Re: [PATCH V2 4/5] idle: Move idle conditions in cpuidle_idle main function

[Peter Zijlstra <peterz@infradead.org>]

On Mon, Feb 24, 2014 at 02:55:50PM +0100, Daniel Lezcano wrote:
> @@ -136,25 +155,8 @@ static void cpu_idle_loop(void)
>  			local_irq_disable();
>  			arch_cpu_idle_enter();
>  
> -			/*
> -			 * In poll mode we reenable interrupts and spin.
> -			 *
> -			 * Also if we detected in the wakeup from idle
> -			 * path that the tick broadcast device expired
> -			 * for us, we don't want to go deep idle as we
> -			 * know that the IPI is going to arrive right
> -			 * away
> -			 */
> -			if (cpu_idle_force_poll || tick_check_broadcast_expired()) {
> -				cpu_idle_poll();
> -			} else {
> -				if (!current_clr_polling_and_test()) {
> -					cpuidle_idle_call();
> -				} else {
> -					local_irq_enable();
> -				}
> -				__current_set_polling();
> -			}
> +			cpuidle_idle_call();
> +

Yeah, not liking that much; you can make it look like:

	if (cpu_idle_force_poll || tick_check_broadcast_expired())
		cpu_idle_poll();
	else
		cpu_idle_call();

Though. That keeps the polling case separate from the actual idle
function.


And when you do that; you can also push down the
current_clr_polling_and_test() muck so it doesn't cover the actual
cpuidle policy code.