KVM: VMX: Properly handle dynamic VM Entry/Exit controls

EFER and PERF_GLOBAL_CTRL MSRs have dedicated VM Entry/Exit controls
that KVM dynamically toggles based on whether or not the guest's value
for each MSRs differs from the host.  Handle the dynamic behavior by
adding a helper that clears the dynamic bits so the bits aren't set
when initializing the VMCS field outside of the dynamic toggling flow.
This makes the handling consistent with similar behavior for other
controls, e.g. pin, exec and sec_exec.  More importantly, it eliminates
two global bools that are stealthily modified by setup_vmcs_config.

Opportunistically clean up a comment and print related to errata for
IA32_PERF_GLOBAL_CTRL.
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

arch/x86/kvm/vmx/vmx.c

@@ -1336,9 +1336,6 @@ static unsigned long *vmx_bitmap[VMX_BITMAP_NR];
 #define vmx_vmread_bitmap                    (vmx_bitmap[VMX_VMREAD_BITMAP])
 #define vmx_vmwrite_bitmap                   (vmx_bitmap[VMX_VMWRITE_BITMAP])
 
-static bool cpu_has_load_ia32_efer;
-static bool cpu_has_load_perf_global_ctrl;
-
 static DECLARE_BITMAP(vmx_vpid_bitmap, VMX_NR_VPIDS);
 static DEFINE_SPINLOCK(vmx_vpid_lock);
 
@@ -1681,6 +1678,18 @@ static inline bool is_icebp(u32 intr_info)
 		== (INTR_TYPE_PRIV_SW_EXCEPTION | INTR_INFO_VALID_MASK);
 }
 
+static inline bool cpu_has_load_ia32_efer(void)
+{
+	return (vmcs_config.vmentry_ctrl & VM_ENTRY_LOAD_IA32_EFER) &&
+	       (vmcs_config.vmexit_ctrl & VM_EXIT_LOAD_IA32_EFER);
+}
+
+static inline bool cpu_has_load_perf_global_ctrl(void)
+{
+	return (vmcs_config.vmentry_ctrl & VM_ENTRY_LOAD_IA32_PERF_GLOBAL_CTRL) &&
+	       (vmcs_config.vmexit_ctrl & VM_EXIT_LOAD_IA32_PERF_GLOBAL_CTRL);
+}
+
 static inline bool cpu_has_vmx_msr_bitmap(void)
 {
 	return vmcs_config.cpu_based_exec_ctrl & CPU_BASED_USE_MSR_BITMAPS;
@@ -2696,7 +2705,7 @@ static void clear_atomic_switch_msr(struct vcpu_vmx *vmx, unsigned msr)
 
 	switch (msr) {
 	case MSR_EFER:
-		if (cpu_has_load_ia32_efer) {
+		if (cpu_has_load_ia32_efer()) {
 			clear_atomic_switch_msr_special(vmx,
 					VM_ENTRY_LOAD_IA32_EFER,
 					VM_EXIT_LOAD_IA32_EFER);
@@ -2704,7 +2713,7 @@ static void clear_atomic_switch_msr(struct vcpu_vmx *vmx, unsigned msr)
 		}
 		break;
 	case MSR_CORE_PERF_GLOBAL_CTRL:
-		if (cpu_has_load_perf_global_ctrl) {
+		if (cpu_has_load_perf_global_ctrl()) {
 			clear_atomic_switch_msr_special(vmx,
 					VM_ENTRY_LOAD_IA32_PERF_GLOBAL_CTRL,
 					VM_EXIT_LOAD_IA32_PERF_GLOBAL_CTRL);
@@ -2749,7 +2758,7 @@ static void add_atomic_switch_msr(struct vcpu_vmx *vmx, unsigned msr,
 
