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

Pull KVM fixes from Paolo Bonzini:
 "ARM:

   - Plug race between enabling MTE and creating vcpus

   - Fix off-by-one bug when checking whether an address range is RAM

  x86:

   - Fixes for the new MMU, especially a memory leak on hosts with <39
     physical address bits

   - Remove bogus EFER.NX checks on 32-bit non-PAE hosts

   - WAITPKG fix"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: x86/mmu: Protect marking SPs unsync when using TDP MMU with spinlock
  KVM: x86/mmu: Don't step down in the TDP iterator when zapping all SPTEs
  KVM: x86/mmu: Don't leak non-leaf SPTEs when zapping all SPTEs
  KVM: nVMX: Use vmx_need_pf_intercept() when deciding if L0 wants a #PF
  kvm: vmx: Sync all matching EPTPs when injecting nested EPT fault
  KVM: x86: remove dead initialization
  KVM: x86: Allow guest to set EFER.NX=1 on non-PAE 32-bit kernels
  KVM: VMX: Use current VMCS to query WAITPKG support for MSR emulation
  KVM: arm64: Fix race when enabling KVM_ARM_CAP_MTE
  KVM: arm64: Fix off-by-one in range_is_memory

Documentation/virt/kvm/locking.rst

@@ -25,10 +25,10 @@ On x86:
 
 - vcpu->mutex is taken outside kvm->arch.hyperv.hv_lock
 
-- kvm->arch.mmu_lock is an rwlock.  kvm->arch.tdp_mmu_pages_lock is
-  taken inside kvm->arch.mmu_lock, and cannot be taken without already
-  holding kvm->arch.mmu_lock (typically with ``read_lock``, otherwise
-  there's no need to take kvm->arch.tdp_mmu_pages_lock at all).
+- kvm->arch.mmu_lock is an rwlock.  kvm->arch.tdp_mmu_pages_lock and
+  kvm->arch.mmu_unsync_pages_lock are taken inside kvm->arch.mmu_lock, and
+  cannot be taken without already holding kvm->arch.mmu_lock (typically with
+  ``read_lock`` for the TDP MMU, thus the need for additional spinlocks).
 
 Everything else is a leaf: no other lock is taken inside the critical
 sections.

arch/arm64/kvm/arm.c

@@ -94,10 +94,14 @@ int kvm_vm_ioctl_enable_cap(struct kvm *kvm,
 		kvm->arch.return_nisv_io_abort_to_user = true;
 		break;
 	case KVM_CAP_ARM_MTE:
-		if (!system_supports_mte() || kvm->created_vcpus)
-			return -EINVAL;
+		mutex_lock(&kvm->lock);
+		if (!system_supports_mte() || kvm->created_vcpus) {
+			r = -EINVAL;
+		} else {
 			r = 0;
 			kvm->arch.mte_enabled = true;
+		}
+		mutex_unlock(&kvm->lock);
 		break;
 	default:
 		r = -EINVAL;

arch/arm64/kvm/hyp/nvhe/mem_protect.c

@@ -193,7 +193,7 @@ static bool range_is_memory(u64 start, u64 end)
 {
 	struct kvm_mem_range r1, r2;
 
-	if (!find_mem_range(start, &r1) || !find_mem_range(end, &r2))
+	if (!find_mem_range(start, &r1) || !find_mem_range(end - 1, &r2))
 		return false;
 	if (r1.start != r2.start)
 		return false;

arch/x86/include/asm/kvm_host.h

@@ -1038,6 +1038,13 @@ struct kvm_arch {
 	struct list_head lpage_disallowed_mmu_pages;
 	struct kvm_page_track_notifier_node mmu_sp_tracker;
 	struct kvm_page_track_notifier_head track_notifier_head;
+	/*
+	 * Protects marking pages unsync during page faults, as TDP MMU page
+	 * faults only take mmu_lock for read.  For simplicity, the unsync
+	 * pages lock is always taken when marking pages unsync regardless of
+	 * whether mmu_lock is held for read or write.
+	 */
+	spinlock_t mmu_unsync_pages_lock;
 
 	struct list_head assigned_dev_head;
 	struct iommu_domain *iommu_domain;

arch/x86/kvm/cpuid.c

@@ -208,30 +208,6 @@ static void kvm_vcpu_after_set_cpuid(struct kvm_vcpu *vcpu)
 	kvm_mmu_after_set_cpuid(vcpu);
 }
 
