[PATCH] kvm: nVMX: VMCLEAR should not cause the vCPU to shut down

[PATCH] kvm: nVMX: VMCLEAR should not cause the vCPU to shut down

[Jim Mattson]

VMCLEAR should silently ignore a failure to clear the launch state of
the VMCS referenced by the operand.

Signed-off-by: Jim Mattson <jmattson@google.com>
---
 arch/x86/kvm/vmx.c | 22 ++++------------------
 1 file changed, 4 insertions(+), 18 deletions(-)

diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index ef4ba71dbb66..bca497a92541 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -7259,9 +7259,8 @@ static int handle_vmoff(struct kvm_vcpu *vcpu)
 static int handle_vmclear(struct kvm_vcpu *vcpu)
 {
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
+	u32 zero = 0;
 	gpa_t vmptr;
-	struct vmcs12 *vmcs12;
-	struct page *page;
 
 	if (!nested_vmx_check_permission(vcpu))
 		return 1;
@@ -7272,22 +7271,9 @@ static int handle_vmclear(struct kvm_vcpu *vcpu)
 	if (vmptr == vmx->nested.current_vmptr)
 		nested_release_vmcs12(vmx);
 
-	page = nested_get_page(vcpu, vmptr);
-	if (page == NULL) {
-		/*
-		 * For accurate processor emulation, VMCLEAR beyond available
-		 * physical memory should do nothing at all. However, it is
-		 * possible that a nested vmx bug, not a guest hypervisor bug,
-		 * resulted in this case, so let's shut down before doing any
-		 * more damage:
-		 */
-		kvm_make_request(KVM_REQ_TRIPLE_FAULT, vcpu);
-		return 1;
-	}
-	vmcs12 = kmap(page);
-	vmcs12->launch_state = 0;
-	kunmap(page);
-	nested_release_page(page);
+	kvm_write_guest(vcpu->kvm,
+			vmptr + offsetof(struct vmcs12, launch_state),
+			&zero, sizeof(zero));
 
 	nested_free_vmcs02(vmx, vmptr);
 
-- 
2.12.0.rc1.440.g5b76565f74-goog

Re: [PATCH] kvm: nVMX: VMCLEAR should not cause the vCPU to shut down

[Radim Krčmář]

2017-03-02 12:41-0800, Jim Mattson:
> VMCLEAR should silently ignore a failure to clear the launch state of
> the VMCS referenced by the operand.
> 
> Signed-off-by: Jim Mattson <jmattson@google.com>
> ---
>  arch/x86/kvm/vmx.c | 22 ++++------------------
>  1 file changed, 4 insertions(+), 18 deletions(-)
> 
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index ef4ba71dbb66..bca497a92541 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -7259,9 +7259,8 @@ static int handle_vmoff(struct kvm_vcpu *vcpu)
>  static int handle_vmclear(struct kvm_vcpu *vcpu)
>  {
>  	struct vcpu_vmx *vmx = to_vmx(vcpu);
> +	u32 zero = 0;
>  	gpa_t vmptr;
> -	struct vmcs12 *vmcs12;
> -	struct page *page;
>  
>  	if (!nested_vmx_check_permission(vcpu))
>  		return 1;
> @@ -7272,22 +7271,9 @@ static int handle_vmclear(struct kvm_vcpu *vcpu)
>  	if (vmptr == vmx->nested.current_vmptr)
>  		nested_release_vmcs12(vmx);
>  
> -	page = nested_get_page(vcpu, vmptr);
> -	if (page == NULL) {
> -		/*
> -		 * For accurate processor emulation, VMCLEAR beyond available
> -		 * physical memory should do nothing at all. However, it is
> -		 * possible that a nested vmx bug, not a guest hypervisor bug,
> -		 * resulted in this case, so let's shut down before doing any
> -		 * more damage:
> -		 */

(Consciously introducing bugs in order to somewhat mitigate possible
 bugs is not desirable.)

