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

Pull kvm fixes from Paolo Bonzini:
 "ARM:

   - Fix MTE shared page detection

   - Enable selftest's use of PMU registers when asked to

  s390:

   - restore 5.13 debugfs names

  x86:

   - fix sizes for vcpu-id indexed arrays

   - fixes for AMD virtualized LAPIC (AVIC)

   - other small bugfixes

  Generic:

   - access tracking performance test

   - dirty_log_perf_test command line parsing fix

   - Fix selftest use of obsolete pthread_yield() in favour of
     sched_yield()

   - use cpu_relax when halt polling

   - fixed missing KVM_CLEAR_DIRTY_LOG compat ioctl"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: add missing compat KVM_CLEAR_DIRTY_LOG
  KVM: use cpu_relax when halt polling
  KVM: SVM: use vmcb01 in svm_refresh_apicv_exec_ctrl
  KVM: SVM: tweak warning about enabled AVIC on nested entry
  KVM: SVM: svm_set_vintr don't warn if AVIC is active but is about to be deactivated
  KVM: s390: restore old debugfs names
  KVM: SVM: delay svm_vcpu_init_msrpm after svm->vmcb is initialized
  KVM: selftests: Introduce access_tracking_perf_test
  KVM: selftests: Fix missing break in dirty_log_perf_test arg parsing
  x86/kvm: fix vcpu-id indexed array sizes
  KVM: x86: Check the right feature bit for MSR_KVM_ASYNC_PF_ACK access
  docs: virt: kvm: api.rst: replace some characters
  KVM: Documentation: Fix KVM_CAP_ENFORCE_PV_FEATURE_CPUID name
  KVM: nSVM: Swap the parameter order for svm_copy_vmrun_state()/svm_copy_vmloadsave_state()
  KVM: nSVM: Rename nested_svm_vmloadsave() to svm_copy_vmloadsave_state()
  KVM: arm64: selftests: get-reg-list: actually enable pmu regs in pmu sublist
  KVM: selftests: change pthread_yield to sched_yield
  KVM: arm64: Fix detection of shared VMAs on guest fault

Documentation/virt/kvm/api.rst

@@ -855,7 +855,7 @@ in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
 use PPIs designated for specific cpus.  The irq field is interpreted
 like this::
 
-  bits:  |  31 ... 28  | 27 ... 24 | 23  ... 16 | 15 ... 0 |
+  bits:  |  31 ... 28  | 27 ... 24 | 23  ... 16 | 15 ... 0 |
   field: | vcpu2_index | irq_type  | vcpu_index |  irq_id  |
 
 The irq_type field has the following values:
@@ -2149,10 +2149,10 @@ prior to calling the KVM_RUN ioctl.
 Errors:
 
   ======   ============================================================
-  ENOENT   no such register
-  EINVAL   invalid register ID, or no such register or used with VMs in
+  ENOENT   no such register
+  EINVAL   invalid register ID, or no such register or used with VMs in
            protected virtualization mode on s390
-  EPERM    (arm64) register access not allowed before vcpu finalization
+  EPERM    (arm64) register access not allowed before vcpu finalization
   ======   ============================================================
 
 (These error codes are indicative only: do not rely on a specific error
@@ -2590,10 +2590,10 @@ following id bit patterns::
 Errors include:
 
   ======== ============================================================
-  ENOENT   no such register
-  EINVAL   invalid register ID, or no such register or used with VMs in
+  ENOENT   no such register
+  EINVAL   invalid register ID, or no such register or used with VMs in
            protected virtualization mode on s390
-  EPERM    (arm64) register access not allowed before vcpu finalization
+  EPERM    (arm64) register access not allowed before vcpu finalization
   ======== ============================================================
 
 (These error codes are indicative only: do not rely on a specific error
@@ -3112,13 +3112,13 @@ current state.  "addr" is ignored.
 Errors:
 
   ======     =================================================================
-  EINVAL     the target is unknown, or the combination of features is invalid.
-  ENOENT     a features bit specified is unknown.
+  EINVAL     the target is unknown, or the combination of features is invalid.
+  ENOENT     a features bit specified is unknown.
   ======     =================================================================
 
 This tells KVM what type of CPU to present to the guest, and what
-optional features it should have.  This will cause a reset of the cpu
-registers to their initial values.  If this is not called, KVM_RUN will
+optional features it should have.  This will cause a reset of the cpu
+registers to their initial values.  If this is not called, KVM_RUN will
 return ENOEXEC for that vcpu.
 
 The initial values are defined as:
@@ -3239,8 +3239,8 @@ VCPU matching underlying host.
 Errors:
 
   =====      ==============================================================
-  E2BIG      the reg index list is too big to fit in the array specified by
-             the user (the number required will be written into n).
+  E2BIG      the reg index list is too big to fit in the array specified by
+             the user (the number required will be written into n).
   =====      ==============================================================
 
 ::
@@ -3288,7 +3288,7 @@ specific device.
 ARM/arm64 divides the id field into two parts, a device id and an
 address type id specific to the individual device::
 
-  bits:  | 63        ...       32 | 31    ...    16 | 15    ...    0 |
+  bits:  | 63        ...       32 | 31    ...    16 | 15    ...    0 |
   field: |        0x00000000      |     device id   |  addr type id  |
 
 ARM/arm64 currently only require this when using the in-kernel GIC
@@ -7049,7 +7049,7 @@ In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to
 trap and emulate MSRs that are outside of the scope of KVM as well as
 limit the attack surface on KVM's MSR emulation code.
 
