On 19/06/20 17:39, Mohammed Gamal wrote: > The last 3 patches (i.e. SVM bits and patch 11) are not intended for > immediate inclusion and probably need more discussion. > We've been noticing some unexpected behavior in handling NPF vmexits > on AMD CPUs (see individual patches for details), and thus we are > proposing a workaround (see last patch) that adds a capability that > userspace can use to decide who to deal with hosts that might have > issues supprting guest MAXPHYADDR < host MAXPHYADDR. I think patch 11 can be committed (more or less). You would actually move it to the beginning of the series and have "allow_smaller_maxphyaddr = !enable_ept;" for VMX. It is then changed to "allow_smaller_maxphyaddr = true;" in "KVM: VMX: Add guest physical address check in EPT violation and misconfig". In fact, it would be a no-brainer to commit patch 11 in that form, so feel free to submit it separately. Thanks, Paolo
