Merge branch kvm-arm64/lock-inversion into kvmarm-master/next

* kvm-arm64/lock-inversion:
  : .
  : vm/vcpu lock inversion fixes, courtesy of Oliver Upton, plus a few
  : extra fixes from both Oliver and Reiji Watanabe.
  :
  : From the initial cover letter:
  :
  : As it so happens, lock ordering in KVM/arm64 is completely backwards.
  : There's a significant amount of VM-wide state that needs to be accessed
  : from the context of a vCPU. Until now, this was accomplished by
  : acquiring the kvm->lock, but that cannot be nested within vcpu->mutex.
  :
  : This series fixes the issue with some fine-grained locking for MP state
  : and a new, dedicated mutex that can nest with both kvm->lock and
  : vcpu->mutex.
  : .
  KVM: arm64: Have kvm_psci_vcpu_on() use WRITE_ONCE() to update mp_state
  KVM: arm64: Acquire mp_state_lock in kvm_arch_vcpu_ioctl_vcpu_init()
  KVM: arm64: vgic: Don't acquire its_lock before config_lock
  KVM: arm64: Use config_lock to protect vgic state
  KVM: arm64: Use config_lock to protect data ordered against KVM_RUN
  KVM: arm64: Avoid lock inversion when setting the VM register width
  KVM: arm64: Avoid vcpu->mutex v. kvm->lock inversion in CPU_ON

Signed-off-by: Marc Zyngier <maz@kernel.org>

arch/arm64/include/asm/kvm_host.h

@@ -199,6 +199,9 @@ struct kvm_arch {
 	/* Mandated version of PSCI */
 	u32 psci_version;
 
+	/* Protects VM-scoped configuration data */
+	struct mutex config_lock;
+
 	/*
 	 * If we encounter a data abort without valid instruction syndrome
 	 * information, report this to user space.  User space can (and
@@ -522,6 +525,7 @@ struct kvm_vcpu_arch {
 
 	/* vcpu power state */
 	struct kvm_mp_state mp_state;
+	spinlock_t mp_state_lock;
 
 	/* Cache some mmu pages needed inside spinlock regions */
 	struct kvm_mmu_memory_cache mmu_page_cache;

arch/arm64/kvm/arm.c

@@ -128,6 +128,16 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
 {
 	int ret;
 
+	mutex_init(&kvm->arch.config_lock);
+
+#ifdef CONFIG_LOCKDEP
+	/* Clue in lockdep that the config_lock must be taken inside kvm->lock */
+	mutex_lock(&kvm->lock);
+	mutex_lock(&kvm->arch.config_lock);
+	mutex_unlock(&kvm->arch.config_lock);
+	mutex_unlock(&kvm->lock);
+#endif
+
 	ret = kvm_share_hyp(kvm, kvm + 1);
 	if (ret)
 		return ret;
@@ -326,6 +336,16 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
 {
 	int err;
 
+	spin_lock_init(&vcpu->arch.mp_state_lock);
+
+#ifdef CONFIG_LOCKDEP
+	/* Inform lockdep that the config_lock is acquired after vcpu->mutex */
+	mutex_lock(&vcpu->mutex);
+	mutex_lock(&vcpu->kvm->arch.config_lock);
+	mutex_unlock(&vcpu->kvm->arch.config_lock);
+	mutex_unlock(&vcpu->mutex);
+#endif
+
 	/* Force users to call KVM_ARM_VCPU_INIT */
 	vcpu->arch.target = -1;
 	bitmap_zero(vcpu->arch.features, KVM_VCPU_MAX_FEATURES);
@@ -443,34 +463,41 @@ void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu)
 	vcpu->cpu = -1;
 }
 
-void kvm_arm_vcpu_power_off(struct kvm_vcpu *vcpu)
+static void __kvm_arm_vcpu_power_off(struct kvm_vcpu *vcpu)
 {
-	vcpu->arch.mp_state.mp_state = KVM_MP_STATE_STOPPED;
+	WRITE_ONCE(vcpu->arch.mp_state.mp_state, KVM_MP_STATE_STOPPED);
 	kvm_make_request(KVM_REQ_SLEEP, vcpu);
 	kvm_vcpu_kick(vcpu);
 }
 
+void kvm_arm_vcpu_power_off(struct kvm_vcpu *vcpu)
+{
+	spin_lock(&vcpu->arch.mp_state_lock);
+	__kvm_arm_vcpu_power_off(vcpu);
+	spin_unlock(&vcpu->arch.mp_state_lock);
+}
+
 bool kvm_arm_vcpu_stopped(struct kvm_vcpu *vcpu)
 {
-	return vcpu->arch.mp_state.mp_state == KVM_MP_STATE_STOPPED;
+	return READ_ONCE(vcpu->arch.mp_state.mp_state) == KVM_MP_STATE_STOPPED;
 }
 
 static void kvm_arm_vcpu_suspend(struct kvm_vcpu *vcpu)
 {
-	vcpu->arch.mp_state.mp_state = KVM_MP_STATE_SUSPENDED;
+	WRITE_ONCE(vcpu->arch.mp_state.mp_state, KVM_MP_STATE_SUSPENDED);
 	kvm_make_request(KVM_REQ_SUSPEND, vcpu);
 	kvm_vcpu_kick(vcpu);
 }
 
