Re: [PATCH v3 1/5] perf/x86: add a function to get the lbr stack

[Peter Zijlstra]

On Thu, Sep 20, 2018 at 06:05:55PM +0800, Wei Wang wrote:
> diff --git a/arch/x86/events/intel/lbr.c b/arch/x86/events/intel/lbr.c
> index c88ed39..c81f160 100644
> --- a/arch/x86/events/intel/lbr.c
> +++ b/arch/x86/events/intel/lbr.c
> @@ -1277,3 +1277,26 @@ void intel_pmu_lbr_init_knl(void)
>  	if (x86_pmu.intel_cap.lbr_format == LBR_FORMAT_LIP)
>  		x86_pmu.intel_cap.lbr_format = LBR_FORMAT_EIP_FLAGS;
>  }
> +
> +/**
> + * perf_get_lbr_stack - get the lbr stack related MSRs
> + *
> + * @stack: the caller's memory to get the lbr stack
> + *
> + * Returns: 0 indicates that the lbr stack has been successfully obtained.
> + */
> +int perf_get_lbr_stack(struct perf_lbr_stack *stack)
> +{
> +	stack->nr = x86_pmu.lbr_nr;
> +	stack->tos = x86_pmu.lbr_tos;
> +	stack->from = x86_pmu.lbr_from;
> +	stack->to = x86_pmu.lbr_to;
> +
> +	if (x86_pmu.intel_cap.lbr_format == LBR_FORMAT_INFO)
> +		stack->info = MSR_LBR_INFO_0;
> +	else
> +		stack->info = 0;
> +
> +	return 0;
> +}
> +EXPORT_SYMBOL_GPL(perf_get_lbr_stack);

Blergh, I know KVM is a module but it really sucks having to export
everything :/

> diff --git a/arch/x86/include/asm/perf_event.h b/arch/x86/include/asm/perf_event.h
> index 12f5408..84cc8cb 100644
> --- a/arch/x86/include/asm/perf_event.h
> +++ b/arch/x86/include/asm/perf_event.h
> @@ -267,7 +267,16 @@ struct perf_guest_switch_msr {
>  	u64 host, guest;
>  };
>  
> +struct perf_lbr_stack {
> +	int		nr;

Do we need a negative number of LBR entries?

> +	unsigned long	tos;
> +	unsigned long	from;
> +	unsigned long	to;
> +	unsigned long	info;

These are all MSR values, that can be 'unsigned int', right? If so,
please also fix struct x86_pmu for that.

> +};
> +
>  extern struct perf_guest_switch_msr *perf_guest_get_msrs(int *nr);
> +extern int perf_get_lbr_stack(struct perf_lbr_stack *stack);
>  extern void perf_get_x86_pmu_capability(struct x86_pmu_capability *cap);
>  extern void perf_check_microcode(void);
>  #else

I would like to use the x86_perf namespace or something along those
lines, this is very much not a generic perf interface -- it is very much
x86 (or rather even Intel) specific.