-8.28 KVM_CAP_ENFORCE_PV_CPUID
+8.28 KVM_CAP_ENFORCE_PV_FEATURE_CPUID
 -----------------------------
 
 Architectures: x86

arch/arm64/kvm/mmu.c

@@ -947,7 +947,7 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
 		vma_shift = get_vma_page_shift(vma, hva);
 	}
 
-	shared = (vma->vm_flags & VM_PFNMAP);
+	shared = (vma->vm_flags & VM_SHARED);
 
 	switch (vma_shift) {
 #ifndef __PAGETABLE_PMD_FOLDED

arch/s390/include/asm/kvm_host.h

@@ -445,15 +445,15 @@ struct kvm_vcpu_stat {
 	u64 instruction_sigp_init_cpu_reset;
 	u64 instruction_sigp_cpu_reset;
 	u64 instruction_sigp_unknown;
-	u64 diagnose_10;
-	u64 diagnose_44;
-	u64 diagnose_9c;
-	u64 diagnose_9c_ignored;
-	u64 diagnose_9c_forward;
-	u64 diagnose_258;
-	u64 diagnose_308;
-	u64 diagnose_500;
-	u64 diagnose_other;
+	u64 instruction_diagnose_10;
+	u64 instruction_diagnose_44;
+	u64 instruction_diagnose_9c;
+	u64 diag_9c_ignored;
+	u64 diag_9c_forward;
+	u64 instruction_diagnose_258;
+	u64 instruction_diagnose_308;
+	u64 instruction_diagnose_500;
+	u64 instruction_diagnose_other;
 	u64 pfault_sync;
 };
 

arch/s390/kvm/diag.c

@@ -24,7 +24,7 @@ static int diag_release_pages(struct kvm_vcpu *vcpu)
 
 	start = vcpu->run->s.regs.gprs[(vcpu->arch.sie_block->ipa & 0xf0) >> 4];
 	end = vcpu->run->s.regs.gprs[vcpu->arch.sie_block->ipa & 0xf] + PAGE_SIZE;
-	vcpu->stat.diagnose_10++;
+	vcpu->stat.instruction_diagnose_10++;
 
 	if (start & ~PAGE_MASK || end & ~PAGE_MASK || start >= end
 	    || start < 2 * PAGE_SIZE)
@@ -74,7 +74,7 @@ static int __diag_page_ref_service(struct kvm_vcpu *vcpu)
 
 	VCPU_EVENT(vcpu, 3, "diag page reference parameter block at 0x%llx",
 		   vcpu->run->s.regs.gprs[rx]);
-	vcpu->stat.diagnose_258++;
+	vcpu->stat.instruction_diagnose_258++;
 	if (vcpu->run->s.regs.gprs[rx] & 7)
 		return kvm_s390_inject_program_int(vcpu, PGM_SPECIFICATION);
 	rc = read_guest(vcpu, vcpu->run->s.regs.gprs[rx], rx, &parm, sizeof(parm));
@@ -145,7 +145,7 @@ static int __diag_page_ref_service(struct kvm_vcpu *vcpu)
 static int __diag_time_slice_end(struct kvm_vcpu *vcpu)
 {
 	VCPU_EVENT(vcpu, 5, "%s", "diag time slice end");
-	vcpu->stat.diagnose_44++;
+	vcpu->stat.instruction_diagnose_44++;
 	kvm_vcpu_on_spin(vcpu, true);
 	return 0;
 }
@@ -169,7 +169,7 @@ static int __diag_time_slice_end_directed(struct kvm_vcpu *vcpu)
 	int tid;
 
 	tid = vcpu->run->s.regs.gprs[(vcpu->arch.sie_block->ipa & 0xf0) >> 4];
-	vcpu->stat.diagnose_9c++;
+	vcpu->stat.instruction_diagnose_9c++;
 
 	/* yield to self */
 	if (tid == vcpu->vcpu_id)
@@ -192,7 +192,7 @@ static int __diag_time_slice_end_directed(struct kvm_vcpu *vcpu)
 		VCPU_EVENT(vcpu, 5,
 			   "diag time slice end directed to %d: yield forwarded",
 			   tid);
-		vcpu->stat.diagnose_9c_forward++;
+		vcpu->stat.diag_9c_forward++;
 		return 0;
 	}
 
@@ -203,7 +203,7 @@ static int __diag_time_slice_end_directed(struct kvm_vcpu *vcpu)
 	return 0;
 no_yield:
 	VCPU_EVENT(vcpu, 5, "diag time slice end directed to %d: ignored", tid);
-	vcpu->stat.diagnose_9c_ignored++;
+	vcpu->stat.diag_9c_ignored++;
 	return 0;
 }
 
@@ -213,7 +213,7 @@ static int __diag_ipl_functions(struct kvm_vcpu *vcpu)
 	unsigned long subcode = vcpu->run->s.regs.gprs[reg] & 0xffff;
 
 	VCPU_EVENT(vcpu, 3, "diag ipl functions, subcode %lx", subcode);
-	vcpu->stat.diagnose_308++;
+	vcpu->stat.instruction_diagnose_308++;
 	switch (subcode) {
 	case 3:
 		vcpu->run->s390_reset_flags = KVM_S390_RESET_CLEAR;
@@ -245,7 +245,7 @@ static int __diag_virtio_hypercall(struct kvm_vcpu *vcpu)
 {
 	int ret;
 