 static bool kvm_arm_vcpu_suspended(struct kvm_vcpu *vcpu)
 {
-	return vcpu->arch.mp_state.mp_state == KVM_MP_STATE_SUSPENDED;
+	return READ_ONCE(vcpu->arch.mp_state.mp_state) == KVM_MP_STATE_SUSPENDED;
 }
 
 int kvm_arch_vcpu_ioctl_get_mpstate(struct kvm_vcpu *vcpu,
 				    struct kvm_mp_state *mp_state)
 {
-	*mp_state = vcpu->arch.mp_state;
+	*mp_state = READ_ONCE(vcpu->arch.mp_state);
 
 	return 0;
 }
@@ -480,12 +507,14 @@ int kvm_arch_vcpu_ioctl_set_mpstate(struct kvm_vcpu *vcpu,
 {
 	int ret = 0;
 
+	spin_lock(&vcpu->arch.mp_state_lock);
+
 	switch (mp_state->mp_state) {
 	case KVM_MP_STATE_RUNNABLE:
-		vcpu->arch.mp_state = *mp_state;
+		WRITE_ONCE(vcpu->arch.mp_state, *mp_state);
 		break;
 	case KVM_MP_STATE_STOPPED:
-		kvm_arm_vcpu_power_off(vcpu);
+		__kvm_arm_vcpu_power_off(vcpu);
 		break;
 	case KVM_MP_STATE_SUSPENDED:
 		kvm_arm_vcpu_suspend(vcpu);
@@ -494,6 +523,8 @@ int kvm_arch_vcpu_ioctl_set_mpstate(struct kvm_vcpu *vcpu,
 		ret = -EINVAL;
 	}
 
+	spin_unlock(&vcpu->arch.mp_state_lock);
+
 	return ret;
 }
 
@@ -593,9 +624,9 @@ int kvm_arch_vcpu_run_pid_change(struct kvm_vcpu *vcpu)
 	if (kvm_vm_is_protected(kvm))
 		kvm_call_hyp_nvhe(__pkvm_vcpu_init_traps, vcpu);
 
-	mutex_lock(&kvm->lock);
+	mutex_lock(&kvm->arch.config_lock);
 	set_bit(KVM_ARCH_FLAG_HAS_RAN_ONCE, &kvm->arch.flags);
-	mutex_unlock(&kvm->lock);
+	mutex_unlock(&kvm->arch.config_lock);
 
 	return ret;
 }
@@ -1210,10 +1241,14 @@ static int kvm_arch_vcpu_ioctl_vcpu_init(struct kvm_vcpu *vcpu,
 	/*
 	 * Handle the "start in power-off" case.
 	 */
+	spin_lock(&vcpu->arch.mp_state_lock);
+
 	if (test_bit(KVM_ARM_VCPU_POWER_OFF, vcpu->arch.features))
-		kvm_arm_vcpu_power_off(vcpu);
+		__kvm_arm_vcpu_power_off(vcpu);
 	else
-		vcpu->arch.mp_state.mp_state = KVM_MP_STATE_RUNNABLE;
+		WRITE_ONCE(vcpu->arch.mp_state.mp_state, KVM_MP_STATE_RUNNABLE);
+
+	spin_unlock(&vcpu->arch.mp_state_lock);
 
 	return 0;
 }

arch/arm64/kvm/guest.c

@@ -957,7 +957,9 @@ int kvm_arm_vcpu_arch_set_attr(struct kvm_vcpu *vcpu,
 
 	switch (attr->group) {
 	case KVM_ARM_VCPU_PMU_V3_CTRL:
+		mutex_lock(&vcpu->kvm->arch.config_lock);
 		ret = kvm_arm_pmu_v3_set_attr(vcpu, attr);
+		mutex_unlock(&vcpu->kvm->arch.config_lock);
 		break;
 	case KVM_ARM_VCPU_TIMER_CTRL:
 		ret = kvm_arm_timer_set_attr(vcpu, attr);

arch/arm64/kvm/hypercalls.c

@@ -377,7 +377,7 @@ static int kvm_arm_set_fw_reg_bmap(struct kvm_vcpu *vcpu, u64 reg_id, u64 val)
 	if (val & ~fw_reg_features)
 		return -EINVAL;
 
-	mutex_lock(&kvm->lock);
+	mutex_lock(&kvm->arch.config_lock);
 
 	if (test_bit(KVM_ARCH_FLAG_HAS_RAN_ONCE, &kvm->arch.flags) &&
 	    val != *fw_reg_bmap) {
@@ -387,7 +387,7 @@ static int kvm_arm_set_fw_reg_bmap(struct kvm_vcpu *vcpu, u64 reg_id, u64 val)
 
 	WRITE_ONCE(*fw_reg_bmap, val);
 out:
-	mutex_unlock(&kvm->lock);
+	mutex_unlock(&kvm->arch.config_lock);
 	return ret;
 }
 

arch/arm64/kvm/pmu-emul.c

@@ -874,7 +874,7 @@ static int kvm_arm_pmu_v3_set_pmu(struct kvm_vcpu *vcpu, int pmu_id)
 	struct arm_pmu *arm_pmu;
 	int ret = -ENXIO;
 
