Skip to contentSkip to conversation

KVM: nVMX: Set X86_CR4_CET in cr4_fixed1_bits if CET IBT is enabled

PATCH In review Yang Weijiang 4 participants 8 messages Feb 25, 2021 – Mar 3, 2021

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 KVM: nVMX: Set X86_CR4_CET in cr4_fixed1_bits if CET IBT is enabled

CET SHSTK and IBT are independently controlled by kernel, set X86_CR4_CET bit in cr4_fixed1_bits if either of them is enabled so that nested guest can enjoy the feature.

---
 arch/x86/kvm/vmx/vmx.c | 1 +
 1 file changed, 1 insertion(+)
arch/x86/kvm/vmx/vmx.c +1 -0
index 5856c5b81084..e92134ee081c 100644
@@ -7258,6 +7258,7 @@ static void nested_vmx_cr_fixed1_bits_update(struct kvm_vcpu *vcpu)
 	cr4_fixed1_update(X86_CR4_UMIP,       ecx, feature_bit(UMIP));
 	cr4_fixed1_update(X86_CR4_LA57,       ecx, feature_bit(LA57));
 	cr4_fixed1_update(X86_CR4_CET,	      ecx, feature_bit(SHSTK));
+	cr4_fixed1_update(X86_CR4_CET,	      edx, feature_bit(IBT));
 
 #undef cr4_fixed1_update
 }
Signature · 1 line
2.26.2
#1 lore raw

PATCH v2 1/1 KVM: nVMX: Sync L2 guest CET states between L1/L2

These fields are rarely updated by L1 QEMU/KVM, sync them when L1 is trying to read/write them and after they're changed. If CET guest entry-load bit is not set by L1 guest, migrate them to L2 manaully.

change in v2:
- Per Sean's review feedback, change CET guest states as rarely-updated fields. And also migrate L1's CET states to L2 if the entry-load bit is not set.
- Opportunistically removed one blank line.

---
 arch/x86/kvm/cpuid.c      |  1 -
 arch/x86/kvm/vmx/nested.c | 29 +++++++++++++++++++++++++++++
 arch/x86/kvm/vmx/vmx.h    |  3 +++
 3 files changed, 32 insertions(+), 1 deletion(-)
arch/x86/kvm/cpuid.c +0 -1
index 46087bca9418..afc97122c05c 100644
@@ -143,7 +143,6 @@ void kvm_update_cpuid_runtime(struct kvm_vcpu *vcpu)
 		}
 		vcpu->arch.guest_supported_xss =
 			(((u64)best->edx << 32) | best->ecx) & supported_xss;
-
 	} else {
 		vcpu->arch.guest_supported_xss = 0;
 	}
arch/x86/kvm/vmx/nested.c +29 -0
index 9728efd529a1..1703b8874fad 100644
@@ -2516,6 +2516,12 @@ static void prepare_vmcs02_rare(struct vcpu_vmx *vmx, struct vmcs12 *vmcs12)
 	vmcs_write32(VM_ENTRY_MSR_LOAD_COUNT, vmx->msr_autoload.guest.nr);
 
 	set_cr4_guest_host_mask(vmx);
+
+	if (kvm_cet_supported()) {
+		vmcs_writel(GUEST_SSP, vmcs12->guest_ssp);
+		vmcs_writel(GUEST_S_CET, vmcs12->guest_s_cet);
+		vmcs_writel(GUEST_INTR_SSP_TABLE, vmcs12->guest_ssp_tbl);
+	}
 }
 