 	switch (msr) {
 	case MSR_EFER:
-		if (cpu_has_load_ia32_efer) {
+		if (cpu_has_load_ia32_efer()) {
 			add_atomic_switch_msr_special(vmx,
 					VM_ENTRY_LOAD_IA32_EFER,
 					VM_EXIT_LOAD_IA32_EFER,
@@ -2760,7 +2769,7 @@ static void add_atomic_switch_msr(struct vcpu_vmx *vmx, unsigned msr,
 		}
 		break;
 	case MSR_CORE_PERF_GLOBAL_CTRL:
-		if (cpu_has_load_perf_global_ctrl) {
+		if (cpu_has_load_perf_global_ctrl()) {
 			add_atomic_switch_msr_special(vmx,
 					VM_ENTRY_LOAD_IA32_PERF_GLOBAL_CTRL,
 					VM_EXIT_LOAD_IA32_PERF_GLOBAL_CTRL,
@@ -2839,7 +2848,7 @@ static bool update_transition_efer(struct vcpu_vmx *vmx, int efer_offset)
 	 * On CPUs that support "load IA32_EFER", always switch EFER
 	 * atomically, since it's faster than switching it manually.
 	 */
-	if (cpu_has_load_ia32_efer ||
+	if (cpu_has_load_ia32_efer() ||
 	    (enable_ept && ((vmx->vcpu.arch.efer ^ host_efer) & EFER_NX))) {
 		if (!(guest_efer & EFER_LMA))
 			guest_efer &= ~EFER_LME;
@@ -4533,14 +4542,6 @@ static __init int adjust_vmx_controls(u32 ctl_min, u32 ctl_opt,
 	return 0;
 }
 
-static __init bool allow_1_setting(u32 msr, u32 ctl)
-{
-	u32 vmx_msr_low, vmx_msr_high;
-
-	rdmsr(msr, vmx_msr_low, vmx_msr_high);
-	return vmx_msr_high & ctl;
-}
-
 static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
 {
 	u32 vmx_msr_low, vmx_msr_high;
@@ -4642,8 +4643,11 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
 #ifdef CONFIG_X86_64
 	min |= VM_EXIT_HOST_ADDR_SPACE_SIZE;
 #endif
-	opt = VM_EXIT_SAVE_IA32_PAT | VM_EXIT_LOAD_IA32_PAT |
-		VM_EXIT_CLEAR_BNDCFGS;
+	opt = VM_EXIT_LOAD_IA32_PERF_GLOBAL_CTRL |
+	      VM_EXIT_SAVE_IA32_PAT |
+	      VM_EXIT_LOAD_IA32_PAT |
+	      VM_EXIT_LOAD_IA32_EFER |
+	      VM_EXIT_CLEAR_BNDCFGS;
 	if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_EXIT_CTLS,
 				&_vmexit_control) < 0)
 		return -EIO;
@@ -4662,11 +4666,38 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
 		_pin_based_exec_control &= ~PIN_BASED_POSTED_INTR;
 
 	min = VM_ENTRY_LOAD_DEBUG_CONTROLS;
-	opt = VM_ENTRY_LOAD_IA32_PAT | VM_ENTRY_LOAD_BNDCFGS;
+	opt = VM_ENTRY_LOAD_IA32_PERF_GLOBAL_CTRL |
+	      VM_ENTRY_LOAD_IA32_PAT |
+	      VM_ENTRY_LOAD_IA32_EFER |
+	      VM_ENTRY_LOAD_BNDCFGS;
 	if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_ENTRY_CTLS,
 				&_vmentry_control) < 0)
 		return -EIO;
 
+	/*
+	 * Some cpus support VM_{ENTRY,EXIT}_IA32_PERF_GLOBAL_CTRL but they
+	 * can't be used due to an errata where VM Exit may incorrectly clear
+	 * IA32_PERF_GLOBAL_CTRL[34:32].  Workaround the errata by using the
+	 * MSR load mechanism to switch IA32_PERF_GLOBAL_CTRL.
+	 */
+	if (boot_cpu_data.x86 == 0x6) {
+		switch (boot_cpu_data.x86_model) {
+		case 26: /* AAK155 */
+		case 30: /* AAP115 */
+		case 37: /* AAT100 */
+		case 44: /* BC86,AAY89,BD102 */
+		case 46: /* BA97 */
+			_vmexit_control &= ~VM_ENTRY_LOAD_IA32_PERF_GLOBAL_CTRL;
+			_vmexit_control &= ~VM_EXIT_LOAD_IA32_PERF_GLOBAL_CTRL;
+			pr_warn_once("kvm: VM_EXIT_LOAD_IA32_PERF_GLOBAL_CTRL "
+					"does not work properly. Using workaround\n");
+			break;
+		default:
+			break;
+		}
+	}
+
+
 	rdmsr(MSR_IA32_VMX_BASIC, vmx_msr_low, vmx_msr_high);
 