-	mutex_lock(&kvm->lock);
+	lockdep_assert_held(&kvm->arch.config_lock);
 	mutex_lock(&arm_pmus_lock);
 
 	list_for_each_entry(entry, &arm_pmus, entry) {
@@ -894,7 +894,6 @@ static int kvm_arm_pmu_v3_set_pmu(struct kvm_vcpu *vcpu, int pmu_id)
 	}
 
 	mutex_unlock(&arm_pmus_lock);
-	mutex_unlock(&kvm->lock);
 	return ret;
 }
 
@@ -902,22 +901,20 @@ int kvm_arm_pmu_v3_set_attr(struct kvm_vcpu *vcpu, struct kvm_device_attr *attr)
 {
 	struct kvm *kvm = vcpu->kvm;
 
+	lockdep_assert_held(&kvm->arch.config_lock);
+
 	if (!kvm_vcpu_has_pmu(vcpu))
 		return -ENODEV;
 
 	if (vcpu->arch.pmu.created)
 		return -EBUSY;
 
-	mutex_lock(&kvm->lock);
 	if (!kvm->arch.arm_pmu) {
 		/* No PMU set, get the default one */
 		kvm->arch.arm_pmu = kvm_pmu_probe_armpmu();
-		if (!kvm->arch.arm_pmu) {
+		if (!kvm->arch.arm_pmu)
-			mutex_unlock(&kvm->lock);
 			return -ENODEV;
-		}
 	}
-	mutex_unlock(&kvm->lock);
 
 	switch (attr->attr) {
 	case KVM_ARM_VCPU_PMU_V3_IRQ: {
@@ -961,19 +958,13 @@ int kvm_arm_pmu_v3_set_attr(struct kvm_vcpu *vcpu, struct kvm_device_attr *attr)
 		     filter.action != KVM_PMU_EVENT_DENY))
 			return -EINVAL;
 
-		mutex_lock(&kvm->lock);
+		if (test_bit(KVM_ARCH_FLAG_HAS_RAN_ONCE, &kvm->arch.flags))
-
-		if (test_bit(KVM_ARCH_FLAG_HAS_RAN_ONCE, &kvm->arch.flags)) {
-			mutex_unlock(&kvm->lock);
 			return -EBUSY;
-		}
 
 		if (!kvm->arch.pmu_filter) {
 			kvm->arch.pmu_filter = bitmap_alloc(nr_events, GFP_KERNEL_ACCOUNT);
-			if (!kvm->arch.pmu_filter) {
+			if (!kvm->arch.pmu_filter)
-				mutex_unlock(&kvm->lock);
 				return -ENOMEM;
-			}
 
 			/*
 			 * The default depends on the first applied filter.
@@ -992,8 +983,6 @@ int kvm_arm_pmu_v3_set_attr(struct kvm_vcpu *vcpu, struct kvm_device_attr *attr)
 		else
 			bitmap_clear(kvm->arch.pmu_filter, filter.base_event, filter.nevents);
 
-		mutex_unlock(&kvm->lock);
-
 		return 0;
 	}
 	case KVM_ARM_VCPU_PMU_V3_SET_PMU: {

arch/arm64/kvm/psci.c

@@ -62,6 +62,7 @@ static unsigned long kvm_psci_vcpu_on(struct kvm_vcpu *source_vcpu)
 	struct vcpu_reset_state *reset_state;
 	struct kvm *kvm = source_vcpu->kvm;
 	struct kvm_vcpu *vcpu = NULL;
+	int ret = PSCI_RET_SUCCESS;
 	unsigned long cpu_id;
 
 	cpu_id = smccc_get_arg1(source_vcpu);
@@ -76,11 +77,15 @@ static unsigned long kvm_psci_vcpu_on(struct kvm_vcpu *source_vcpu)
 	 */
 	if (!vcpu)
 		return PSCI_RET_INVALID_PARAMS;
+
+	spin_lock(&vcpu->arch.mp_state_lock);
 	if (!kvm_arm_vcpu_stopped(vcpu)) {
 		if (kvm_psci_version(source_vcpu) != KVM_ARM_PSCI_0_1)
-			return PSCI_RET_ALREADY_ON;
+			ret = PSCI_RET_ALREADY_ON;
 		else
-			return PSCI_RET_INVALID_PARAMS;
+			ret = PSCI_RET_INVALID_PARAMS;
+
+		goto out_unlock;
 	}
 
 	reset_state = &vcpu->arch.reset_state;
@@ -96,7 +101,7 @@ static unsigned long kvm_psci_vcpu_on(struct kvm_vcpu *source_vcpu)
 	 */
 	reset_state->r0 = smccc_get_arg3(source_vcpu);
 
-	WRITE_ONCE(reset_state->reset, true);
+	reset_state->reset = true;
 	kvm_make_request(KVM_REQ_VCPU_RESET, vcpu);
 
 	/*
@@ -105,10 +110,12 @@ static unsigned long kvm_psci_vcpu_on(struct kvm_vcpu *source_vcpu)
 	 */
 	smp_wmb();
 
-	vcpu->arch.mp_state.mp_state = KVM_MP_STATE_RUNNABLE;
+	WRITE_ONCE(vcpu->arch.mp_state.mp_state, KVM_MP_STATE_RUNNABLE);
 	kvm_vcpu_wake_up(vcpu);
 