 /*
@@ -2556,6 +2562,15 @@ static int prepare_vmcs02(struct kvm_vcpu *vcpu, struct vmcs12 *vmcs12,
 	if (kvm_mpx_supported() && (!vmx->nested.nested_run_pending ||
 	    !(vmcs12->vm_entry_controls & VM_ENTRY_LOAD_BNDCFGS)))
 		vmcs_write64(GUEST_BNDCFGS, vmx->nested.vmcs01_guest_bndcfgs);
+
+	if (kvm_cet_supported() && (!vmx->nested.nested_run_pending ||
+	    !(vmcs12->vm_entry_controls & VM_ENTRY_LOAD_CET_STATE))) {
+		vmcs_writel(GUEST_SSP, vmx->nested.vmcs01_guest_ssp);
+		vmcs_writel(GUEST_S_CET, vmx->nested.vmcs01_guest_s_cet);
+		vmcs_writel(GUEST_INTR_SSP_TABLE,
+			    vmx->nested.vmcs01_guest_ssp_tbl);
+	}
+
 	vmx_set_rflags(vcpu, vmcs12->guest_rflags);
 
 	/* EXCEPTION_BITMAP and CR0_GUEST_HOST_MASK should basically be the
@@ -3375,6 +3390,12 @@ enum nvmx_vmentry_status nested_vmx_enter_non_root_mode(struct kvm_vcpu *vcpu,
 	if (kvm_mpx_supported() &&
 		!(vmcs12->vm_entry_controls & VM_ENTRY_LOAD_BNDCFGS))
 		vmx->nested.vmcs01_guest_bndcfgs = vmcs_read64(GUEST_BNDCFGS);
+	if (kvm_cet_supported() &&
+		!(vmcs12->vm_entry_controls & VM_ENTRY_LOAD_CET_STATE)) {
+		vmx->nested.vmcs01_guest_ssp = vmcs_readl(GUEST_SSP);
+		vmx->nested.vmcs01_guest_s_cet = vmcs_readl(GUEST_S_CET);
+		vmx->nested.vmcs01_guest_ssp_tbl = vmcs_readl(GUEST_INTR_SSP_TABLE);
+	}
 
 	/*
 	 * Overwrite vmcs01.GUEST_CR3 with L1's CR3 if EPT is disabled *and*
@@ -4001,6 +4022,9 @@ static bool is_vmcs12_ext_field(unsigned long field)
 	case GUEST_IDTR_BASE:
 	case GUEST_PENDING_DBG_EXCEPTIONS:
 	case GUEST_BNDCFGS:
+	case GUEST_SSP:
+	case GUEST_INTR_SSP_TABLE:
+	case GUEST_S_CET:
 		return true;
 	default:
 		break;
@@ -4052,6 +4076,11 @@ static void sync_vmcs02_to_vmcs12_rare(struct kvm_vcpu *vcpu,
 		vmcs_readl(GUEST_PENDING_DBG_EXCEPTIONS);
 	if (kvm_mpx_supported())
 		vmcs12->guest_bndcfgs = vmcs_read64(GUEST_BNDCFGS);
+	if (kvm_cet_supported()) {
+		vmcs12->guest_ssp = vmcs_readl(GUEST_SSP);
+		vmcs12->guest_s_cet = vmcs_readl(GUEST_S_CET);
+		vmcs12->guest_ssp_tbl = vmcs_readl(GUEST_INTR_SSP_TABLE);
+	}
 
 	vmx->nested.need_sync_vmcs02_to_vmcs12_rare = false;
 }
arch/x86/kvm/vmx/vmx.h +3 -0
index 9d3a557949ac..36dc4fdb0909 100644
@@ -155,6 +155,9 @@ struct nested_vmx {
 	/* to migrate it to L2 if VM_ENTRY_LOAD_DEBUG_CONTROLS is off */
 	u64 vmcs01_debugctl;
 	u64 vmcs01_guest_bndcfgs;
+	u64 vmcs01_guest_ssp;
+	u64 vmcs01_guest_s_cet;
+	u64 vmcs01_guest_ssp_tbl;
 
 	/* to migrate it to L1 if L2 writes to L1's CR8 directly */
 	int l1_tpr_threshold;
Signature · 1 line
2.26.2
#2 lore raw

Re: [PATCH v2] KVM: nVMX: Sync L2 guest CET states between L1/L2

+Vitaly

Quoted 34 lines · Yang Weijiang

These fields are rarely updated by L1 QEMU/KVM, sync them when L1 is trying to read/write them and after they're changed. If CET guest entry-load bit is not set by L1 guest, migrate them to L2 manaully.

change in v2:
- Per Sean's review feedback, change CET guest states as rarely-updated fields. And also migrate L1's CET states to L2 if the entry-load bit is not set.
- Opportunistically removed one blank line.