-static int is_efer_nx(void)
-{
-	return host_efer & EFER_NX;
-}
-
-static void cpuid_fix_nx_cap(struct kvm_vcpu *vcpu)
-{
-	int i;
-	struct kvm_cpuid_entry2 *e, *entry;
-
-	entry = NULL;
-	for (i = 0; i < vcpu->arch.cpuid_nent; ++i) {
-		e = &vcpu->arch.cpuid_entries[i];
-		if (e->function == 0x80000001) {
-			entry = e;
-			break;
-		}
-	}
-	if (entry && cpuid_entry_has(entry, X86_FEATURE_NX) && !is_efer_nx()) {
-		cpuid_entry_clear(entry, X86_FEATURE_NX);
-		printk(KERN_INFO "kvm: guest NX capability removed\n");
-	}
-}
-
 int cpuid_query_maxphyaddr(struct kvm_vcpu *vcpu)
 {
 	struct kvm_cpuid_entry2 *best;
@@ -302,7 +278,6 @@ int kvm_vcpu_ioctl_set_cpuid(struct kvm_vcpu *vcpu,
 	vcpu->arch.cpuid_entries = e2;
 	vcpu->arch.cpuid_nent = cpuid->nent;
 
-	cpuid_fix_nx_cap(vcpu);
 	kvm_update_cpuid_runtime(vcpu);
 	kvm_vcpu_after_set_cpuid(vcpu);
 
@@ -401,7 +376,6 @@ static __always_inline void kvm_cpu_cap_mask(enum cpuid_leafs leaf, u32 mask)
 
 void kvm_set_cpu_caps(void)
 {
-	unsigned int f_nx = is_efer_nx() ? F(NX) : 0;
 #ifdef CONFIG_X86_64
 	unsigned int f_gbpages = F(GBPAGES);
 	unsigned int f_lm = F(LM);
@@ -515,7 +489,7 @@ void kvm_set_cpu_caps(void)
 		F(CX8) | F(APIC) | 0 /* Reserved */ | F(SYSCALL) |
 		F(MTRR) | F(PGE) | F(MCA) | F(CMOV) |
 		F(PAT) | F(PSE36) | 0 /* Reserved */ |
-		f_nx | 0 /* Reserved */ | F(MMXEXT) | F(MMX) |
+		F(NX) | 0 /* Reserved */ | F(MMXEXT) | F(MMX) |
 		F(FXSR) | F(FXSR_OPT) | f_gbpages | F(RDTSCP) |
 		0 /* Reserved */ | f_lm | F(3DNOWEXT) | F(3DNOW)
 	);

arch/x86/kvm/hyperv.c

@@ -1933,7 +1933,7 @@ static u64 kvm_hv_send_ipi(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc, bool
 void kvm_hv_set_cpuid(struct kvm_vcpu *vcpu)
 {
 	struct kvm_cpuid_entry2 *entry;
-	struct kvm_vcpu_hv *hv_vcpu = to_hv_vcpu(vcpu);
+	struct kvm_vcpu_hv *hv_vcpu;
 
 	entry = kvm_find_cpuid_entry(vcpu, HYPERV_CPUID_INTERFACE, 0);
 	if (entry && entry->eax == HYPERV_CPUID_SIGNATURE_EAX) {

arch/x86/kvm/mmu/mmu.c

@@ -2535,6 +2535,7 @@ static void kvm_unsync_page(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp)
 int mmu_try_to_unsync_pages(struct kvm_vcpu *vcpu, gfn_t gfn, bool can_unsync)
 {
 	struct kvm_mmu_page *sp;
+	bool locked = false;
 