-	return PSCI_RET_SUCCESS;
+out_unlock:
+	spin_unlock(&vcpu->arch.mp_state_lock);
+	return ret;
 }
 
 static unsigned long kvm_psci_vcpu_affinity_info(struct kvm_vcpu *vcpu)
@@ -168,8 +175,11 @@ static void kvm_prepare_system_event(struct kvm_vcpu *vcpu, u32 type, u64 flags)
 	 * after this call is handled and before the VCPUs have been
 	 * re-initialized.
 	 */
-	kvm_for_each_vcpu(i, tmp, vcpu->kvm)
+	kvm_for_each_vcpu(i, tmp, vcpu->kvm) {
-		tmp->arch.mp_state.mp_state = KVM_MP_STATE_STOPPED;
+		spin_lock(&tmp->arch.mp_state_lock);
+		WRITE_ONCE(tmp->arch.mp_state.mp_state, KVM_MP_STATE_STOPPED);
+		spin_unlock(&tmp->arch.mp_state_lock);
+	}
 	kvm_make_all_cpus_request(vcpu->kvm, KVM_REQ_SLEEP);
 
 	memset(&vcpu->run->system_event, 0, sizeof(vcpu->run->system_event));
@@ -229,7 +239,6 @@ static unsigned long kvm_psci_check_allowed_function(struct kvm_vcpu *vcpu, u32
 
 static int kvm_psci_0_2_call(struct kvm_vcpu *vcpu)
 {
-	struct kvm *kvm = vcpu->kvm;
 	u32 psci_fn = smccc_get_function(vcpu);
 	unsigned long val;
 	int ret = 1;
@@ -254,9 +263,7 @@ static int kvm_psci_0_2_call(struct kvm_vcpu *vcpu)
 		kvm_psci_narrow_to_32bit(vcpu);
 		fallthrough;
 	case PSCI_0_2_FN64_CPU_ON:
-		mutex_lock(&kvm->lock);
 		val = kvm_psci_vcpu_on(vcpu);
-		mutex_unlock(&kvm->lock);
 		break;
 	case PSCI_0_2_FN_AFFINITY_INFO:
 		kvm_psci_narrow_to_32bit(vcpu);
@@ -395,7 +402,6 @@ static int kvm_psci_1_x_call(struct kvm_vcpu *vcpu, u32 minor)
 
 static int kvm_psci_0_1_call(struct kvm_vcpu *vcpu)
 {
-	struct kvm *kvm = vcpu->kvm;
 	u32 psci_fn = smccc_get_function(vcpu);
 	unsigned long val;
 
@@ -405,9 +411,7 @@ static int kvm_psci_0_1_call(struct kvm_vcpu *vcpu)
 		val = PSCI_RET_SUCCESS;
 		break;
 	case KVM_PSCI_FN_CPU_ON:
-		mutex_lock(&kvm->lock);
 		val = kvm_psci_vcpu_on(vcpu);
-		mutex_unlock(&kvm->lock);
 		break;
 	default:
 		val = PSCI_RET_NOT_SUPPORTED;

arch/arm64/kvm/reset.c

@@ -205,7 +205,7 @@ static int kvm_set_vm_width(struct kvm_vcpu *vcpu)
 
 	is32bit = vcpu_has_feature(vcpu, KVM_ARM_VCPU_EL1_32BIT);
 
-	lockdep_assert_held(&kvm->lock);
+	lockdep_assert_held(&kvm->arch.config_lock);
 
 	if (test_bit(KVM_ARCH_FLAG_REG_WIDTH_CONFIGURED, &kvm->arch.flags)) {
 		/*
@@ -262,17 +262,18 @@ int kvm_reset_vcpu(struct kvm_vcpu *vcpu)
 	bool loaded;
 	u32 pstate;
 
-	mutex_lock(&vcpu->kvm->lock);
+	mutex_lock(&vcpu->kvm->arch.config_lock);
 	ret = kvm_set_vm_width(vcpu);
-	if (!ret) {
+	mutex_unlock(&vcpu->kvm->arch.config_lock);
-		reset_state = vcpu->arch.reset_state;
-		WRITE_ONCE(vcpu->arch.reset_state.reset, false);
-	}
-	mutex_unlock(&vcpu->kvm->lock);
 
 	if (ret)
 		return ret;
 
+	spin_lock(&vcpu->arch.mp_state_lock);
+	reset_state = vcpu->arch.reset_state;
+	vcpu->arch.reset_state.reset = false;
+	spin_unlock(&vcpu->arch.mp_state_lock);
+
 	/* Reset PMU outside of the non-preemptible section */
 	kvm_pmu_vcpu_reset(vcpu);
 

arch/arm64/kvm/vgic/vgic-debug.c

@@ -85,7 +85,7 @@ static void *vgic_debug_start(struct seq_file *s, loff_t *pos)
 	struct kvm *kvm = s->private;
 	struct vgic_state_iter *iter;
 
-	mutex_lock(&kvm->lock);
+	mutex_lock(&kvm->arch.config_lock);
 	iter = kvm->arch.vgic.iter;
 	if (iter) {
 		iter = ERR_PTR(-EBUSY);
@@ -104,7 +104,7 @@ static void *vgic_debug_start(struct seq_file *s, loff_t *pos)
 	if (end_of_vgic(iter))
 		iter = NULL;
 out:
-	mutex_unlock(&kvm->lock);
+	mutex_unlock(&kvm->arch.config_lock);
 	return iter;
 }
 
