Rate Group Slipping

[noahrose18]

Hello!

I am trying to create a rate group that supports various interrupt frequencies (1/4Hz, 1/2Hz, 1Hz, 10Hz, and 25Hz). I thus set the clock of the overall system to the Lowest common multiple (50Hz), and I created an event within the 10Hz handler to validate that it is working correctly.

I am noticing that the timing of the 10Hz event (as seen within the fprime-gds popup events tab) is slipping, aka it is interrupting slighlty late every iteration. I am wondering if this is the expected behavior and, if not, how to make these interrupts hit at the same exact interval every time.

[timcanham]

@noahrose18 Just some clarification: The terminology would be rate groups, at 1/4Hz, 1/2Hz, etc. The typical approach would be to drive the Svc/RateGroupDriver with your base 50Hz and then use the dividers to call your Svc/ActiveRateGroup instances at the rates you want. Are you doing this?

[noahrose18]

I'm using Linux. I'm seeing that when I ask for a ISR at a frequency of 10Hz, it is actually executing slightly less than 10Hz (say 9.5Hz). So, this is slightly different than the error that you are describing, right?

[timcanham]

Yes, although what I'm describing isn't an error in an RTOS, it's how priority scheduling of tasks works. What is your "ISR" in this context?

[noahrose18]

The ISR is the Handler executing at 10Hz. It logs one event and sends out data to two ports.

[timcanham]

Is the ISR driven by a hardware timer? Something else?

If you are doing the pattern:

while (1) {
   doSomething();
   sleep(100ms);
}

It will slow the loop since you are executing the logic then sleeping.

[noahrose18]

Looks like this is what is happening in the main.cpp code. Thanks!

[timcanham]

There is a Svc/LinuxTimer implementation in the core repo based on the Linux timerfd feature that you can use to get repeatable cycles.

[timcanham]

https://github.com/nasa/fprime/tree/devel/Svc/LinuxTimer