 	/*
 	 * Force write-protection if the page is being tracked.  Note, the page
@@ -2557,9 +2558,34 @@ int mmu_try_to_unsync_pages(struct kvm_vcpu *vcpu, gfn_t gfn, bool can_unsync)
 		if (sp->unsync)
 			continue;
 
+		/*
+		 * TDP MMU page faults require an additional spinlock as they
+		 * run with mmu_lock held for read, not write, and the unsync
+		 * logic is not thread safe.  Take the spinklock regardless of
+		 * the MMU type to avoid extra conditionals/parameters, there's
+		 * no meaningful penalty if mmu_lock is held for write.
+		 */
+		if (!locked) {
+			locked = true;
+			spin_lock(&vcpu->kvm->arch.mmu_unsync_pages_lock);
+
+			/*
+			 * Recheck after taking the spinlock, a different vCPU
+			 * may have since marked the page unsync.  A false
+			 * positive on the unprotected check above is not
+			 * possible as clearing sp->unsync _must_ hold mmu_lock
+			 * for write, i.e. unsync cannot transition from 0->1
+			 * while this CPU holds mmu_lock for read (or write).
+			 */
+			if (READ_ONCE(sp->unsync))
+				continue;
+		}
+
 		WARN_ON(sp->role.level != PG_LEVEL_4K);
 		kvm_unsync_page(vcpu, sp);
 	}
+	if (locked)
+		spin_unlock(&vcpu->kvm->arch.mmu_unsync_pages_lock);
 
 	/*
 	 * We need to ensure that the marking of unsync pages is visible
@@ -5537,6 +5563,8 @@ void kvm_mmu_init_vm(struct kvm *kvm)
 {
 	struct kvm_page_track_notifier_node *node = &kvm->arch.mmu_sp_tracker;
 
+	spin_lock_init(&kvm->arch.mmu_unsync_pages_lock);
+
 	if (!kvm_mmu_init_tdp_mmu(kvm))
 		/*
 		 * No smp_load/store wrappers needed here as we are in

arch/x86/kvm/mmu/tdp_mmu.c

@@ -43,6 +43,7 @@ void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm)
 	if (!kvm->arch.tdp_mmu_enabled)
 		return;
 
+	WARN_ON(!list_empty(&kvm->arch.tdp_mmu_pages));
 	WARN_ON(!list_empty(&kvm->arch.tdp_mmu_roots));
 
 	/*
@@ -81,8 +82,6 @@ static void tdp_mmu_free_sp_rcu_callback(struct rcu_head *head)
 void kvm_tdp_mmu_put_root(struct kvm *kvm, struct kvm_mmu_page *root,
 			  bool shared)
 {
-	gfn_t max_gfn = 1ULL << (shadow_phys_bits - PAGE_SHIFT);
-
 	kvm_lockdep_assert_mmu_lock_held(kvm, shared);
 
 	if (!refcount_dec_and_test(&root->tdp_mmu_root_count))
@@ -94,7 +93,7 @@ void kvm_tdp_mmu_put_root(struct kvm *kvm, struct kvm_mmu_page *root,
 	list_del_rcu(&root->link);
 	spin_unlock(&kvm->arch.tdp_mmu_pages_lock);
 
-	zap_gfn_range(kvm, root, 0, max_gfn, false, false, shared);
+	zap_gfn_range(kvm, root, 0, -1ull, false, false, shared);
 
 	call_rcu(&root->rcu_head, tdp_mmu_free_sp_rcu_callback);
 }
@@ -724,13 +723,29 @@ static bool zap_gfn_range(struct kvm *kvm, struct kvm_mmu_page *root,
 			  gfn_t start, gfn_t end, bool can_yield, bool flush,
 			  bool shared)
 {
+	gfn_t max_gfn_host = 1ULL << (shadow_phys_bits - PAGE_SHIFT);
+	bool zap_all = (start == 0 && end >= max_gfn_host);
 	struct tdp_iter iter;
 
