Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm

Pull kvm fixes from Paolo Bonzini:
 "x86:

   - Miscellaneous bugfixes

   - A small cleanup for the new workqueue code

   - Documentation syntax fix

  RISC-V:

   - Remove hgatp zeroing in kvm_arch_vcpu_put()

   - Fix alignment of the guest_hang() in KVM selftest

   - Fix PTE A and D bits in KVM selftest

   - Missing #include in vcpu_fp.c

  ARM:

   - Some PSCI fixes after introducing PSCIv1.1 and SYSTEM_RESET2

   - Fix the MMU write-lock not being taken on THP split

   - Fix mixed-width VM handling

   - Fix potential UAF when debugfs registration fails

   - Various selftest updates for all of the above"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (24 commits)
  KVM: x86: hyper-v: Avoid writing to TSC page without an active vCPU
  KVM: SVM: Do not activate AVIC for SEV-enabled guest
  Documentation: KVM: Add SPDX-License-Identifier tag
  selftests: kvm: add tsc_scaling_sync to .gitignore
  RISC-V: KVM: include missing hwcap.h into vcpu_fp
  KVM: selftests: riscv: Fix alignment of the guest_hang() function
  KVM: selftests: riscv: Set PTE A and D bits in VS-stage page table
  RISC-V: KVM: Don't clear hgatp CSR in kvm_arch_vcpu_put()
  selftests: KVM: Free the GIC FD when cleaning up in arch_timer
  selftests: KVM: Don't leak GIC FD across dirty log test iterations
  KVM: Don't create VM debugfs files outside of the VM directory
  KVM: selftests: get-reg-list: Add KVM_REG_ARM_FW_REG(3)
  KVM: avoid NULL pointer dereference in kvm_dirty_ring_push
  KVM: arm64: selftests: Introduce vcpu_width_config
  KVM: arm64: mixed-width check should be skipped for uninitialized vCPUs
  KVM: arm64: vgic: Remove unnecessary type castings
  KVM: arm64: Don't split hugepages outside of MMU write lock
  KVM: arm64: Drop unneeded minor version check from PSCI v1.x handler
  KVM: arm64: Actually prevent SMC64 SYSTEM_RESET2 from AArch32
  KVM: arm64: Generally disallow SMC64 for AArch32 guests
  ...

8 files changed, 57 insertions, 50 deletions

diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index d23e80a56eb8..e0c0f0e1f754 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -974,12 +974,10 @@ enum hv_tsc_page_status {
 	HV_TSC_PAGE_UNSET = 0,
 	/* TSC page MSR was written by the guest, update pending */
 	HV_TSC_PAGE_GUEST_CHANGED,
-	/* TSC page MSR was written by KVM userspace, update pending */
+	/* TSC page update was triggered from the host side */
 	HV_TSC_PAGE_HOST_CHANGED,
 	/* TSC page was properly set up and is currently active  */
 	HV_TSC_PAGE_SET,
-	/* TSC page is currently being updated and therefore is inactive */
-	HV_TSC_PAGE_UPDATING,
 	/* TSC page was set up with an inaccessible GPA */
 	HV_TSC_PAGE_BROKEN,
 };
@@ -1052,6 +1050,7 @@ enum kvm_apicv_inhibit {
 	APICV_INHIBIT_REASON_X2APIC,
 	APICV_INHIBIT_REASON_BLOCKIRQ,
 	APICV_INHIBIT_REASON_ABSENT,
+	APICV_INHIBIT_REASON_SEV,
 };
 
 struct kvm_arch {
@@ -1585,8 +1584,9 @@ static inline int kvm_arch_flush_remote_tlb(struct kvm *kvm)
 #define kvm_arch_pmi_in_guest(vcpu) \
 	((vcpu) && (vcpu)->arch.handling_intr_from_guest)
 
-int kvm_mmu_module_init(void);
-void kvm_mmu_module_exit(void);
+void kvm_mmu_x86_module_init(void);
+int kvm_mmu_vendor_module_init(void);
+void kvm_mmu_vendor_module_exit(void);
 
