| Message ID | 20240809073120.250974-2-aboorvad@linux.ibm.com (mailing list archive) |
|---|---|
| State | Changes Requested, archived |
| Headers | show |
| Series | cpuidle/menu: Address performance drop from favoring physical over polling cpuidle state | expand |
On 8/9/24 08:31, Aboorva Devarajan wrote: > Update the cpuidle menu governor to avoid prioritizing physical states > over polling states when predicted idle duration is lesser than the > physical states target residency duration for performance gains. > > Signed-off-by: Aboorva Devarajan <aboorvad@linux.ibm.com> > --- > drivers/cpuidle/governors/menu.c | 11 ----------- > 1 file changed, 11 deletions(-) > > diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c > index f3c9d49f0f2a..cf99ca103f9b 100644 > --- a/drivers/cpuidle/governors/menu.c > +++ b/drivers/cpuidle/governors/menu.c > @@ -354,17 +354,6 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev, > idx = i; /* first enabled state */ > > if (s->target_residency_ns > predicted_ns) { > - /* > - * Use a physical idle state, not busy polling, unless > - * a timer is going to trigger soon enough. > - */ > - if ((drv->states[idx].flags & CPUIDLE_FLAG_POLLING) && > - s->exit_latency_ns <= latency_req && > - s->target_residency_ns <= data->next_timer_ns) { > - predicted_ns = s->target_residency_ns; > - idx = i; > - break; > - } > if (predicted_ns < TICK_NSEC) > break; > How about this? data->next_timer_ns is set to KTIME_MAX in case the last few intervals were very short, which might be the case for you. -->8-- diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c index f3c9d49f0f2a..77b40201c446 100644 --- a/drivers/cpuidle/governors/menu.c +++ b/drivers/cpuidle/governors/menu.c @@ -360,6 +360,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev, */ if ((drv->states[idx].flags & CPUIDLE_FLAG_POLLING) && s->exit_latency_ns <= latency_req && + data->next_timer_ns != KTIME_MAX && s->target_residency_ns <= data->next_timer_ns) { predicted_ns = s->target_residency_ns; idx = i;
diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c index f3c9d49f0f2a..cf99ca103f9b 100644 --- a/drivers/cpuidle/governors/menu.c +++ b/drivers/cpuidle/governors/menu.c @@ -354,17 +354,6 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev, idx = i; /* first enabled state */ if (s->target_residency_ns > predicted_ns) { - /* - * Use a physical idle state, not busy polling, unless - * a timer is going to trigger soon enough. - */ - if ((drv->states[idx].flags & CPUIDLE_FLAG_POLLING) && - s->exit_latency_ns <= latency_req && - s->target_residency_ns <= data->next_timer_ns) { - predicted_ns = s->target_residency_ns; - idx = i; - break; - } if (predicted_ns < TICK_NSEC) break;
Update the cpuidle menu governor to avoid prioritizing physical states over polling states when predicted idle duration is lesser than the physical states target residency duration for performance gains. Signed-off-by: Aboorva Devarajan <aboorvad@linux.ibm.com> --- drivers/cpuidle/governors/menu.c | 11 ----------- 1 file changed, 11 deletions(-)