>> why ? we would stop claiming IPIs in spapr_irq_init() and so >> they won't appear as being valid anymore, at boot time or >> restore time. >> > > If we don't claim the IPIs in spapr_irq_init() anymore then > we must at least claim them on the path of H_INT_GET_SOURCE_INFO > otherwise it will fail with H_P2 and the guest won't even > try to setup the IPI.
IPIs could be handled a special case as we know in which range they are. > Even if we do that, we still have a > window where the source is valid in QEMU but not yet at > the KVM level. May be some kind of lazy binding with KVM would help us. The problem is that the guest ESB pages won't be backed by HW pages which could lead to a crash. We could overlap the emulated one until we are bound to KVM. Just a thought. C.
