On 30/07/2017 11:42, Wanpeng Li wrote: > From: Wanpeng Li <wanpeng.li@xxxxxxxxxxx> > > When SMP VM start, AP may lost INIT because of receiving INIT between > kvm_vcpu_ioctl_x86_get/set_vcpu_events. > > vcpu 0 vcpu 1 > kvm_vcpu_ioctl_x86_get_vcpu_events > events->smi.latched_init = 0 > send INIT to vcpu1 > set vcpu1's pending_events > kvm_vcpu_ioctl_x86_set_vcpu_events > if (events->smi.latched_init == 0) > clear INIT in pending_events > > This patch fixes it by not touching INIT pending if INIT is not latched > which in SMM and just restoring the latched INIT after SET_VCPU_EVENTS. > > Thanks Peng Hao for the report and original commit message. > > Reported-by: Peng Hao <peng.hao2@xxxxxxxxxx> > Cc: Paolo Bonzini <pbonzini@xxxxxxxxxx> > Cc: Radim Krčmář <rkrcmar@xxxxxxxxxx> > Signed-off-by: Wanpeng Li <wanpeng.li@xxxxxxxxxxx> > --- > arch/x86/kvm/x86.c | 8 ++------ > 1 file changed, 2 insertions(+), 6 deletions(-) > > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c > index 1a4c84d..eaca86d 100644 > --- a/arch/x86/kvm/x86.c > +++ b/arch/x86/kvm/x86.c > @@ -3164,12 +3164,8 @@ static int kvm_vcpu_ioctl_x86_set_vcpu_events(struct kvm_vcpu *vcpu, > vcpu->arch.hflags |= HF_SMM_INSIDE_NMI_MASK; > else > vcpu->arch.hflags &= ~HF_SMM_INSIDE_NMI_MASK; > - if (lapic_in_kernel(vcpu)) { > - if (events->smi.latched_init) > - set_bit(KVM_APIC_INIT, &vcpu->arch.apic->pending_events); > - else > - clear_bit(KVM_APIC_INIT, &vcpu->arch.apic->pending_events); > - } > + if (events->smi.latched_init && lapic_in_kernel(vcpu)) > + set_bit(KVM_APIC_INIT, &vcpu->arch.apic->pending_events); All this block: if (events->smi.smm_inside_nmi) vcpu->arch.hflags |= HF_SMM_INSIDE_NMI_MASK; else vcpu->arch.hflags &= ~HF_SMM_INSIDE_NMI_MASK; if (lapic_in_kernel(vcpu)) { if (events->smi.latched_init) set_bit(KVM_APIC_INIT, &vcpu->arch.apic->pending_events); else clear_bit(KVM_APIC_INIT, &vcpu->arch.apic->pending_events); } only applies if you're in SMM, so perhaps you can move it under if (events->smi.smm) ? Paolo > } > > kvm_make_request(KVM_REQ_EVENT, vcpu); >