 	/* IA-32 SDM Vol 3B: VMCS size is never greater than 4kB. */
@@ -4698,48 +4729,6 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
 	if (static_branch_unlikely(&enable_evmcs))
 		evmcs_sanitize_exec_ctrls(vmcs_conf);
 
-	cpu_has_load_ia32_efer =
-		allow_1_setting(MSR_IA32_VMX_ENTRY_CTLS,
-				VM_ENTRY_LOAD_IA32_EFER)
-		&& allow_1_setting(MSR_IA32_VMX_EXIT_CTLS,
-				   VM_EXIT_LOAD_IA32_EFER);
-
-	cpu_has_load_perf_global_ctrl =
-		allow_1_setting(MSR_IA32_VMX_ENTRY_CTLS,
-				VM_ENTRY_LOAD_IA32_PERF_GLOBAL_CTRL)
-		&& allow_1_setting(MSR_IA32_VMX_EXIT_CTLS,
-				   VM_EXIT_LOAD_IA32_PERF_GLOBAL_CTRL);
-
-	/*
-	 * Some cpus support VM_ENTRY_(LOAD|SAVE)_IA32_PERF_GLOBAL_CTRL
-	 * but due to errata below it can't be used. Workaround is to use
-	 * msr load mechanism to switch IA32_PERF_GLOBAL_CTRL.
-	 *
-	 * VM Exit May Incorrectly Clear IA32_PERF_GLOBAL_CTRL [34:32]
-	 *
-	 * AAK155             (model 26)
-	 * AAP115             (model 30)
-	 * AAT100             (model 37)
-	 * BC86,AAY89,BD102   (model 44)
-	 * BA97               (model 46)
-	 *
-	 */
-	if (cpu_has_load_perf_global_ctrl && boot_cpu_data.x86 == 0x6) {
-		switch (boot_cpu_data.x86_model) {
-		case 26:
-		case 30:
-		case 37:
-		case 44:
-		case 46:
-			cpu_has_load_perf_global_ctrl = false;
-			printk_once(KERN_WARNING"kvm: VM_EXIT_LOAD_IA32_PERF_GLOBAL_CTRL "
-					"does not work properly. Using workaround\n");
-			break;
-		default:
-			break;
-		}
-	}
-
 	return 0;
 }
 
@@ -6375,7 +6364,7 @@ static void vmx_set_constant_host_state(struct vcpu_vmx *vmx)
 		vmcs_write64(HOST_IA32_PAT, low32 | ((u64) high32 << 32));
 	}
 
-	if (cpu_has_load_ia32_efer)
+	if (cpu_has_load_ia32_efer())
 		vmcs_write64(HOST_IA32_EFER, host_efer);
 }
 
@@ -6425,6 +6414,20 @@ static void vmx_refresh_apicv_exec_ctrl(struct kvm_vcpu *vcpu)
 		vmx_update_msr_bitmap(vcpu);
 }
 
+static u32 vmx_vmentry_ctrl(void)
+{
+	/* Loading of EFER and PERF_GLOBAL_CTRL are toggled dynamically */
+	return vmcs_config.vmentry_ctrl &
+		~(VM_ENTRY_LOAD_IA32_PERF_GLOBAL_CTRL | VM_ENTRY_LOAD_IA32_EFER);
+}
+
+static u32 vmx_vmexit_ctrl(void)
+{
+	/* Loading of EFER and PERF_GLOBAL_CTRL are toggled dynamically */
+	return vmcs_config.vmexit_ctrl &
+		~(VM_EXIT_LOAD_IA32_PERF_GLOBAL_CTRL | VM_EXIT_LOAD_IA32_EFER);
+}
+
 static u32 vmx_exec_control(struct vcpu_vmx *vmx)
 {
 	u32 exec_control = vmcs_config.cpu_based_exec_ctrl;
@@ -6690,10 +6693,10 @@ static void vmx_vcpu_setup(struct vcpu_vmx *vmx)
 
 	vmx->arch_capabilities = kvm_get_arch_capabilities();
 
