Re: Dell XPS 13 9360 (Kaby Lake): Power button LED stays sporadically trying to S0ix suspend

["Zhang, Rui" <rui.zhang@intel.com>]

Hi, Paul,

Thanks for reporting this.

The problem is that, the system may fail to enter S0ix if it is
suspended with high temperature.
So a delay loop in the driver suspend code is introduced to allow the
system to cool down, ef63b043ac86 ("thermal: intel: pch: fix S0ix
failure due to PCH temperature above threshold"). And then enhanced by
this commit 92923028e979 ("thermal: intel: pch: enhance overheat
handling")

> 
> The difference is:
> 
> 1.  LED stays on:
> 
>          intel_pch_thermal 0000:00:14.2: Wakeup event detected, abort
> cooling

     [ 2918.091771] ACPI: EC: interrupt blocked
     [ 2956.317686] intel_pch_thermal 0000:00:14.2: Wakeup event
detected, abort cooling

so the system doesn't go to a safe temperature after 40+ seconds, and a
wakeup event is fired and s2idle is aborted.

> 
> 2.  LED turns off:
> 
>          intel_pch_thermal 0000:00:14.2: CPU-PCH is cool [44C]
> 
     [ 7977.637987] ACPI: EC: interrupt blocked
     [ 7977.641470] intel_pch_thermal 0000:00:14.2: CPU-PCH is cool
[44C]

In this case, system is not overheat and we suspend immediately.

> Does intel_pch_thermal prevent the system from S0ix suspending, so
> the 
> LED stays on until the device is cooled down?

yes, because we are still in driver .suspend_noirq() callbacks at this
time. And we have to block suspend, or else the system can only reach
PCx after suspended.

>  If so, could this be 
> better logged, so users know, what is going on.

hmmm, any suggestions?
or how about the change below?

Thanks,
rui

---
 drivers/thermal/intel/intel_pch_thermal.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/thermal/intel/intel_pch_thermal.c
b/drivers/thermal/intel/intel_pch_thermal.c
index f5be2c389351..91a6149df350 100644
--- a/drivers/thermal/intel/intel_pch_thermal.c
+++ b/drivers/thermal/intel/intel_pch_thermal.c
@@ -298,6 +298,11 @@ static int intel_pch_thermal_suspend_noirq(struct
device *device)
 	/* Get the PCH current temperature value */
 	pch_cur_temp = GET_PCH_TEMP(WPT_TEMP_TSR & readw(ptd->hw_base
+ WPT_TEMP));
 
+	if (pch_cur_temp >= pch_thr_temp)
+		dev_info(&ptd->pdev->dev,
+			"CPU-PCH overheats [%dC], S0ix might fail.
Start cooling...\n",
+			pch_cur_temp);
+
 	/*
 	 * If current PCH temperature is higher than configured PCH
threshold
 	 * value, run some delay loop with sleep to let the current
temperature
-- 
2.34.1