Re: [PATCH v7 2/7] kvm: vmx: Define CET VMCS fields and CPUID flags

[Yang Weijiang]

On Wed, Oct 02, 2019 at 11:04:07AM -0700, Jim Mattson wrote:
> On Thu, Sep 26, 2019 at 7:17 PM Yang Weijiang <weijiang.yang@xxxxxxxxx> wrote:
> >
> > CET(Control-flow Enforcement Technology) is an upcoming Intel(R)
> > processor feature that blocks Return/Jump-Oriented Programming(ROP)
> > attacks. It provides the following capabilities to defend
> > against ROP/JOP style control-flow subversion attacks:
> >  /*
> > diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
> > index 9d282fec0a62..1aa86b87b6ab 100644
> > --- a/arch/x86/kvm/cpuid.c
> > +++ b/arch/x86/kvm/cpuid.c
> > @@ -365,13 +365,13 @@ static inline void do_cpuid_7_mask(struct kvm_cpuid_entry2 *entry, int index)
> >                 F(AVX512VBMI) | F(LA57) | F(PKU) | 0 /*OSPKE*/ |
> >                 F(AVX512_VPOPCNTDQ) | F(UMIP) | F(AVX512_VBMI2) | F(GFNI) |
> >                 F(VAES) | F(VPCLMULQDQ) | F(AVX512_VNNI) | F(AVX512_BITALG) |
> > -               F(CLDEMOTE) | F(MOVDIRI) | F(MOVDIR64B);
> > +               F(CLDEMOTE) | F(MOVDIRI) | F(MOVDIR64B) | F(SHSTK);
> >
> >         /* cpuid 7.0.edx*/
> >         const u32 kvm_cpuid_7_0_edx_x86_features =
> >                 F(AVX512_4VNNIW) | F(AVX512_4FMAPS) | F(SPEC_CTRL) |
> >                 F(SPEC_CTRL_SSBD) | F(ARCH_CAPABILITIES) | F(INTEL_STIBP) |
> > -               F(MD_CLEAR);
> > +               F(MD_CLEAR) | F(IBT);
> 
> Claiming that SHSTK and IBT are supported in the guest seems premature
> as of this change, since you haven't actually done anything to
> virtualize the features yet.
>
OK, will put the flags in other patch.
> >         /* cpuid 7.1.eax */
> >         const u32 kvm_cpuid_7_1_eax_x86_features =
> > diff --git a/arch/x86/kvm/x86.h b/arch/x86/kvm/x86.h
> > index fbffabad0370..a85800b23e6e 100644
> > --- a/arch/x86/kvm/x86.h
> > +++ b/arch/x86/kvm/x86.h
> > @@ -298,7 +298,8 @@ int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
> >   * Right now, no XSS states are used on x86 platform,
> >   * expand the macro for new features.
> >   */
> > -#define KVM_SUPPORTED_XSS      (0)
> > +#define KVM_SUPPORTED_XSS      (XFEATURE_MASK_CET_USER \
> > +                               | XFEATURE_MASK_CET_KERNEL)
> 
> If IA32_XSS can dynamically change within the guest, it will have to
> be enumerated by KVM_GET_MSR_INDEX_LIST.
thanks for pointing it out, need to add IA32_XSS to msrs_to_save list.

>(Note that Aaron Lewis is
> working on a series which will include that enumeration, if you'd like
> to wait.) I'm also not convinced that there is sufficient
> virtualization of these features to allow these bits to be set in the
> guest IA32_XSS at this point.
> 
It's true CET is working in guest after I added XSS/XSAVES support in
KVM and QEMU, but I'd like to look at Aaron's new patch...

> >  extern u64 host_xcr0;
> >
> > --
> > 2.17.2
> >