On Thu, Feb 27, 2020 at 04:28:33PM -0800, Sean Christopherson wrote: > On Tue, Feb 25, 2020 at 04:10:18PM +0100, Paolo Bonzini wrote: > > On 01/02/20 19:51, Sean Christopherson wrote: > > > + /* CPUID 0x8000000A */ > > > + /* Support next_rip if host supports it */ > > > + if (boot_cpu_has(X86_FEATURE_NRIPS)) > > > + kvm_cpu_cap_set(X86_FEATURE_NRIPS); > > > > Should this also be conditional on "nested"? > > I think that makes sense? AFAICT it should probably be conditional on > "nrips" as well. X86_FEATURE_NPT should also be conditional on "nested". > I'll tack on a patch to make those changes, the cleanup is easier without > the things spread across different case statements, e.g. wrap the entire > SVM feature leaf in "if (nested)". Regarding NRIPS, the original commit added the "Support next_rip if host supports it" comment, but I can't tell is "host supports" means "supported in hardware" or "supported by KVM". In other words, should I make the cap dependent "nrips" or leave it as is?