+	/*
+	 * No need to try to step down in the iterator when zapping all SPTEs,
+	 * zapping the top-level non-leaf SPTEs will recurse on their children.
+	 */
+	int min_level = zap_all ? root->role.level : PG_LEVEL_4K;
+
+	/*
+	 * Bound the walk at host.MAXPHYADDR, guest accesses beyond that will
+	 * hit a #PF(RSVD) and never get to an EPT Violation/Misconfig / #NPF,
+	 * and so KVM will never install a SPTE for such addresses.
+	 */
+	end = min(end, max_gfn_host);
+
 	kvm_lockdep_assert_mmu_lock_held(kvm, shared);
 
 	rcu_read_lock();
 
-	tdp_root_for_each_pte(iter, root, start, end) {
+	for_each_tdp_pte_min_level(iter, root->spt, root->role.level,
+				   min_level, start, end) {
 retry:
 		if (can_yield &&
 		    tdp_mmu_iter_cond_resched(kvm, &iter, flush, shared)) {
@@ -744,9 +759,10 @@ static bool zap_gfn_range(struct kvm *kvm, struct kvm_mmu_page *root,
 		/*
 		 * If this is a non-last-level SPTE that covers a larger range
 		 * than should be zapped, continue, and zap the mappings at a
-		 * lower level.
+		 * lower level, except when zapping all SPTEs.
 		 */
-		if ((iter.gfn < start ||
+		if (!zap_all &&
+		    (iter.gfn < start ||
 		     iter.gfn + KVM_PAGES_PER_HPAGE(iter.level) > end) &&
 		    !is_last_spte(iter.old_spte, iter.level))
 			continue;
@@ -794,12 +810,11 @@ bool __kvm_tdp_mmu_zap_gfn_range(struct kvm *kvm, int as_id, gfn_t start,
 
 void kvm_tdp_mmu_zap_all(struct kvm *kvm)
 {
-	gfn_t max_gfn = 1ULL << (shadow_phys_bits - PAGE_SHIFT);
 	bool flush = false;
 	int i;
 
 	for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++)
-		flush = kvm_tdp_mmu_zap_gfn_range(kvm, i, 0, max_gfn,
+		flush = kvm_tdp_mmu_zap_gfn_range(kvm, i, 0, -1ull,
 						  flush, false);
 
 	if (flush)
@@ -838,7 +853,6 @@ static struct kvm_mmu_page *next_invalidated_root(struct kvm *kvm,
  */
 void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm)
 {
-	gfn_t max_gfn = 1ULL << (shadow_phys_bits - PAGE_SHIFT);
 	struct kvm_mmu_page *next_root;
 	struct kvm_mmu_page *root;
 	bool flush = false;
@@ -854,8 +868,7 @@ void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm)
 
 		rcu_read_unlock();
 
-		flush = zap_gfn_range(kvm, root, 0, max_gfn, true, flush,
-				      true);
+		flush = zap_gfn_range(kvm, root, 0, -1ull, true, flush, true);
 
 		/*
 		 * Put the reference acquired in

arch/x86/kvm/vmx/nested.c

@@ -330,6 +330,31 @@ void nested_vmx_free_vcpu(struct kvm_vcpu *vcpu)
 	vcpu_put(vcpu);
 }
 
