[PATCH] KVM: nVMX: remove side effects from nested_vmx_exit_reflected

[PATCH] KVM: nVMX: remove side effects from nested_vmx_exit_reflected

[Paolo Bonzini]

The name of nested_vmx_exit_reflected suggests that it's purely
a test, but it actually marks VMCS12 pages as dirty.  Move this to
vmx_handle_exit, observing that the initial nested_run_pending check in
nested_vmx_exit_reflected is pointless---nested_run_pending has just
been cleared in vmx_vcpu_run and won't be set until handle_vmlaunch
or handle_vmresume.

Suggested-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 arch/x86/kvm/vmx/nested.c | 18 ++----------------
 arch/x86/kvm/vmx/nested.h |  1 +
 arch/x86/kvm/vmx/vmx.c    | 19 +++++++++++++++++--
 3 files changed, 20 insertions(+), 18 deletions(-)

diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
index 8578513907d7..4ff859c99946 100644
--- a/arch/x86/kvm/vmx/nested.c
+++ b/arch/x86/kvm/vmx/nested.c
@@ -3527,7 +3527,7 @@ static void vmcs12_save_pending_event(struct kvm_vcpu *vcpu,
 }
 
 
-static void nested_mark_vmcs12_pages_dirty(struct kvm_vcpu *vcpu)
+void nested_mark_vmcs12_pages_dirty(struct kvm_vcpu *vcpu)
 {
 	struct vmcs12 *vmcs12 = get_vmcs12(vcpu);
 	gfn_t gfn;
@@ -5543,8 +5543,7 @@ bool nested_vmx_exit_reflected(struct kvm_vcpu *vcpu, u32 exit_reason)
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
 	struct vmcs12 *vmcs12 = get_vmcs12(vcpu);
 
-	if (vmx->nested.nested_run_pending)
-		return false;
+	WARN_ON_ONCE(vmx->nested.nested_run_pending);
 
 	if (unlikely(vmx->fail)) {
 		trace_kvm_nested_vmenter_failed(
@@ -5553,19 +5552,6 @@ bool nested_vmx_exit_reflected(struct kvm_vcpu *vcpu, u32 exit_reason)
 		return true;
 	}
 
-	/*
-	 * The host physical addresses of some pages of guest memory
-	 * are loaded into the vmcs02 (e.g. vmcs12's Virtual APIC
-	 * Page). The CPU may write to these pages via their host
-	 * physical address while L2 is running, bypassing any
-	 * address-translation-based dirty tracking (e.g. EPT write
-	 * protection).
-	 *
-	 * Mark them dirty on every exit from L2 to prevent them from
-	 * getting out of sync with dirty tracking.
-	 */
-	nested_mark_vmcs12_pages_dirty(vcpu);
-
 	trace_kvm_nested_vmexit(kvm_rip_read(vcpu), exit_reason,
 				vmcs_readl(EXIT_QUALIFICATION),
 				vmx->idt_vectoring_info,
diff --git a/arch/x86/kvm/vmx/nested.h b/arch/x86/kvm/vmx/nested.h
index 21d36652f213..f70968b76d33 100644
--- a/arch/x86/kvm/vmx/nested.h
+++ b/arch/x86/kvm/vmx/nested.h
@@ -33,6 +33,7 @@ void nested_vmx_vmexit(struct kvm_vcpu *vcpu, u32 exit_reason,
 int get_vmx_mem_address(struct kvm_vcpu *vcpu, unsigned long exit_qualification,
 			u32 vmx_instruction_info, bool wr, int len, gva_t *ret);
 void nested_vmx_pmu_entry_exit_ctls_update(struct kvm_vcpu *vcpu);
+void nested_mark_vmcs12_pages_dirty(struct kvm_vcpu *vcpu);
 bool nested_vmx_check_io_bitmaps(struct kvm_vcpu *vcpu, unsigned int port,
 				 int size);
 
diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
index b447d66f44e6..07299a957d4a 100644
--- a/arch/x86/kvm/vmx/vmx.c
+++ b/arch/x86/kvm/vmx/vmx.c
@@ -5851,8 +5851,23 @@ static int vmx_handle_exit(struct kvm_vcpu *vcpu,
 	if (vmx->emulation_required)
 		return handle_invalid_guest_state(vcpu);
 
