KVM: x86: hyper-v: Make Hyper-V emulation enablement conditional

Hyper-V emulation is enabled in KVM unconditionally. This is bad at least
from security standpoint as it is an extra attack surface. Ideally, there
should be a per-VM capability explicitly enabled by VMM but currently it
is not the case and we can't mandate one without breaking backwards
compatibility. We can, however, check guest visible CPUIDs and only enable
Hyper-V emulation when "Hv#1" interface was exposed in
HYPERV_CPUID_INTERFACE.

Note, VMMs are free to act in any sequence they like, e.g. they can try
to set MSRs first and CPUIDs later so we still need to allow the host
to read/write Hyper-V specific MSRs unconditionally.

Signed-off-by: Vitaly Kuznetsov <[e-mail]>
Message-Id: <[e-mail]>
[Add selftest vcpu_set_hv_cpuid API to avoid breaking xen_vmcall_test. - Paolo]
Signed-off-by: Paolo Bonzini <[e-mail]>

arch/x86/include/asm/kvm_host.h

···
 	/* used for guest single stepping over the given code position */
 	unsigned long singlestep_rip;
 
+	bool hyperv_enabled;
 	struct kvm_vcpu_hv *hyperv;
 	struct kvm_vcpu_xen xen;
 

arch/x86/kvm/cpuid.c

···
 	vcpu->arch.cr4_guest_rsvd_bits =
 	    __cr4_reserved_bits(guest_cpuid_has, vcpu);
 
+	kvm_hv_set_cpuid(vcpu);
+
 	/* Invoke the vendor callback only after the above state is updated. */
 	static_call(kvm_x86_vcpu_after_set_cpuid)(vcpu);
 

arch/x86/kvm/hyperv.c

···
 #include "trace.h"
 #include "irq.h"
 
+/* "Hv#1" signature */
+#define HYPERV_CPUID_SIGNATURE_EAX 0x31237648
+
 #define KVM_HV_MAX_SPARSE_VCPU_SET_BITS DIV_ROUND_UP(KVM_MAX_VCPUS, 64)
 
 static void stimer_mark_pending(struct kvm_vcpu_hv_stimer *stimer,
···
 {
 	struct kvm_hv *hv = to_kvm_hv(vcpu->kvm);
 
+	if (!host && !vcpu->arch.hyperv_enabled)
+		return 1;
+
 	if (kvm_hv_msr_partition_wide(msr)) {
 		int r;
 
···
 int kvm_hv_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata, bool host)
 {
 	struct kvm_hv *hv = to_kvm_hv(vcpu->kvm);
+
+	if (!host && !vcpu->arch.hyperv_enabled)
+		return 1;
 
 	if (kvm_hv_msr_partition_wide(msr)) {
 		int r;
···
 	return HV_STATUS_SUCCESS;
 }
 
-bool kvm_hv_hypercall_enabled(struct kvm *kvm)
+void kvm_hv_set_cpuid(struct kvm_vcpu *vcpu)
 {
-	return to_kvm_hv(kvm)->hv_guest_os_id != 0;
+	struct kvm_cpuid_entry2 *entry;
+
+	entry = kvm_find_cpuid_entry(vcpu, HYPERV_CPUID_INTERFACE, 0);
+	if (entry && entry->eax == HYPERV_CPUID_SIGNATURE_EAX)
+		vcpu->arch.hyperv_enabled = true;
+	else
+		vcpu->arch.hyperv_enabled = false;
+}
+
+bool kvm_hv_hypercall_enabled(struct kvm_vcpu *vcpu)
+{
+	return vcpu->arch.hyperv_enabled && to_kvm_hv(vcpu->kvm)->hv_guest_os_id;
 }
 
 static void kvm_hv_hypercall_set_result(struct kvm_vcpu *vcpu, u64 result)
···
 			break;
 
 		case HYPERV_CPUID_INTERFACE:
-			memcpy(signature, "Hv#1\0\0\0\0\0\0\0\0", 12);
-			ent->eax = signature[0];
+			ent->eax = HYPERV_CPUID_SIGNATURE_EAX;
 			break;
 
 		case HYPERV_CPUID_VERSION:

arch/x86/kvm/hyperv.h

···
 int kvm_hv_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data, bool host);
 int kvm_hv_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata, bool host);
 
-bool kvm_hv_hypercall_enabled(struct kvm *kvm);
+bool kvm_hv_hypercall_enabled(struct kvm_vcpu *vcpu);
 int kvm_hv_hypercall(struct kvm_vcpu *vcpu);
 
 void kvm_hv_irq_routing_update(struct kvm *kvm);
···
 
 void kvm_hv_init_vm(struct kvm *kvm);
 void kvm_hv_destroy_vm(struct kvm *kvm);
+void kvm_hv_set_cpuid(struct kvm_vcpu *vcpu);
 int kvm_vm_ioctl_hv_eventfd(struct kvm *kvm, struct kvm_hyperv_eventfd *args);
 int kvm_get_hv_cpuid(struct kvm_vcpu *vcpu, struct kvm_cpuid2 *cpuid,
 		     struct kvm_cpuid_entry2 __user *entries);

arch/x86/kvm/x86.c

···
 	if (kvm_xen_hypercall_enabled(vcpu->kvm))
 		return kvm_xen_hypercall(vcpu);
 
