Vision Processing and Periodic Tasks stops working

[Tom Line]

We are having an intermittent issue that is increasing in frequency.

What we are seeing is that when we are running the code from our laptop (i530 quad core, lots of memory, nothing other than labview running, CPU not maxed out) using our classmate as the driver station (basic dashboard with no modifications), our vision processing .vi and periodic tasks .vi will stop functioning, and probing of either will result in the probe showing 'not executed'.

Our code normally runs at about 70% CPU utilization until the fire button is pressed, then the vision runs and the cpu jumps to 100% until the trigger is released.

Each of our periodic tasks has a wait in it, with no weight shorter than 20 ms.

Why would periodic / vision processing stop executing? Is there some built in safety that shuts them off? This does not happen that we've seen with a permanent deploy to the robot.

[Greg McKaskle]

About an hour ago I filed a bug report that demonstrated probes not working when run on the RT target. It is a particular team's periodic tasks. I have never seen this before, and I don't have details yet as to what contributes to the bug. Before you believe that your code isn't running, place an indicator or a global in the loop and verify.

Greg McKaskle

[RyanN]

Originally Posted by Greg McKaskle

About an hour ago I filed a bug report that demonstrated probes not working when run on the RT target. It is a particular team's periodic tasks. I have never seen this before, and I don't have details yet as to what contributes to the bug. Before you believe that your code isn't running, place an indicator or a global in the loop and verify.

Greg McKaskle

I'll add to this.

It seems like probes won't work unless the parts of the code are probed before running/temporarily deploying the code. I think this has been a bug for some time, but I saw it as just a limitation of the system.

[Tom Line]

I can verify that the tasks are no longer running, because the targetting information no longer gets passed to the turret. I.e. - we lose turret tracking and our velocity PID control when periodic tasks and the vision .vi stop executing.

[androb4]

Originally Posted by Tom Line

I can verify that the tasks are no longer running, because the targetting information no longer gets passed to the turret. I.e. - we lose turret tracking and our velocity PID control when periodic tasks and the vision .vi stop executing.

The same exact thing happend to us today!

What's going on!?!!?! The world is going to end!!

[Greg McKaskle]

Probes should not need to be added to the VI before running in order to work, and that didn't impact the issue in the bug report I filed. It appeared that the node identifier used to describe where the probe or breakpoint was added didn't match where the compiled code's offsets were. As I said, I've never seen this before and it currently seems to be RT specific. I'll know more in a few days. In the meantime, probes in most VIs work fine. If they claim not to be updated, that could be a false statement. I also saw issues with breakpoints not working.

I'll be happy to try and reproduce and explain what is going on with the code that is no longer running, but I'll need something to reproduce it. I asked Tom for his code. If you also have a repeatable case, either go through support or PM me.

Greg McKaskle