>>> On 22.12.15 at 03:54, <huaitong....@intel.com> wrote: > On Mon, 2015-12-21 at 08:07 -0700, Jan Beulich wrote: >> > > > On 21.12.15 at 08:21, <huaitong....@intel.com> wrote: >> > @@ -4600,6 +4600,14 @@ void hvm_cpuid(unsigned int input, unsigned >> > int *eax, unsigned int *ebx, >> > /* Don't expose INVPCID to non-hap hvm. */ >> > if ( (count == 0) && !hap_enabled(d) ) >> > *ebx &= ~cpufeat_mask(X86_FEATURE_INVPCID); >> > + >> > + /* X86_FEATURE_PKU is not yet implemented for shadow >> > paging. */ >> > + if ( (count == 0) && !hap_enabled(d) ) >> > + *ecx &= ~cpufeat_mask(X86_FEATURE_PKU); >> >> I'm still missing the xsave dependency here. > Xsave dependency deletion becasue we use RDPKRU to get PKRU register > value instead of XSAVE now.
What the hypervisor does doesn't matter here. The question is whether from an architectural standpoint XSAVE is a prerequsite. If it is, then you need to clear PKU when _guest_ XSAVE is clear. Jan _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel
