Re: [PATCH v5 2/5] KVM: x86: Add IBPB support

[David Woodhouse]

On Wed, 2018-01-31 at 11:45 -0800, Jim Mattson wrote:
> On Wed, Jan 31, 2018 at 11:37 AM, KarimAllah Ahmed  wrote:
> 
> > +ÂÂÂÂÂÂ nested_vmx_disable_intercept_for_msr(msr_bitmap_l1, msr_bitmap_l0,
> > +ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ MSR_IA32_PRED_CMD,
> > +ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ MSR_TYPE_W);
> > +
> 
> I still think this should be predicated on L1 having
> guest_cpuid_has(vcpu, X86_FEATURE_IBPB) or guest_cpuid_has(vcpu,
> X86_FEATURE_SPEC_CTRL), because of the potential impact to the
> hypertwin. If L0 denies the feature to L1 by clearing those CPUID
> bits, L1 shouldn't be able to bypass that restriction by launching L2.

Rather than doing the expensive guest_cpu_has() every time (which is
worse now as we realised we need two of them) perhaps we should
introduce a local flag for that too?

Attachment: smime.p7s
Description: S/MIME cryptographic signature