I have now opened a bug upstream (KDE bug 434922 —
https://bugs.kde.org/show_bug.cgi?id=434922 — cannot link to it), but it would really be important to first find out if the issue is specific to ARM, specific to RPI4 within ARM, or specific to manjaro.
To the best of my understanding the issue seems to be with the ksystemstats daemon. This daemon is meant to be autostarted whenever there is something needing its services to look into system statistics (like CPU load). This is a relatively new daemon, so that currently plasma ships with some tools that rely on it (modern cpu load plasmoids) and tools that do not ksysguard…
On RPI4 manjaro stable, ksystemstats does not seem to work. It gets started correctly, but it fails to interact with its clients. This is why the clients cannot see any sensor nor display any data. At the same time ksystemstats locks into a situation that causes it to use 100% of the CPU.
@joe-77, I doubt this has to do with the lm_sensors up. I think that KDE has its own framework for accessing this kind of system information (the ksystemstats daemon).
@0n0w1c, can you please expand a little? On my system you can get the graphs correctly by using ksysguard (that does not appear to use ksystemstats yet) or older plasmoids (that do not use ksystemstats), but all the modern plasmoids (e.g., individual CPU load) fail and lead to a 100% resource usage from ksystemstats. What are you using to get the graphs?
As a side note… do people in the KDE plasma support forum see this, is the KDE label sufficient for that?