@@ -132,12 +132,12 @@ static void vgic_debug_stop(struct seq_file *s, void *v)
 	if (IS_ERR(v))
 		return;
 
-	mutex_lock(&kvm->lock);
+	mutex_lock(&kvm->arch.config_lock);
 	iter = kvm->arch.vgic.iter;
 	kfree(iter->lpi_array);
 	kfree(iter);
 	kvm->arch.vgic.iter = NULL;
-	mutex_unlock(&kvm->lock);
+	mutex_unlock(&kvm->arch.config_lock);
 }
 
 static void print_dist_state(struct seq_file *s, struct vgic_dist *dist)

arch/arm64/kvm/vgic/vgic-init.c

@@ -74,9 +74,6 @@ int kvm_vgic_create(struct kvm *kvm, u32 type)
 	unsigned long i;
 	int ret;
 
-	if (irqchip_in_kernel(kvm))
-		return -EEXIST;
-
 	/*
 	 * This function is also called by the KVM_CREATE_IRQCHIP handler,
 	 * which had no chance yet to check the availability of the GICv2
@@ -87,10 +84,20 @@ int kvm_vgic_create(struct kvm *kvm, u32 type)
 		!kvm_vgic_global_state.can_emulate_gicv2)
 		return -ENODEV;
 
+	/* Must be held to avoid race with vCPU creation */
+	lockdep_assert_held(&kvm->lock);
+
 	ret = -EBUSY;
 	if (!lock_all_vcpus(kvm))
 		return ret;
 
+	mutex_lock(&kvm->arch.config_lock);
+
+	if (irqchip_in_kernel(kvm)) {
+		ret = -EEXIST;
+		goto out_unlock;
+	}
+
 	kvm_for_each_vcpu(i, vcpu, kvm) {
 		if (vcpu_has_run_once(vcpu))
 			goto out_unlock;
@@ -118,6 +125,7 @@ int kvm_vgic_create(struct kvm *kvm, u32 type)
 		INIT_LIST_HEAD(&kvm->arch.vgic.rd_regions);
 
 out_unlock:
+	mutex_unlock(&kvm->arch.config_lock);
 	unlock_all_vcpus(kvm);
 	return ret;
 }
@@ -227,9 +235,9 @@ int kvm_vgic_vcpu_init(struct kvm_vcpu *vcpu)
 	 * KVM io device for the redistributor that belongs to this VCPU.
 	 */
 	if (dist->vgic_model == KVM_DEV_TYPE_ARM_VGIC_V3) {
-		mutex_lock(&vcpu->kvm->lock);
+		mutex_lock(&vcpu->kvm->arch.config_lock);
 		ret = vgic_register_redist_iodev(vcpu);
-		mutex_unlock(&vcpu->kvm->lock);
+		mutex_unlock(&vcpu->kvm->arch.config_lock);
 	}
 	return ret;
 }
@@ -250,7 +258,6 @@ static void kvm_vgic_vcpu_enable(struct kvm_vcpu *vcpu)
  * The function is generally called when nr_spis has been explicitly set
  * by the guest through the KVM DEVICE API. If not nr_spis is set to 256.
  * vgic_initialized() returns true when this function has succeeded.
- * Must be called with kvm->lock held!
  */
 int vgic_init(struct kvm *kvm)
 {
@@ -259,6 +266,8 @@ int vgic_init(struct kvm *kvm)
 	int ret = 0, i;
 	unsigned long idx;
 
+	lockdep_assert_held(&kvm->arch.config_lock);
+
 	if (vgic_initialized(kvm))
 		return 0;
 
@@ -373,12 +382,13 @@ void kvm_vgic_vcpu_destroy(struct kvm_vcpu *vcpu)
 	vgic_cpu->rd_iodev.base_addr = VGIC_ADDR_UNDEF;
 }
 
-/* To be called with kvm->lock held */
 static void __kvm_vgic_destroy(struct kvm *kvm)
 {
 	struct kvm_vcpu *vcpu;
 	unsigned long i;
 
+	lockdep_assert_held(&kvm->arch.config_lock);
+
 	vgic_debug_destroy(kvm);
 
 	kvm_for_each_vcpu(i, vcpu, kvm)
@@ -389,9 +399,9 @@ static void __kvm_vgic_destroy(struct kvm *kvm)
 
 void kvm_vgic_destroy(struct kvm *kvm)
 {
-	mutex_lock(&kvm->lock);
+	mutex_lock(&kvm->arch.config_lock);
 	__kvm_vgic_destroy(kvm);
-	mutex_unlock(&kvm->lock);
+	mutex_unlock(&kvm->arch.config_lock);
 }
 
 /**
@@ -414,9 +424,9 @@ int vgic_lazy_init(struct kvm *kvm)
 		if (kvm->arch.vgic.vgic_model != KVM_DEV_TYPE_ARM_VGIC_V2)
 			return -EBUSY;
 
-		mutex_lock(&kvm->lock);
+		mutex_lock(&kvm->arch.config_lock);
 		ret = vgic_init(kvm);
-		mutex_unlock(&kvm->lock);
+		mutex_unlock(&kvm->arch.config_lock);
 	}
 
 	return ret;
@@ -441,7 +451,7 @@ int kvm_vgic_map_resources(struct kvm *kvm)
 	if (likely(vgic_ready(kvm)))
 		return 0;
 
