High fan & cpu when screen locked in Fedora 44 & KDE Plasma

All;

I recently did a clean install of Fedora 44 KDE Plasma on a Lenovo Legion 9

When the screen locker kicks in the fan and cpu spike and dont stop / return to normal until I log back in. When I d login I see the apps on my desktop but the background and panel are all black for anywhere from 10 to about 45seconds, then the desktop background and panel show up.

I connected via ssh while this was happening and ran a top -n 1 to a file, here is what I got:

I do have an Nvidia card and I have akmod-nvidia installed.

Thanks in advance

Have a look in both the user and system journals for nvidia, plasma and kwin logs.

Also I see you have a firefox tab working hard. Does tne problem go away if you quit firefox?

This is a known pattern with the lock screen and Nvidia. When the screen locks, the greeter or a wallpaper effect can get stuck spinning on the GPU. Try switching your lock screen wallpaper to a plain color or image instead of any animated or Plasma effect type, that often kills the spike. Also make sure kscreenlocker and your Nvidia drivers are fully updated. The black delay on login points the same direction.

Is there a bug ticket for the known issue?

I tried changing the screen locker background to a solid color, this is much better but it does not completely resolve the issue, the fan/cpu still spike but more to a mid speed based on the sound of the fan as opposed to as high as it can go.

The only real fix I have found so far is to do the following every time I walk away from the computer:

  • unplug the docking station (which disconnects the 2 external screens)
  • close the laptop lid

Note this does not put it to sleep as I have the setting “When Laptop lid closed” (in system settings → Power management) to “Lock Screen”

And to answer the above question about firefox, I still see this issue even if I close all firefox windows before letting the laptop go idle

The issue has changed as of the last few updates, now if I forget to unplug my monitors before I step away then when I unlock (fans blasting) then the monitors never recover, I just get a set of blank screens, I can see the windows on the current virtual desktop but they are completely unresponsive

I suspect reverting back to F43 is my best next step since I need a working laptop for work

I seem to have found a work around for now, I set these settings in the System Settings:

Screen Locking → Lock Screen automatically: Never

Power Management ( in all profiles ):slight_smile:
Turn Off Screen : 5minutes When locked and unlocked

preliminary tests show that I can wake the screens that are turned off with a mouse movement and all screens wake up properly

The obvious drawback is I get no actual screen locking but I work remote in my own home office so I can live with this for now.

Maybe this can help someone else fighting this issue as well

More info:

it seems the main issue with the unlocker is with my biggest monitor, a Samsung LS57CG952NNXZAOdyssey Neo G9 57" Dual 4K UHD (7680 x 2160) 240Hz Curved Screen UltraWide Gaming Monitor

I think this because I have another location I use and there i have 2 of these monitors:
Dell U4025QW 39.7" 5K WQHD (5120 x 2160) 120Hz Wide Curved Screen Monitor

and in the location with the 2 dell U4025QW monitors I dont have this issue.

Also, it seems that if I lock the desktop manually (right click on the desktop and choose ‘Lock Screen’ then I don’t have the unlocking issue or the high cpu/fan issue or the blank screens when unlocking.

Hope this info is helpful

All;

I was wrong with respect to finding a fix, even without the screen locker I get the same black screen behavior and have to reboot if I let the screens turn off for some time. I thought I could avoid going back to F43 but I was wrong.

F43 users the same kernel and kde plasma version as f44.

It’s therefore less likely you will see this fixed by that downgrade.

thanks for the info, saved me from doing all the work to downgrade. Any thoughts on how to debug this?

There is a Fedora archive of old kernels that you could try to see if it is a kernel regression.

I do not have notes on how to install the old kernels

But maybe @pg-tips knows?

Try this. It’s slightly awkward on a system with Nvidia drivers because you need the additional packages to allow the driver kernel module to be rebuilt.

  1. If you don’t already have the updates-archive repo available, set it up:
sudo dnf install fedora-repos-archive
sudo dnf config-manager setopt updates-archive.enabled=no
  1. Install the older kernel and relevant related packages:
sudo dnf install kernel{,-core,-devel,-modules-core,-modules-extra}-7.0.12-200.fc44 --enable-repo=updates-archive
  1. Wait a few minutes for Nvidia driver module build to complete.

  2. Check the driver was rebuilt:

dnf list --installed "kmod-nvidia*"

In the output you’re expecting to see a version of kmod-nvidia-... specific to the kernel version you installed.

  1. Reboot.

That example uses kernel 7.0.12. To see other available kernels, check the update history on Koji. You can replace the string 7.0.12-200.fc44 in step 2 with the corresponding version for a different fc44 kernel.