>> >> On Sun, 2018-01-28 at 12:40 -0800, Andy Lutomirski wrote: >> > >> > Do you mean that the host would intercept the guest WRMSR and do >> > WRMSR itself? I would suggest that doing so is inconsistent with the >> > docs. As specified, doing WRMSR to write 1 to IBRS does *not* >> > protect the guest. >> >> I believe it does. Guest kernel is protected from any guest userspace >> predictions learned before IBRS was last set to 1 in *any* mode, >> including host. > > the specification requires you to write a 1 on each transition to higher > privilege. Right. Andy's concern was about VMX non-root (i.e. guest) ring 0 attempting to write IBRS but it's trapped and actually happens in the host. As long as it *remains* set when the host re-enters the vCPU that should be fine. -- dwmw2
