On LoongArch system, guest PMU hardware is shared by guest and host and PMU interrupt is separated. PMU is pass-through to VM, and there is PMU context switch when exit to host and return to guest. There is optimiation to check whether PMU is enabled by guest. If not, it is not necessary to return to guest. However it is enabled, PMU context for guest need switch on. Now KVM_REQ_PMU notification is set on vcpu context switch, however it is missing if there is no vcpu context switch and PMU is used by guest VM. Fixes: f4e40ea9f78f ("LoongArch: KVM: Add PMU support for guest") Cc: Signed-off-by: Bibo Mao --- arch/loongarch/kvm/vcpu.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/arch/loongarch/kvm/vcpu.c b/arch/loongarch/kvm/vcpu.c index 30e3b089a596..bf56ad29ac15 100644 --- a/arch/loongarch/kvm/vcpu.c +++ b/arch/loongarch/kvm/vcpu.c @@ -132,6 +132,9 @@ static void kvm_lose_pmu(struct kvm_vcpu *vcpu) * Clear KVM_LARCH_PMU if the guest is not using PMU CSRs when * exiting the guest, so that the next time trap into the guest. * We don't need to deal with PMU CSRs contexts. + * + * Otherwise set request bit KVM_REQ_PMU to restore guest PMU + * before entering guest VM */ val = kvm_read_sw_gcsr(csr, LOONGARCH_CSR_PERFCTRL0); val |= kvm_read_sw_gcsr(csr, LOONGARCH_CSR_PERFCTRL1); @@ -139,6 +142,8 @@ static void kvm_lose_pmu(struct kvm_vcpu *vcpu) val |= kvm_read_sw_gcsr(csr, LOONGARCH_CSR_PERFCTRL3); if (!(val & KVM_PMU_EVENT_ENABLED)) vcpu->arch.aux_inuse &= ~KVM_LARCH_PMU; + else + kvm_make_request(KVM_REQ_PMU, vcpu); kvm_restore_host_pmu(vcpu); } -- 2.39.3 PMU hardware about VM is switched on VM exit to host rather than vCPU context sched off, PMU is checked and restored on return to VM. It is not necessary to check PMU on vCPU context sched on callback, since the request is made on VM exit entry or VM PMU CSR access abort routine already. Signed-off-by: Bibo Mao --- arch/loongarch/kvm/vcpu.c | 14 ++++---------- 1 file changed, 4 insertions(+), 10 deletions(-) diff --git a/arch/loongarch/kvm/vcpu.c b/arch/loongarch/kvm/vcpu.c index bf56ad29ac15..dc35bf526fe9 100644 --- a/arch/loongarch/kvm/vcpu.c +++ b/arch/loongarch/kvm/vcpu.c @@ -148,12 +148,6 @@ static void kvm_lose_pmu(struct kvm_vcpu *vcpu) kvm_restore_host_pmu(vcpu); } -static void kvm_restore_pmu(struct kvm_vcpu *vcpu) -{ - if ((vcpu->arch.aux_inuse & KVM_LARCH_PMU)) - kvm_make_request(KVM_REQ_PMU, vcpu); -} - static void kvm_check_pmu(struct kvm_vcpu *vcpu) { if (kvm_check_request(KVM_REQ_PMU, vcpu)) { @@ -304,7 +298,10 @@ static int kvm_pre_enter_guest(struct kvm_vcpu *vcpu) vcpu->arch.aux_inuse &= ~KVM_LARCH_SWCSR_LATEST; if (kvm_request_pending(vcpu) || xfer_to_guest_mode_work_pending()) { - kvm_lose_pmu(vcpu); + if (vcpu->arch.aux_inuse & KVM_LARCH_PMU) { + kvm_lose_pmu(vcpu); + kvm_make_request(KVM_REQ_PMU, vcpu); + } /* make sure the vcpu mode has been written */ smp_store_mb(vcpu->mode, OUTSIDE_GUEST_MODE); local_irq_enable(); @@ -1609,9 +1606,6 @@ static int _kvm_vcpu_load(struct kvm_vcpu *vcpu, int cpu) kvm_restore_timer(vcpu); kvm_make_request(KVM_REQ_STEAL_UPDATE, vcpu); - /* Restore hardware PMU CSRs */ - kvm_restore_pmu(vcpu); - /* Don't bother restoring registers multiple times unless necessary */ if (vcpu->arch.aux_inuse & KVM_LARCH_HWCSR_USABLE) return 0; -- 2.39.3