-	vcpu->stat.diagnose_500++;
+	vcpu->stat.instruction_diagnose_500++;
 	/* No virtio-ccw notification? Get out quickly. */
 	if (!vcpu->kvm->arch.css_support ||
 	    (vcpu->run->s.regs.gprs[1] != KVM_S390_VIRTIO_CCW_NOTIFY))
@@ -299,7 +299,7 @@ int kvm_s390_handle_diag(struct kvm_vcpu *vcpu)
 	case 0x500:
 		return __diag_virtio_hypercall(vcpu);
 	default:
-		vcpu->stat.diagnose_other++;
+		vcpu->stat.instruction_diagnose_other++;
 		return -EOPNOTSUPP;
 	}
 }

arch/s390/kvm/kvm-s390.c

@@ -163,15 +163,15 @@ const struct _kvm_stats_desc kvm_vcpu_stats_desc[] = {
 	STATS_DESC_COUNTER(VCPU, instruction_sigp_init_cpu_reset),
 	STATS_DESC_COUNTER(VCPU, instruction_sigp_cpu_reset),
 	STATS_DESC_COUNTER(VCPU, instruction_sigp_unknown),
-	STATS_DESC_COUNTER(VCPU, diagnose_10),
-	STATS_DESC_COUNTER(VCPU, diagnose_44),
-	STATS_DESC_COUNTER(VCPU, diagnose_9c),
-	STATS_DESC_COUNTER(VCPU, diagnose_9c_ignored),
-	STATS_DESC_COUNTER(VCPU, diagnose_9c_forward),
-	STATS_DESC_COUNTER(VCPU, diagnose_258),
-	STATS_DESC_COUNTER(VCPU, diagnose_308),
-	STATS_DESC_COUNTER(VCPU, diagnose_500),
-	STATS_DESC_COUNTER(VCPU, diagnose_other),
+	STATS_DESC_COUNTER(VCPU, instruction_diagnose_10),
+	STATS_DESC_COUNTER(VCPU, instruction_diagnose_44),
+	STATS_DESC_COUNTER(VCPU, instruction_diagnose_9c),
+	STATS_DESC_COUNTER(VCPU, diag_9c_ignored),
+	STATS_DESC_COUNTER(VCPU, diag_9c_forward),
+	STATS_DESC_COUNTER(VCPU, instruction_diagnose_258),
+	STATS_DESC_COUNTER(VCPU, instruction_diagnose_308),
+	STATS_DESC_COUNTER(VCPU, instruction_diagnose_500),
+	STATS_DESC_COUNTER(VCPU, instruction_diagnose_other),
 	STATS_DESC_COUNTER(VCPU, pfault_sync)
 };
 static_assert(ARRAY_SIZE(kvm_vcpu_stats_desc) ==

arch/x86/kvm/ioapic.c

@@ -96,7 +96,7 @@ static unsigned long ioapic_read_indirect(struct kvm_ioapic *ioapic,
 static void rtc_irq_eoi_tracking_reset(struct kvm_ioapic *ioapic)
 {
 	ioapic->rtc_status.pending_eoi = 0;
-	bitmap_zero(ioapic->rtc_status.dest_map.map, KVM_MAX_VCPU_ID);
+	bitmap_zero(ioapic->rtc_status.dest_map.map, KVM_MAX_VCPU_ID + 1);
 }
 
 static void kvm_rtc_eoi_tracking_restore_all(struct kvm_ioapic *ioapic);

arch/x86/kvm/ioapic.h

@@ -43,13 +43,13 @@ struct kvm_vcpu;
 
 struct dest_map {
 	/* vcpu bitmap where IRQ has been sent */
-	DECLARE_BITMAP(map, KVM_MAX_VCPU_ID);
+	DECLARE_BITMAP(map, KVM_MAX_VCPU_ID + 1);
 
 	/*
 	 * Vector sent to a given vcpu, only valid when
 	 * the vcpu's bit in map is set
 	 */
-	u8 vectors[KVM_MAX_VCPU_ID];
+	u8 vectors[KVM_MAX_VCPU_ID + 1];
 };
 
 

arch/x86/kvm/svm/avic.c

@@ -646,7 +646,7 @@ static int svm_set_pi_irte_mode(struct kvm_vcpu *vcpu, bool activate)
 void svm_refresh_apicv_exec_ctrl(struct kvm_vcpu *vcpu)
 {
 	struct vcpu_svm *svm = to_svm(vcpu);
-	struct vmcb *vmcb = svm->vmcb;
+	struct vmcb *vmcb = svm->vmcb01.ptr;
 	bool activated = kvm_vcpu_apicv_active(vcpu);
 
 	if (!enable_apicv)

arch/x86/kvm/svm/nested.c

@@ -515,7 +515,7 @@ static void nested_vmcb02_prepare_control(struct vcpu_svm *svm)
 	 * Also covers avic_vapic_bar, avic_backing_page, avic_logical_id,
 	 * avic_physical_id.
 	 */
-	WARN_ON(svm->vmcb01.ptr->control.int_ctl & AVIC_ENABLE_MASK);
+	WARN_ON(kvm_apicv_activated(svm->vcpu.kvm));
 
 	/* Copied from vmcb01.  msrpm_base can be overwritten later.  */
 	svm->vmcb->control.nested_ctl = svm->vmcb01.ptr->control.nested_ctl;
@@ -702,8 +702,8 @@ int nested_svm_vmrun(struct kvm_vcpu *vcpu)
 }
 
 /* Copy state save area fields which are handled by VMRUN */
-void svm_copy_vmrun_state(struct vmcb_save_area *from_save,
-			  struct vmcb_save_area *to_save)
+void svm_copy_vmrun_state(struct vmcb_save_area *to_save,
+			  struct vmcb_save_area *from_save)
 {
 	to_save->es = from_save->es;
 	to_save->cs = from_save->cs;
@@ -722,7 +722,7 @@ void svm_copy_vmrun_state(struct vmcb_save_area *from_save,
 	to_save->cpl = 0;
 }
 
