[PATCH 0/3] KVM: x86: fix a couple of issues with Enlightened VMCS enablement

[PATCH 0/3] KVM: x86: fix a couple of issues with Enlightened VMCS enablement

[Vitaly Kuznetsov]

It was discovered that:
- hyperv_cpuid test fails on AMD
- hyperv_cpuid test crashes kernel on Intel when nested=0
The test itself is good, we need to fix the issues.

Vitaly Kuznetsov (3):
  KVM: x86: hyper-v: don't crash on KVM_GET_SUPPORTED_HV_CPUID when
    kvm_intel.nested is disabled
  KVM: x86: svm: remove unneeded nested_enable_evmcs() hook
  KVM: x86: announce KVM_CAP_HYPERV_ENLIGHTENED_VMCS support only when
    it is available

 arch/x86/kvm/hyperv.c  |  5 ++++-
 arch/x86/kvm/svm.c     | 17 ++---------------
 arch/x86/kvm/vmx/vmx.c |  1 +
 arch/x86/kvm/x86.c     |  3 ++-
 4 files changed, 9 insertions(+), 17 deletions(-)

-- 
2.20.1

[PATCH 1/3] KVM: x86: hyper-v: don't crash on KVM_GET_SUPPORTED_HV_CPUID when kvm_intel.nested is disabled

[Vitaly Kuznetsov]

If kvm_intel is loaded with nested=0 parameter an attempt to perform
KVM_GET_SUPPORTED_HV_CPUID results in OOPS as nested_get_evmcs_version hook
in kvm_x86_ops is NULL (we assign it in nested_vmx_hardware_setup() and
this only happens in case nested is enabled).

Check that kvm_x86_ops->nested_get_evmcs_version is not NULL before
calling it. With this, we can remove the stub from svm as it is no
longer needed.

Fixes: e2e871ab2f02 ("x86/kvm/hyper-v: Introduce nested_get_evmcs_version() helper")
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
---
 arch/x86/kvm/hyperv.c  | 5 ++++-
 arch/x86/kvm/svm.c     | 8 +-------
 arch/x86/kvm/vmx/vmx.c | 1 +
 3 files changed, 6 insertions(+), 8 deletions(-)

diff --git a/arch/x86/kvm/hyperv.c b/arch/x86/kvm/hyperv.c
index 58bf61b17431..3f5ad84853fb 100644
--- a/arch/x86/kvm/hyperv.c
+++ b/arch/x86/kvm/hyperv.c
@@ -1785,7 +1785,7 @@ int kvm_vm_ioctl_hv_eventfd(struct kvm *kvm, struct kvm_hyperv_eventfd *args)
 int kvm_vcpu_ioctl_get_hv_cpuid(struct kvm_vcpu *vcpu, struct kvm_cpuid2 *cpuid,
 				struct kvm_cpuid_entry2 __user *entries)
 {
-	uint16_t evmcs_ver = kvm_x86_ops->nested_get_evmcs_version(vcpu);
+	uint16_t evmcs_ver = 0;
 	struct kvm_cpuid_entry2 cpuid_entries[] = {
 		{ .function = HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS },
 		{ .function = HYPERV_CPUID_INTERFACE },
@@ -1797,6 +1797,9 @@ int kvm_vcpu_ioctl_get_hv_cpuid(struct kvm_vcpu *vcpu, struct kvm_cpuid2 *cpuid,
 	};
 	int i, nent = ARRAY_SIZE(cpuid_entries);
 
+	if (kvm_x86_ops->nested_get_evmcs_version)
+		evmcs_ver = kvm_x86_ops->nested_get_evmcs_version(vcpu);
+
 	/* Skip NESTED_FEATURES if eVMCS is not supported */
 	if (!evmcs_ver)
 		--nent;
diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
index 572c7c4ca974..40175c42f116 100644
--- a/arch/x86/kvm/svm.c
+++ b/arch/x86/kvm/svm.c
@@ -7122,12 +7122,6 @@ static int svm_unregister_enc_region(struct kvm *kvm,
 	return ret;
 }
 
-static uint16_t nested_get_evmcs_version(struct kvm_vcpu *vcpu)
-{
-	/* Not supported */
-	return 0;
-}
-
 static int nested_enable_evmcs(struct kvm_vcpu *vcpu,
 				   uint16_t *vmcs_version)
 {
@@ -7344,7 +7338,7 @@ static struct kvm_x86_ops svm_x86_ops __ro_after_init = {
 	.mem_enc_unreg_region = svm_unregister_enc_region,
 
