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

[Peter Zijlstra <peterz@infradead.org>]

On Thu, Jul 09, 2015 at 12:30:30PM +0300, Adrian Hunter wrote:
> On 09/07/15 11:50, Peter Zijlstra wrote:

> > 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.

> >> 	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?

Because if we limit known to be what we're tried to probe, you simply do
not know about PMUs for the wrong arch or not build by the kernel etc.

> 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),
> };

So I really don't like this hard-coded table much, that's just going to
be a pain to maintain.