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

Pull kvm fixes from Paolo Bonzini:
 "s390:

   - PCI interpretation compile fixes

  RISC-V:

   - fix unused variable warnings in vcpu_timer.c

   - move extern sbi_ext declarations to a header

  x86:

   - check validity of argument to KVM_SET_MP_STATE

   - use guest's global_ctrl to completely disable guest PEBS

   - fix a memory leak on memory allocation failure

   - mask off unsupported and unknown bits of IA32_ARCH_CAPABILITIES

   - fix build failure with Clang integrated assembler

   - fix MSR interception

   - always flush TLBs when enabling dirty logging"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: x86: check validity of argument to KVM_SET_MP_STATE
  perf/x86/core: Completely disable guest PEBS via guest's global_ctrl
  KVM: x86: fix memoryleak in kvm_arch_vcpu_create()
  KVM: x86: Mask off unsupported and unknown bits of IA32_ARCH_CAPABILITIES
  KVM: s390: pci: Hook to access KVM lowlevel from VFIO
  riscv: kvm: move extern sbi_ext declarations to a header
  riscv: kvm: vcpu_timer: fix unused variable warnings
  KVM: selftests: Fix ambiguous mov in KVM_ASM_SAFE()
  KVM: selftests: Fix KVM_EXCEPTION_MAGIC build with Clang
  KVM: VMX: Heed the 'msr' argument in msr_write_intercepted()
  kvm: x86: mmu: Always flush TLBs when enabling dirty logging
  kvm: x86: mmu: Drop the need_remote_flush() function

Author: torvalds

arch/riscv/include/asm/kvm_vcpu_sbi.h

@@ -33,4 +33,16 @@ void kvm_riscv_vcpu_sbi_system_reset(struct kvm_vcpu *vcpu,
 				     u32 type, u64 flags);
 const struct kvm_vcpu_sbi_extension *kvm_vcpu_sbi_find_ext(unsigned long extid);
 
+#ifdef CONFIG_RISCV_SBI_V01
+extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_v01;
+#endif
+extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_base;
+extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_time;
+extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_ipi;
+extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_rfence;
+extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_srst;
+extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_hsm;
+extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_experimental;
+extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_vendor;
+
 #endif /* __RISCV_KVM_VCPU_SBI_H__ */

arch/riscv/kvm/vcpu_sbi.c

@@ -32,23 +32,13 @@ static int kvm_linux_err_map_sbi(int err)
 	};
 }
 
-#ifdef CONFIG_RISCV_SBI_V01
-extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_v01;
-#else
+#ifndef CONFIG_RISCV_SBI_V01
 static const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_v01 = {
 	.extid_start = -1UL,
 	.extid_end = -1UL,
 	.handler = NULL,
 };
 #endif
-extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_base;
-extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_time;
-extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_ipi;
-extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_rfence;
-extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_srst;
-extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_hsm;
-extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_experimental;
-extern const struct kvm_vcpu_sbi_extension vcpu_sbi_ext_vendor;
 
 static const struct kvm_vcpu_sbi_extension *sbi_ext[] = {
 	&vcpu_sbi_ext_v01,

arch/riscv/kvm/vcpu_timer.c

@@ -299,15 +299,13 @@ static void kvm_riscv_vcpu_update_timedelta(struct kvm_vcpu *vcpu)
 
 void kvm_riscv_vcpu_timer_restore(struct kvm_vcpu *vcpu)
 {
-	struct kvm_vcpu_csr *csr;
 	struct kvm_vcpu_timer *t = &vcpu->arch.timer;
 
 	kvm_riscv_vcpu_update_timedelta(vcpu);
 
 	if (!t->sstc_enabled)
 		return;
 
-	csr = &vcpu->arch.guest_csr;
 #if defined(CONFIG_32BIT)
 	csr_write(CSR_VSTIMECMP, (u32)t->next_cycles);
 	csr_write(CSR_VSTIMECMPH, (u32)(t->next_cycles >> 32));
@@ -324,13 +322,11 @@ void kvm_riscv_vcpu_timer_restore(struct kvm_vcpu *vcpu)
 
 void kvm_riscv_vcpu_timer_save(struct kvm_vcpu *vcpu)
 {
-	struct kvm_vcpu_csr *csr;
 	struct kvm_vcpu_timer *t = &vcpu->arch.timer;
 
 	if (!t->sstc_enabled)
 		return;
 