 	.nested_enable_evmcs = nested_enable_evmcs,
-	.nested_get_evmcs_version = nested_get_evmcs_version,
+	.nested_get_evmcs_version = NULL,
 
 	.need_emulation_on_page_fault = svm_need_emulation_on_page_fault,
 
diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
index 3c936e7366b9..c81e5210b159 100644
--- a/arch/x86/kvm/vmx/vmx.c
+++ b/arch/x86/kvm/vmx/vmx.c
@@ -7812,6 +7812,7 @@ static struct kvm_x86_ops vmx_x86_ops __ro_after_init = {
 	.set_nested_state = NULL,
 	.get_vmcs12_pages = NULL,
 	.nested_enable_evmcs = NULL,
+	.nested_get_evmcs_version = NULL,
 	.need_emulation_on_page_fault = vmx_need_emulation_on_page_fault,
 	.apic_init_signal_blocked = vmx_apic_init_signal_blocked,
 };
-- 
2.20.1

[PATCH 2/3] KVM: x86: svm: remove unneeded nested_enable_evmcs() hook

[Vitaly Kuznetsov]

Since commit 5158917c7b019 ("KVM: x86: nVMX: Allow nested_enable_evmcs to
be NULL") the code in x86.c is prepared to see nested_enable_evmcs being
NULL and in VMX case it actually is when nesting is disabled. Remove the
unneeded stub from SVM code.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
---
 arch/x86/kvm/svm.c | 9 +--------
 1 file changed, 1 insertion(+), 8 deletions(-)

diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
index 40175c42f116..6d52c65d625b 100644
--- a/arch/x86/kvm/svm.c
+++ b/arch/x86/kvm/svm.c
@@ -7122,13 +7122,6 @@ static int svm_unregister_enc_region(struct kvm *kvm,
 	return ret;
 }
 
-static int nested_enable_evmcs(struct kvm_vcpu *vcpu,
-				   uint16_t *vmcs_version)
-{
-	/* Intel-only feature */
-	return -ENODEV;
-}
-
 static bool svm_need_emulation_on_page_fault(struct kvm_vcpu *vcpu)
 {
 	unsigned long cr4 = kvm_read_cr4(vcpu);
@@ -7337,7 +7330,7 @@ static struct kvm_x86_ops svm_x86_ops __ro_after_init = {
 	.mem_enc_reg_region = svm_register_enc_region,
 	.mem_enc_unreg_region = svm_unregister_enc_region,
 
-	.nested_enable_evmcs = nested_enable_evmcs,
+	.nested_enable_evmcs = NULL,
 	.nested_get_evmcs_version = NULL,
 
 	.need_emulation_on_page_fault = svm_need_emulation_on_page_fault,
-- 
2.20.1

[PATCH 3/3] KVM: x86: announce KVM_CAP_HYPERV_ENLIGHTENED_VMCS support only when it is available

[Vitaly Kuznetsov]

It was discovered that after commit 65efa61dc0d5 ("selftests: kvm: provide
common function to enable eVMCS") hyperv_cpuid selftest is failing on AMD.
The reason is that the commit changed _vcpu_ioctl() to vcpu_ioctl() in the
test and this one can't fail.

Instead of fixing the test is seems to make more sense to not announce
KVM_CAP_HYPERV_ENLIGHTENED_VMCS support if it is definitely missing
(on svm and in case kvm_intel.nested=0).

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
---
 arch/x86/kvm/x86.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index d1cd0fcff9e7..ef2e8b138300 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -3106,7 +3106,6 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 	case KVM_CAP_HYPERV_EVENTFD:
 	case KVM_CAP_HYPERV_TLBFLUSH:
 	case KVM_CAP_HYPERV_SEND_IPI:
-	case KVM_CAP_HYPERV_ENLIGHTENED_VMCS:
 	case KVM_CAP_HYPERV_CPUID:
 	case KVM_CAP_PCI_SEGMENT:
 	case KVM_CAP_DEBUGREGS:
@@ -3183,6 +3182,8 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 		r = kvm_x86_ops->get_nested_state ?
 			kvm_x86_ops->get_nested_state(NULL, NULL, 0) : 0;
 		break;
+	case KVM_CAP_HYPERV_ENLIGHTENED_VMCS:
+		r = kvm_x86_ops->nested_enable_evmcs != NULL;
 	default:
 		break;
 	}
-- 
2.20.1

Re: [PATCH 3/3] KVM: x86: announce KVM_CAP_HYPERV_ENLIGHTENED_VMCS support only when it is available

[Jim Mattson]