-void nested_svm_vmloadsave(struct vmcb *from_vmcb, struct vmcb *to_vmcb)
+void svm_copy_vmloadsave_state(struct vmcb *to_vmcb, struct vmcb *from_vmcb)
 {
 	to_vmcb->save.fs = from_vmcb->save.fs;
 	to_vmcb->save.gs = from_vmcb->save.gs;
@@ -1385,7 +1385,7 @@ static int svm_set_nested_state(struct kvm_vcpu *vcpu,
 
 	svm->nested.vmcb12_gpa = kvm_state->hdr.svm.vmcb_pa;
 
-	svm_copy_vmrun_state(save, &svm->vmcb01.ptr->save);
+	svm_copy_vmrun_state(&svm->vmcb01.ptr->save, save);
 	nested_load_control_from_vmcb12(svm, ctl);
 
 	svm_switch_vmcb(svm, &svm->nested.vmcb02);

arch/x86/kvm/svm/svm.c

@@ -1406,8 +1406,6 @@ static int svm_create_vcpu(struct kvm_vcpu *vcpu)
 		goto error_free_vmsa_page;
 	}
 
-	svm_vcpu_init_msrpm(vcpu, svm->msrpm);
-
 	svm->vmcb01.ptr = page_address(vmcb01_page);
 	svm->vmcb01.pa = __sme_set(page_to_pfn(vmcb01_page) << PAGE_SHIFT);
 
@@ -1419,6 +1417,8 @@ static int svm_create_vcpu(struct kvm_vcpu *vcpu)
 	svm_switch_vmcb(svm, &svm->vmcb01);
 	init_vmcb(vcpu);
 
+	svm_vcpu_init_msrpm(vcpu, svm->msrpm);
+
 	svm_init_osvw(vcpu);
 	vcpu->arch.microcode_version = 0x01000065;
 
@@ -1568,8 +1568,11 @@ static void svm_set_vintr(struct vcpu_svm *svm)
 {
 	struct vmcb_control_area *control;
 
-	/* The following fields are ignored when AVIC is enabled */
-	WARN_ON(kvm_vcpu_apicv_active(&svm->vcpu));
+	/*
+	 * The following fields are ignored when AVIC is enabled
+	 */
+	WARN_ON(kvm_apicv_activated(svm->vcpu.kvm));
+
 	svm_set_intercept(svm, INTERCEPT_VINTR);
 
 	/*
@@ -2147,11 +2150,12 @@ static int vmload_vmsave_interception(struct kvm_vcpu *vcpu, bool vmload)
 	ret = kvm_skip_emulated_instruction(vcpu);
 
 	if (vmload) {
-		nested_svm_vmloadsave(vmcb12, svm->vmcb);
+		svm_copy_vmloadsave_state(svm->vmcb, vmcb12);
 		svm->sysenter_eip_hi = 0;
 		svm->sysenter_esp_hi = 0;
-	} else
-		nested_svm_vmloadsave(svm->vmcb, vmcb12);
+	} else {
+		svm_copy_vmloadsave_state(vmcb12, svm->vmcb);
+	}
 
 	kvm_vcpu_unmap(vcpu, &map, true);
 
@@ -4344,8 +4348,8 @@ static int svm_enter_smm(struct kvm_vcpu *vcpu, char *smstate)
 
 		BUILD_BUG_ON(offsetof(struct vmcb, save) != 0x400);
 
-		svm_copy_vmrun_state(&svm->vmcb01.ptr->save,
-				     map_save.hva + 0x400);
+		svm_copy_vmrun_state(map_save.hva + 0x400,
+				     &svm->vmcb01.ptr->save);
 
 		kvm_vcpu_unmap(vcpu, &map_save, true);
 	}
@@ -4393,8 +4397,8 @@ static int svm_leave_smm(struct kvm_vcpu *vcpu, const char *smstate)
 					 &map_save) == -EINVAL)
 				return 1;
 
-			svm_copy_vmrun_state(map_save.hva + 0x400,
-					     &svm->vmcb01.ptr->save);
+			svm_copy_vmrun_state(&svm->vmcb01.ptr->save,
+					     map_save.hva + 0x400);
 
 			kvm_vcpu_unmap(vcpu, &map_save, true);
 		}

arch/x86/kvm/svm/svm.h

@@ -464,9 +464,9 @@ void svm_leave_nested(struct vcpu_svm *svm);
 void svm_free_nested(struct vcpu_svm *svm);
 int svm_allocate_nested(struct vcpu_svm *svm);
 int nested_svm_vmrun(struct kvm_vcpu *vcpu);
-void svm_copy_vmrun_state(struct vmcb_save_area *from_save,
-			  struct vmcb_save_area *to_save);
-void nested_svm_vmloadsave(struct vmcb *from_vmcb, struct vmcb *to_vmcb);
+void svm_copy_vmrun_state(struct vmcb_save_area *to_save,
+			  struct vmcb_save_area *from_save);
+void svm_copy_vmloadsave_state(struct vmcb *to_vmcb, struct vmcb *from_vmcb);
 int nested_svm_vmexit(struct vcpu_svm *svm);
 
 static inline int nested_svm_simple_vmexit(struct vcpu_svm *svm, u32 exit_code)