+#define EPTP_PA_MASK   GENMASK_ULL(51, 12)
+
+static bool nested_ept_root_matches(hpa_t root_hpa, u64 root_eptp, u64 eptp)
+{
+	return VALID_PAGE(root_hpa) &&
+	       ((root_eptp & EPTP_PA_MASK) == (eptp & EPTP_PA_MASK));
+}
+
+static void nested_ept_invalidate_addr(struct kvm_vcpu *vcpu, gpa_t eptp,
+				       gpa_t addr)
+{
+	uint i;
+	struct kvm_mmu_root_info *cached_root;
+
+	WARN_ON_ONCE(!mmu_is_nested(vcpu));
+
+	for (i = 0; i < KVM_MMU_NUM_PREV_ROOTS; i++) {
+		cached_root = &vcpu->arch.mmu->prev_roots[i];
+
+		if (nested_ept_root_matches(cached_root->hpa, cached_root->pgd,
+					    eptp))
+			vcpu->arch.mmu->invlpg(vcpu, addr, cached_root->hpa);
+	}
+}
+
 static void nested_ept_inject_page_fault(struct kvm_vcpu *vcpu,
 		struct x86_exception *fault)
 {
@@ -342,11 +367,23 @@ static void nested_ept_inject_page_fault(struct kvm_vcpu *vcpu,
 		vm_exit_reason = EXIT_REASON_PML_FULL;
 		vmx->nested.pml_full = false;
 		exit_qualification &= INTR_INFO_UNBLOCK_NMI;
-	} else if (fault->error_code & PFERR_RSVD_MASK)
+	} else {
+		if (fault->error_code & PFERR_RSVD_MASK)
 			vm_exit_reason = EXIT_REASON_EPT_MISCONFIG;
 		else
 			vm_exit_reason = EXIT_REASON_EPT_VIOLATION;
 
+		/*
+		 * Although the caller (kvm_inject_emulated_page_fault) would
+		 * have already synced the faulting address in the shadow EPT
+		 * tables for the current EPTP12, we also need to sync it for
+		 * any other cached EPTP02s based on the same EP4TA, since the
+		 * TLB associates mappings to the EP4TA rather than the full EPTP.
+		 */
+		nested_ept_invalidate_addr(vcpu, vmcs12->ept_pointer,
+					   fault->address);
+	}
+
 	nested_vmx_vmexit(vcpu, vm_exit_reason, 0, exit_qualification);
 	vmcs12->guest_physical_address = fault->address;
 }
@@ -5325,14 +5362,6 @@ static int handle_vmptrst(struct kvm_vcpu *vcpu)
 	return nested_vmx_succeed(vcpu);
 }
 
-#define EPTP_PA_MASK   GENMASK_ULL(51, 12)
-
-static bool nested_ept_root_matches(hpa_t root_hpa, u64 root_eptp, u64 eptp)
-{
-	return VALID_PAGE(root_hpa) &&
-		((root_eptp & EPTP_PA_MASK) == (eptp & EPTP_PA_MASK));
-}
-
 /* Emulate the INVEPT instruction */
 static int handle_invept(struct kvm_vcpu *vcpu)
 {
@@ -5826,7 +5855,8 @@ static bool nested_vmx_l0_wants_exit(struct kvm_vcpu *vcpu,
 		if (is_nmi(intr_info))
 			return true;
 		else if (is_page_fault(intr_info))
-			return vcpu->arch.apf.host_apf_flags || !enable_ept;
+			return vcpu->arch.apf.host_apf_flags ||
+			       vmx_need_pf_intercept(vcpu);
 		else if (is_debug(intr_info) &&
 			 vcpu->guest_debug &
 			 (KVM_GUESTDBG_SINGLESTEP | KVM_GUESTDBG_USE_HW_BP))

arch/x86/kvm/vmx/vmx.h

@@ -522,7 +522,7 @@ static inline struct vmcs *alloc_vmcs(bool shadow)
 
 static inline bool vmx_has_waitpkg(struct vcpu_vmx *vmx)
 {
-	return vmx->secondary_exec_control &
+	return secondary_exec_controls_get(vmx) &
 		SECONDARY_EXEC_ENABLE_USR_WAIT_PAUSE;
 }