Re: [PATCH] KVM: nVMX: Fix the NMI IDT-vectoring info handling

From: Jan Kiszka
Date: Thu Sep 22 2016 - 06:04:46 EST


On 2016-09-22 11:45, Wanpeng Li wrote:
> 2016-09-22 17:37 GMT+08:00 Paolo Bonzini <pbonzini@xxxxxxxxxx>:
>>
>>
>> On 22/09/2016 05:34, Wanpeng Li wrote:
>>> - if (vmx->rmode.vm86_active) {
>>> - if (kvm_inject_realmode_interrupt(vcpu, NMI_VECTOR, 0) != EMULATE_DONE)
>>> - kvm_make_request(KVM_REQ_TRIPLE_FAULT, vcpu);
>>> - return;
>>> + ++vcpu->stat.nmi_injections;
>>> + vmx->nmi_known_unmasked = false;
>>> + if (vmx->rmode.vm86_active) {
>>> + if (kvm_inject_realmode_interrupt(vcpu, NMI_VECTOR, 0) != EMULATE_DONE)
>>> + kvm_make_request(KVM_REQ_TRIPLE_FAULT, vcpu);
>>> + return;
>>> + }
>>> }
>>> vmcs_write32(VM_ENTRY_INTR_INFO_FIELD,
>>> INTR_TYPE_NMI_INTR | INTR_INFO_VALID_MASK | NMI_VECTOR);
>>
>> Hi,
>>
>> the patch is mostly okay but the "if (vmx->rmode.vm86_active)" part
>> should be done also if the VCPU is in guest mode. See
>> vmx_queue_exception for a similar example.
>
> Thanks for pointing out. :)
>
>>
>> I would like to know which processors lack virtual NMI support. I'd
>> rather rip that code out...
>
> At least the Sandy Bridge server on my hand supports Virtual NMIs, I
> don't have machine older than this.

IIRC, my last-but-one notebook was lacking vNMI, and that was a Core 2
Duo. Maybe there are embedded product out there relying on this
emulation, but I don't have any evidence.

Jan

--
Siemens AG, Corporate Technology, CT RDA ITP SES-DE
Corporate Competence Center Embedded Linux