| Message ID | 20200317075315.70933-1-like.xu@linux.intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | kvm/x86: Reduce counter period change overhead and delay the effective time | expand |
On 2020/3/17 15:53, Like Xu wrote: > The cost of perf_event_period() is unstable, and when the guest samples > multiple events, the overhead increases dramatically (5378 ns on E5-2699). > > For a non-running counter, the effective time of the new period is when > its corresponding enable bit is enabled. Calling perf_event_period() > in advance is superfluous. For a running counter, it's safe to delay the > effective time until the KVM_REQ_PMU event is handled. If there are > multiple perf_event_period() calls before handling KVM_REQ_PMU, > it helps to reduce the total cost. > > Signed-off-by: Like Xu <like.xu@linux.intel.com> > > --- > arch/x86/kvm/pmu.c | 11 ----------- > arch/x86/kvm/pmu.h | 11 +++++++++++ > arch/x86/kvm/vmx/pmu_intel.c | 10 ++++------ > 3 files changed, 15 insertions(+), 17 deletions(-) > > diff --git a/arch/x86/kvm/pmu.c b/arch/x86/kvm/pmu.c > index d1f8ca57d354..527a8bb85080 100644 > --- a/arch/x86/kvm/pmu.c > +++ b/arch/x86/kvm/pmu.c > @@ -437,17 +437,6 @@ void kvm_pmu_init(struct kvm_vcpu *vcpu) > kvm_pmu_refresh(vcpu); > } > > -static inline bool pmc_speculative_in_use(struct kvm_pmc *pmc) > -{ > - struct kvm_pmu *pmu = pmc_to_pmu(pmc); > - > - if (pmc_is_fixed(pmc)) > - return fixed_ctrl_field(pmu->fixed_ctr_ctrl, > - pmc->idx - INTEL_PMC_IDX_FIXED) & 0x3; > - > - return pmc->eventsel & ARCH_PERFMON_EVENTSEL_ENABLE; > -} > - > /* Release perf_events for vPMCs that have been unused for a full time slice. */ > void kvm_pmu_cleanup(struct kvm_vcpu *vcpu) > { > diff --git a/arch/x86/kvm/pmu.h b/arch/x86/kvm/pmu.h > index d7da2b9e0755..cd112e825d2c 100644 > --- a/arch/x86/kvm/pmu.h > +++ b/arch/x86/kvm/pmu.h > @@ -138,6 +138,17 @@ static inline u64 get_sample_period(struct kvm_pmc *pmc, u64 counter_value) > return sample_period; > } > > +static inline bool pmc_speculative_in_use(struct kvm_pmc *pmc) > +{ > + struct kvm_pmu *pmu = pmc_to_pmu(pmc); > + > + if (pmc_is_fixed(pmc)) > + return fixed_ctrl_field(pmu->fixed_ctr_ctrl, > + pmc->idx - INTEL_PMC_IDX_FIXED) & 0x3; > + > + return pmc->eventsel & ARCH_PERFMON_EVENTSEL_ENABLE; > +} > + > void reprogram_gp_counter(struct kvm_pmc *pmc, u64 eventsel); > void reprogram_fixed_counter(struct kvm_pmc *pmc, u8 ctrl, int fixed_idx); > void reprogram_counter(struct kvm_pmu *pmu, int pmc_idx); > diff --git a/arch/x86/kvm/vmx/pmu_intel.c b/arch/x86/kvm/vmx/pmu_intel.c > index 7c857737b438..4e689273eb05 100644 > --- a/arch/x86/kvm/vmx/pmu_intel.c > +++ b/arch/x86/kvm/vmx/pmu_intel.c > @@ -263,15 +263,13 @@ static int intel_pmu_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info) > if (!msr_info->host_initiated) > data = (s64)(s32)data; > pmc->counter += data - pmc_read_counter(pmc); > - if (pmc->perf_event) > - perf_event_period(pmc->perf_event, > - get_sample_period(pmc, data)); > + if (pmc_speculative_in_use(pmc)) { Oops, the "{" is a shameful mistake. > + kvm_make_request(KVM_REQ_PMU, pmc->vcpu); > return 0; > } else if ((pmc = get_fixed_pmc(pmu, msr))) { > pmc->counter += data - pmc_read_counter(pmc); > - if (pmc->perf_event) > - perf_event_period(pmc->perf_event, > - get_sample_period(pmc, data)); > + if (pmc_speculative_in_use(pmc)) { > + kvm_make_request(KVM_REQ_PMU, pmc->vcpu); > return 0; > } else if ((pmc = get_gp_pmc(pmu, msr, MSR_P6_EVNTSEL0))) { > if (data == pmc->eventsel) >