arch/x86/kvm/svm/svm_onhyperv.h

@@ -89,7 +89,7 @@ static inline void svm_hv_vmcb_dirty_nested_enlightenments(
 	 * as we mark it dirty unconditionally towards end of vcpu
 	 * init phase.
 	 */
-	if (vmcb && vmcb_is_clean(vmcb, VMCB_HV_NESTED_ENLIGHTENMENTS) &&
+	if (vmcb_is_clean(vmcb, VMCB_HV_NESTED_ENLIGHTENMENTS) &&
 	    hve->hv_enlightenments_control.msr_bitmap)
 		vmcb_mark_dirty(vmcb, VMCB_HV_NESTED_ENLIGHTENMENTS);
 }

arch/x86/kvm/x86.c

@@ -3407,7 +3407,7 @@ int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
 			return 1;
 		break;
 	case MSR_KVM_ASYNC_PF_ACK:
-		if (!guest_pv_has(vcpu, KVM_FEATURE_ASYNC_PF))
+		if (!guest_pv_has(vcpu, KVM_FEATURE_ASYNC_PF_INT))
 			return 1;
 		if (data & 0x1) {
 			vcpu->arch.apf.pageready_pending = false;
@@ -3746,7 +3746,7 @@ int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
 		msr_info->data = vcpu->arch.apf.msr_int_val;
 		break;
 	case MSR_KVM_ASYNC_PF_ACK:
-		if (!guest_pv_has(vcpu, KVM_FEATURE_ASYNC_PF))
+		if (!guest_pv_has(vcpu, KVM_FEATURE_ASYNC_PF_INT))
 			return 1;
 
 		msr_info->data = 0;

tools/testing/selftests/kvm/.gitignore

@@ -38,6 +38,7 @@
 /x86_64/xen_vmcall_test
 /x86_64/xss_msr_test
 /x86_64/vmx_pmu_msrs_test
+/access_tracking_perf_test
 /demand_paging_test
 /dirty_log_test
 /dirty_log_perf_test

tools/testing/selftests/kvm/Makefile

@@ -71,6 +71,7 @@ TEST_GEN_PROGS_x86_64 += x86_64/tsc_msrs_test
 TEST_GEN_PROGS_x86_64 += x86_64/vmx_pmu_msrs_test
 TEST_GEN_PROGS_x86_64 += x86_64/xen_shinfo_test
 TEST_GEN_PROGS_x86_64 += x86_64/xen_vmcall_test
+TEST_GEN_PROGS_x86_64 += access_tracking_perf_test
 TEST_GEN_PROGS_x86_64 += demand_paging_test
 TEST_GEN_PROGS_x86_64 += dirty_log_test
 TEST_GEN_PROGS_x86_64 += dirty_log_perf_test

tools/testing/selftests/kvm/aarch64/get-reg-list.c

@@ -1019,7 +1019,8 @@ static __u64 sve_rejects_set[] = {
 #define VREGS_SUBLIST \
 	{ "vregs", .regs = vregs, .regs_n = ARRAY_SIZE(vregs), }
 #define PMU_SUBLIST \
-	{ "pmu", .regs = pmu_regs, .regs_n = ARRAY_SIZE(pmu_regs), }
+	{ "pmu", .capability = KVM_CAP_ARM_PMU_V3, .feature = KVM_ARM_VCPU_PMU_V3, \
+	  .regs = pmu_regs, .regs_n = ARRAY_SIZE(pmu_regs), }
 #define SVE_SUBLIST \
 	{ "sve", .capability = KVM_CAP_ARM_SVE, .feature = KVM_ARM_VCPU_SVE, .finalize = true, \
 	  .regs = sve_regs, .regs_n = ARRAY_SIZE(sve_regs), \

