Hi, It looks like the 4.14 linux kernel has something which breaks postcopy; 4.13 and 4.15 both seem OK, but both Peter and myself ran into problems with double faults after postcopy migration on a Fedora 27 host, and I've recreated the same problem on upstream kernels from v4.14.0 through v4.14.0. The stand alone userfaultfd tests seem ok though.
I might try some further bisecting, but for now I think I'll just flip to a bleeding edge 4.15. Dave -- Dr. David Alan Gilbert / dgilb...@redhat.com / Manchester, UK
