Re: [PATCH 3/4] KVM: async_pf: Force a nested vmexit if the injected #PF is async_pf

[Wanpeng Li]

2017-06-14 2:55 GMT+08:00 Radim KrÄmÃÅ <rkrcmar@xxxxxxxxxx>:
> 2017-06-12 23:08-0700, Wanpeng Li:
>> From: Wanpeng Li <wanpeng.li@xxxxxxxxxxx>
>>
>> Add an async_page_fault field to vcpu->arch.exception to identify an async
>> page fault, and constructs the expected vm-exit information fields. Force
>> a nested VM exit from nested_vmx_check_exception() if the injected #PF
>> is async page fault.
>>
>> Cc: Paolo Bonzini <pbonzini@xxxxxxxxxx>
>> Cc: Radim KrÄmÃÅ <rkrcmar@xxxxxxxxxx>
>> Signed-off-by: Wanpeng Li <wanpeng.li@xxxxxxxxxxx>
>> ---
>> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
>> @@ -2422,13 +2422,28 @@ static void skip_emulated_instruction(struct kvm_vcpu *vcpu)
>>  static int nested_vmx_check_exception(struct kvm_vcpu *vcpu, unsigned nr)
>
> This function could use the same treatment as vmx_queue_exception(), so
> we are not mixing 'nr' with 'vcpu->arch.exception.*'.
>
>>  {
>>       struct vmcs12 *vmcs12 = get_vmcs12(vcpu);
>> +     u32 intr_info = 0;
>> +     unsigned long exit_qualification = 0;
>>
>> -     if (!(vmcs12->exception_bitmap & (1u << nr)))
>> +     if (!((vmcs12->exception_bitmap & (1u << nr)) ||
>> +             (nr == PF_VECTOR && vcpu->arch.exception.async_page_fault)))
>>               return 0;
>>
>> +     intr_info = nr | INTR_INFO_VALID_MASK;
>> +     exit_qualification = vmcs_readl(EXIT_QUALIFICATION);
>
> This part still uses EXIT_QUALIFICATION(), which means it is not general
> and I think it would be nicer to just do simple special case on top:
>
>         if (vcpu->arch.exception.async_page_fault) {
>                 vmcs_write32(VM_EXIT_INTR_ERROR_CODE, vcpu->arch.exception.error_code);
>                 nested_vmx_vmexit(vcpu, EXIT_REASON_EXCEPTION_NMI,
>                                   PF_VECTOR | INTR_TYPE_HARD_EXCEPTION |
>                                   INTR_INFO_DELIVER_CODE_MASK | INTR_INFO_VALID_MASK,
>                                   vcpu->arch.cr2);
>                 return 1;
>         }

Good point.

>
> Using vcpu->arch.cr2 is suspicious as VMX doesn't update CR2 on VM
> exits;  isn't this going to change the CR2 visible in L2 guest after a
> nested VM entry?

Sorry, I don't fully understand the question. As you know this
vcpu->arch.cr2 which includes token is set before async pf injection,
and L1 will intercept it from EXIT_QUALIFICATION during nested vmexit,
why it can change the CR2 visible in L2 guest after a nested VM entry?

>
> Btw. nested_vmx_check_exception() didn't support emulated exceptions at
> all (it only passed through ones we got from hardware),

I think so.

Regards,
Wanpeng Li