On Tue, Aug 27, 2019 at 9:04 AM Vitaly Kuznetsov <vkuznets@redhat.com> wrote:
>
> It was discovered that after commit 65efa61dc0d5 ("selftests: kvm: provide
> common function to enable eVMCS") hyperv_cpuid selftest is failing on AMD.
> The reason is that the commit changed _vcpu_ioctl() to vcpu_ioctl() in the
> test and this one can't fail.
>
> Instead of fixing the test is seems to make more sense to not announce
> KVM_CAP_HYPERV_ENLIGHTENED_VMCS support if it is definitely missing
> (on svm and in case kvm_intel.nested=0).
>
> Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
> ---
>  arch/x86/kvm/x86.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index d1cd0fcff9e7..ef2e8b138300 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -3106,7 +3106,6 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
>         case KVM_CAP_HYPERV_EVENTFD:
>         case KVM_CAP_HYPERV_TLBFLUSH:
>         case KVM_CAP_HYPERV_SEND_IPI:
> -       case KVM_CAP_HYPERV_ENLIGHTENED_VMCS:
>         case KVM_CAP_HYPERV_CPUID:
>         case KVM_CAP_PCI_SEGMENT:
>         case KVM_CAP_DEBUGREGS:
> @@ -3183,6 +3182,8 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
>                 r = kvm_x86_ops->get_nested_state ?
>                         kvm_x86_ops->get_nested_state(NULL, NULL, 0) : 0;
>                 break;
> +       case KVM_CAP_HYPERV_ENLIGHTENED_VMCS:
> +               r = kvm_x86_ops->nested_enable_evmcs != NULL;

You should probably have an explicit break here, in case someone later
adds another case below.

>         default:
>                 break;
>         }
> --
> 2.20.1
>

Re: [PATCH 2/3] KVM: x86: svm: remove unneeded nested_enable_evmcs() hook

[Jim Mattson]

On Tue, Aug 27, 2019 at 9:04 AM Vitaly Kuznetsov <vkuznets@redhat.com> wrote:
>
> Since commit 5158917c7b019 ("KVM: x86: nVMX: Allow nested_enable_evmcs to
> be NULL") the code in x86.c is prepared to see nested_enable_evmcs being
> NULL and in VMX case it actually is when nesting is disabled. Remove the
> unneeded stub from SVM code.
>
> Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: Jim Mattson <jmattson@google.com>

Re: [PATCH 1/3] KVM: x86: hyper-v: don't crash on KVM_GET_SUPPORTED_HV_CPUID when kvm_intel.nested is disabled

[Jim Mattson]

On Tue, Aug 27, 2019 at 9:04 AM Vitaly Kuznetsov <vkuznets@redhat.com> wrote:
>
> If kvm_intel is loaded with nested=0 parameter an attempt to perform
> KVM_GET_SUPPORTED_HV_CPUID results in OOPS as nested_get_evmcs_version hook
> in kvm_x86_ops is NULL (we assign it in nested_vmx_hardware_setup() and
> this only happens in case nested is enabled).
>
> Check that kvm_x86_ops->nested_get_evmcs_version is not NULL before
> calling it. With this, we can remove the stub from svm as it is no
> longer needed.
>
> Fixes: e2e871ab2f02 ("x86/kvm/hyper-v: Introduce nested_get_evmcs_version() helper")
> Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: Jim Mattson <jmattson@google.com>

Re: [PATCH 1/3] KVM: x86: hyper-v: don't crash on KVM_GET_SUPPORTED_HV_CPUID when kvm_intel.nested is disabled

[Radim Krčmář]

2019-08-27 18:04+0200, Vitaly Kuznetsov:
> If kvm_intel is loaded with nested=0 parameter an attempt to perform
> KVM_GET_SUPPORTED_HV_CPUID results in OOPS as nested_get_evmcs_version hook
> in kvm_x86_ops is NULL (we assign it in nested_vmx_hardware_setup() and
> this only happens in case nested is enabled).
> 
> Check that kvm_x86_ops->nested_get_evmcs_version is not NULL before
> calling it. With this, we can remove the stub from svm as it is no
> longer needed.
> 

Added

Cc: <stable@vger.kernel.org>

> Fixes: e2e871ab2f02 ("x86/kvm/hyper-v: Introduce nested_get_evmcs_version() helper")
> Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>

and applied, thanks.

