Re: [PATCH RFC] i386/kvm: fix enlightened VMCS with fine-grained VMX feature enablement

[Vitaly Kuznetsov <vkuznets@redhat.com>]

Paolo Bonzini <pbonzini@redhat.com> writes:

> SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES
>>
>> the problem with it is that we don't have 'apic_access_addr' field in
>> eVMCS ('virtual_apic_page_addr' is there). By running the same setup
>> with eVMCS disabled I figured out which address can be hardcoded to make
>> it boot.
>>
>
> Maybe it's really hard coded (what is the value? Is it consistent across
> Hyper-V version?) Can you try changing KVM to enable it and see if the
> hardcoded APIC access address works?

Unfortunately, it's not the same even between BIOS and UEFI booted
WS2016 (Gen1/Gen2). It is, however, the same across reboot for the same
image so for example for WS2016Gen1 it is '0x294000' so if I do:

diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
index fab7451a5793..8366b2a02b3b 100644
--- a/arch/x86/kvm/vmx/nested.c
+++ b/arch/x86/kvm/vmx/nested.c
@@ -1433,6 +1433,8 @@ static int copy_enlightened_to_vmcs12(struct vcpu_vmx *vmx)
        vmcs12->tpr_threshold = evmcs->tpr_threshold;
        vmcs12->guest_rip = evmcs->guest_rip;
 
+       vmcs12->apic_access_addr = 0x294000;

it all works. I'm really puzzled.

-- 
Vitaly