KVM: CPUID: add support for supervisor states

Current CPUID 0xd enumeration code does not support supervisor
states, because KVM only supports setting IA32_XSS to zero.
Change it instead to use a new variable supported_xss, to be
set from the hardware_setup callback which is in charge of CPU
capabilities.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

arch/x86/kvm/cpuid.c

@@ -642,15 +642,22 @@ static inline int __do_cpuid_func(struct kvm_cpuid_array *array, u32 function)
 
 		cpuid_entry_override(entry, CPUID_D_1_EAX);
 		if (entry->eax & (F(XSAVES)|F(XSAVEC)))
-			entry->ebx = xstate_required_size(supported_xcr0, true);
-		else
+			entry->ebx = xstate_required_size(supported_xcr0 | supported_xss,
+							  true);
+		else {
+			WARN_ON_ONCE(supported_xss != 0);
 			entry->ebx = 0;
-		/* Saving XSS controlled state via XSAVES isn't supported. */
-		entry->ecx = 0;
-		entry->edx = 0;
+		}
+		entry->ecx &= supported_xss;
+		entry->edx &= supported_xss >> 32;
 
 		for (i = 2; i < 64; ++i) {
-			if (!(supported_xcr0 & BIT_ULL(i)))
+			bool s_state;
+			if (supported_xcr0 & BIT_ULL(i))
+				s_state = false;
+			else if (supported_xss & BIT_ULL(i))
+				s_state = true;
+			else
 				continue;
 
 			entry = do_host_cpuid(array, function, i);
@@ -659,17 +666,16 @@ static inline int __do_cpuid_func(struct kvm_cpuid_array *array, u32 function)
 
 			/*
 			 * The supported check above should have filtered out
-			 * invalid sub-leafs as well as sub-leafs managed by
-			 * IA32_XSS MSR.  Only XCR0-managed sub-leafs should
+			 * invalid sub-leafs.  Only valid sub-leafs should
 			 * reach this point, and they should have a non-zero
-			 * save state size.
+			 * save state size.  Furthermore, check whether the
+			 * processor agrees with supported_xcr0/supported_xss
+			 * on whether this is an XCR0- or IA32_XSS-managed area.
 			 */
-			if (WARN_ON_ONCE(!entry->eax || (entry->ecx & 1))) {
+			if (WARN_ON_ONCE(!entry->eax || (entry->ecx & 0x1) != s_state)) {
 				--array->nent;
 				continue;
 			}
-
-			entry->ecx = 0;
 			entry->edx = 0;
 		}
 		break;

arch/x86/kvm/svm.c

@@ -1371,6 +1371,8 @@ static __init void svm_set_cpu_caps(void)
 {
 	kvm_set_cpu_caps();
 
+	supported_xss = 0;
+
 	/* CPUID 0x80000001 and 0x8000000A (SVM features) */
 	if (nested) {
 		kvm_cpu_cap_set(X86_FEATURE_SVM);

arch/x86/kvm/vmx/vmx.c

@@ -7126,6 +7126,7 @@ static __init void vmx_set_cpu_caps(void)
 		kvm_cpu_cap_set(X86_FEATURE_UMIP);
 
 	/* CPUID 0xD.1 */
+	supported_xss = 0;
 	if (!vmx_xsaves_supported())
 		kvm_cpu_cap_clear(X86_FEATURE_XSAVES);
 

arch/x86/kvm/x86.c

@@ -190,6 +190,8 @@ u64 __read_mostly host_efer;
 EXPORT_SYMBOL_GPL(host_efer);
 
 static u64 __read_mostly host_xss;
+u64 __read_mostly supported_xss;
+EXPORT_SYMBOL_GPL(supported_xss);
 
 struct kvm_stats_debugfs_item debugfs_entries[] = {
 	{ "pf_fixed", VCPU_STAT(pf_fixed) },
@@ -2827,7 +2829,7 @@ int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
 		 * IA32_XSS[bit 8]. Guests have to use RDMSR/WRMSR rather than
 		 * XSAVES/XRSTORS to save/restore PT MSRs.
 		 */
-		if (data != 0)
+		if (data & ~supported_xss)
 			return 1;
 		vcpu->arch.ia32_xss = data;
 		break;
@@ -9617,10 +9619,16 @@ int kvm_arch_hardware_setup(void)
 
 	rdmsrl_safe(MSR_EFER, &host_efer);
 
+	if (boot_cpu_has(X86_FEATURE_XSAVES))
+		rdmsrl(MSR_IA32_XSS, host_xss);
+
 	r = kvm_x86_ops->hardware_setup();
 	if (r != 0)
 		return r;
 
+	if (!kvm_cpu_cap_has(X86_FEATURE_XSAVES))
+		supported_xss = 0;
+
 	cr4_reserved_bits = kvm_host_cr4_reserved_bits(&boot_cpu_data);
 
 	if (kvm_has_tsc_control) {
@@ -9637,9 +9645,6 @@ int kvm_arch_hardware_setup(void)
 		kvm_default_tsc_scaling_ratio = 1ULL << kvm_tsc_scaling_ratio_frac_bits;
 	}
 
-	if (boot_cpu_has(X86_FEATURE_XSAVES))
-		rdmsrl(MSR_IA32_XSS, host_xss);
-
 	kvm_init_msr_list();
 	return 0;
 }

arch/x86/kvm/x86.h

@@ -272,6 +272,7 @@ enum exit_fastpath_completion handle_fastpath_set_msr_irqoff(struct kvm_vcpu *vc
 
 extern u64 host_xcr0;
 extern u64 supported_xcr0;
+extern u64 supported_xss;
 
 static inline bool kvm_mpx_supported(void)
 {