-	mutex_lock(&kvm->lock);
+	mutex_lock(&kvm->arch.config_lock);
 	if (vgic_ready(kvm))
 		goto out;
 
@@ -459,7 +469,7 @@ int kvm_vgic_map_resources(struct kvm *kvm)
 		dist->ready = true;
 
 out:
-	mutex_unlock(&kvm->lock);
+	mutex_unlock(&kvm->arch.config_lock);
 	return ret;
 }
 

arch/arm64/kvm/vgic/vgic-its.c

@@ -1958,6 +1958,16 @@ static int vgic_its_create(struct kvm_device *dev, u32 type)
 	mutex_init(&its->its_lock);
 	mutex_init(&its->cmd_lock);
 
+	/* Yep, even more trickery for lock ordering... */
+#ifdef CONFIG_LOCKDEP
+	mutex_lock(&dev->kvm->arch.config_lock);
+	mutex_lock(&its->cmd_lock);
+	mutex_lock(&its->its_lock);
+	mutex_unlock(&its->its_lock);
+	mutex_unlock(&its->cmd_lock);
+	mutex_unlock(&dev->kvm->arch.config_lock);
+#endif
+
 	its->vgic_its_base = VGIC_ADDR_UNDEF;
 
 	INIT_LIST_HEAD(&its->device_list);
@@ -2045,6 +2055,13 @@ static int vgic_its_attr_regs_access(struct kvm_device *dev,
 
 	mutex_lock(&dev->kvm->lock);
 
+	if (!lock_all_vcpus(dev->kvm)) {
+		mutex_unlock(&dev->kvm->lock);
+		return -EBUSY;
+	}
+
+	mutex_lock(&dev->kvm->arch.config_lock);
+
 	if (IS_VGIC_ADDR_UNDEF(its->vgic_its_base)) {
 		ret = -ENXIO;
 		goto out;
@@ -2058,11 +2075,6 @@ static int vgic_its_attr_regs_access(struct kvm_device *dev,
 		goto out;
 	}
 
-	if (!lock_all_vcpus(dev->kvm)) {
-		ret = -EBUSY;
-		goto out;
-	}
-
 	addr = its->vgic_its_base + offset;
 
 	len = region->access_flags & VGIC_ACCESS_64bit ? 8 : 4;
@@ -2076,8 +2088,9 @@ static int vgic_its_attr_regs_access(struct kvm_device *dev,
 	} else {
 		*reg = region->its_read(dev->kvm, its, addr, len);
 	}
-	unlock_all_vcpus(dev->kvm);
 out:
+	mutex_unlock(&dev->kvm->arch.config_lock);
+	unlock_all_vcpus(dev->kvm);
 	mutex_unlock(&dev->kvm->lock);
 	return ret;
 }
@@ -2749,14 +2762,15 @@ static int vgic_its_ctrl(struct kvm *kvm, struct vgic_its *its, u64 attr)
 		return 0;
 
 	mutex_lock(&kvm->lock);
-	mutex_lock(&its->its_lock);
 
 	if (!lock_all_vcpus(kvm)) {
-		mutex_unlock(&its->its_lock);
 		mutex_unlock(&kvm->lock);
 		return -EBUSY;
 	}
 
+	mutex_lock(&kvm->arch.config_lock);
+	mutex_lock(&its->its_lock);
+
 	switch (attr) {
 	case KVM_DEV_ARM_ITS_CTRL_RESET:
 		vgic_its_reset(kvm, its);
@@ -2769,8 +2783,9 @@ static int vgic_its_ctrl(struct kvm *kvm, struct vgic_its *its, u64 attr)
 		break;
 	}
 
-	unlock_all_vcpus(kvm);
 	mutex_unlock(&its->its_lock);
+	mutex_unlock(&kvm->arch.config_lock);
+	unlock_all_vcpus(kvm);
 	mutex_unlock(&kvm->lock);
 	return ret;
 }

arch/arm64/kvm/vgic/vgic-kvm-device.c

@@ -46,7 +46,7 @@ int kvm_set_legacy_vgic_v2_addr(struct kvm *kvm, struct kvm_arm_device_addr *dev
 	struct vgic_dist *vgic = &kvm->arch.vgic;
 	int r;
 
-	mutex_lock(&kvm->lock);
+	mutex_lock(&kvm->arch.config_lock);
 	switch (FIELD_GET(KVM_ARM_DEVICE_TYPE_MASK, dev_addr->id)) {
 	case KVM_VGIC_V2_ADDR_TYPE_DIST:
 		r = vgic_check_type(kvm, KVM_DEV_TYPE_ARM_VGIC_V2);
@@ -68,7 +68,7 @@ int kvm_set_legacy_vgic_v2_addr(struct kvm *kvm, struct kvm_arm_device_addr *dev
 		r = -ENODEV;
 	}
 
-	mutex_unlock(&kvm->lock);
+	mutex_unlock(&kvm->arch.config_lock);
 
 	return r;
 }
@@ -102,7 +102,7 @@ static int kvm_vgic_addr(struct kvm *kvm, struct kvm_device_attr *attr, bool wri
 		if (get_user(addr, uaddr))
 			return -EFAULT;
 