-	if (is_guest_mode(vcpu) && nested_vmx_exit_reflected(vcpu, exit_reason))
-		return nested_vmx_reflect_vmexit(vcpu, exit_reason);
+	if (is_guest_mode(vcpu)) {
+		/*
+		 * The host physical addresses of some pages of guest memory
+		 * are loaded into the vmcs02 (e.g. vmcs12's Virtual APIC
+		 * Page). The CPU may write to these pages via their host
+		 * physical address while L2 is running, bypassing any
+		 * address-translation-based dirty tracking (e.g. EPT write
+		 * protection).
+		 *
+		 * Mark them dirty on every exit from L2 to prevent them from
+		 * getting out of sync with dirty tracking.
+		 */
+		nested_mark_vmcs12_pages_dirty(vcpu);
+
+		if (nested_vmx_exit_reflected(vcpu, exit_reason))
+			return nested_vmx_reflect_vmexit(vcpu, exit_reason);
+	}
 
 	if (exit_reason & VMX_EXIT_REASONS_FAILED_VMENTRY) {
 		dump_vmcs();
-- 
1.8.3.1

Re: [PATCH] KVM: nVMX: remove side effects from nested_vmx_exit_reflected

[Vitaly Kuznetsov]

Paolo Bonzini <pbonzini@redhat.com> writes:

> The name of nested_vmx_exit_reflected suggests that it's purely
> a test, but it actually marks VMCS12 pages as dirty.  Move this to
> vmx_handle_exit, observing that the initial nested_run_pending check in
> nested_vmx_exit_reflected is pointless---nested_run_pending has just
> been cleared in vmx_vcpu_run and won't be set until handle_vmlaunch
> or handle_vmresume.
>
> Suggested-by: Vitaly Kuznetsov <vkuznets@redhat.com>
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
>  arch/x86/kvm/vmx/nested.c | 18 ++----------------
>  arch/x86/kvm/vmx/nested.h |  1 +
>  arch/x86/kvm/vmx/vmx.c    | 19 +++++++++++++++++--
>  3 files changed, 20 insertions(+), 18 deletions(-)
>
> diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
> index 8578513907d7..4ff859c99946 100644
> --- a/arch/x86/kvm/vmx/nested.c
> +++ b/arch/x86/kvm/vmx/nested.c
> @@ -3527,7 +3527,7 @@ static void vmcs12_save_pending_event(struct kvm_vcpu *vcpu,
>  }
>  
>  
> -static void nested_mark_vmcs12_pages_dirty(struct kvm_vcpu *vcpu)
> +void nested_mark_vmcs12_pages_dirty(struct kvm_vcpu *vcpu)
>  {
>  	struct vmcs12 *vmcs12 = get_vmcs12(vcpu);
>  	gfn_t gfn;
> @@ -5543,8 +5543,7 @@ bool nested_vmx_exit_reflected(struct kvm_vcpu *vcpu, u32 exit_reason)
>  	struct vcpu_vmx *vmx = to_vmx(vcpu);
>  	struct vmcs12 *vmcs12 = get_vmcs12(vcpu);
>  
> -	if (vmx->nested.nested_run_pending)
> -		return false;
> +	WARN_ON_ONCE(vmx->nested.nested_run_pending);
>  
>  	if (unlikely(vmx->fail)) {
>  		trace_kvm_nested_vmenter_failed(
> @@ -5553,19 +5552,6 @@ bool nested_vmx_exit_reflected(struct kvm_vcpu *vcpu, u32 exit_reason)
>  		return true;
>  	}
>  
> -	/*
> -	 * The host physical addresses of some pages of guest memory
> -	 * are loaded into the vmcs02 (e.g. vmcs12's Virtual APIC
> -	 * Page). The CPU may write to these pages via their host
> -	 * physical address while L2 is running, bypassing any
> -	 * address-translation-based dirty tracking (e.g. EPT write
> -	 * protection).
> -	 *
> -	 * Mark them dirty on every exit from L2 to prevent them from
> -	 * getting out of sync with dirty tracking.
> -	 */
> -	nested_mark_vmcs12_pages_dirty(vcpu);
> -
>  	trace_kvm_nested_vmexit(kvm_rip_read(vcpu), exit_reason,
>  				vmcs_readl(EXIT_QUALIFICATION),
>  				vmx->idt_vectoring_info,
> diff --git a/arch/x86/kvm/vmx/nested.h b/arch/x86/kvm/vmx/nested.h
> index 21d36652f213..f70968b76d33 100644
> --- a/arch/x86/kvm/vmx/nested.h
> +++ b/arch/x86/kvm/vmx/nested.h
> @@ -33,6 +33,7 @@ void nested_vmx_vmexit(struct kvm_vcpu *vcpu, u32 exit_reason,
>  int get_vmx_mem_address(struct kvm_vcpu *vcpu, unsigned long exit_qualification,
>  			u32 vmx_instruction_info, bool wr, int len, gva_t *ret);
>  void nested_vmx_pmu_entry_exit_ctls_update(struct kvm_vcpu *vcpu);
> +void nested_mark_vmcs12_pages_dirty(struct kvm_vcpu *vcpu);
>  bool nested_vmx_check_io_bitmaps(struct kvm_vcpu *vcpu, unsigned int port,
>  				 int size);
>  
> diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
> index b447d66f44e6..07299a957d4a 100644
> --- a/arch/x86/kvm/vmx/vmx.c
> +++ b/arch/x86/kvm/vmx/vmx.c
> @@ -5851,8 +5851,23 @@ static int vmx_handle_exit(struct kvm_vcpu *vcpu,
>  	if (vmx->emulation_required)
>  		return handle_invalid_guest_state(vcpu);
>  
> -	if (is_guest_mode(vcpu) && nested_vmx_exit_reflected(vcpu, exit_reason))
> -		return nested_vmx_reflect_vmexit(vcpu, exit_reason);
> +	if (is_guest_mode(vcpu)) {
> +		/*
> +		 * The host physical addresses of some pages of guest memory
> +		 * are loaded into the vmcs02 (e.g. vmcs12's Virtual APIC
> +		 * Page). The CPU may write to these pages via their host
> +		 * physical address while L2 is running, bypassing any
> +		 * address-translation-based dirty tracking (e.g. EPT write
> +		 * protection).
> +		 *
> +		 * Mark them dirty on every exit from L2 to prevent them from
> +		 * getting out of sync with dirty tracking.
> +		 */
> +		nested_mark_vmcs12_pages_dirty(vcpu);
> +
> +		if (nested_vmx_exit_reflected(vcpu, exit_reason))
> +			return nested_vmx_reflect_vmexit(vcpu, exit_reason);
> +	}
>  
>  	if (exit_reason & VMX_EXIT_REASONS_FAILED_VMENTRY) {
>  		dump_vmcs();

