Re: [Qemu-devel] [PATCH v3 1/2] target-i386: KVM: add basic Intel LMCE support

[Haozhong Zhang <haozhong.zhang@intel.com>]

On 06/03/16 17:57, Radim Krčmář wrote:
> 2016-06-03 14:09+0800, Haozhong Zhang:
> > This patch adds the support to inject SRAR and SRAO as LMCE, i.e. they
> > will be injected to only one VCPU rather than broadcast to all
> > VCPUs. As KVM reports LMCE support on Intel platforms, this features is
> > only available on Intel platforms.
> > 
> > Signed-off-by: Ashok Raj <ashok.raj@intel.com>
> > Signed-off-by: Haozhong Zhang <haozhong.zhang@intel.com>
> > ---
> > diff --git a/target-i386/cpu.c b/target-i386/cpu.c
> > @@ -2786,6 +2798,20 @@ static void mce_init(X86CPU *cpu)
> >          && (cenv->features[FEAT_1_EDX] & (CPUID_MCE | CPUID_MCA)) ==
> >              (CPUID_MCE | CPUID_MCA)) {
> >          cenv->mcg_cap = MCE_CAP_DEF | MCE_BANKS_DEF;
> > +
> > +        if (cpu->enable_lmce) {
> > +            if (lmce_supported()) {
> > +                cenv->mcg_cap |= MCG_LMCE_P;
> > +                cenv->msr_ia32_feature_control |=
> > +                    MSR_IA32_FEATURE_CONTROL_LMCE |
> > +                    MSR_IA32_FEATURE_CONTROL_LOCKED;
> 
> Locking right from the start breaks nested KVM, because nested relies on
> setting VMXON feature from inside of the guest.
> 
> Do we keep it unlocked, or move everything into QEMU?
> 
> (The latter seems simpler.)
>

Setting guest MSR_IA32_FEATURE_CONTROL is not necessary here, it's
instead the guest BIOS/OS duty to enable and lock corresponding
features. I'll remove this in the next version.

Thanks,
Haozhong