 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
 int kvm_mmu_create(struct kvm_vcpu *vcpu);
diff --git a/arch/x86/kvm/hyperv.c b/arch/x86/kvm/hyperv.c
index 123b677111c5..46f9dfb60469 100644
--- a/arch/x86/kvm/hyperv.c
+++ b/arch/x86/kvm/hyperv.c
@@ -1135,11 +1135,13 @@ void kvm_hv_setup_tsc_page(struct kvm *kvm,
 	BUILD_BUG_ON(sizeof(tsc_seq) != sizeof(hv->tsc_ref.tsc_sequence));
 	BUILD_BUG_ON(offsetof(struct ms_hyperv_tsc_page, tsc_sequence) != 0);
 
+	mutex_lock(&hv->hv_lock);
+
 	if (hv->hv_tsc_page_status == HV_TSC_PAGE_BROKEN ||
+	    hv->hv_tsc_page_status == HV_TSC_PAGE_SET ||
 	    hv->hv_tsc_page_status == HV_TSC_PAGE_UNSET)
-		return;
+		goto out_unlock;
 
-	mutex_lock(&hv->hv_lock);
 	if (!(hv->hv_tsc_page & HV_X64_MSR_TSC_REFERENCE_ENABLE))
 		goto out_unlock;
 
@@ -1201,45 +1203,19 @@ out_unlock:
 	mutex_unlock(&hv->hv_lock);
 }
 
-void kvm_hv_invalidate_tsc_page(struct kvm *kvm)
+void kvm_hv_request_tsc_page_update(struct kvm *kvm)
 {
 	struct kvm_hv *hv = to_kvm_hv(kvm);
-	u64 gfn;
-	int idx;
-
-	if (hv->hv_tsc_page_status == HV_TSC_PAGE_BROKEN ||
-	    hv->hv_tsc_page_status == HV_TSC_PAGE_UNSET ||
-	    tsc_page_update_unsafe(hv))
-		return;
 
 	mutex_lock(&hv->hv_lock);
 
-	if (!(hv->hv_tsc_page & HV_X64_MSR_TSC_REFERENCE_ENABLE))
-		goto out_unlock;
-
-	/* Preserve HV_TSC_PAGE_GUEST_CHANGED/HV_TSC_PAGE_HOST_CHANGED states */
-	if (hv->hv_tsc_page_status == HV_TSC_PAGE_SET)
-		hv->hv_tsc_page_status = HV_TSC_PAGE_UPDATING;
+	if (hv->hv_tsc_page_status == HV_TSC_PAGE_SET &&
+	    !tsc_page_update_unsafe(hv))
+		hv->hv_tsc_page_status = HV_TSC_PAGE_HOST_CHANGED;
 
-	gfn = hv->hv_tsc_page >> HV_X64_MSR_TSC_REFERENCE_ADDRESS_SHIFT;
-
-	hv->tsc_ref.tsc_sequence = 0;
-
-	/*
-	 * Take the srcu lock as memslots will be accessed to check the gfn
-	 * cache generation against the memslots generation.
-	 */
-	idx = srcu_read_lock(&kvm->srcu);
-	if (kvm_write_guest(kvm, gfn_to_gpa(gfn),
-			    &hv->tsc_ref, sizeof(hv->tsc_ref.tsc_sequence)))
-		hv->hv_tsc_page_status = HV_TSC_PAGE_BROKEN;
-	srcu_read_unlock(&kvm->srcu, idx);
-
-out_unlock:
 	mutex_unlock(&hv->hv_lock);
 }
 