The only functional difference seems to be that we're now doing
nested_mark_vmcs12_pages_dirty() in vmx->fail case too and this seems
superfluous: we failed to enter L2 so 'special' pages should remain
intact (right?) but this should be an uncommon case.

Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>

-- 
Vitaly

Re: [PATCH] KVM: nVMX: remove side effects from nested_vmx_exit_reflected

[Paolo Bonzini]

On 18/03/20 11:52, Vitaly Kuznetsov wrote:
> The only functional difference seems to be that we're now doing
> nested_mark_vmcs12_pages_dirty() in vmx->fail case too and this seems
> superfluous: we failed to enter L2 so 'special' pages should remain
> intact (right?) but this should be an uncommon case.
> 
> Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>

I'm not entirely sure if the PID could be written before the processor
decrees a vmfail.  It doesn't really hurt anyway as you say though.
Thanks for the review!

Paolo

Re: [PATCH] KVM: nVMX: remove side effects from nested_vmx_exit_reflected

[Sean Christopherson]

On Wed, Mar 18, 2020 at 11:59:13AM +0100, Paolo Bonzini wrote:
> On 18/03/20 11:52, Vitaly Kuznetsov wrote:
> > The only functional difference seems to be that we're now doing
> > nested_mark_vmcs12_pages_dirty() in vmx->fail case too and this seems
> > superfluous: we failed to enter L2 so 'special' pages should remain
> > intact (right?) but this should be an uncommon case.
> > 
> > Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
> 
> I'm not entirely sure if the PID could be written before the processor
> decrees a vmfail.  It doesn't really hurt anyway as you say though.

I would expect that writing special pages on VM-Fail would be classified
as a CPU bug.

Re: [PATCH] KVM: nVMX: remove side effects from nested_vmx_exit_reflected

[Krish Sadhukhan]

