Re: [PATCH v2 1/2] x86/hvm: allow guest to use clflushopt and clwb

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

On 01/07/16 07:34, Jan Beulich wrote:
> >>> On 07.01.16 at 15:01, <andrew.cooper3@citrix.com> wrote:
> > On 07/01/16 13:49, Jan Beulich wrote:
> >>>>> On 30.12.15 at 12:48, <haozhong.zhang@intel.com> wrote:
> >>> --- a/xen/arch/x86/hvm/hvm.c
> >>> +++ b/xen/arch/x86/hvm/hvm.c
> >>> @@ -4583,21 +4583,30 @@ void hvm_cpuid(unsigned int input, unsigned int *eax, 
> > unsigned int *ebx,
> >>>              *edx &= ~cpufeat_mask(X86_FEATURE_PSE36);
> >>>          break;
> >>>      case 0x7:
> >>> -        if ( (count == 0) && !cpu_has_smep )
> >>> -            *ebx &= ~cpufeat_mask(X86_FEATURE_SMEP);
> >>> +        if ( count == 0 )
> >>> +        {
> >>> +            if ( !cpu_has_smep )
> >>> +                *ebx &= ~cpufeat_mask(X86_FEATURE_SMEP);
> >>> +
> >>> +            if ( !cpu_has_smap )
> >>> +                *ebx &= ~cpufeat_mask(X86_FEATURE_SMAP);
> >>> +
> >>> +            /* Don't expose MPX to hvm when VMX support is not available */
> >>> +            if ( !(vmx_vmexit_control & VM_EXIT_CLEAR_BNDCFGS) ||
> >>> +                 !(vmx_vmentry_control & VM_ENTRY_LOAD_BNDCFGS) )
> >>> +                *ebx &= ~cpufeat_mask(X86_FEATURE_MPX);
> >>>  
> >>> -        if ( (count == 0) && !cpu_has_smap )
> >>> -            *ebx &= ~cpufeat_mask(X86_FEATURE_SMAP);
> >>> +            /* Don't expose INVPCID to non-hap hvm. */
> >>> +            if ( !hap_enabled(d) )
> >>> +                *ebx &= ~cpufeat_mask(X86_FEATURE_INVPCID);
> >>>  
> >>> -        /* Don't expose MPX to hvm when VMX support is not available */
> >>> -        if ( (count == 0) &&
> >>> -             (!(vmx_vmexit_control & VM_EXIT_CLEAR_BNDCFGS) ||
> >>> -              !(vmx_vmentry_control & VM_ENTRY_LOAD_BNDCFGS)) )
> >>> -            *ebx &= ~cpufeat_mask(X86_FEATURE_MPX);
> >>> +            if ( !cpu_has_clflushopt )
> >>> +                *ebx &= ~cpufeat_mask(X86_FEATURE_CLFLUSHOPT);
> >>> +
> >>> +            if ( !cpu_has_clwb )
> >>> +                *ebx &= ~cpufeat_mask(X86_FEATURE_CLWB);
> >> I don't think we need this: Other than other things adjusted here,
> >> there's nothing disabling these two features when found available,
> >> and there are no extra conditions to consider. Otherwise, if we
> >> were to follow this route, quite a bit of code would need to be
> >> added to other case statements in this function. But that's all (I
> >> think) going to be taken care of by Andrew's CPUID leveling series.
> > 
> > My series does take care of all of this.
> > 
> > I would prefer that these two changes get taken as soon as they are
> > ready, so I can rebase.
> 
> If we don't need the change quoted above, your rebase will actually
> be easier to do.
> 
> Jan
>
I'll remove changes in hvm_cpuid() in the next version. Changes in
xen/include/asm-x86/cpufeature.h will be removed as well, because
there will be no use of them.

Haozhong