On Wed, Aug 10, 2016 at 09:04:13PM +0200, Radim Krčmář wrote: > 2016-08-10 15:27-0300, Eduardo Habkost: > > Hi, > > > > I got a bug report[1] that seems to be caused by kvm_pv_unhalt > > not working with kernel_irqchip=off. > > > > Is kvm_pv_unhalt supposed to work at all without kernel_irqchip? > > No. Guest interface of current PV unhalt is based on APIC ID and KVM > does not ask userspace to handle the unknown mapping to VCPU. > > The code in kernel just somehow managed not to BUG. :) > > > Should QEMU prevent it from being enabled if kernel_irqchip=off? > > Yes. We'd need a new guest or userspace interface to make it work. Thanks for the answers! Now I have another question: are features that require the in-kernel irqchip supposed to be present in GET_SUPPORTED_CPUID? We have examples of both cases in KVM: * TSC_DEADLINE_TIMER is _not_ present in GET_SUPPORTED_CPUID, and is reported through KVM_CAP_TSC_DEADLINE_TIMER. * X2APIC is present in GET_SUPPORTED_CPUID, but the bit makes sense only if the in-kernel irqchip is used. * KVM_PV_UNHALT is present in GET_SUPPORTED_CPUID, but requires the in-kernel irqchip to work. Should userspace expect more cases like x2apic and kvm_pv_unhalt in the future? -- Eduardo -- 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