-	vm_exit_controls_init(vmx, vmcs_config.vmexit_ctrl);
+	vm_exit_controls_init(vmx, vmx_vmexit_ctrl());
 
 	/* 22.2.1, 20.8.1 */
-	vm_entry_controls_init(vmx, vmcs_config.vmentry_ctrl);
+	vm_entry_controls_init(vmx, vmx_vmentry_ctrl());
 
 	vmx->vcpu.arch.cr0_guest_owned_bits = X86_CR0_TS;
 	vmcs_writel(CR0_GUEST_HOST_MASK, ~X86_CR0_TS);
@@ -10468,7 +10471,7 @@ static void dump_vmcs(void)
 	pr_err("DebugCtl = 0x%016llx  DebugExceptions = 0x%016lx\n",
 	       vmcs_read64(GUEST_IA32_DEBUGCTL),
 	       vmcs_readl(GUEST_PENDING_DBG_EXCEPTIONS));
-	if (cpu_has_load_perf_global_ctrl &&
+	if (cpu_has_load_perf_global_ctrl() &&
 	    vmentry_ctl & VM_ENTRY_LOAD_IA32_PERF_GLOBAL_CTRL)
 		pr_err("PerfGlobCtl = 0x%016llx\n",
 		       vmcs_read64(GUEST_IA32_PERF_GLOBAL_CTRL));
@@ -10505,7 +10508,7 @@ static void dump_vmcs(void)
 		pr_err("EFER = 0x%016llx  PAT = 0x%016llx\n",
 		       vmcs_read64(HOST_IA32_EFER),
 		       vmcs_read64(HOST_IA32_PAT));
-	if (cpu_has_load_perf_global_ctrl &&
+	if (cpu_has_load_perf_global_ctrl() &&
 	    vmexit_ctl & VM_EXIT_LOAD_IA32_PERF_GLOBAL_CTRL)
 		pr_err("PerfGlobCtl = 0x%016llx\n",
 		       vmcs_read64(HOST_IA32_PERF_GLOBAL_CTRL));
@@ -12730,9 +12733,9 @@ static void prepare_vmcs02_early(struct vcpu_vmx *vmx, struct vmcs12 *vmcs12)
 	 * on the related bits (if supported by the CPU) in the hope that
 	 * we can avoid VMWrites during vmx_set_efer().
 	 */
-	exec_control = (vmcs12->vm_entry_controls | vmcs_config.vmentry_ctrl) &
+	exec_control = (vmcs12->vm_entry_controls | vmx_vmentry_ctrl()) &
 			~VM_ENTRY_IA32E_MODE & ~VM_ENTRY_LOAD_IA32_EFER;
-	if (cpu_has_load_ia32_efer) {
+	if (cpu_has_load_ia32_efer()) {
 		if (guest_efer & EFER_LMA)
 			exec_control |= VM_ENTRY_IA32E_MODE;
 		if (guest_efer != host_efer)
@@ -12747,8 +12750,8 @@ static void prepare_vmcs02_early(struct vcpu_vmx *vmx, struct vmcs12 *vmcs12)
 	 * we should use its exit controls. Note that VM_EXIT_LOAD_IA32_EFER
 	 * bits may be modified by vmx_set_efer() in prepare_vmcs02().
 	 */
-	exec_control = vmcs_config.vmexit_ctrl;
-	if (cpu_has_load_ia32_efer && guest_efer != host_efer)
+	exec_control = vmx_vmexit_ctrl();
+	if (cpu_has_load_ia32_efer() && guest_efer != host_efer)
 		exec_control |= VM_EXIT_LOAD_IA32_EFER;
 	vm_exit_controls_init(vmx, exec_control);
 
@@ -14073,7 +14076,7 @@ static inline u64 nested_vmx_get_vmcs01_guest_efer(struct vcpu_vmx *vmx)
 	if (vm_entry_controls_get(vmx) & VM_ENTRY_LOAD_IA32_EFER)
 		return vmcs_read64(GUEST_IA32_EFER);
 
-	if (cpu_has_load_ia32_efer)
+	if (cpu_has_load_ia32_efer())
 		return host_efer;
 
 	for (i = 0; i < vmx->msr_autoload.guest.nr; ++i) {