---
 arch/x86/kvm/cpuid.c      |  1 -
 arch/x86/kvm/vmx/nested.c | 29 +++++++++++++++++++++++++++++
 arch/x86/kvm/vmx/vmx.h    |  3 +++
 3 files changed, 32 insertions(+), 1 deletion(-)
arch/x86/kvm/cpuid.c +0 -1
index 46087bca9418..afc97122c05c 100644
@@ -143,7 +143,6 @@ void kvm_update_cpuid_runtime(struct kvm_vcpu *vcpu)
 		}
 		vcpu->arch.guest_supported_xss =
 			(((u64)best->edx << 32) | best->ecx) & supported_xss;
-
 	} else {
 		vcpu->arch.guest_supported_xss = 0;
 	}
arch/x86/kvm/vmx/nested.c +0 -0
index 9728efd529a1..1703b8874fad 100644

arch/x86/kvm/vmx/nested.c static void prepare_vmcs02_rare(struct vcpu_vmx *vmx, struct vmcs12 *vmcs12) · line 2520

Quoted 5 lines
arch/x86/kvm/vmx/nested.c +2 -0
@@ -2516,6 +2516,12 @@ static void prepare_vmcs02_rare(struct vcpu_vmx *vmx, struct vmcs12 *vmcs12)
 	vmcs_write32(VM_ENTRY_MSR_LOAD_COUNT, vmx->msr_autoload.guest.nr);
 
 	set_cr4_guest_host_mask(vmx);
+
+	if (kvm_cet_supported()) {

This needs to be conditioned on CET coming from vmcs12, it's on the loading of host state on VM-Exit that is unconditional (if CET is supported).

	if (kvm_cet_supported() && vmx->nested.nested_run_pending &&
	    (vmcs12->vm_entry_controls & VM_ENTRY_LOAD_CET_STATE)) {

I also assume these should be guarded by one of the eVMCS fields, though a quick search of the public docs didn't provide a hit on the CET fields.

Vitaly, any idea if these will be GUEST_GRP2 or something else?

Quoted 5 lines
arch/x86/kvm/vmx/nested.c +4 -0
+		vmcs_writel(GUEST_SSP, vmcs12->guest_ssp);
+		vmcs_writel(GUEST_S_CET, vmcs12->guest_s_cet);
+		vmcs_writel(GUEST_INTR_SSP_TABLE, vmcs12->guest_ssp_tbl);
+	}
 }
#3 lore raw

Re: [PATCH v2] KVM: nVMX: Sync L2 guest CET states between L1/L2

Quoted 46 lines · Sean Christopherson

+Vitaly

Quoted 34 lines · Yang Weijiang

These fields are rarely updated by L1 QEMU/KVM, sync them when L1 is trying to read/write them and after they're changed. If CET guest entry-load bit is not set by L1 guest, migrate them to L2 manaully.

change in v2:
- Per Sean's review feedback, change CET guest states as rarely-updated fields. And also migrate L1's CET states to L2 if the entry-load bit is not set.
- Opportunistically removed one blank line.