-	csr = &vcpu->arch.guest_csr;
 	t = &vcpu->arch.timer;
 #if defined(CONFIG_32BIT)
 	t->next_cycles = csr_read(CSR_VSTIMECMP);

arch/s390/include/asm/kvm_host.h

@@ -1038,16 +1038,11 @@ static inline void kvm_arch_vcpu_unblocking(struct kvm_vcpu *vcpu) {}
 #define __KVM_HAVE_ARCH_VM_FREE
 void kvm_arch_free_vm(struct kvm *kvm);
 
-#ifdef CONFIG_VFIO_PCI_ZDEV_KVM
-int kvm_s390_pci_register_kvm(struct zpci_dev *zdev, struct kvm *kvm);
-void kvm_s390_pci_unregister_kvm(struct zpci_dev *zdev);
-#else
-static inline int kvm_s390_pci_register_kvm(struct zpci_dev *dev,
-					    struct kvm *kvm)
-{
-	return -EPERM;
-}
-static inline void kvm_s390_pci_unregister_kvm(struct zpci_dev *dev) {}
-#endif
+struct zpci_kvm_hook {
+	int (*kvm_register)(void *opaque, struct kvm *kvm);
+	void (*kvm_unregister)(void *opaque);
+};
+
+extern struct zpci_kvm_hook zpci_kvm_hook;
 
 #endif

arch/s390/kvm/pci.c

@@ -431,8 +431,9 @@ static void kvm_s390_pci_dev_release(struct zpci_dev *zdev)
  * available, enable them and let userspace indicate whether or not they will
  * be used (specify SHM bit to disable).
  */
-int kvm_s390_pci_register_kvm(struct zpci_dev *zdev, struct kvm *kvm)
+static int kvm_s390_pci_register_kvm(void *opaque, struct kvm *kvm)
 {
+	struct zpci_dev *zdev = opaque;
 	int rc;
 
 	if (!zdev)
@@ -510,10 +511,10 @@ int kvm_s390_pci_register_kvm(struct zpci_dev *zdev, struct kvm *kvm)
 	kvm_put_kvm(kvm);
 	return rc;
 }
-EXPORT_SYMBOL_GPL(kvm_s390_pci_register_kvm);
 
-void kvm_s390_pci_unregister_kvm(struct zpci_dev *zdev)
+static void kvm_s390_pci_unregister_kvm(void *opaque)
 {
+	struct zpci_dev *zdev = opaque;
 	struct kvm *kvm;
 
 	if (!zdev)
@@ -566,7 +567,6 @@ void kvm_s390_pci_unregister_kvm(struct zpci_dev *zdev)
 
 	kvm_put_kvm(kvm);
 }
-EXPORT_SYMBOL_GPL(kvm_s390_pci_unregister_kvm);
 
 void kvm_s390_pci_init_list(struct kvm *kvm)
 {
@@ -678,13 +678,17 @@ int kvm_s390_pci_init(void)
 
 	spin_lock_init(&aift->gait_lock);
 	mutex_init(&aift->aift_lock);
+	zpci_kvm_hook.kvm_register = kvm_s390_pci_register_kvm;
+	zpci_kvm_hook.kvm_unregister = kvm_s390_pci_unregister_kvm;
 
 	return 0;
 }
 
 void kvm_s390_pci_exit(void)
 {
 	mutex_destroy(&aift->aift_lock);
+	zpci_kvm_hook.kvm_register = NULL;
+	zpci_kvm_hook.kvm_unregister = NULL;
 
 	kfree(aift);
 }

arch/s390/pci/Makefile

@@ -5,5 +5,5 @@
 
 obj-$(CONFIG_PCI)	+= pci.o pci_irq.o pci_dma.o pci_clp.o pci_sysfs.o \
 			   pci_event.o pci_debug.o pci_insn.o pci_mmio.o \
-			   pci_bus.o
+			   pci_bus.o pci_kvm_hook.o
 obj-$(CONFIG_PCI_IOV)	+= pci_iov.o

arch/s390/pci/pci_kvm_hook.c

@@ -0,0 +1,11 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * VFIO ZPCI devices support
+ *
+ * Copyright (C) IBM Corp. 2022.  All rights reserved.
+ *	Author(s): Pierre Morel <[email protected]>
+ */
+#include <linux/kvm_host.h>
+
+struct zpci_kvm_hook zpci_kvm_hook;
+EXPORT_SYMBOL_GPL(zpci_kvm_hook);