-	mutex_lock(&kvm->lock);
+	mutex_lock(&kvm->arch.config_lock);
 	switch (attr->attr) {
 	case KVM_VGIC_V2_ADDR_TYPE_DIST:
 		r = vgic_check_type(kvm, KVM_DEV_TYPE_ARM_VGIC_V2);
@@ -191,7 +191,7 @@ static int kvm_vgic_addr(struct kvm *kvm, struct kvm_device_attr *attr, bool wri
 	}
 
 out:
-	mutex_unlock(&kvm->lock);
+	mutex_unlock(&kvm->arch.config_lock);
 
 	if (!r && !write)
 		r =  put_user(addr, uaddr);
@@ -227,7 +227,7 @@ static int vgic_set_common_attr(struct kvm_device *dev,
 		    (val & 31))
 			return -EINVAL;
 
-		mutex_lock(&dev->kvm->lock);
+		mutex_lock(&dev->kvm->arch.config_lock);
 
 		if (vgic_ready(dev->kvm) || dev->kvm->arch.vgic.nr_spis)
 			ret = -EBUSY;
@@ -235,16 +235,16 @@ static int vgic_set_common_attr(struct kvm_device *dev,
 			dev->kvm->arch.vgic.nr_spis =
 				val - VGIC_NR_PRIVATE_IRQS;
 
-		mutex_unlock(&dev->kvm->lock);
+		mutex_unlock(&dev->kvm->arch.config_lock);
 
 		return ret;
 	}
 	case KVM_DEV_ARM_VGIC_GRP_CTRL: {
 		switch (attr->attr) {
 		case KVM_DEV_ARM_VGIC_CTRL_INIT:
-			mutex_lock(&dev->kvm->lock);
+			mutex_lock(&dev->kvm->arch.config_lock);
 			r = vgic_init(dev->kvm);
-			mutex_unlock(&dev->kvm->lock);
+			mutex_unlock(&dev->kvm->arch.config_lock);
 			return r;
 		case KVM_DEV_ARM_VGIC_SAVE_PENDING_TABLES:
 			/*
@@ -260,7 +260,10 @@ static int vgic_set_common_attr(struct kvm_device *dev,
 				mutex_unlock(&dev->kvm->lock);
 				return -EBUSY;
 			}
+
+			mutex_lock(&dev->kvm->arch.config_lock);
 			r = vgic_v3_save_pending_tables(dev->kvm);
+			mutex_unlock(&dev->kvm->arch.config_lock);
 			unlock_all_vcpus(dev->kvm);
 			mutex_unlock(&dev->kvm->lock);
 			return r;
@@ -411,15 +414,17 @@ static int vgic_v2_attr_regs_access(struct kvm_device *dev,
 
 	mutex_lock(&dev->kvm->lock);
 
+	if (!lock_all_vcpus(dev->kvm)) {
+		mutex_unlock(&dev->kvm->lock);
+		return -EBUSY;
+	}
+
+	mutex_lock(&dev->kvm->arch.config_lock);
+
 	ret = vgic_init(dev->kvm);
 	if (ret)
 		goto out;
 
-	if (!lock_all_vcpus(dev->kvm)) {
-		ret = -EBUSY;
-		goto out;
-	}
-
 	switch (attr->group) {
 	case KVM_DEV_ARM_VGIC_GRP_CPU_REGS:
 		ret = vgic_v2_cpuif_uaccess(vcpu, is_write, addr, &val);
@@ -432,8 +437,9 @@ static int vgic_v2_attr_regs_access(struct kvm_device *dev,
 		break;
 	}
 
-	unlock_all_vcpus(dev->kvm);
 out:
+	mutex_unlock(&dev->kvm->arch.config_lock);
+	unlock_all_vcpus(dev->kvm);
 	mutex_unlock(&dev->kvm->lock);
 
 	if (!ret && !is_write)
@@ -569,12 +575,14 @@ static int vgic_v3_attr_regs_access(struct kvm_device *dev,
 
 	mutex_lock(&dev->kvm->lock);
 
-	if (unlikely(!vgic_initialized(dev->kvm))) {
+	if (!lock_all_vcpus(dev->kvm)) {
-		ret = -EBUSY;
+		mutex_unlock(&dev->kvm->lock);
-		goto out;
+		return -EBUSY;
 	}
 
-	if (!lock_all_vcpus(dev->kvm)) {
+	mutex_lock(&dev->kvm->arch.config_lock);
+
+	if (unlikely(!vgic_initialized(dev->kvm))) {
 		ret = -EBUSY;
 		goto out;
 	}
@@ -609,8 +617,9 @@ static int vgic_v3_attr_regs_access(struct kvm_device *dev,
 		break;
 	}
 
-	unlock_all_vcpus(dev->kvm);
 out:
+	mutex_unlock(&dev->kvm->arch.config_lock);
+	unlock_all_vcpus(dev->kvm);
 	mutex_unlock(&dev->kvm->lock);
 
 	if (!ret && uaccess && !is_write) {

arch/arm64/kvm/vgic/vgic-mmio-v3.c

@@ -111,7 +111,7 @@ static void vgic_mmio_write_v3_misc(struct kvm_vcpu *vcpu,
 	case GICD_CTLR: {
 		bool was_enabled, is_hwsgi;
 
-		mutex_lock(&vcpu->kvm->lock);
+		mutex_lock(&vcpu->kvm->arch.config_lock);
 
 		was_enabled = dist->enabled;
 		is_hwsgi = dist->nassgireq;
@@ -139,7 +139,7 @@ static void vgic_mmio_write_v3_misc(struct kvm_vcpu *vcpu,
 		else if (!was_enabled && dist->enabled)
 			vgic_kick_vcpus(vcpu->kvm);
 
