On Wed, Dec 11, 2019 at 04:27:42PM +1100, David Gibson wrote: > Ah, right. We'll need to check for -ENOTTY specifically and ignore > it, then. We don't want this spewing warnings on every non-secure > guest.
I am posting v2 with explicit check for -ENOTTY. > > > It looks like we may need a new KVM capability to advertise the presence > > of KVM_PPC_SVM_OFF ioctl (or more generally, to advertise host kernel's > > capability to support secure guests). > > Actually, that's probably a better idea still. If and when we decide to have this KVM capability and that goes upstream, we can update the QEMU accordingly? Regards, Bharata.
