On Thu, 2018-03-22 at 18:09 +0100, Rafael J. Wysocki wrote:
>
> +++ linux-pm/drivers/cpuidle/poll_state.c
> @@ -10,6 +10,7 @@
> #include <linux/sched/idle.h>
>
> #define POLL_IDLE_TIME_LIMIT (TICK_NSEC / 16)
> +#define POLL_IDLE_COUNT 1000
>
> static int __cpuidle poll_idle(struct cpuidle_device *dev,
> struct cpuidle_driver *drv, int
> index)
> @@ -18,9 +19,14 @@ static int __cpuidle poll_idle(struct cp
>
> local_irq_enable();
> if (!current_set_polling_and_test()) {
> + unsigned int loop_count = 0;
> +
> while (!need_resched()) {
> cpu_relax();
> + if (loop_count++ < POLL_IDLE_COUNT)
> + continue;
>
> + loop_count = 0;
> if (local_clock() - time_start >
> POLL_IDLE_TIME_LIMIT)
> break;
> }
OK, I am still seeing a performance
degradation with the above, though
not throughout the entire workload.
It appears that making the idle loop
do anything besides cpu_relax() for
a significant amount of time slows
things down.
I plan to try two more things:
1) Disable polling on SMT systems, with
the idea that putting one thread to
sleep with monitor/mwait in C1 will
allow the other thread to run faster.
2) Insert more cpu_relax() calls into the
main loop, so the CPU core spends more
of its time in cpu_relax() and less
time doing other things:
static int __cpuidle poll_idle(struct cpuidle_device *dev,
struct cpuidle_driver *drv, int index)
{
u64 time_start = local_clock();
local_irq_enable();
if (!current_set_polling_and_test()) {
unsigned int loop_count = 0;
while (!need_resched()) {
cpu_relax();
cpu_relax();
cpu_relax();
cpu_relax();
cpu_relax();
cpu_relax();
cpu_relax();
cpu_relax();
if (loop_count++ < POLL_IDLE_COUNT)
continue;
loop_count = 0;
if (local_clock() - time_start > POLL_IDLE_TIME_LIMIT)
break;
}
}
current_clr_polling();
return index;
}
I will let you know how they perform.
--
All Rights Reversed.
signature.asc
Description: This is a digitally signed message part
