On Tue, Mar 12, 2019 at 9:13 PM Xiaoyao Li <xiaoyao.li@xxxxxxxxxxxxxxx> wrote: > Yes, I found your fix patches in the maillist just now. > I am so sorry about sending the fix patches without checking you have already > sent out fix serires, and so sorry about the reply yesterday. I might use the > wrong word, anyway, no offense. No problem. It's a word I imagine you're hearing a lot at Intel these days. :-) > And now, as you said, kvm *should* emulate IA32_ARCH_CAPABILITIES for AMD as > well as Intel. Sean's this patch fixes the most part. Also we need your patch > https://patchwork.kernel.org/patch/10784297/. to move IA32_ARCH_CAPABILITIES > from msrs_to_save[] to emulated_msrs[], while removing the 'case > MSR_IA32_ARCH_CAPABILITIES' part. > > That's enough, right? At the risk of beating a dead horse, yes, with the caveat that userspace should not be restricted from setting bits in the virtual MSR that aren't set in the physical MSR.
