Re: [RFC PATCH] perf: Provide status of known PMUs

[Adrian Hunter <adrian.hunter@intel.com>]

On 09/07/15 11:50, Peter Zijlstra wrote:
> On Thu, Jul 09, 2015 at 10:48:00AM +0300, Adrian Hunter wrote:
>> Known PMUs may not be present for various reasons.
>> Provide a way for the user to know what the reason
>> is.
> 
> Not a bad idea, but I do wonder where we should draw the line on what is
> 'known'. The patch as proposed will have bts/pt listed as 'known' for
> every arch out there.
> 
> By that logic, x86 should list the ppc/sparc/mips/arm/etc.. PMUs as
> known and wrong_arch too, which might be a tad excessive.
> 
> Can we limit it to PMUs for which we've (attempted to) load the drivers?
> That would obviously make a few of your status bits redundant, but then
> you've not explained why we're interested in it.
> 
>> 	Supported
>> 	Driver error
>> 	Driver not loaded
>> 	Not supported by hardware
>> 	Wrong vendor
>> 	Unknown status
> 
> There would work.
> 
>> 	Driver not in kernel config
>> 	Not supported by kernel
>> 	Wrong architecture
> 
> These will be hard, for if we don't load the driver we don't 'know' of
> them.

Are they that hard?

The architecture one is done by perf core at the moment when the initial
status is defined e.g.

#if defined(CONFIG_X86)
#define PERF_PMU_STATUS_ARCH_X86	PERF_PMU_STATUS_UNKNOWN
#else
#define PERF_PMU_STATUS_ARCH_X86	PERF_PMU_STATUS_WRONG_ARCH
#endif

static struct known_pmu known_pmus[] = {
	KNOWN_PMU("intel_pt", "Intel", PERF_PMU_STATUS_ARCH_X86),
	KNOWN_PMU("intel_bts", "Intel", PERF_PMU_STATUS_ARCH_X86),
	KNOWN_PMU(NULL, NULL, 0),
};

For config and module the arch perf init code can check e.g.

#if !defined(CONFIG_MY_PMU) && !defined(CONFIG_MY_PMU_MODULE)
	perf_pmu_update_status("my_pmu", PERF_PMU_STATUS_NOT_CONFIG, NULL);
#elif defined(CONFIG_MY_PMU_MODULE)
	perf_pmu_update_status("my_pmu", PERF_PMU_STATUS_NOT_LOADED, NULL);
#endif