-
 static bool hv_check_msr_access(struct kvm_vcpu_hv *hv_vcpu, u32 msr)
 {
 	if (!hv_vcpu->enforce_cpuid)
diff --git a/arch/x86/kvm/hyperv.h b/arch/x86/kvm/hyperv.h
index e19c00ee9ab3..da2737f2a956 100644
--- a/arch/x86/kvm/hyperv.h
+++ b/arch/x86/kvm/hyperv.h
@@ -137,7 +137,7 @@ void kvm_hv_process_stimers(struct kvm_vcpu *vcpu);
 
 void kvm_hv_setup_tsc_page(struct kvm *kvm,
 			   struct pvclock_vcpu_time_info *hv_clock);
-void kvm_hv_invalidate_tsc_page(struct kvm *kvm);
+void kvm_hv_request_tsc_page_update(struct kvm *kvm);
 
 void kvm_hv_init_vm(struct kvm *kvm);
 void kvm_hv_destroy_vm(struct kvm *kvm);
diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
index 8f19ea752704..f9080ee50ffa 100644
--- a/arch/x86/kvm/mmu/mmu.c
+++ b/arch/x86/kvm/mmu/mmu.c
@@ -6237,12 +6237,24 @@ static int set_nx_huge_pages(const char *val, const struct kernel_param *kp)
 	return 0;
 }
 
-int kvm_mmu_module_init(void)
+/*
+ * nx_huge_pages needs to be resolved to true/false when kvm.ko is loaded, as
+ * its default value of -1 is technically undefined behavior for a boolean.
+ */
+void kvm_mmu_x86_module_init(void)
 {
-	int ret = -ENOMEM;
-
 	if (nx_huge_pages == -1)
 		__set_nx_huge_pages(get_nx_auto_mode());
+}
+
+/*
+ * The bulk of the MMU initialization is deferred until the vendor module is
+ * loaded as many of the masks/values may be modified by VMX or SVM, i.e. need
+ * to be reset when a potentially different vendor module is loaded.
+ */
+int kvm_mmu_vendor_module_init(void)
+{
+	int ret = -ENOMEM;
 
 	/*
 	 * MMU roles use union aliasing which is, generally speaking, an
@@ -6290,7 +6302,7 @@ void kvm_mmu_destroy(struct kvm_vcpu *vcpu)
 	mmu_free_memory_caches(vcpu);
 }
 
-void kvm_mmu_module_exit(void)
+void kvm_mmu_vendor_module_exit(void)
 {
 	mmu_destroy_caches();
 	percpu_counter_destroy(&kvm_total_used_mmu_pages);
diff --git a/arch/x86/kvm/mmu/tdp_mmu.c b/arch/x86/kvm/mmu/tdp_mmu.c
index d71d177ae6b8..c472769e0300 100644
--- a/arch/x86/kvm/mmu/tdp_mmu.c
+++ b/arch/x86/kvm/mmu/tdp_mmu.c
@@ -51,7 +51,7 @@ void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm)
 	if (!kvm->arch.tdp_mmu_enabled)
 		return;
 
-	flush_workqueue(kvm->arch.tdp_mmu_zap_wq);
+	/* Also waits for any queued work items.  */
 	destroy_workqueue(kvm->arch.tdp_mmu_zap_wq);
 
 	WARN_ON(!list_empty(&kvm->arch.tdp_mmu_pages));
diff --git a/arch/x86/kvm/svm/avic.c b/arch/x86/kvm/svm/avic.c
index a1cf9c31273b..421619540ff9 100644
--- a/arch/x86/kvm/svm/avic.c
+++ b/arch/x86/kvm/svm/avic.c
@@ -837,7 +837,8 @@ bool avic_check_apicv_inhibit_reasons(enum kvm_apicv_inhibit reason)
 			  BIT(APICV_INHIBIT_REASON_IRQWIN) |
 			  BIT(APICV_INHIBIT_REASON_PIT_REINJ) |
 			  BIT(APICV_INHIBIT_REASON_X2APIC) |
-			  BIT(APICV_INHIBIT_REASON_BLOCKIRQ);
+			  BIT(APICV_INHIBIT_REASON_BLOCKIRQ) |
+			  BIT(APICV_INHIBIT_REASON_SEV);
 
 	return supported & BIT(reason);
 }
diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c
index 75fa6dd268f0..537aaddc852f 100644
--- a/arch/x86/kvm/svm/sev.c
+++ b/arch/x86/kvm/svm/sev.c
@@ -260,6 +260,8 @@ static int sev_guest_init(struct kvm *kvm, struct kvm_sev_cmd *argp)
 	INIT_LIST_HEAD(&sev->regions_list);
 	INIT_LIST_HEAD(&sev->mirror_vms);
 
