On 03/01/2018 03:36 PM, David Hildenbrand wrote: >> I have a concern with this proposition and with the original code: > Very good, this is exactly what I talked to Conny about yesterday. > > Short version: CPU model is guest ABI, everything else is configuration. >
Nod. > > So I think for now we are fine. Later, this might be tricky to check but > not impossible. > I think we are all in agreement on the important stuff. I think, we also all agree, that certain things need to be improved. E.g. the KVM code manipulating ECA.28, -cpu xxx,ap=on needs to imply no Operation Exception when guest executes an AP instruction (this is currently not the case as we can have a vm a vfio-ap device -- so open won't get called -- but with ap=on). Hope Tony will address these in the next version. Regards, Halil
