Paolo Bonzini <pbonzini@xxxxxxxxxx> writes:
> On 18/02/20 18:08, Vitaly Kuznetsov wrote:
>> Paolo Bonzini <pbonzini@xxxxxxxxxx> writes:
>>
>>> On 18/02/20 15:44, Vitaly Kuznetsov wrote:
>>>> Signed-off-by: Vitaly Kuznetsov <vkuznets@xxxxxxxxxx>
>>>> ---
>>>> RFC: This is somewhat similar to eVMCS breakage and it is likely possible
>>>> to fix this in KVM. I decided to try QEMU first as this is a single
>>>> control and unlike eVMCS we don't need to keep a list of things to disable.
>>>
>>> I think you should disable "virtual-interrupt delivery" instead (which
>>> in turn requires "process posted interrupts" to be zero). That is the
>>> one that is incompatible with AutoEOI interrupts.
>>
>> I'm fighting the symptoms, not the cause :-) My understanding is that
>> when SynIC is enabled for CPU0 KVM does
>>
>> kvm_vcpu_update_apicv()
>> vmx_refresh_apicv_exec_ctrl()
>> pin_controls_set()
>>
>> for *all* vCPUs (KVM_REQ_APICV_UPDATE). I'm not sure why
>> SECONDARY_EXEC_APIC_REGISTER_VIRT/SECONDARY_EXEC_VIRTUAL_INTR_DELIVERY
>> are not causing problems and only PIN_BASED_POSTED_INTR does as we clear
>> them all (not very important atm).
>
> Let's take a step back, what is the symptom, i.e. how does it fail?
I just do
~/qemu/x86_64-softmmu/qemu-system-x86_64 -machine q35,accel=kvm -cpu host,hv_vpindex,hv_synic -smp 2 -m 16384 -vnc :0
and get
qemu-system-x86_64: error: failed to set MSR 0x48d to 0xff00000016
qemu-system-x86_64: /root/qemu/target/i386/kvm.c:2684: kvm_buf_set_msrs: Assertion `ret == cpu->kvm_msr_buf->nmsrs' failed.
Aborted
(it works with '-smp 1' or without 'hv_synic')
> Because thinking more about it, since we have separate VMCS we can set
> PIN_BASED_POSTED_INTR and SECONDARY_EXEC_VIRTUAL_INTR_DELIVERY just fine
> in the vmcs02.
> The important part is to unconditionally call
> vmx_deliver_nested_posted_interrupt.
>
> Something like
>
> if (kvm_x86_ops->deliver_posted_interrupt(vcpu, vector)) {
> kvm_lapic_set_irr(vector, apic);
> kvm_make_request(KVM_REQ_EVENT, vcpu);
> kvm_vcpu_kick(vcpu);
> }
>
> and in vmx_deliver_posted_interrupt
>
> r = vmx_deliver_nested_posted_interrupt(vcpu, vector);
> if (!r)
> return 0;
>
> if (!vcpu->arch.apicv_active)
> return -1;
> ...
> return 0;
Sound like a plan, let me try playing with it.
--
Vitaly