-	if (kvm_hv_hypercall_enabled(vcpu->kvm))
+	if (kvm_hv_hypercall_enabled(vcpu))
 		return kvm_hv_hypercall(vcpu);
 
 	nr = kvm_rax_read(vcpu);

arch/x86/kvm/xen.c

···
 
 	/* Hyper-V hypercalls get bit 31 set in EAX */
 	if ((input & 0x80000000) &&
-	    kvm_hv_hypercall_enabled(vcpu->kvm))
+	    kvm_hv_hypercall_enabled(vcpu))
 		return kvm_hv_hypercall(vcpu);
 
 	longmode = is_64_bit_mode(vcpu);

tools/testing/selftests/kvm/include/x86_64/processor.h

···
 		       uint64_t a3);
 
 struct kvm_cpuid2 *kvm_get_supported_hv_cpuid(void);
+void vcpu_set_hv_cpuid(struct kvm_vm *vm, uint32_t vcpuid);
 struct kvm_cpuid2 *vcpu_get_supported_hv_cpuid(struct kvm_vm *vm, uint32_t vcpuid);
 
 /*

tools/testing/selftests/kvm/lib/x86_64/processor.c

···
 	return cpuid;
 }
 
+void vcpu_set_hv_cpuid(struct kvm_vm *vm, uint32_t vcpuid)
+{
+	static struct kvm_cpuid2 *cpuid_full;
+	struct kvm_cpuid2 *cpuid_sys, *cpuid_hv;
+	int i, nent = 0;
+
+	if (!cpuid_full) {
+		cpuid_sys = kvm_get_supported_cpuid();
+		cpuid_hv = kvm_get_supported_hv_cpuid();
+
+		cpuid_full = malloc(sizeof(*cpuid_full) +
+				    (cpuid_sys->nent + cpuid_hv->nent) *
+				    sizeof(struct kvm_cpuid_entry2));
+		if (!cpuid_full) {
+			perror("malloc");
+			abort();
+		}
+
+		/* Need to skip KVM CPUID leaves 0x400000xx */
+		for (i = 0; i < cpuid_sys->nent; i++) {
+			if (cpuid_sys->entries[i].function >= 0x40000000 &&
+			    cpuid_sys->entries[i].function < 0x40000100)
+				continue;
+			cpuid_full->entries[nent] = cpuid_sys->entries[i];
+			nent++;
+		}
+
+		memcpy(&cpuid_full->entries[nent], cpuid_hv->entries,
+		       cpuid_hv->nent * sizeof(struct kvm_cpuid_entry2));
+		cpuid_full->nent = nent + cpuid_hv->nent;
+	}
+
+	vcpu_set_cpuid(vm, vcpuid, cpuid_full);
+}
+
 struct kvm_cpuid2 *vcpu_get_supported_hv_cpuid(struct kvm_vm *vm, uint32_t vcpuid)
 {
 	static struct kvm_cpuid2 *cpuid;

tools/testing/selftests/kvm/x86_64/evmcs_test.c

···
 	GUEST_ASSERT(vmlaunch());
 }
 
-struct kvm_cpuid2 *guest_get_cpuid(void)
-{
-	static struct kvm_cpuid2 *cpuid_full;
-	struct kvm_cpuid2 *cpuid_sys, *cpuid_hv;
-	int i, nent = 0;
-
-	if (cpuid_full)
-		return cpuid_full;
-
-	cpuid_sys = kvm_get_supported_cpuid();
-	cpuid_hv = kvm_get_supported_hv_cpuid();
-
-	cpuid_full = malloc(sizeof(*cpuid_full) +
-			    (cpuid_sys->nent + cpuid_hv->nent) *
-			    sizeof(struct kvm_cpuid_entry2));
-	if (!cpuid_full) {
-		perror("malloc");
-		abort();
-	}
-
-	/* Need to skip KVM CPUID leaves 0x400000xx */
-	for (i = 0; i < cpuid_sys->nent; i++) {
-		if (cpuid_sys->entries[i].function >= 0x40000000 &&
-		    cpuid_sys->entries[i].function < 0x40000100)
-			continue;
-		cpuid_full->entries[nent] = cpuid_sys->entries[i];
-		nent++;
-	}
-
-	memcpy(&cpuid_full->entries[nent], cpuid_hv->entries,
-	       cpuid_hv->nent * sizeof(struct kvm_cpuid_entry2));
-	cpuid_full->nent = nent + cpuid_hv->nent;
-
-	return cpuid_full;
-}
-
 int main(int argc, char *argv[])
 {
 	vm_vaddr_t vmx_pages_gva = 0;
···
 		exit(KSFT_SKIP);
 	}
 
-	vcpu_set_cpuid(vm, VCPU_ID, guest_get_cpuid());
+	vcpu_set_hv_cpuid(vm, VCPU_ID);
 	vcpu_enable_evmcs(vm, VCPU_ID);
 
 	run = vcpu_state(vm, VCPU_ID);
···
 		/* Restore state in a new VM.  */
 		kvm_vm_restart(vm, O_RDWR);
 		vm_vcpu_add(vm, VCPU_ID);
-		vcpu_set_cpuid(vm, VCPU_ID, guest_get_cpuid());
+		vcpu_set_hv_cpuid(vm, VCPU_ID);
 		vcpu_enable_evmcs(vm, VCPU_ID);
 		vcpu_load_state(vm, VCPU_ID, state);
 		run = vcpu_state(vm, VCPU_ID);

tools/testing/selftests/kvm/x86_64/xen_vmcall_test.c

···
 	}
 
 	vm = vm_create_default(VCPU_ID, 0, (void *) guest_code);
-	vcpu_set_cpuid(vm, VCPU_ID, kvm_get_supported_cpuid());
+	vcpu_set_hv_cpuid(vm, VCPU_ID);
 
 	struct kvm_xen_hvm_config hvmc = {
 		.flags = KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL,