I keep the system monitor open on a second monitor, and don't often restart, so
it will be open for serveral days/weeks at a time. I noticed that after just a
day or two, it starts taking noticably more CPU time, eventually getting to the
point it is constantly using 100% of a single core. During this, the UI also
becomes less responsive. Shortly after startup, it uses more like 10-30% of a
core.

The overview page I have open when this happens consists of several different
line charts tracking things such as CPU/GPU usage, memory usage, and network
usage, as well as a "Process Table" section.

I attached a thread trace using gdb when it is in the state of constantly
running at 100% CPU (I took several traces - they are all similar, with the
same call stack going through `dataChanged`). I'm guessing that Thread 1 is the
culprit, and that it is possibly due to storing sensor history for the line
chart and never removing that history, so that it continuously grows and
requires more processing power.


I believe a similar (perhaps the same) issue affects the system monitor
plasmoid desktop widgets - except in that case, it is plasmashell that
increases CPU usage and becomes unresponsive.