---
 arch/x86/kvm/cpuid.c      |  1 -
 arch/x86/kvm/vmx/nested.c | 29 +++++++++++++++++++++++++++++
 arch/x86/kvm/vmx/vmx.h    |  3 +++
 3 files changed, 32 insertions(+), 1 deletion(-)
arch/x86/kvm/cpuid.c +0 -1
index 46087bca9418..afc97122c05c 100644
@@ -143,7 +143,6 @@ void kvm_update_cpuid_runtime(struct kvm_vcpu *vcpu)
 		}
 		vcpu->arch.guest_supported_xss =
 			(((u64)best->edx << 32) | best->ecx) & supported_xss;
-
 	} else {
 		vcpu->arch.guest_supported_xss = 0;
 	}
arch/x86/kvm/vmx/nested.c +0 -0
index 9728efd529a1..1703b8874fad 100644

arch/x86/kvm/vmx/nested.c static void prepare_vmcs02_rare(struct vcpu_vmx *vmx, struct vmcs12 *vmcs12) · line 2520

Quoted 5 lines
arch/x86/kvm/vmx/nested.c +2 -0
@@ -2516,6 +2516,12 @@ static void prepare_vmcs02_rare(struct vcpu_vmx *vmx, struct vmcs12 *vmcs12)
 	vmcs_write32(VM_ENTRY_MSR_LOAD_COUNT, vmx->msr_autoload.guest.nr);
 
 	set_cr4_guest_host_mask(vmx);
+
+	if (kvm_cet_supported()) {

This needs to be conditioned on CET coming from vmcs12, it's on the loading of host state on VM-Exit that is unconditional (if CET is supported).

Quoted 3 lines
	if (kvm_cet_supported() && vmx->nested.nested_run_pending &&
	    (vmcs12->vm_entry_controls & VM_ENTRY_LOAD_CET_STATE)) {

Thanks Sean! Will change it.

Quoted 3 lines

I also assume these should be guarded by one of the eVMCS fields, though a quick search of the public docs didn't provide a hit on the CET fields.

I got some description from MSFT as below, do you mean that:

GuestSsp uses clean field GUEST_BASIC (bit 10) GuestSCet/GuestInterruptSspTableAddr uses GUEST_GRP1 (bit 11) HostSCet/HostSsp/HostInterruptSspTableAddr uses HOST_GRP1 (bit 14)

If it is, should these go into separate patch series for Hyper-v nested support? I have some pending patches for the enabling.

Quoted 7 lines

Vitaly, any idea if these will be GUEST_GRP2 or something else?

Quoted 5 lines
arch/x86/kvm/vmx/nested.c +4 -0
+		vmcs_writel(GUEST_SSP, vmcs12->guest_ssp);
+		vmcs_writel(GUEST_S_CET, vmcs12->guest_s_cet);
+		vmcs_writel(GUEST_INTR_SSP_TABLE, vmcs12->guest_ssp_tbl);
+	}
 }
maintainer
#4 lore raw

Re: [PATCH v2] KVM: nVMX: Sync L2 guest CET states between L1/L2

Quoted 8 lines · Yang Weijiang

I got some description from MSFT as below, do you mean that:

GuestSsp uses clean field GUEST_BASIC (bit 10) GuestSCet/GuestInterruptSspTableAddr uses GUEST_GRP1 (bit 11) HostSCet/HostSsp/HostInterruptSspTableAddr uses HOST_GRP1 (bit 14)

If it is, should these go into separate patch series for Hyper-v nested support? I have some pending patches for the enabling.

Yes, it should be a separate patch. The main patch however should add the CET fields to EVMCS1_UNSUPPORTED_VMENTRY_CTRL and EVMCS1_UNSUPPORTED_VMEXIT_CTRL.

Thanks,

Paolo

#5 lore raw

Re: [PATCH v2] KVM: nVMX: Sync L2 guest CET states between L1/L2

In reply to Sean Christopherson
Quoted 46 lines · Sean Christopherson

+Vitaly

Quoted 34 lines · Yang Weijiang

These fields are rarely updated by L1 QEMU/KVM, sync them when L1 is trying to read/write them and after they're changed. If CET guest entry-load bit is not set by L1 guest, migrate them to L2 manaully.

change in v2:
- Per Sean's review feedback, change CET guest states as rarely-updated fields. And also migrate L1's CET states to L2 if the entry-load bit is not set.
- Opportunistically removed one blank line.

---
 arch/x86/kvm/cpuid.c      |  1 -
 arch/x86/kvm/vmx/nested.c | 29 +++++++++++++++++++++++++++++
 arch/x86/kvm/vmx/vmx.h    |  3 +++
 3 files changed, 32 insertions(+), 1 deletion(-)
arch/x86/kvm/cpuid.c +0 -1
index 46087bca9418..afc97122c05c 100644
@@ -143,7 +143,6 @@ void kvm_update_cpuid_runtime(struct kvm_vcpu *vcpu)
 		}
 		vcpu->arch.guest_supported_xss =
 			(((u64)best->edx << 32) | best->ecx) & supported_xss;
-
 	} else {
 		vcpu->arch.guest_supported_xss = 0;
 	}
arch/x86/kvm/vmx/nested.c +0 -0
index 9728efd529a1..1703b8874fad 100644

arch/x86/kvm/vmx/nested.c static void prepare_vmcs02_rare(struct vcpu_vmx *vmx, struct vmcs12 *vmcs12) · line 2520

Quoted 5 lines
arch/x86/kvm/vmx/nested.c +2 -0
@@ -2516,6 +2516,12 @@ static void prepare_vmcs02_rare(struct vcpu_vmx *vmx, struct vmcs12 *vmcs12)
 	vmcs_write32(VM_ENTRY_MSR_LOAD_COUNT, vmx->msr_autoload.guest.nr);
 
 	set_cr4_guest_host_mask(vmx);
+
+	if (kvm_cet_supported()) {

This needs to be conditioned on CET coming from vmcs12, it's on the loading of host state on VM-Exit that is unconditional (if CET is supported).

Quoted 8 lines
	if (kvm_cet_supported() && vmx->nested.nested_run_pending &&
	    (vmcs12->vm_entry_controls & VM_ENTRY_LOAD_CET_STATE)) {

I also assume these should be guarded by one of the eVMCS fields, though a quick search of the public docs didn't provide a hit on the CET fields.

Vitaly, any idea if these will be GUEST_GRP2 or something else?

The latest published TLFS I see is 6.0b and it doesn't list anything CET related in eVMCS v1.0 :-( So I agree with Paolo: we just need to adjust EVMCS1_UNSUPPORTED_VMENTRY_CTRL/ EVMCS1_UNSUPPORTED_VMEXIT_CTRL for now and enable it separately later.

Quoted 6 lines
Quoted 5 lines
arch/x86/kvm/vmx/nested.c +4 -0
+		vmcs_writel(GUEST_SSP, vmcs12->guest_ssp);
+		vmcs_writel(GUEST_S_CET, vmcs12->guest_s_cet);
+		vmcs_writel(GUEST_INTR_SSP_TABLE, vmcs12->guest_ssp_tbl);
+	}
 }
Signature · 1 line
Vitaly
#6 lore raw

Re: [PATCH v2] KVM: nVMX: Sync L2 guest CET states between L1/L2

Quoted 54 lines · Vitaly Kuznetsov
Quoted 52 lines · Sean Christopherson

+Vitaly

Quoted 34 lines · Yang Weijiang

These fields are rarely updated by L1 QEMU/KVM, sync them when L1 is trying to read/write them and after they're changed. If CET guest entry-load bit is not set by L1 guest, migrate them to L2 manaully.

change in v2:
- Per Sean's review feedback, change CET guest states as rarely-updated fields. And also migrate L1's CET states to L2 if the entry-load bit is not set.
- Opportunistically removed one blank line.

