Re: [PATCH v1 3/8] KVM: x86: Allocate performance counter for PEBS event

[Peter Zijlstra]

On Wed, Oct 30, 2019 at 08:42:10AM +0200, Alexander Shishkin wrote:
> "Kang, Luwei" <luwei.kang@xxxxxxxxx> writes:
> 
> >> >  static void pmc_reprogram_counter(struct kvm_pmc *pmc, u32 type,
> >> >  				  unsigned config, bool exclude_user,
> >> >  				  bool exclude_kernel, bool intr,
> >> > -				  bool in_tx, bool in_tx_cp)
> >> > +				  bool in_tx, bool in_tx_cp, bool pebs)
> >> >  {
> >> >  	struct perf_event *event;
> >> >  	struct perf_event_attr attr = {
> >> > @@ -111,9 +111,12 @@ static void pmc_reprogram_counter(struct kvm_pmc *pmc, u32 type,
> >> >  		.exclude_user = exclude_user,
> >> >  		.exclude_kernel = exclude_kernel,
> >> >  		.config = config,
> >> > +		.precise_ip = pebs ? 1 : 0,
> >> > +		.aux_output = pebs ? 1 : 0,
> >> 
> >> srsly?
> >
> > Hi Peter,
> >     Thanks for review. For aux_output, I think it should be set 1 when the guest wants to enabled PEBS by Intel PT.
> 
> attr.aux_output==1 means your group leader should be an intel_pt event
> for this to succeed. Luckily for this instance,
> perf_event_create_kernel_counter() doesn't actually check the
> attr.aux_output.
> 
> Also, does 'bool pebs' mean PEBS-via-PT or just a PEBS counter? Or does
> it mean that in kvm it's the same thing?

It means pebs-over-pt, they only allow pebs if the output to pt bit is
set.