CPU fan stuck at high speed sometimes


#41

Hi all,

Maybe these steps will help reproduce.

I use my laptop plugged in via HDMI with a USB-C hub (keyboard/mouse/headset).

If i plug in my charger the CPU drops to 400MHz, i then have to use a gnome plugin called “CPU-Z” to tell the CPU governor to use 100% of the CPU, everything is fine then.

If i then let the laptop go to sleep for around 15 minutes (eating lunch) and return, when i wake the laptop, the fan then spins at 100% and the CPU is then throttled to 400MHz, however this time the “CPU-Z” plugin can not recover this, so i have to unplug the power adapter & HDMI, shut the laptop lid and let it go into hibernation.

When i then open the laptop, the fan is throttled down again and “CPU-Z” can then tell the governor to use 100% of the CPU again.

I hope this information can help.

Adam


#42

I can reproduce these steps without fail on a librem 13 v4 every time

Possibly related with: https://github.com/erpalma/throttled


#43

Surprisingly, after I wrote my comment here I didn’t get that error again. But I also played around with some distros and ended up with Manjaro, using the 5.2 kernel. I’m not sure if this was the solution or maybe that I installed TLP and did some settings for the CPU_HWP. Either way, no weired fan actions and no freezes anymore!


#44

So, if it is interesting for anyone. I got some trouble with Manjaro and switched back to Linux Mint. I also updated Coreboot to 4.11 - After a while I got the weired fan back and it seems again that a kernel update (from 5.0 to 5.3) solved this issue for me. Because I did the same TLP CPU_HWP edit (to get sure that this won’t happen) but afterwards the fan got still mad. With the following kernel update everything is fine, till now.


#45

Ok, kernel update didn’t help. Occasionally the fan is stlll getting weird after sleep…


#46

@ollrich can you still produce this with some consistency? If so, I have a test build I’d like you to try, so shoot me a PM


#47

@MrChromebox It still happens occasionally but not really with consistency. Every 2 to 5 wake ups from suspend, sometimes it needs more to trigger that problem.
I’m using a small workaround and put my machine to suspend via systemctl suspend not by closing the lid. That is helping a little bit (maybe after 12 to 15 wake ups the error than happens) but doesn’t solved it.

Is this a test build I could use on my daily driver? or could this be problematic?


#48

can you do me a favor and install the lm-sensors package, and run watch -n 1 sensors to get an updated temp output, and then reproduce the issue. I’m curious if any of the temp readings are incorrect/invalid when the issue occurs. I’ve suspended 100+ times over the past few days and unable to reproduce it here


#49

@MrChromebox it took a while but I got the error just now. Seems like nothing happend to the temp output

Screenshot%20from%202020-02-14%2009-30-35


#50

@MrChromebox is there anything else I can help with?


#51

I was wondering if the CPU temp was being reported incorrectly / too high in cases where the fan speed is stuck at max, but that appears to not be the case. Need to do some more digging it seems


#52

May it be helpful:
Had it a lot with my L13 two years ago, but it got better with PureOS updates. Still the same coreboot etc. it was shipped with.
Now I’d say I get it every twentieth time or less.