AM2434: AM2434, Industrial Communication SDK for Profinet (and EtherNet/IP), embedded "HWALWatchDog" Task Behavior

Hi Vit,

PN_setWatchDogTimer() sets the watchdog timer that turns off the PPM transmission if the Profinet application stops responding. For more details, please check https://software-dl.ti.com/processor-industrial-sw/esd/ind_comms_sdk/am243x/latest/docs/api_guide_am243x/PROFINET_DEVICE_FWHAL.html#autotoc_md209.

Regarding HWALWatchDog, can you please elaborate on where do you see this timer? what application are you running? 

Thanks.
Kind regards,
Kamil

The list of Profinet tasks:


Or similar Ethernet/IP from ROV view:


What does your embedded "HWALWatchDog" task do?

- When watchdog is activated?
- How often is watchdog cleared?
- What does HWALWatchDog check (because too high priority, tasks below this priority cannot be checked not beeing stucked...)?
Still not sure whether I need my application watchdog (named WATCHDOGTask) in fact...

Hello,

please, tell me more about your watchdog design and solution in your Industrial Communication SDK (especially Profinet, then EtherNet/IP...)

Any answers on previous questions?
How to distinguish between stuck and overloaded tasks?

Can you provide me with a view of the source code that would answer many of the questions?
Can you respond faster?

Regards, Vit

Hello Vit,

HWALWatchDog is an internal task used to protect our intellectual property. It controls the service duration of our firmware based on the chip type and license obtained by the customer. it's execution time is relatively short and its periodicity is high, so it should not cause any CPU load issues in general.

Regarding task monitoring I can see you're using the Run-time Object View (ROV) which is the main way to do that on CCS.

If you're facing a problem, I would advise you to describe it here in details so we can try to help you solve it.

Source code is not fully available for the public, and due to the current high load of tasks, our responses might be a bit delayed. If you require a higher priority support, please contact your regional TI sales representative for additional details.

Kind regards,
Kamil

Hello Kamil,

During the stress tests (lasting 2 minutes), the "RxTask" (Profinet, EtherNet/IP) is unsurprisingly overloaded.
Since this task has a high priority, tasks with lower priority are not processed (i.e. for longer than 0.5 seconds).

How (by monitoring what) can I determine from the application perspective that the system is only overloaded by communication, not stuck?

The problem is not a temporary loss of communication, but a lack of detection between overloaded and stuck states.

Best regards,
Vit

Hello Vit,

ROV is the only available way to check task statuses for now. Other RTOS monitoring features are on our roadmap but still not available. However, if you wanna know if your application task is working, why don't you just add a visual function (e.g. UART print, GPIO signal, etc.)? I would also like to have more details about the issue you're facing:
1. What kind of a stress test are you doing?
2. Are you using a specific application to do it?
3. What are the main differences between your application/setup and the default one delivered in the SDK?
4. For PROFINET, have you tried to run Netload test that comes with the official "Automated RT Tester"?

Kind regards,
Kamil