvm_arch_irq_routing_update(struct kvm *kvm) +{ + kvm_hv_irq_routing_update(kvm); +} -- 2.9.3 _______________________________________________ Devel mailing list Devel@openvz.org https://lists.openvz.org/mailman/listinfo/devel