[PATCH] cpuidle: undelaying cpuidle in dpm_{suspend|resume}()

[PATCH] cpuidle: undelaying cpuidle in dpm_{suspend|resume}()

[Ikjoon Jang]

cpuidle is paused only during dpm_suspend_noirq() ~ dpm_resume_noirq().
But some device drivers need random sized IOs in dpm_{suspend|resume}()
stage (e.g. re-downloading firmware in resume).
And with such a device, cpuidle's latencies could be critical to
response time of system suspend/resume.

To minimize those latencies, we could apply pm_qos to such device drivers,
but simply undelaying cpuidle from dpm_suspend_noirq() to dpm suspend()
seems no harm.
---
 drivers/base/power/main.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/drivers/base/power/main.c b/drivers/base/power/main.c
index 134a8af51511..5928dd2139e8 100644
--- a/drivers/base/power/main.c
+++ b/drivers/base/power/main.c
@@ -773,8 +773,6 @@ void dpm_resume_noirq(pm_message_t state)
 
 	resume_device_irqs();
 	device_wakeup_disarm_wake_irqs();
-
-	cpuidle_resume();
 }
 
 static pm_callback_t dpm_subsys_resume_early_cb(struct device *dev,
@@ -1069,6 +1067,7 @@ void dpm_resume(pm_message_t state)
 
 	cpufreq_resume();
 	devfreq_resume();
+	cpuidle_resume();
 	trace_suspend_resume(TPS("dpm_resume"), state.event, false);
 }
 
@@ -1411,8 +1410,6 @@ int dpm_suspend_noirq(pm_message_t state)
 {
 	int ret;
 
-	cpuidle_pause();
-
 	device_wakeup_arm_wake_irqs();
 	suspend_device_irqs();
 
@@ -1830,6 +1827,7 @@ int dpm_suspend(pm_message_t state)
 	trace_suspend_resume(TPS("dpm_suspend"), state.event, true);
 	might_sleep();
 
+	cpuidle_pause();
 	devfreq_suspend();
 	cpufreq_suspend();
 
-- 
2.24.0.rc0.303.g954a862665-goog

Re: [PATCH] cpuidle: undelaying cpuidle in dpm_{suspend|resume}()

[Greg Kroah-Hartman]

On Tue, Oct 29, 2019 at 03:31:45PM +0800, Ikjoon Jang wrote:
> cpuidle is paused only during dpm_suspend_noirq() ~ dpm_resume_noirq().
> But some device drivers need random sized IOs in dpm_{suspend|resume}()
> stage (e.g. re-downloading firmware in resume).
> And with such a device, cpuidle's latencies could be critical to
> response time of system suspend/resume.
> 
> To minimize those latencies, we could apply pm_qos to such device drivers,
> but simply undelaying cpuidle from dpm_suspend_noirq() to dpm suspend()
> seems no harm.
> ---
>  drivers/base/power/main.c | 6 ++----
>  1 file changed, 2 insertions(+), 4 deletions(-)

no signed-off-by?

:(