Re: [PATCH 3/3] KVM: x86: announce KVM_CAP_HYPERV_ENLIGHTENED_VMCS support only when it is available

[Sean Christopherson]

On Tue, Aug 27, 2019 at 09:54:39AM -0700, Jim Mattson wrote:
> On Tue, Aug 27, 2019 at 9:04 AM Vitaly Kuznetsov <vkuznets@redhat.com> wrote:
> >
> > It was discovered that after commit 65efa61dc0d5 ("selftests: kvm: provide
> > common function to enable eVMCS") hyperv_cpuid selftest is failing on AMD.
> > The reason is that the commit changed _vcpu_ioctl() to vcpu_ioctl() in the
> > test and this one can't fail.
> >
> > Instead of fixing the test is seems to make more sense to not announce
> > KVM_CAP_HYPERV_ENLIGHTENED_VMCS support if it is definitely missing
> > (on svm and in case kvm_intel.nested=0).
> >
> > Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
> > ---
> >  arch/x86/kvm/x86.c | 3 ++-
> >  1 file changed, 2 insertions(+), 1 deletion(-)
> >
> > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> > index d1cd0fcff9e7..ef2e8b138300 100644
> > --- a/arch/x86/kvm/x86.c
> > +++ b/arch/x86/kvm/x86.c
> > @@ -3106,7 +3106,6 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
> >         case KVM_CAP_HYPERV_EVENTFD:
> >         case KVM_CAP_HYPERV_TLBFLUSH:
> >         case KVM_CAP_HYPERV_SEND_IPI:
> > -       case KVM_CAP_HYPERV_ENLIGHTENED_VMCS:
> >         case KVM_CAP_HYPERV_CPUID:
> >         case KVM_CAP_PCI_SEGMENT:
> >         case KVM_CAP_DEBUGREGS:
> > @@ -3183,6 +3182,8 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
> >                 r = kvm_x86_ops->get_nested_state ?
> >                         kvm_x86_ops->get_nested_state(NULL, NULL, 0) : 0;
> >                 break;
> > +       case KVM_CAP_HYPERV_ENLIGHTENED_VMCS:
> > +               r = kvm_x86_ops->nested_enable_evmcs != NULL;
> 
> You should probably have an explicit break here, in case someone later
> adds another case below.

Yep, this will trigger a warning on compilers with -Wimplicit-fallthrough.

Re: [PATCH 3/3] KVM: x86: announce KVM_CAP_HYPERV_ENLIGHTENED_VMCS support only when it is available

[Vitaly Kuznetsov]

Sean Christopherson <sean.j.christopherson@intel.com> writes:

> On Tue, Aug 27, 2019 at 09:54:39AM -0700, Jim Mattson wrote:
>> On Tue, Aug 27, 2019 at 9:04 AM Vitaly Kuznetsov <vkuznets@redhat.com> wrote:
>> >
>> > It was discovered that after commit 65efa61dc0d5 ("selftests: kvm: provide
>> > common function to enable eVMCS") hyperv_cpuid selftest is failing on AMD.
>> > The reason is that the commit changed _vcpu_ioctl() to vcpu_ioctl() in the
>> > test and this one can't fail.
>> >
>> > Instead of fixing the test is seems to make more sense to not announce
>> > KVM_CAP_HYPERV_ENLIGHTENED_VMCS support if it is definitely missing
>> > (on svm and in case kvm_intel.nested=0).
>> >
>> > Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
>> > ---
>> >  arch/x86/kvm/x86.c | 3 ++-
>> >  1 file changed, 2 insertions(+), 1 deletion(-)
>> >
>> > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
>> > index d1cd0fcff9e7..ef2e8b138300 100644
>> > --- a/arch/x86/kvm/x86.c
>> > +++ b/arch/x86/kvm/x86.c
>> > @@ -3106,7 +3106,6 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
>> >         case KVM_CAP_HYPERV_EVENTFD:
>> >         case KVM_CAP_HYPERV_TLBFLUSH:
>> >         case KVM_CAP_HYPERV_SEND_IPI:
>> > -       case KVM_CAP_HYPERV_ENLIGHTENED_VMCS:
>> >         case KVM_CAP_HYPERV_CPUID:
>> >         case KVM_CAP_PCI_SEGMENT:
>> >         case KVM_CAP_DEBUGREGS:
>> > @@ -3183,6 +3182,8 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
>> >                 r = kvm_x86_ops->get_nested_state ?
>> >                         kvm_x86_ops->get_nested_state(NULL, NULL, 0) : 0;
>> >                 break;
>> > +       case KVM_CAP_HYPERV_ENLIGHTENED_VMCS:
>> > +               r = kvm_x86_ops->nested_enable_evmcs != NULL;
>> 
>> You should probably have an explicit break here, in case someone later
>> adds another case below.
>
> Yep, this will trigger a warning on compilers with -Wimplicit-fallthrough.

I always forget there's more than checkpatch.pl :-) Thanks you your
reviews guys, I'm going to send v2 of this patchset without PATCH1 which
was already queued by Radim.

-- 
Vitaly