arch/x86/events/intel/core.c

@@ -4052,8 +4052,9 @@ static struct perf_guest_switch_msr *intel_guest_get_msrs(int *nr, void *data)
 		/* Disable guest PEBS if host PEBS is enabled. */
 		arr[pebs_enable].guest = 0;
 	} else {
-		/* Disable guest PEBS for cross-mapped PEBS counters. */
+		/* Disable guest PEBS thoroughly for cross-mapped PEBS counters. */
 		arr[pebs_enable].guest &= ~kvm_pmu->host_cross_mapped_mask;
+		arr[global_ctrl].guest &= ~kvm_pmu->host_cross_mapped_mask;
 		/* Set hw GLOBAL_CTRL bits for PEBS counter when it runs for guest */
 		arr[global_ctrl].guest |= arr[pebs_enable].guest;
 	}

arch/x86/kvm/mmu/mmu.c

@@ -5361,19 +5361,6 @@ void kvm_mmu_free_obsolete_roots(struct kvm_vcpu *vcpu)
 	__kvm_mmu_free_obsolete_roots(vcpu->kvm, &vcpu->arch.guest_mmu);
 }
 
-static bool need_remote_flush(u64 old, u64 new)
-{
-	if (!is_shadow_present_pte(old))
-		return false;
-	if (!is_shadow_present_pte(new))
-		return true;
-	if ((old ^ new) & SPTE_BASE_ADDR_MASK)
-		return true;
-	old ^= shadow_nx_mask;
-	new ^= shadow_nx_mask;
-	return (old & ~new & SPTE_PERM_MASK) != 0;
-}
-
 static u64 mmu_pte_write_fetch_gpte(struct kvm_vcpu *vcpu, gpa_t *gpa,
 				    int *bytes)
 {
@@ -5519,7 +5506,7 @@ static void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
 			mmu_page_zap_pte(vcpu->kvm, sp, spte, NULL);
 			if (gentry && sp->role.level != PG_LEVEL_4K)
 				++vcpu->kvm->stat.mmu_pde_zapped;
-			if (need_remote_flush(entry, *spte))
+			if (is_shadow_present_pte(entry))
 				flush = true;
 			++spte;
 		}
@@ -6085,47 +6072,18 @@ void kvm_mmu_slot_remove_write_access(struct kvm *kvm,
 				      const struct kvm_memory_slot *memslot,
 				      int start_level)
 {
-	bool flush = false;
-
 	if (kvm_memslots_have_rmaps(kvm)) {
 		write_lock(&kvm->mmu_lock);
-		flush = slot_handle_level(kvm, memslot, slot_rmap_write_protect,
-					  start_level, KVM_MAX_HUGEPAGE_LEVEL,
-					  false);
+		slot_handle_level(kvm, memslot, slot_rmap_write_protect,
+				  start_level, KVM_MAX_HUGEPAGE_LEVEL, false);
 		write_unlock(&kvm->mmu_lock);
 	}
 
 	if (is_tdp_mmu_enabled(kvm)) {
 		read_lock(&kvm->mmu_lock);
-		flush |= kvm_tdp_mmu_wrprot_slot(kvm, memslot, start_level);
+		kvm_tdp_mmu_wrprot_slot(kvm, memslot, start_level);
 		read_unlock(&kvm->mmu_lock);
 	}
-
-	/*
-	 * Flush TLBs if any SPTEs had to be write-protected to ensure that
-	 * guest writes are reflected in the dirty bitmap before the memslot
-	 * update completes, i.e. before enabling dirty logging is visible to
-	 * userspace.
-	 *
-	 * Perform the TLB flush outside the mmu_lock to reduce the amount of
-	 * time the lock is held. However, this does mean that another CPU can
-	 * now grab mmu_lock and encounter a write-protected SPTE while CPUs
-	 * still have a writable mapping for the associated GFN in their TLB.
-	 *
-	 * This is safe but requires KVM to be careful when making decisions
-	 * based on the write-protection status of an SPTE. Specifically, KVM
-	 * also write-protects SPTEs to monitor changes to guest page tables
-	 * during shadow paging, and must guarantee no CPUs can write to those
-	 * page before the lock is dropped. As mentioned in the previous
-	 * paragraph, a write-protected SPTE is no guarantee that CPU cannot
-	 * perform writes. So to determine if a TLB flush is truly required, KVM
-	 * will clear a separate software-only bit (MMU-writable) and skip the
-	 * flush if-and-only-if this bit was already clear.
-	 *
-	 * See is_writable_pte() for more details.
-	 */
-	if (flush)
-		kvm_arch_flush_remote_tlbs_memslot(kvm, memslot);
 }
 
