| Message ID | 20220123184541.993212-2-daviddunn@google.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | KVM: x86: Provide per VM capability for disabling PMU virtualization | expand |
On Sun, Jan 23, 2022 at 10:45 AM David Dunn <daviddunn@google.com> wrote: > > KVM_CAP_PMU_DISABLE is used to disable PMU virtualization on individual > x86 VMs. PMU configuration must be done prior to creating VCPUs. > > To enable future extension, KVM_CAP_PMU_CAPABILITY reports available > settings via bitmask when queried via check_extension. > > For VMs that have PMU virtualization disabled, usermode will need to > clear CPUID leaf 0xA to notify guests. > > Signed-off-by: David Dunn <daviddunn@google.com> Nit: The two references to CPUID leaf 0xA should be qualified as applying only to Intel VMs. Reviewed-by: Jim Mattson <jmattson@google.com>
cc LKML and full list of KVM reviewers. On 25/1/2022 2:39 am, Jim Mattson wrote: > On Sun, Jan 23, 2022 at 10:45 AM David Dunn <daviddunn@google.com> wrote: >> >> KVM_CAP_PMU_DISABLE is used to disable PMU virtualization on individual >> x86 VMs. PMU configuration must be done prior to creating VCPUs. >> >> To enable future extension, KVM_CAP_PMU_CAPABILITY reports available >> settings via bitmask when queried via check_extension. >> >> For VMs that have PMU virtualization disabled, usermode will need to >> clear CPUID leaf 0xA to notify guests. >> >> Signed-off-by: David Dunn <daviddunn@google.com> > > Nit: The two references to CPUID leaf 0xA should be qualified as > applying only to Intel VMs. > > Reviewed-by: Jim Mattson <jmattson@google.com> > Nit: It looks like we already have "#define KVM_CAP_SYS_ATTRIBUTES 209". Hope it helps a little: Reviewed-by: Like Xu <likexu@tencent.com>
Thanks Like. I just sent a new v6 series with these minor merge conflicts resolved. Dave On Wed, Feb 9, 2022 at 2:33 AM Like Xu <like.xu.linux@gmail.com> wrote: > > cc LKML and full list of KVM reviewers. > > On 25/1/2022 2:39 am, Jim Mattson wrote: > > On Sun, Jan 23, 2022 at 10:45 AM David Dunn <daviddunn@google.com> wrote: > >> > >> KVM_CAP_PMU_DISABLE is used to disable PMU virtualization on individual > >> x86 VMs. PMU configuration must be done prior to creating VCPUs. > >> > >> To enable future extension, KVM_CAP_PMU_CAPABILITY reports available > >> settings via bitmask when queried via check_extension. > >> > >> For VMs that have PMU virtualization disabled, usermode will need to > >> clear CPUID leaf 0xA to notify guests. > >> > >> Signed-off-by: David Dunn <daviddunn@google.com> > > > > Nit: The two references to CPUID leaf 0xA should be qualified as > > applying only to Intel VMs. > > > > Reviewed-by: Jim Mattson <jmattson@google.com> > > > > Nit: It looks like we already have "#define KVM_CAP_SYS_ATTRIBUTES 209". > > Hope it helps a little: > > Reviewed-by: Like Xu <likexu@tencent.com>
diff --git a/Documentation/virt/kvm/api.rst b/Documentation/virt/kvm/api.rst index bb8cfddbb22d..ae0d52c5359e 100644 --- a/Documentation/virt/kvm/api.rst +++ b/Documentation/virt/kvm/api.rst @@ -7559,3 +7559,25 @@ The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset of the result of KVM_CHECK_EXTENSION. KVM will forward to userspace the hypercalls whose corresponding bit is in the argument, and return ENOSYS for the others. + +8.35 KVM_CAP_PMU_CAPABILITY +--------------------------- + +:Capability KVM_CAP_PMU_CAPABILITY +:Architectures: x86 +:Type: vm +:Parameters: arg[0] is bitmask of PMU virtualization capabilities. +:Returns 0 on success, -EINVAL when arg[0] contains invalid bits + +This capability alters PMU virtualization in KVM. + +Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of +PMU virtualization capabilities that can be adjusted on a VM. + +The argument to KVM_ENABLE_CAP is also a bitmask and selects specific +PMU virtualization capabilities to be applied to the VM. This can +only be invoked on a VM prior to the creation of VCPUs. + +At this time, KVM_CAP_PMU_DISABLE is the only capability. Setting +this capability will disable PMU virtualization for that VM. Usermode +should adjust CPUID leaf 0xA to reflect that the PMU is disabled. diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h index 682ad02a4e58..5cdcd4a7671b 100644 --- a/arch/x86/include/asm/kvm_host.h +++ b/arch/x86/include/asm/kvm_host.h @@ -1232,6 +1232,7 @@ struct kvm_arch { hpa_t