Re: [V2 PATCH 0/6] x86, NMI: give NMI handler a face-lift

[Peter Zijlstra <peterz@infradead.org>]

On Fri, 2010-11-19 at 17:59 -0500, Don Zickus wrote:
> +++ b/arch/x86/kernel/cpu/perf_event.c
> @@ -381,6 +381,19 @@ static void release_pmc_hardware(void) {}
>  
>  #endif
>  
> +static bool check_hw_exists(void)
> +{
> +       u64 val, val_new;
> +
> +       val = 0xabcdUL;
> +       (void) checking_wrmsrl(x86_pmu.perfctr, val);
> +       rdmsrl(x86_pmu.perfctr, val_new);


Yeah, this looks about right, although for extreme prudence I'd also use
a checking_rdmsrl().


> +       if (val != val_new)
> +               return false;
> +
> +       return true;
> +}
> +
>  static void reserve_ds_buffers(void);
>  static void release_ds_buffers(void);
>  
> @@ -1371,6 +1385,12 @@ void __init init_hw_perf_events(void)
>  
>         pmu_check_apic();
>  
> +       /* sanity check that the hardware exists or is emulated */
> +       if (!check_hw_exists()) {
> +               pr_cont("no PMU driver, software events only.\n");
> +               return;
> +       } 

Maybe report something like this: 
  "Broken PMU hardware detected, software events only."

Because this is really not something that's supposed to happen.