"System error KERNEL.446571: 1528: Internal error detected by SMP multi-tasker" in LabVIEW RT 8.5

[Sima]

Hi Robert,

Having experienced the "Internal Error Message ThEvent.cpp line 192" error, I'd say this is a different issue you are having. We have experienced that error several times and NI is working on it. For us it was an issue related to network shared variables on quad core machines. Is this relevant to you?

Sima

[FlamingYawn]

I am using a quad core machine, but not shared variables.  I'm afraid I'm experiencing multiple errors from multiple problems.  I'm just not sure how to track them down any better.

[Sima]

That's very interesting. No shared variables?

The way I started the debugging process was to disable chunks of the code one at a time and see if what would crash. Eventually, I ended up with a single loop that contained some networked shared variables writing to single process variables. NI said they believed the crash was related to networked shared variables on quad core machines, and they were able to reproduce this error (i.e. Thevent.cpp line 192) using the VI I sent them.

Since you don't have shared variables, I wonder if something else is going on. Could you briefly describe what your code is doing?

Message Edited by Sima on 06-19-2008 09:40 AM

[FlamingYawn]

I have a main calculation loop that is talking to lots of other loops over RT FIFOs.  I have a CAN loop, a TCP loop, a DAQ loop, a VISA loop that talks to a custom PCI device, and several other minor loops.  There's a lot going on. 

I wish I could spend more time tracking down crashes and LabVIEW bugs, but it feels like if I did that, I wouldn't have time to do any actual development.

[FlamingYawn]

I was just running the app in the dev environment targeted to my quad core RT desktop target.  When I hit the stop button, the RT box crashed with the following error:

"System error KERNEL.446571: 344: Internal error detected by SMP multi-tasker."

Is this 344 error related to the 1528?  What does 344 mean?

No error info was logged to the error log file.  This is the first time I've seen the error, and I've already followed the same procedure (without crashing) about 10 times with the same code.

Any ideas, Mike?  NI?

[Sima]

Robert, I'm curious. Are you using timed loops?

[Sima]

---

FlamingYawn wrote:

I have a main calculation loop that is talking to lots of other loops over RT FIFOs.  I have a CAN loop, a TCP loop, a DAQ loop, a VISA loop that talks to a custom PCI device, and several other minor loops.  There's a lot going on. 

I wish I could spend more time tracking down crashes and LabVIEW bugs, but it feels like if I did that, I wouldn't have time to do any actual development.

---

Unfortunately, we've had to do a lot of that as it was the only way we could get closer to a solution. Try a binary search: disable half your code and find out the result. If it passes, do a binary search on your other half, and so on.

I'm interested to find out if the "ThEvent.cpp line 192" error you experienced is related to this kernel crash or whether they are 2 separate problems. I'd also like to know what else other than networked shared variables on a quad core can cause that crash (which happened in your code).

[FlamingYawn]

I was using timed loops until I realized they have way too much overhead and jitter if you want to work in the microsecond range.  Now I used timed structures to set the priority of certain parts of code and roll my own timing.

Message Edited by FlamingYawn on 06-19-2008 01:01 PM

[FlamingYawn]

Unfortuantely most of these errors aren't reliable.  If I took out half my code and didn't get the error, I wouldn't know if it was due to the missing code or random luck.  In other words, how do I reliably know if it "passes"?

I appreciate your thoughts on these frustrations!

[Sima]

How long do you have to run before you see the error, if it does occur? On some occasions, when I've thought I'd fixed the problem, I ran for 30 hours before I saw another crash. It can be a slow and painful process.