Re: [XEN PATCH] x86/msr: Allow hardware domain to read package C-state residency counters

[Marek Marczykowski-Górecki <marmarek@xxxxxxxxxxxxxxxxxxxxxx>]

On Wed, Jul 19, 2023 at 12:38:39AM +0200, Simon Gaiser wrote:
> Jan Beulich:
> > On 18.07.2023 15:17, Simon Gaiser wrote:
> >> --- a/xen/arch/x86/pv/emul-priv-op.c
> >> +++ b/xen/arch/x86/pv/emul-priv-op.c
> >> @@ -965,6 +965,20 @@ static int cf_check read_msr(
> >>          *val = 0;
> >>          return X86EMUL_OKAY;
> >>  
> >> +    case MSR_PKG_C2_RESIDENCY:
> >> +    case MSR_PKG_C3_RESIDENCY:
> >> +    case MSR_PKG_C6_RESIDENCY:
> >> +    case MSR_PKG_C7_RESIDENCY:
> >> +    case MSR_PKG_C8_RESIDENCY:
> >> +    case MSR_PKG_C9_RESIDENCY:
> >> +    case MSR_PKG_C10_RESIDENCY:
> >> +        if ( boot_cpu_data.x86_vendor != X86_VENDOR_INTEL )
> >> +            break;
> >> +        if ( is_hardware_domain(currd) )
> >> +            goto normal;
> >> +        *val = 0;
> >> +        return X86EMUL_OKAY;
> > 
> > In addition to what Andrew said: Why would we suddenly allow these
> > reads to succeed for DomU-s?
> 
> That patch wouldn't actually allow those reads, but fake a 0 response,
> or do I miss something. If you mean that behavior: I just mirrored what
> is done there in some of the other cases. If you prefer something else,
> for example treating it as unimplemented, I can change that.

I think Jan meant exactly this difference - faking 0, instead of
failing the read.

-- 
Best Regards,
Marek Marczykowski-Górecki
Invisible Things Lab

Attachment: signature.asc
Description: PGP signature