On 3/17/20 11:00 AM, Paolo Bonzini wrote:
> The name of nested_vmx_exit_reflected suggests that it's purely
> a test, but it actually marks VMCS12 pages as dirty.  Move this to
> vmx_handle_exit, observing that the initial nested_run_pending check in
> nested_vmx_exit_reflected is pointless---nested_run_pending has just
> been cleared in vmx_vcpu_run and won't be set until handle_vmlaunch
> or handle_vmresume.
>
> Suggested-by: Vitaly Kuznetsov <vkuznets@redhat.com>
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
>   arch/x86/kvm/vmx/nested.c | 18 ++----------------
>   arch/x86/kvm/vmx/nested.h |  1 +
>   arch/x86/kvm/vmx/vmx.c    | 19 +++++++++++++++++--
>   3 files changed, 20 insertions(+), 18 deletions(-)
>
> diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
> index 8578513907d7..4ff859c99946 100644
> --- a/arch/x86/kvm/vmx/nested.c
> +++ b/arch/x86/kvm/vmx/nested.c
> @@ -3527,7 +3527,7 @@ static void vmcs12_save_pending_event(struct kvm_vcpu *vcpu,
>   }
>   
>   
> -static void nested_mark_vmcs12_pages_dirty(struct kvm_vcpu *vcpu)
> +void nested_mark_vmcs12_pages_dirty(struct kvm_vcpu *vcpu)
>   {
>   	struct vmcs12 *vmcs12 = get_vmcs12(vcpu);
>   	gfn_t gfn;
> @@ -5543,8 +5543,7 @@ bool nested_vmx_exit_reflected(struct kvm_vcpu *vcpu, u32 exit_reason)
>   	struct vcpu_vmx *vmx = to_vmx(vcpu);
>   	struct vmcs12 *vmcs12 = get_vmcs12(vcpu);
>   
> -	if (vmx->nested.nested_run_pending)
> -		return false;
> +	WARN_ON_ONCE(vmx->nested.nested_run_pending);
>   
>   	if (unlikely(vmx->fail)) {
>   		trace_kvm_nested_vmenter_failed(
> @@ -5553,19 +5552,6 @@ bool nested_vmx_exit_reflected(struct kvm_vcpu *vcpu, u32 exit_reason)
>   		return true;
>   	}
>   
> -	/*
> -	 * The host physical addresses of some pages of guest memory
> -	 * are loaded into the vmcs02 (e.g. vmcs12's Virtual APIC
> -	 * Page). The CPU may write to these pages via their host
> -	 * physical address while L2 is running, bypassing any
> -	 * address-translation-based dirty tracking (e.g. EPT write
> -	 * protection).
> -	 *
> -	 * Mark them dirty on every exit from L2 to prevent them from
> -	 * getting out of sync with dirty tracking.
> -	 */
> -	nested_mark_vmcs12_pages_dirty(vcpu);
> -
>   	trace_kvm_nested_vmexit(kvm_rip_read(vcpu), exit_reason,
>   				vmcs_readl(EXIT_QUALIFICATION),
>   				vmx->idt_vectoring_info,
> diff --git a/arch/x86/kvm/vmx/nested.h b/arch/x86/kvm/vmx/nested.h
> index 21d36652f213..f70968b76d33 100644
> --- a/arch/x86/kvm/vmx/nested.h
> +++ b/arch/x86/kvm/vmx/nested.h
> @@ -33,6 +33,7 @@ void nested_vmx_vmexit(struct kvm_vcpu *vcpu, u32 exit_reason,
>   int get_vmx_mem_address(struct kvm_vcpu *vcpu, unsigned long exit_qualification,
>   			u32 vmx_instruction_info, bool wr, int len, gva_t *ret);
>   void nested_vmx_pmu_entry_exit_ctls_update(struct kvm_vcpu *vcpu);
> +void nested_mark_vmcs12_pages_dirty(struct kvm_vcpu *vcpu);
>   bool nested_vmx_check_io_bitmaps(struct kvm_vcpu *vcpu, unsigned int port,
>   				 int size);
>   
> diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
> index b447d66f44e6..07299a957d4a 100644
> --- a/arch/x86/kvm/vmx/vmx.c
> +++ b/arch/x86/kvm/vmx/vmx.c
> @@ -5851,8 +5851,23 @@ static int vmx_handle_exit(struct kvm_vcpu *vcpu,
>   	if (vmx->emulation_required)
>   		return handle_invalid_guest_state(vcpu);
>   
> -	if (is_guest_mode(vcpu) && nested_vmx_exit_reflected(vcpu, exit_reason))
> -		return nested_vmx_reflect_vmexit(vcpu, exit_reason);
> +	if (is_guest_mode(vcpu)) {
> +		/*
> +		 * The host physical addresses of some pages of guest memory
> +		 * are loaded into the vmcs02 (e.g. vmcs12's Virtual APIC
> +		 * Page). The CPU may write to these pages via their host
> +		 * physical address while L2 is running, bypassing any
> +		 * address-translation-based dirty tracking (e.g. EPT write
> +		 * protection).
> +		 *
> +		 * Mark them dirty on every exit from L2 to prevent them from
> +		 * getting out of sync with dirty tracking.
> +		 */
> +		nested_mark_vmcs12_pages_dirty(vcpu);
> +
> +		if (nested_vmx_exit_reflected(vcpu, exit_reason))
> +			return nested_vmx_reflect_vmexit(vcpu, exit_reason);
> +	}
>   
>   	if (exit_reason & VMX_EXIT_REASONS_FAILED_VMENTRY) {
>   		dump_vmcs();
Reviewed-by: Krish Sadhukhan <krish.sadhukhan@oracle.com>