tools/testing/selftests/kvm/access_tracking_perf_test.c

+// SPDX-License-Identifier: GPL-2.0
+/*
+ * access_tracking_perf_test
+ *
+ * Copyright (C) 2021, Google, Inc.
+ *
+ * This test measures the performance effects of KVM's access tracking.
+ * Access tracking is driven by the MMU notifiers test_young, clear_young, and
+ * clear_flush_young. These notifiers do not have a direct userspace API,
+ * however the clear_young notifier can be triggered by marking a pages as idle
+ * in /sys/kernel/mm/page_idle/bitmap. This test leverages that mechanism to
+ * enable access tracking on guest memory.
+ *
+ * To measure performance this test runs a VM with a configurable number of
+ * vCPUs that each touch every page in disjoint regions of memory. Performance
+ * is measured in the time it takes all vCPUs to finish touching their
+ * predefined region.
+ *
+ * Note that a deterministic correctness test of access tracking is not possible
+ * by using page_idle as it exists today. This is for a few reasons:
+ *
+ * 1. page_idle only issues clear_young notifiers, which lack a TLB flush. This
+ *    means subsequent guest accesses are not guaranteed to see page table
+ *    updates made by KVM until some time in the future.
+ *
+ * 2. page_idle only operates on LRU pages. Newly allocated pages are not
+ *    immediately allocated to LRU lists. Instead they are held in a "pagevec",
+ *    which is drained to LRU lists some time in the future. There is no
+ *    userspace API to force this drain to occur.
+ *
+ * These limitations are worked around in this test by using a large enough
+ * region of memory for each vCPU such that the number of translations cached in
+ * the TLB and the number of pages held in pagevecs are a small fraction of the
+ * overall workload. And if either of those conditions are not true this test
+ * will fail rather than silently passing.
+ */
+#include <inttypes.h>
+#include <limits.h>
+#include <pthread.h>
+#include <sys/mman.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+
+#include "kvm_util.h"
+#include "test_util.h"
+#include "perf_test_util.h"
+#include "guest_modes.h"
+
+/* Global variable used to synchronize all of the vCPU threads. */
+static int iteration = -1;
+
+/* Defines what vCPU threads should do during a given iteration. */
+static enum {
+	/* Run the vCPU to access all its memory. */
+	ITERATION_ACCESS_MEMORY,
+	/* Mark the vCPU's memory idle in page_idle. */
+	ITERATION_MARK_IDLE,
+} iteration_work;
+
+/* Set to true when vCPU threads should exit. */
+static bool done;
+
+/* The iteration that was last completed by each vCPU. */
+static int vcpu_last_completed_iteration[KVM_MAX_VCPUS];
+
+/* Whether to overlap the regions of memory vCPUs access. */
+static bool overlap_memory_access;
+
+struct test_params {
+	/* The backing source for the region of memory. */
+	enum vm_mem_backing_src_type backing_src;
+
+	/* The amount of memory to allocate for each vCPU. */
+	uint64_t vcpu_memory_bytes;
+
+	/* The number of vCPUs to create in the VM. */
+	int vcpus;
+};
+
+static uint64_t pread_uint64(int fd, const char *filename, uint64_t index)
+{
+	uint64_t value;
+	off_t offset = index * sizeof(value);
+
+	TEST_ASSERT(pread(fd, &value, sizeof(value), offset) == sizeof(value),
+		    "pread from %s offset 0x%" PRIx64 " failed!",
+		    filename, offset);
+
+	return value;
+
+}
+
+#define PAGEMAP_PRESENT (1ULL << 63)
+#define PAGEMAP_PFN_MASK ((1ULL << 55) - 1)
+
+static uint64_t lookup_pfn(int pagemap_fd, struct kvm_vm *vm, uint64_t gva)
+{
+	uint64_t hva = (uint64_t) addr_gva2hva(vm, gva);
+	uint64_t entry;
+	uint64_t pfn;
+
+	entry = pread_uint64(pagemap_fd, "pagemap", hva / getpagesize());
+	if (!(entry & PAGEMAP_PRESENT))
+		return 0;
+
+	pfn = entry & PAGEMAP_PFN_MASK;
+	if (!pfn) {
+		print_skip("Looking up PFNs requires CAP_SYS_ADMIN");
+		exit(KSFT_SKIP);
+	}
+
+	return pfn;
+}
+
+static bool is_page_idle(int page_idle_fd, uint64_t pfn)
+{
+	uint64_t bits = pread_uint64(page_idle_fd, "page_idle", pfn / 64);
+
+	return !!((bits >> (pfn % 64)) & 1);
+}
+
+static void mark_page_idle(int page_idle_fd, uint64_t pfn)
+{
+	uint64_t bits = 1ULL << (pfn % 64);
+
+	TEST_ASSERT(pwrite(page_idle_fd, &bits, 8, 8 * (pfn / 64)) == 8,
+		    "Set page_idle bits for PFN 0x%" PRIx64, pfn);
+}
+
+static void mark_vcpu_memory_idle(struct kvm_vm *vm, int vcpu_id)
+{
+	uint64_t base_gva = perf_test_args.vcpu_args[vcpu_id].gva;
+	uint64_t pages = perf_test_args.vcpu_args[vcpu_id].pages;
+	uint64_t page;
+	uint64_t still_idle = 0;
+	uint64_t no_pfn = 0;
+	int page_idle_fd;
+	int pagemap_fd;
+
+	/* If vCPUs are using an overlapping region, let vCPU 0 mark it idle. */
+	if (overlap_memory_access && vcpu_id)
+		return;
+
+	page_idle_fd = open("/sys/kernel/mm/page_idle/bitmap", O_RDWR);
+	TEST_ASSERT(page_idle_fd > 0, "Failed to open page_idle.");
+
+	pagemap_fd = open("/proc/self/pagemap", O_RDONLY);
+	TEST_ASSERT(pagemap_fd > 0, "Failed to open pagemap.");
+
+	for (page = 0; page < pages; page++) {
+		uint64_t gva = base_gva + page * perf_test_args.guest_page_size;
+		uint64_t pfn = lookup_pfn(pagemap_fd, vm, gva);
+
+		if (!pfn) {
+			no_pfn++;
+			continue;
+		}
+
+		if (is_page_idle(page_idle_fd, pfn)) {
+			still_idle++;
+			continue;
+		}
+
+		mark_page_idle(page_idle_fd, pfn);
+	}
+
+	/*
+	 * Assumption: Less than 1% of pages are going to be swapped out from
+	 * under us during this test.
+	 */
+	TEST_ASSERT(no_pfn < pages / 100,
+		    "vCPU %d: No PFN for %" PRIu64 " out of %" PRIu64 " pages.",
+		    vcpu_id, no_pfn, pages);
+
+	/*
+	 * Test that at least 90% of memory has been marked idle (the rest might
+	 * not be marked idle because the pages have not yet made it to an LRU
+	 * list or the translations are still cached in the TLB). 90% is
+	 * arbitrary; high enough that we ensure most memory access went through
+	 * access tracking but low enough as to not make the test too brittle
+	 * over time and across architectures.
+	 */
+	TEST_ASSERT(still_idle < pages / 10,
+		    "vCPU%d: Too many pages still idle (%"PRIu64 " out of %"
+		    PRIu64 ").\n",
+		    vcpu_id, still_idle, pages);
+
+	close(page_idle_fd);
+	close(pagemap_fd);
+}
+
+static void assert_ucall(struct kvm_vm *vm, uint32_t vcpu_id,
+			 uint64_t expected_ucall)
+{
+	struct ucall uc;