+	kvm_set_apicv_inhibit(kvm, APICV_INHIBIT_REASON_SEV);
+
 	return 0;
 
 e_free:
@@ -465,6 +467,7 @@ static void sev_clflush_pages(struct page *pages[], unsigned long npages)
 		page_virtual = kmap_atomic(pages[i]);
 		clflush_cache_range(page_virtual, PAGE_SIZE);
 		kunmap_atomic(page_virtual);
+		cond_resched();
 	}
 }
 
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 0c0ca599a353..547ba00ef64f 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -2901,7 +2901,7 @@ static void kvm_end_pvclock_update(struct kvm *kvm)
 
 static void kvm_update_masterclock(struct kvm *kvm)
 {
-	kvm_hv_invalidate_tsc_page(kvm);
+	kvm_hv_request_tsc_page_update(kvm);
 	kvm_start_pvclock_update(kvm);
 	pvclock_update_vm_gtod_copy(kvm);
 	kvm_end_pvclock_update(kvm);
@@ -3113,8 +3113,7 @@ static int kvm_guest_time_update(struct kvm_vcpu *v)
 				       offsetof(struct compat_vcpu_info, time));
 	if (vcpu->xen.vcpu_time_info_set)
 		kvm_setup_pvclock_page(v, &vcpu->xen.vcpu_time_info_cache, 0);
-	if (!v->vcpu_idx)
-		kvm_hv_setup_tsc_page(v->kvm, &vcpu->hv_clock);
+	kvm_hv_setup_tsc_page(v->kvm, &vcpu->hv_clock);
 	return 0;
 }
 
@@ -6241,7 +6240,7 @@ static int kvm_vm_ioctl_set_clock(struct kvm *kvm, void __user *argp)
 	if (data.flags & ~KVM_CLOCK_VALID_FLAGS)
 		return -EINVAL;
 
-	kvm_hv_invalidate_tsc_page(kvm);
+	kvm_hv_request_tsc_page_update(kvm);
 	kvm_start_pvclock_update(kvm);
 	pvclock_update_vm_gtod_copy(kvm);
 
@@ -8926,7 +8925,7 @@ int kvm_arch_init(void *opaque)
 	}
 	kvm_nr_uret_msrs = 0;
 
-	r = kvm_mmu_module_init();
+	r = kvm_mmu_vendor_module_init();
 	if (r)
 		goto out_free_percpu;
 
@@ -8974,7 +8973,7 @@ void kvm_arch_exit(void)
 	cancel_work_sync(&pvclock_gtod_work);
 #endif
 	kvm_x86_ops.hardware_enable = NULL;
-	kvm_mmu_module_exit();
+	kvm_mmu_vendor_module_exit();
 	free_percpu(user_return_msrs);
 	kmem_cache_destroy(x86_emulator_cache);
 #ifdef CONFIG_KVM_XEN
@@ -12986,3 +12985,19 @@ EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_vmgexit_enter);
 EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_vmgexit_exit);
 EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_vmgexit_msr_protocol_enter);
 EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_vmgexit_msr_protocol_exit);
+
+static int __init kvm_x86_init(void)
+{
+	kvm_mmu_x86_module_init();
+	return 0;
+}
+module_init(kvm_x86_init);
+
+static void __exit kvm_x86_exit(void)
+{
+	/*
+	 * If module_init() is implemented, module_exit() must also be
+	 * implemented to allow module unload.
+	 */
+}
+module_exit(kvm_x86_exit);