On Tue, May 25, 2021 at 12:24 PM Reiji Watanabe <reijiw@xxxxxxxxxx> wrote:
>
> > @@ -9880,11 +9888,16 @@ int kvm_arch_vcpu_ioctl_get_sregs(struct kvm_vcpu *vcpu,
> > int kvm_arch_vcpu_ioctl_get_mpstate(struct kvm_vcpu *vcpu,
> > struct kvm_mp_state *mp_state)
> > {
> > + int r = 0;
> > +
> > vcpu_load(vcpu);
> > if (kvm_mpx_supported())
> > kvm_load_guest_fpu(vcpu);
> >
> > - kvm_apic_accept_events(vcpu);
> > + r = kvm_apic_accept_events(vcpu);
> > + if (r < 0)
> > + goto out;
> > +
> > if ((vcpu->arch.mp_state == KVM_MP_STATE_HALTED ||
> > vcpu->arch.mp_state == KVM_MP_STATE_AP_RESET_HOLD) &&
> > vcpu->arch.pv.pv_unhalted)
> > @@ -9892,6 +9905,7 @@ int kvm_arch_vcpu_ioctl_get_mpstate(struct kvm_vcpu *vcpu,
> > else
> > mp_state->mp_state = vcpu->arch.mp_state;
> >
> > +out:
> > if (kvm_mpx_supported())
> > kvm_put_guest_fpu(vcpu);
> > vcpu_put(vcpu);
>
> With the change, if the return value from kvm_apic_accept_events()
> is < 0, kvm_arch_vcpu_ioctl_get_mpstate(), which is called from
> KVM_GET_MP_STATE ioctl, doesn't set mp_state returning 0 (success).
> It leads KVM_GET_MP_STATE ioctl to return an undefined mp_state for
> the success case.
Yikes! I think I intended to return 'r' when it is less than 0 (e.g.
the -ENXIO I introduce later in the series). However, I'm not quite
sure what to do with values of r > 0. I'll look into it and send out
v2.
