[Letux-kernel] Strange things happening with latest kernels

[H. Nikolaus Schaller]

> Am 18.06.2019 um 19:22 schrieb Andreas Kemnade <andreas at kemnade.info>:
> 
> On Tue, 18 Jun 2019 15:06:49 +0200
> "H. Nikolaus Schaller" <hns at goldelico.com> wrote:
> 
>>> Am 17.06.2019 um 21:56 schrieb H. Nikolaus Schaller <hns at goldelico.com>:
>>> 
>>> Hi Andreas,
>>> 
>>>> Am 15.06.2019 um 18:46 schrieb H. Nikolaus Schaller <hns at goldelico.com>:
>>>> 
>>> 
>>> I have just made another strange observation:
>>> 
>>> login: root
>>> password:
>>> 
>>> root at letux:~# ./temperatures 
>>> Sat Jan 1 00:50:10 UTC 2000 58° 3648mV 800MHz
>>> root at letux:~# ./temperatures 
>>> Sat Jan 1 00:50:16 UTC 2000 67° 3642mV 800MHz
>>> root at letux:~# ./temperatures 
>>> Sat Jan 1 00:50:18 UTC 2000 67° 3642mV 800MHz
>>> root at letux:~# ./temperatures 
>>> Sat Jan 1 00:50:20 UTC 2000 67° 3648mV 800MHz
>>> root at letux:~# 
>>> 
>>> The first call shows a lower temperature and
>>> with the second the temperature jumps up suddenly
>>> by 9° and stays there.
>>> 
>>> I can try to check with the Thermocamera if it
>>> really gets hotter...
>>> 
>>> Or try this in a loop without some seconds delay
>>> to see if there is a gradual increase.
>>> 
>>> This is with letux-5.2-rc5 on gta04. Please note that
>>> I forgot to merge the twl4030-opp-fix so that
>>> I will rebuild letux-5.2-rc5 sources and binaries
>>> again.  
>> 
>> I have done some tests and it seems as if the
>> first temperature test (even 5 minutes after boot)
>> is significantly lower than those after:
>> 
>> letux login: root
>> Password: 
>> Last login: Sat Jan  1 14:58:38 UTC 2000 on console
>> Linux letux 5.2.0-rc5-letux+ #296 SMP PREEMPT Tue Jun 18 09:36:52 CEST 2019 armv7l
>> 
>> The programs included with the Debian GNU/Linux system are free software;
>> the exact distribution terms for each program are described in the
>> individual files in /usr/share/doc/*/copyright.
>> 
>> Debian GNU/Linux comes with ABSOLUTELY NO WARRANTY, to the extent
>> permitted by applicable law.
>> 
>> !!!  LetuxOS WARNING                                                      !!!
>> !!!  Root access through ssh and console with empty password is enabled.  !!!
>> !!!  This is a big security hole.                                         !!!
>> !!!  To disable, run 'apt-get remove letux-ssh-root'                      !!!
>> 
>> root at letux:~# cat /sys/class/thermal/thermal_zone0/temp
>> 53500
>> root at letux:~# cat /sys/class/thermal/thermal_zone0/temp
>> 62000
>> root at letux:~# cat /sys/class/thermal/thermal_zone0/temp
>> 62000
>> root at letux:~# cat /sys/class/thermal/thermal_zone0/temp
>> 62000
>> root at letux:~# cat /sys/class/thermal/thermal_zone0/temp
>> 62000
>> root at letux:~# cat /sys/class/thermal/thermal_zone0/temp
>> 62000
>> root at letux:~# cat /sys/class/thermal/thermal_zone0/temp
>> 62000
>> root at letux:~# cat /sys/class/thermal/thermal_zone0/temp
>> 62000
>> root at letux:~# cat /sys/class/thermal/thermal_zone0/temp
>> 63500
>> root at letux:~# cat /sys/class/thermal/thermal_zone0/temp
>> 62000
>> 
>> There is a more or less systematic and immediate jump of +9°
>> for the second request. Maybe a bug in the hwmon readout?
> 
> Maybe we are using old output and then reading triggers a new
> measurement.

I have run another test:

while true; do cat /sys/class/thermal/thermal_zone0/temp; done

This gradually goes up from ca. 53° to 62° within the first 5-10
loops and then temperature goes higher. Of course this makes
100% system load.

So it may just be an artifact that the processor is idling long
enough at 300MHz or so until we start reading out the temperature.

If I look at the core clock frequency through a cat command
it is rarely at 300MHz.


> hmm, I tested with a letux 3704 with almost no modules loaded,
> openmoko battery connected.
> Readout with letux 5.2-rc4 + my regulator patch: quite stable:
> 32mA, 25 C Battery temp (=room temperature), 47C thermal_zone0.
> 
> I think we should start with the simplest system possible and
> analyze things.

Yes, indeed. I still have too many influencing factors.

> I have the impression that something with display pm got
> broken.
> 
> About unplanned heating: I think there might be some connection with
> bad wifi reception (on GTA04A5 and Pyra), firmware trouble and kernel
> going mad. BUt these two things are still on my list to investigate
> things.
> For the display testing I have to fix my scripts, so that they work
> properly with backlight integrated into panel driver and without.
> I think I have seen values like 30-35mA for display modules loaded
> but display blank once. But last measurement seems to be higher,
> so there seems to be something rotten.

BR,
Nikolaus