Re: [PATCH v3 7/9] KVM-GST: KVM Steal time accounting

[Peter Zijlstra <peterz@xxxxxxxxxxxxx>]

On Wed, 2011-06-29 at 11:29 -0400, Glauber Costa wrote:
> +static noinline bool touch_steal_time(int is_idle)

That noinline is very unlucky there,

> +{
> +       u64 steal, st = 0;
> +
> +       if (static_branch(&paravirt_steal_enabled)) {
> +
> +               steal = paravirt_steal_clock(smp_processor_id());
> +
> +               steal -= this_rq()->prev_steal_time;
> +
> +               st = steal_ticks(steal);
> +               this_rq()->prev_steal_time += st * TICK_NSEC;
> +
> +               if (is_idle || st == 0)
> +                       return false;
> +
> +               account_steal_time(st);
> +               return true;
> +       }
> +       return false;
> +}
> +
>  static void update_rq_clock_task(struct rq *rq, s64 delta)
>  {
>         s64 irq_delta;
> @@ -3716,6 +3760,9 @@ void account_user_time(struct task_struct *p,
> cputime_t cputime,
>         struct cpu_usage_stat *cpustat = &kstat_this_cpu.cpustat;
>         cputime64_t tmp;
>  
> +       if (touch_steal_time(0))
> +               return; 

Means we have an unconditional call here, even if the static_branch() is
patched out.
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html