hv_root_tdp; spinlock_t hv_root_tdp_lock; #endif + bool enable_pmu; }; struct kvm_vm_stat { diff --git a/arch/x86/kvm/svm/pmu.c b/arch/x86/kvm/svm/pmu.c index 5aa45f13b16d..d4de52409335 100644 --- a/arch/x86/kvm/svm/pmu.c +++ b/arch/x86/kvm/svm/pmu.c @@ -101,7 +101,7 @@ static inline struct kvm_pmc *get_gp_pmc_amd(struct kvm_pmu *pmu, u32 msr, { struct kvm_vcpu *vcpu = pmu_to_vcpu(pmu); - if (!enable_pmu) + if (!vcpu->kvm->arch.enable_pmu) return NULL; switch (msr) { diff --git a/arch/x86/kvm/vmx/pmu_intel.c b/arch/x86/kvm/vmx/pmu_intel.c index 466d18fc0c5d..2c5868d77268 100644 --- a/arch/x86/kvm/vmx/pmu_intel.c +++ b/arch/x86/kvm/vmx/pmu_intel.c @@ -487,7 +487,7 @@ static void intel_pmu_refresh(struct kvm_vcpu *vcpu) pmu->reserved_bits = 0xffffffff00200000ull; entry = kvm_find_cpuid_entry(vcpu, 0xa, 0); - if (!entry || !enable_pmu) + if (!entry || !vcpu->kvm->arch.enable_pmu) return; eax.full = entry->eax; edx.full = entry->edx; diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index 55518b7d3b96..a033f019a3f0 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -4326,6 +4326,9 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext) if (r < sizeof(struct kvm_xsave)) r = sizeof(struct kvm_xsave); break; + case KVM_CAP_PMU_CAPABILITY: + r = enable_pmu ? KVM_CAP_PMU_MASK : 0; + break; } default: break; @@ -5937,6 +5940,14 @@ int kvm_vm_ioctl_enable_cap(struct kvm *kvm, kvm->arch.exit_on_emulation_error = cap->args[0]; r = 0; break; + case KVM_CAP_PMU_CAPABILITY: + r = -EINVAL; + if (!enable_pmu || kvm->created_vcpus > 0 || + cap->args[0] & ~KVM_CAP_PMU_MASK) + break; + kvm->arch.enable_pmu = !(cap->args[0] & KVM_CAP_PMU_DISABLE); + r = 0; + break; default: r = -EINVAL; break; @@ -11562,6 +11573,7 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type) raw_spin_unlock_irqrestore(&kvm->arch.tsc_write_lock, flags); kvm->arch.guest_can_read_msr_platform_info = true; + kvm->arch.enable_pmu = enable_pmu; #if IS_ENABLED(CONFIG_HYPERV) spin_lock_init(&kvm->arch.hv_root_tdp_lock); diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h index 9563d294f181..1c5e6e172817 100644 --- a/include/uapi/linux/kvm.h +++ b/include/uapi/linux/kvm.h @@ -1133,6 +1133,7 @@ struct kvm_ppc_resize_hpt { #define KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM 206 #define KVM_CAP_VM_GPA_BITS 207 #define KVM_CAP_XSAVE2 208 +#define KVM_CAP_PMU_CAPABILITY 209 #ifdef KVM_CAP_IRQ_ROUTING @@ -1972,6 +1973,9 @@ struct kvm_dirty_gfn { #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0) #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1) +#define KVM_CAP_PMU_DISABLE (1 << 0) +#define KVM_CAP_PMU_MASK (KVM_CAP_PMU_DISABLE) + /** * struct kvm_stats_header - Header of per vm/vcpu binary statistics data. * @flags: Some extra information for header, always 0 for now. diff --git a/tools/include/uapi/linux/kvm.h b/tools/include/uapi/linux/kvm.h index 9563d294f181..a361cf6e8604 100644 --- a/tools/include/uapi/linux/kvm.h +++ b/tools/include/uapi/linux/kvm.h @@ -1133,6 +1133,7 @@ struct kvm_ppc_resize_hpt { #define KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM 206 #define KVM_CAP_VM_GPA_BITS 207 #define KVM_CAP_XSAVE2 208 +#define KVM_CAP_PMU_CAPABILITY 209 #ifdef KVM_CAP_IRQ_ROUTING @@ -1972,6 +1973,9 @@ struct kvm_dirty_gfn { #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0) #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1) +#define KVM_CAP_PMU_DISABLE (1 << 0) +#define KVM_CAP_PMU_MASK (KVM_PMU_CONFIG_DISABLE) + /** * struct kvm_stats_header - Header of per vm/vcpu binary statistics data. * @flags: Some extra information for header, always 0 for now.
KVM_CAP_PMU_DISABLE is used to disable PMU virtualization on individual x86 VMs. PMU configuration must be done prior to creating VCPUs. To enable future extension, KVM_CAP_PMU_CAPABILITY reports available settings via bitmask when queried via check_extension. For VMs that have PMU virtualization disabled, usermode will need to clear CPUID leaf 0xA to notify guests. Signed-off-by: David Dunn <daviddunn@google.com> --- Documentation/virt/kvm/api.rst | 22 ++++++++++++++++++++++ arch/x86/include/asm/kvm_host.h | 1 + arch/x86/kvm/svm/pmu.c | 2 +- arch/x86/kvm/vmx/pmu_intel.c | 2 +- arch/x86/kvm/x86.c | 12 ++++++++++++ include/uapi/linux/kvm.h | 4 ++++ tools/include/uapi/linux/kvm.h | 4 ++++ 7 files changed, 45 insertions(+), 2 deletions(-)