 static inline bool need_topup(struct kvm_mmu_memory_cache *cache, int min)
@@ -6493,32 +6451,30 @@ void kvm_arch_flush_remote_tlbs_memslot(struct kvm *kvm,
 void kvm_mmu_slot_leaf_clear_dirty(struct kvm *kvm,
 				   const struct kvm_memory_slot *memslot)
 {
-	bool flush = false;
-
 	if (kvm_memslots_have_rmaps(kvm)) {
 		write_lock(&kvm->mmu_lock);
 		/*
 		 * Clear dirty bits only on 4k SPTEs since the legacy MMU only
 		 * support dirty logging at a 4k granularity.
 		 */
-		flush = slot_handle_level_4k(kvm, memslot, __rmap_clear_dirty, false);
+		slot_handle_level_4k(kvm, memslot, __rmap_clear_dirty, false);
 		write_unlock(&kvm->mmu_lock);
 	}
 
 	if (is_tdp_mmu_enabled(kvm)) {
 		read_lock(&kvm->mmu_lock);
-		flush |= kvm_tdp_mmu_clear_dirty_slot(kvm, memslot);
+		kvm_tdp_mmu_clear_dirty_slot(kvm, memslot);
 		read_unlock(&kvm->mmu_lock);
 	}
 
 	/*
+	 * The caller will flush the TLBs after this function returns.
+	 *
 	 * It's also safe to flush TLBs out of mmu lock here as currently this
 	 * function is only used for dirty logging, in which case flushing TLB
 	 * out of mmu lock also guarantees no dirty pages will be lost in
 	 * dirty_bitmap.
 	 */
-	if (flush)
-		kvm_arch_flush_remote_tlbs_memslot(kvm, memslot);
 }
 
 void kvm_mmu_zap_all(struct kvm *kvm)

arch/x86/kvm/mmu/spte.h

@@ -343,7 +343,7 @@ static __always_inline bool is_rsvd_spte(struct rsvd_bits_validate *rsvd_check,
 }
 
 /*
- * An shadow-present leaf SPTE may be non-writable for 3 possible reasons:
+ * A shadow-present leaf SPTE may be non-writable for 4 possible reasons:
  *
  *  1. To intercept writes for dirty logging. KVM write-protects huge pages
  *     so that they can be split be split down into the dirty logging
@@ -361,8 +361,13 @@ static __always_inline bool is_rsvd_spte(struct rsvd_bits_validate *rsvd_check,
  *     read-only memslot or guest memory backed by a read-only VMA. Writes to
  *     such pages are disallowed entirely.
  *
- * To keep track of why a given SPTE is write-protected, KVM uses 2
- * software-only bits in the SPTE:
+ *  4. To emulate the Accessed bit for SPTEs without A/D bits.  Note, in this
+ *     case, the SPTE is access-protected, not just write-protected!
+ *
+ * For cases #1 and #4, KVM can safely make such SPTEs writable without taking
+ * mmu_lock as capturing the Accessed/Dirty state doesn't require taking it.
+ * To differentiate #1 and #4 from #2 and #3, KVM uses two software-only bits
+ * in the SPTE:
  *
  *  shadow_mmu_writable_mask, aka MMU-writable -
  *    Cleared on SPTEs that KVM is currently write-protecting for shadow paging
@@ -391,7 +396,8 @@ static __always_inline bool is_rsvd_spte(struct rsvd_bits_validate *rsvd_check,
  * shadow page tables between vCPUs. Write-protecting an SPTE for dirty logging
  * (which does not clear the MMU-writable bit), does not flush TLBs before
  * dropping the lock, as it only needs to synchronize guest writes with the
- * dirty bitmap.
+ * dirty bitmap. Similarly, making the SPTE inaccessible (and non-writable) for
+ * access-tracking via the clear_young() MMU notifier also does not flush TLBs.
  *
  * So, there is the problem: clearing the MMU-writable bit can encounter a
  * write-protected SPTE while CPUs still have writable mappings for that SPTE