Another unexpected realtime delay on task 0 with Mesa 7i76E

Hi everyone

I'm keep having the warning "unexpected realtime delay on task 0" with my Mesa 7i76E setup. I know there are many topics about possible causes in the forum, like power saving features, virtualization, Intel network adapter and poor network connectivity. Power saving features and an Intel adapter I can rule out, but I rely on CPU-Virtualization on this system to run my CAD/CAM software and to generate G-Code. I'm aware, that this might cause spikes in the jitter of the servo-thread (currently up to 350'000 ns when a VM is running).

Right now, I'm trying to understand, what is triggering this message, what values are being compared and what thresholds are in place. My current assumption is, that as soon as the servo-thread.time exceeds the specified servo period (currently 1 ms), the warning will be triggered. Or am I completely wrong here? If however correct, would that mean, that changing the servo period to let's say 2 ms, the chances of getting the error would be less likely? Is the servo-thread.time the sum of all time requirements for operations within the servo thread including read and write time from the Mesa card? And servo-thread.tmax would be the maximum servo-thread.time that has occurred so far?

Thank you for any help in advance!

Yes, it means the servo thread could not meet its deadline. increasing the servo thread period to 2 ms should help (statistically) but you may actually have unbounded latency if virtualization is enabled.

Thank you a lot

So in this case I will monitor the servo-thread.time to find out, when exactly these spikes occur and try to get to the bottom of it. And the servo-thread.tmax should point me towards a feasible servo period.

Yes, note that on X86, all times are in units of CPU clocks.