+	uint64_t actual_ucall = get_ucall(vm, vcpu_id, &uc);
+
+	TEST_ASSERT(expected_ucall == actual_ucall,
+		    "Guest exited unexpectedly (expected ucall %" PRIu64
+		    ", got %" PRIu64 ")",
+		    expected_ucall, actual_ucall);
+}
+
+static bool spin_wait_for_next_iteration(int *current_iteration)
+{
+	int last_iteration = *current_iteration;
+
+	do {
+		if (READ_ONCE(done))
+			return false;
+
+		*current_iteration = READ_ONCE(iteration);
+	} while (last_iteration == *current_iteration);
+
+	return true;
+}
+
+static void *vcpu_thread_main(void *arg)
+{
+	struct perf_test_vcpu_args *vcpu_args = arg;
+	struct kvm_vm *vm = perf_test_args.vm;
+	int vcpu_id = vcpu_args->vcpu_id;
+	int current_iteration = -1;
+
+	vcpu_args_set(vm, vcpu_id, 1, vcpu_id);
+
+	while (spin_wait_for_next_iteration(&current_iteration)) {
+		switch (READ_ONCE(iteration_work)) {
+		case ITERATION_ACCESS_MEMORY:
+			vcpu_run(vm, vcpu_id);
+			assert_ucall(vm, vcpu_id, UCALL_SYNC);
+			break;
+		case ITERATION_MARK_IDLE:
+			mark_vcpu_memory_idle(vm, vcpu_id);
+			break;
+		};
+
+		vcpu_last_completed_iteration[vcpu_id] = current_iteration;
+	}
+
+	return NULL;
+}
+
+static void spin_wait_for_vcpu(int vcpu_id, int target_iteration)
+{
+	while (READ_ONCE(vcpu_last_completed_iteration[vcpu_id]) !=
+	       target_iteration) {
+		continue;
+	}
+}
+
+/* The type of memory accesses to perform in the VM. */
+enum access_type {
+	ACCESS_READ,
+	ACCESS_WRITE,
+};
+
+static void run_iteration(struct kvm_vm *vm, int vcpus, const char *description)
+{
+	struct timespec ts_start;
+	struct timespec ts_elapsed;
+	int next_iteration;
+	int vcpu_id;
+
+	/* Kick off the vCPUs by incrementing iteration. */
+	next_iteration = ++iteration;
+
+	clock_gettime(CLOCK_MONOTONIC, &ts_start);
+
+	/* Wait for all vCPUs to finish the iteration. */
+	for (vcpu_id = 0; vcpu_id < vcpus; vcpu_id++)
+		spin_wait_for_vcpu(vcpu_id, next_iteration);
+
+	ts_elapsed = timespec_elapsed(ts_start);
+	pr_info("%-30s: %ld.%09lds\n",
+		description, ts_elapsed.tv_sec, ts_elapsed.tv_nsec);
+}
+
+static void access_memory(struct kvm_vm *vm, int vcpus, enum access_type access,
+			  const char *description)
+{
+	perf_test_args.wr_fract = (access == ACCESS_READ) ? INT_MAX : 1;
+	sync_global_to_guest(vm, perf_test_args);
+	iteration_work = ITERATION_ACCESS_MEMORY;
+	run_iteration(vm, vcpus, description);
+}
+
+static void mark_memory_idle(struct kvm_vm *vm, int vcpus)
+{
+	/*
+	 * Even though this parallelizes the work across vCPUs, this is still a
+	 * very slow operation because page_idle forces the test to mark one pfn
+	 * at a time and the clear_young notifier serializes on the KVM MMU
+	 * lock.
+	 */
+	pr_debug("Marking VM memory idle (slow)...\n");
+	iteration_work = ITERATION_MARK_IDLE;
+	run_iteration(vm, vcpus, "Mark memory idle");
+}
+
+static pthread_t *create_vcpu_threads(int vcpus)
+{
+	pthread_t *vcpu_threads;
+	int i;
+
+	vcpu_threads = malloc(vcpus * sizeof(vcpu_threads[0]));
+	TEST_ASSERT(vcpu_threads, "Failed to allocate vcpu_threads.");
+
+	for (i = 0; i < vcpus; i++) {
+		vcpu_last_completed_iteration[i] = iteration;
+		pthread_create(&vcpu_threads[i], NULL, vcpu_thread_main,
+			       &perf_test_args.vcpu_args[i]);
+	}
+
+	return vcpu_threads;
+}
+
+static void terminate_vcpu_threads(pthread_t *vcpu_threads, int vcpus)
+{
+	int i;
+
+	/* Set done to signal the vCPU threads to exit */
+	done = true;
+
+	for (i = 0; i < vcpus; i++)
+		pthread_join(vcpu_threads[i], NULL);
+}
+
+static void run_test(enum vm_guest_mode mode, void *arg)
+{
+	struct test_params *params = arg;
+	struct kvm_vm *vm;
+	pthread_t *vcpu_threads;
+	int vcpus = params->vcpus;
+
+	vm = perf_test_create_vm(mode, vcpus, params->vcpu_memory_bytes,
+				 params->backing_src);
+
+	perf_test_setup_vcpus(vm, vcpus, params->vcpu_memory_bytes,
+			      !overlap_memory_access);
+
+	vcpu_threads = create_vcpu_threads(vcpus);
+
+	pr_info("\n");
+	access_memory(vm, vcpus, ACCESS_WRITE, "Populating memory");
+
+	/* As a control, read and write to the populated memory first. */
+	access_memory(vm, vcpus, ACCESS_WRITE, "Writing to populated memory");
+	access_memory(vm, vcpus, ACCESS_READ, "Reading from populated memory");
+
+	/* Repeat on memory that has been marked as idle. */
+	mark_memory_idle(vm, vcpus);
+	access_memory(vm, vcpus, ACCESS_WRITE, "Writing to idle memory");
+	mark_memory_idle(vm, vcpus);
+	access_memory(vm, vcpus, ACCESS_READ, "Reading from idle memory");
+
+	terminate_vcpu_threads(vcpu_threads, vcpus);
+	free(vcpu_threads);
+	perf_test_destroy_vm(vm);
+}
+
+static void help(char *name)
+{
+	puts("");
+	printf("usage: %s [-h] [-m mode] [-b vcpu_bytes] [-v vcpus] [-o]  [-s mem_type]\n",
+	       name);
+	puts("");
+	printf(" -h: Display this help message.");
+	guest_modes_help();
+	printf(" -b: specify the size of the memory region which should be\n"
+	       "     dirtied by each vCPU. e.g. 10M or 3G.\n"
+	       "     (default: 1G)\n");
+	printf(" -v: specify the number of vCPUs to run.\n");
+	printf(" -o: Overlap guest memory accesses instead of partitioning\n"
+	       "     them into a separate region of memory for each vCPU.\n");
+	printf(" -s: specify the type of memory that should be used to\n"
+	       "     back the guest data region.\n\n");
+	backing_src_help();
+	puts("");
+	exit(0);
+}
+
+int main(int argc, char *argv[])
+{
+	struct test_params params = {
+		.backing_src = VM_MEM_SRC_ANONYMOUS,
+		.vcpu_memory_bytes = DEFAULT_PER_VCPU_MEM_SIZE,
+		.vcpus = 1,
+	};
+	int page_idle_fd;
+	int opt;
+
+	guest_modes_append_default();
+
+	while ((opt = getopt(argc, argv, "hm:b:v:os:")) != -1) {
+		switch (opt) {
+		case 'm':
+			guest_modes_cmdline(optarg);
+			break;
+		case 'b':
+			params.vcpu_memory_bytes = parse_size(optarg);
+			break;
+		case 'v':
+			params.vcpus = atoi(optarg);
+			break;
+		case 'o':
+			overlap_memory_access = true;
+			break;
+		case 's':
+			params.backing_src = parse_backing_src_type(optarg);
+			break;
+		case 'h':
+		default:
+			help(argv[0]);
+			break;
+		}
+	}
+
+	page_idle_fd = open("/sys/kernel/mm/page_idle/bitmap", O_RDWR);
+	if (page_idle_fd < 0) {
+		print_skip("CONFIG_IDLE_PAGE_TRACKING is not enabled");
+		exit(KSFT_SKIP);
+	}
+	close(page_idle_fd);
+
+	for_each_guest_mode(run_test, &params);
+
+	return 0;
+}