> -		kvm_make_request(KVM_REQ_TRIPLE_FAULT, vcpu);
> -		return 1;
> -	}
> -	vmcs12 = kmap(page);
> -	vmcs12->launch_state = 0;
> -	kunmap(page);
> -	nested_release_page(page);
> +	kvm_write_guest(vcpu->kvm,

I have changed this line to

   	kvm_vcpu_write_guest(vcpu

and applied to kvm/queue (planning to merge in 4.11-rc2).

The change makes a difference if the guest is using VMX inside SMM and I
don't see a reason to avoid the SMM address space in that case.
Let me know what you think,

thanks.

> +			vmptr + offsetof(struct vmcs12, launch_state),
> +			&zero, sizeof(zero));

(Could have also been kvm_vcpu_write_guest_page(), but that is only
 performance gain.)

>  
>  	nested_free_vmcs02(vmx, vmptr);
>  
> -- 
> 2.12.0.rc1.440.g5b76565f74-goog
> 

Re: [PATCH] kvm: nVMX: VMCLEAR should not cause the vCPU to shut down

[Jim Mattson]

That sounds good to me.

Thanks!

--jim

On Fri, Mar 3, 2017 at 8:12 AM, Radim Krčmář <rkrcmar@redhat.com> wrote:
> 2017-03-02 12:41-0800, Jim Mattson:
>> VMCLEAR should silently ignore a failure to clear the launch state of
>> the VMCS referenced by the operand.
>>
>> Signed-off-by: Jim Mattson <jmattson@google.com>
>> ---
>>  arch/x86/kvm/vmx.c | 22 ++++------------------
>>  1 file changed, 4 insertions(+), 18 deletions(-)
>>
>> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
>> index ef4ba71dbb66..bca497a92541 100644
>> --- a/arch/x86/kvm/vmx.c
>> +++ b/arch/x86/kvm/vmx.c
>> @@ -7259,9 +7259,8 @@ static int handle_vmoff(struct kvm_vcpu *vcpu)
>>  static int handle_vmclear(struct kvm_vcpu *vcpu)
>>  {
>>       struct vcpu_vmx *vmx = to_vmx(vcpu);
>> +     u32 zero = 0;
>>       gpa_t vmptr;
>> -     struct vmcs12 *vmcs12;
>> -     struct page *page;
>>
>>       if (!nested_vmx_check_permission(vcpu))
>>               return 1;
>> @@ -7272,22 +7271,9 @@ static int handle_vmclear(struct kvm_vcpu *vcpu)
>>       if (vmptr == vmx->nested.current_vmptr)
>>               nested_release_vmcs12(vmx);
>>
>> -     page = nested_get_page(vcpu, vmptr);
>> -     if (page == NULL) {
>> -             /*
>> -              * For accurate processor emulation, VMCLEAR beyond available
>> -              * physical memory should do nothing at all. However, it is
>> -              * possible that a nested vmx bug, not a guest hypervisor bug,
>> -              * resulted in this case, so let's shut down before doing any
>> -              * more damage:
>> -              */
>
> (Consciously introducing bugs in order to somewhat mitigate possible
>  bugs is not desirable.)
>
>> -             kvm_make_request(KVM_REQ_TRIPLE_FAULT, vcpu);
>> -             return 1;
>> -     }
>> -     vmcs12 = kmap(page);
>> -     vmcs12->launch_state = 0;
>> -     kunmap(page);
>> -     nested_release_page(page);
>> +     kvm_write_guest(vcpu->kvm,
>
> I have changed this line to
>
>         kvm_vcpu_write_guest(vcpu
>
> and applied to kvm/queue (planning to merge in 4.11-rc2).
>
> The change makes a difference if the guest is using VMX inside SMM and I
> don't see a reason to avoid the SMM address space in that case.
> Let me know what you think,
>
> thanks.
>
>> +                     vmptr + offsetof(struct vmcs12, launch_state),
>> +                     &zero, sizeof(zero));
>
> (Could have also been kvm_vcpu_write_guest_page(), but that is only
>  performance gain.)
>
>>
>>       nested_free_vmcs02(vmx, vmptr);
>>
>> --
>> 2.12.0.rc1.440.g5b76565f74-goog
>>