Re: [Xen-devel] [PATCH v12 for-xen-4.5 11/20] x86/VPMU: Interface for setting PMU mode and flags

[Boris Ostrovsky <boris.ostrovsky@xxxxxxxxxx>]

On 09/26/2014 06:00 PM, Daniel De Graaf wrote:
> On 09/25/2014 03:28 PM, Boris Ostrovsky wrote:
> > Add runtime interface for setting PMU mode and flags. Three main 
> > modes are
> > provided:
> > * XENPMU_MODE_OFF:  PMU is not virtualized
> > * XENPMU_MODE_SELF: Guests can access PMU MSRs and receive PMU 
> > interrupts.
> > * XENPMU_MODE_HV: Same as XENPMU_MODE_SELF for non-proviledged 
> > guests, dom0
> >    can profile itself and the hypervisor.
> >
> > Note that PMU modes are different from what can be provided at Xen's 
> > boot line
> > with 'vpmu' argument. An 'off' (or '0') value is equivalent to 
> > XENPMU_MODE_OFF.
> > Any other value, on the other hand, will cause VPMU mode to be set to
> > XENPMU_MODE_SELF during boot.
> >
> > For feature flags only Intel's BTS is currently supported.
> >
> > Mode and flags are set via HYPERVISOR_xenpmu_op hypercall.
> >
> > Signed-off-by: Boris Ostrovsky <boris.ostrovsky@xxxxxxxxxx>
> > Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>
> Do you think it would be useful for some service domain in a 
> disaggregated
> system to be able to query but not modify these PMU settings (i.e. 
> only use
> XENPMU_mode_get and XENPMU_feature_get)?  If this might be useful, then
> splitting up the permission checks to use two bits (pmu_get, pmu_set) is
> preferred.  However, I don't want to suggest this split if it will 
> never be
> useful - and I think that's the case here.  If you agree, then:
>
> Acked-by: Daniel De Graaf <dgdegra@xxxxxxxxxxxxx>

I can't think of a need to use PMU in such a scenario, to be honest.

At least now. Eventually, when we extend its use to profile full system, 
that might be useful.

-boris


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel