Re: [PATCH 1/2] cpu/hotplug: Add cpuhp_latency trace event

[Thomas Gleixner]

On Wed, Sep 23 2020 at 16:37, Prasad Sodagudi wrote:
> Add ftrace event trace_cpuhp_latency to track cpu
> hotplug latency. It helps to track the hotplug latency
> impact by firmware changes and kernel cpu hotplug callbacks.

Why?

And even if that makes sense, the implementation makes absolutely no
sense at all.

      trace_cpuhp_callback_enter();
      callback();
      trace_cpuhp_callback_exit();

No point in all this start time and conditional sched_clock() muck.

But then again, hotplug is slow by definition and nobody cares.

Thanks,

        tglx