[PATCH] kvm: vmx: Use the hardware provided GPA instead of page walk

[PATCH] kvm: vmx: Use the hardware provided GPA instead of page walk

[Paolo Bonzini]

As in the SVM patch, the guest physical address is passed by
VMX to x86_emulate_instruction already, so mark the GPA as available
in vcpu->arch.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 arch/x86/kvm/vmx.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 7c3e42623090..0828b02b5af2 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -6382,6 +6382,7 @@ static int handle_ept_violation(struct kvm_vcpu *vcpu)
 			EPT_VIOLATION_EXECUTABLE))
 		      ? PFERR_PRESENT_MASK : 0;
 
+	vcpu->arch.gpa_available = true;
 	vcpu->arch.exit_qualification = exit_qualification;
 
 	return kvm_mmu_page_fault(vcpu, gpa, error_code, NULL, 0);
@@ -6399,6 +6400,7 @@ static int handle_ept_misconfig(struct kvm_vcpu *vcpu)
 	}
 
 	ret = handle_mmio_page_fault(vcpu, gpa, true);
+	vcpu->arch.gpa_available = true;
 	if (likely(ret == RET_MMIO_PF_EMULATE))
 		return x86_emulate_instruction(vcpu, gpa, 0, NULL, 0) ==
 					      EMULATE_DONE;
@@ -8517,6 +8519,7 @@ static int vmx_handle_exit(struct kvm_vcpu *vcpu)
 	u32 vectoring_info = vmx->idt_vectoring_info;
 
 	trace_kvm_exit(exit_reason, vcpu, KVM_ISA_VMX);
+	vcpu->arch.gpa_available = false;
 
 	/*
 	 * Flush logged GPAs PML buffer, this will make dirty_bitmap more
-- 
1.8.3.1

Re: [PATCH] kvm: vmx: Use the hardware provided GPA instead of page walk

[Paolo Bonzini]

On 27/01/2017 19:34, Paolo Bonzini wrote:
> As in the SVM patch, the guest physical address is passed by
> VMX to x86_emulate_instruction already, so mark the GPA as available
> in vcpu->arch.
> 
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
>  arch/x86/kvm/vmx.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index 7c3e42623090..0828b02b5af2 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -6382,6 +6382,7 @@ static int handle_ept_violation(struct kvm_vcpu *vcpu)
>  			EPT_VIOLATION_EXECUTABLE))
>  		      ? PFERR_PRESENT_MASK : 0;
>  
> +	vcpu->arch.gpa_available = true;
>  	vcpu->arch.exit_qualification = exit_qualification;
>  
>  	return kvm_mmu_page_fault(vcpu, gpa, error_code, NULL, 0);
> @@ -6399,6 +6400,7 @@ static int handle_ept_misconfig(struct kvm_vcpu *vcpu)
>  	}
>  
>  	ret = handle_mmio_page_fault(vcpu, gpa, true);
> +	vcpu->arch.gpa_available = true;
>  	if (likely(ret == RET_MMIO_PF_EMULATE))
>  		return x86_emulate_instruction(vcpu, gpa, 0, NULL, 0) ==
>  					      EMULATE_DONE;
> @@ -8517,6 +8519,7 @@ static int vmx_handle_exit(struct kvm_vcpu *vcpu)
>  	u32 vectoring_info = vmx->idt_vectoring_info;
>  
>  	trace_kvm_exit(exit_reason, vcpu, KVM_ISA_VMX);
> +	vcpu->arch.gpa_available = false;
>  
>  	/*
>  	 * Flush logged GPAs PML buffer, this will make dirty_bitmap more
> 

Ping?

Paolo

Re: [PATCH] kvm: vmx: Use the hardware provided GPA instead of page walk

[Radim Krčmář]

2017-01-27 19:34+0100, Paolo Bonzini:
> As in the SVM patch, the guest physical address is passed by
> VMX to x86_emulate_instruction already, so mark the GPA as available
> in vcpu->arch.
> 
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---

In line with the SVM patch,

Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>

Still, I don't see why we need a variable that is touched on every VM
exit.  Is this a mere shortcut to propagating the gpa_available bit with
the gpa argument to kvm_mmu_page_fault() and x86_emulate_instruction()?

Thanks.

Re: [PATCH] kvm: vmx: Use the hardware provided GPA instead of page walk

[Paolo Bonzini]

On 07/02/2017 21:52, Radim Krčmář wrote:
> 2017-01-27 19:34+0100, Paolo Bonzini:
>> As in the SVM patch, the guest physical address is passed by
>> VMX to x86_emulate_instruction already, so mark the GPA as available
>> in vcpu->arch.
>>
>> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
>> ---
> 
> In line with the SVM patch,
> 
> Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>
> 
> Still, I don't see why we need a variable that is touched on every VM
> exit.  Is this a mere shortcut to propagating the gpa_available bit with
> the gpa argument to kvm_mmu_page_fault() and x86_emulate_instruction()?

Yes, pretty much.

Paolo