---
 arch/x86/kvm/cpuid.c      |  1 -
 arch/x86/kvm/vmx/nested.c | 29 +++++++++++++++++++++++++++++
 arch/x86/kvm/vmx/vmx.h    |  3 +++
 3 files changed, 32 insertions(+), 1 deletion(-)
arch/x86/kvm/cpuid.c +0 -1
index 46087bca9418..afc97122c05c 100644
@@ -143,7 +143,6 @@ void kvm_update_cpuid_runtime(struct kvm_vcpu *vcpu)
 		}
 		vcpu->arch.guest_supported_xss =
 			(((u64)best->edx << 32) | best->ecx) & supported_xss;
-
 	} else {
 		vcpu->arch.guest_supported_xss = 0;
 	}
arch/x86/kvm/vmx/nested.c +0 -0
index 9728efd529a1..1703b8874fad 100644

arch/x86/kvm/vmx/nested.c static void prepare_vmcs02_rare(struct vcpu_vmx *vmx, struct vmcs12 *vmcs12) · line 2520

Quoted 5 lines
arch/x86/kvm/vmx/nested.c +2 -0
@@ -2516,6 +2516,12 @@ static void prepare_vmcs02_rare(struct vcpu_vmx *vmx, struct vmcs12 *vmcs12)
 	vmcs_write32(VM_ENTRY_MSR_LOAD_COUNT, vmx->msr_autoload.guest.nr);
 
 	set_cr4_guest_host_mask(vmx);
+
+	if (kvm_cet_supported()) {

This needs to be conditioned on CET coming from vmcs12, it's on the loading of host state on VM-Exit that is unconditional (if CET is supported).

	if (kvm_cet_supported() && vmx->nested.nested_run_pending &&
	    (vmcs12->vm_entry_controls & VM_ENTRY_LOAD_CET_STATE)) {

I also assume these should be guarded by one of the eVMCS fields, though a quick search of the public docs didn't provide a hit on the CET fields.

Quoted 8 lines
Quoted 2 lines

Vitaly, any idea if these will be GUEST_GRP2 or something else?

The latest published TLFS I see is 6.0b and it doesn't list anything CET related in eVMCS v1.0 :-( So I agree with Paolo: we just need to adjust EVMCS1_UNSUPPORTED_VMENTRY_CTRL/ EVMCS1_UNSUPPORTED_VMEXIT_CTRL for now and enable it separately later.

Thanks Vitaly and Paolo.
New patches have been sent to community.

Quoted 9 lines
Quoted 6 lines
Quoted 5 lines
arch/x86/kvm/vmx/nested.c +4 -0
+		vmcs_writel(GUEST_SSP, vmcs12->guest_ssp);
+		vmcs_writel(GUEST_S_CET, vmcs12->guest_s_cet);
+		vmcs_writel(GUEST_INTR_SSP_TABLE, vmcs12->guest_ssp_tbl);
+	}
 }
Signature · 1 line
Vitaly
maintainer Reviewed-by on v1 1/1
#7 lore raw
In reply to Yang Weijiang
Quoted 14 lines · Yang Weijiang

CET SHSTK and IBT are independently controlled by kernel, set X86_CR4_CET bit in cr4_fixed1_bits if either of them is enabled so that nested guest can enjoy the feature.

---
 arch/x86/kvm/vmx/vmx.c | 1 +
 1 file changed, 1 insertion(+)
arch/x86/kvm/vmx/vmx.c +0 -0
index 5856c5b81084..e92134ee081c 100644

arch/x86/kvm/vmx/vmx.c static void nested_vmx_cr_fixed1_bits_update(struct kvm_vcpu *vcpu) · line 7261

Quoted 4 lines
arch/x86/kvm/vmx/vmx.c +1 -0
@@ -7258,6 +7258,7 @@ static void nested_vmx_cr_fixed1_bits_update(struct kvm_vcpu *vcpu)
 	cr4_fixed1_update(X86_CR4_UMIP,       ecx, feature_bit(UMIP));
 	cr4_fixed1_update(X86_CR4_LA57,       ecx, feature_bit(LA57));
 	cr4_fixed1_update(X86_CR4_CET,	      ecx, feature_bit(SHSTK));
+	cr4_fixed1_update(X86_CR4_CET,	      edx, feature_bit(IBT));

Ugh, what sadist put SHSTK and IBT in separate output registers.

Quoted 6 lines
arch/x86/kvm/vmx/vmx.c +0 -0
 #undef cr4_fixed1_update
 }
Signature · 1 line
2.26.2