tools/testing/selftests/kvm/dirty_log_perf_test.c

@@ -312,6 +312,7 @@ int main(int argc, char *argv[])
 			break;
 		case 'o':
 			p.partition_vcpu_memory_access = false;
+			break;
 		case 's':
 			p.backing_src = parse_backing_src_type(optarg);
 			break;

tools/testing/selftests/kvm/steal_time.c

@@ -320,7 +320,7 @@ int main(int ac, char **av)
 		run_delay = get_run_delay();
 		pthread_create(&thread, &attr, do_steal_time, NULL);
 		do
-			pthread_yield();
+			sched_yield();
 		while (get_run_delay() - run_delay < MIN_RUN_DELAY_NS);
 		pthread_join(thread, NULL);
 		run_delay = get_run_delay() - run_delay;

virt/kvm/kvm_main.c

@@ -3110,6 +3110,7 @@ void kvm_vcpu_block(struct kvm_vcpu *vcpu)
 					++vcpu->stat.generic.halt_poll_invalid;
 				goto out;
 			}
+			cpu_relax();
 			poll_end = cur = ktime_get();
 		} while (kvm_vcpu_can_poll(cur, stop));
 	}
@@ -4390,6 +4391,16 @@ struct compat_kvm_dirty_log {
 	};
 };
 
+struct compat_kvm_clear_dirty_log {
+	__u32 slot;
+	__u32 num_pages;
+	__u64 first_page;
+	union {
+		compat_uptr_t dirty_bitmap; /* one bit per page */
+		__u64 padding2;
+	};
+};
+
 static long kvm_vm_compat_ioctl(struct file *filp,
 			   unsigned int ioctl, unsigned long arg)
 {
@@ -4399,6 +4410,24 @@ static long kvm_vm_compat_ioctl(struct file *filp,
 	if (kvm->mm != current->mm)
 		return -EIO;
 	switch (ioctl) {
+#ifdef CONFIG_KVM_GENERIC_DIRTYLOG_READ_PROTECT
+	case KVM_CLEAR_DIRTY_LOG: {
+		struct compat_kvm_clear_dirty_log compat_log;
+		struct kvm_clear_dirty_log log;
+
+		if (copy_from_user(&compat_log, (void __user *)arg,
+				   sizeof(compat_log)))
+			return -EFAULT;
+		log.slot	 = compat_log.slot;
+		log.num_pages	 = compat_log.num_pages;
+		log.first_page	 = compat_log.first_page;
+		log.padding2	 = compat_log.padding2;
+		log.dirty_bitmap = compat_ptr(compat_log.dirty_bitmap);
+
+		r = kvm_vm_ioctl_clear_dirty_log(kvm, &log);
+		break;
+	}
+#endif
 	case KVM_GET_DIRTY_LOG: {
 		struct compat_kvm_dirty_log compat_log;
 		struct kvm_dirty_log log;