diff --git a/arch/x86/kvm/pmu.c b/arch/x86/kvm/pmu.c index d1f8ca57d354..527a8bb85080 100644 --- a/arch/x86/kvm/pmu.c +++ b/arch/x86/kvm/pmu.c @@ -437,17 +437,6 @@ void kvm_pmu_init(struct kvm_vcpu *vcpu) kvm_pmu_refresh(vcpu); } -static inline bool pmc_speculative_in_use(struct kvm_pmc *pmc) -{ - struct kvm_pmu *pmu = pmc_to_pmu(pmc); - - if (pmc_is_fixed(pmc)) - return fixed_ctrl_field(pmu->fixed_ctr_ctrl, - pmc->idx - INTEL_PMC_IDX_FIXED) & 0x3; - - return pmc->eventsel & ARCH_PERFMON_EVENTSEL_ENABLE; -} - /* Release perf_events for vPMCs that have been unused for a full time slice. */ void kvm_pmu_cleanup(struct kvm_vcpu *vcpu) { diff --git a/arch/x86/kvm/pmu.h b/arch/x86/kvm/pmu.h index d7da2b9e0755..cd112e825d2c 100644 --- a/arch/x86/kvm/pmu.h +++ b/arch/x86/kvm/pmu.h @@ -138,6 +138,17 @@ static inline u64 get_sample_period(struct kvm_pmc *pmc, u64 counter_value) return sample_period; } +static inline bool pmc_speculative_in_use(struct kvm_pmc *pmc) +{ + struct kvm_pmu *pmu = pmc_to_pmu(pmc); + + if (pmc_is_fixed(pmc)) + return fixed_ctrl_field(pmu->fixed_ctr_ctrl, + pmc->idx - INTEL_PMC_IDX_FIXED) & 0x3; + + return pmc->eventsel & ARCH_PERFMON_EVENTSEL_ENABLE; +} + void reprogram_gp_counter(struct kvm_pmc *pmc, u64 eventsel); void reprogram_fixed_counter(struct kvm_pmc *pmc, u8 ctrl, int fixed_idx); void reprogram_counter(struct kvm_pmu *pmu, int pmc_idx); diff --git a/arch/x86/kvm/vmx/pmu_intel.c b/arch/x86/kvm/vmx/pmu_intel.c index 7c857737b438..4e689273eb05 100644 --- a/arch/x86/kvm/vmx/pmu_intel.c +++ b/arch/x86/kvm/vmx/pmu_intel.c @@ -263,15 +263,13 @@ static int intel_pmu_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info) if (!msr_info->host_initiated) data = (s64)(s32)data; pmc->counter += data - pmc_read_counter(pmc); - if (pmc->perf_event) - perf_event_period(pmc->perf_event, - get_sample_period(pmc, data)); + if (pmc_speculative_in_use(pmc)) { + kvm_make_request(KVM_REQ_PMU, pmc->vcpu); return 0; } else if ((pmc = get_fixed_pmc(pmu, msr))) { pmc->counter += data - pmc_read_counter(pmc); - if (pmc->perf_event) - perf_event_period(pmc->perf_event, - get_sample_period(pmc, data)); + if (pmc_speculative_in_use(pmc)) { + kvm_make_request(KVM_REQ_PMU, pmc->vcpu); return 0; } else if ((pmc = get_gp_pmc(pmu, msr, MSR_P6_EVNTSEL0))) { if (data == pmc->eventsel)
The cost of perf_event_period() is unstable, and when the guest samples multiple events, the overhead increases dramatically (5378 ns on E5-2699). For a non-running counter, the effective time of the new period is when its corresponding enable bit is enabled. Calling perf_event_period() in advance is superfluous. For a running counter, it's safe to delay the effective time until the KVM_REQ_PMU event is handled. If there are multiple perf_event_period() calls before handling KVM_REQ_PMU, it helps to reduce the total cost. Signed-off-by: Like Xu <like.xu@linux.intel.com> --- arch/x86/kvm/pmu.c | 11 ----------- arch/x86/kvm/pmu.h | 11 +++++++++++ arch/x86/kvm/vmx/pmu_intel.c | 10 ++++------ 3 files changed, 15 insertions(+), 17 deletions(-)