-		mutex_unlock(&vcpu->kvm->lock);
+		mutex_unlock(&vcpu->kvm->arch.config_lock);
 		break;
 	}
 	case GICD_TYPER:

arch/arm64/kvm/vgic/vgic-mmio.c

@@ -530,13 +530,13 @@ unsigned long vgic_mmio_read_active(struct kvm_vcpu *vcpu,
 	u32 intid = VGIC_ADDR_TO_INTID(addr, 1);
 	u32 val;
 
-	mutex_lock(&vcpu->kvm->lock);
+	mutex_lock(&vcpu->kvm->arch.config_lock);
 	vgic_access_active_prepare(vcpu, intid);
 
 	val = __vgic_mmio_read_active(vcpu, addr, len);
 
 	vgic_access_active_finish(vcpu, intid);
-	mutex_unlock(&vcpu->kvm->lock);
+	mutex_unlock(&vcpu->kvm->arch.config_lock);
 
 	return val;
 }
@@ -625,13 +625,13 @@ void vgic_mmio_write_cactive(struct kvm_vcpu *vcpu,
 {
 	u32 intid = VGIC_ADDR_TO_INTID(addr, 1);
 
-	mutex_lock(&vcpu->kvm->lock);
+	mutex_lock(&vcpu->kvm->arch.config_lock);
 	vgic_access_active_prepare(vcpu, intid);
 
 	__vgic_mmio_write_cactive(vcpu, addr, len, val);
 
 	vgic_access_active_finish(vcpu, intid);
-	mutex_unlock(&vcpu->kvm->lock);
+	mutex_unlock(&vcpu->kvm->arch.config_lock);
 }
 
 int vgic_mmio_uaccess_write_cactive(struct kvm_vcpu *vcpu,
@@ -662,13 +662,13 @@ void vgic_mmio_write_sactive(struct kvm_vcpu *vcpu,
 {
 	u32 intid = VGIC_ADDR_TO_INTID(addr, 1);
 
-	mutex_lock(&vcpu->kvm->lock);
+	mutex_lock(&vcpu->kvm->arch.config_lock);
 	vgic_access_active_prepare(vcpu, intid);
 
 	__vgic_mmio_write_sactive(vcpu, addr, len, val);
 
 	vgic_access_active_finish(vcpu, intid);
-	mutex_unlock(&vcpu->kvm->lock);
+	mutex_unlock(&vcpu->kvm->arch.config_lock);
 }
 
 int vgic_mmio_uaccess_write_sactive(struct kvm_vcpu *vcpu,

arch/arm64/kvm/vgic/vgic-v4.c

@@ -232,9 +232,8 @@ int vgic_v4_request_vpe_irq(struct kvm_vcpu *vcpu, int irq)
  * @kvm:	Pointer to the VM being initialized
  *
  * We may be called each time a vITS is created, or when the
- * vgic is initialized. This relies on kvm->lock to be
+ * vgic is initialized. In both cases, the number of vcpus
- * held. In both cases, the number of vcpus should now be
+ * should now be fixed.
- * fixed.
  */
 int vgic_v4_init(struct kvm *kvm)
 {
@@ -243,6 +242,8 @@ int vgic_v4_init(struct kvm *kvm)
 	int nr_vcpus, ret;
 	unsigned long i;
 
+	lockdep_assert_held(&kvm->arch.config_lock);
+
 	if (!kvm_vgic_global_state.has_gicv4)
 		return 0; /* Nothing to see here... move along. */
 
@@ -309,14 +310,14 @@ int vgic_v4_init(struct kvm *kvm)
 /**
  * vgic_v4_teardown - Free the GICv4 data structures
  * @kvm:	Pointer to the VM being destroyed
- *
- * Relies on kvm->lock to be held.
  */
 void vgic_v4_teardown(struct kvm *kvm)
 {
 	struct its_vm *its_vm = &kvm->arch.vgic.its_vm;
 	int i;
 
+	lockdep_assert_held(&kvm->arch.config_lock);
+
 	if (!its_vm->vpes)
 		return;
 

arch/arm64/kvm/vgic/vgic.c

@@ -24,11 +24,13 @@ struct vgic_global kvm_vgic_global_state __ro_after_init = {
 /*
  * Locking order is always:
  * kvm->lock (mutex)
- *   its->cmd_lock (mutex)
+ *   vcpu->mutex (mutex)
- *     its->its_lock (mutex)
+ *     kvm->arch.config_lock (mutex)
- *       vgic_cpu->ap_list_lock		must be taken with IRQs disabled
+ *       its->cmd_lock (mutex)
- *         kvm->lpi_list_lock		must be taken with IRQs disabled
+ *         its->its_lock (mutex)
- *           vgic_irq->irq_lock		must be taken with IRQs disabled
+ *           vgic_cpu->ap_list_lock		must be taken with IRQs disabled
+ *             kvm->lpi_list_lock		must be taken with IRQs disabled
+ *               vgic_irq->irq_lock		must be taken with IRQs disabled
  *
  * As the ap_list_lock might be taken from the timer interrupt handler,
  * we have to disable IRQs before taking this lock and everything lower