TMS570LC4357: Watchdog forces MCU restart

shuo liu

Prodigy 80 points

Part Number: TMS570LC4357

Hi, gundavarapu

I've run into another new problem.

First, my configuration is as follows:

1.Initialize the watchdog and RTI modules

2. Feed the dog regularly

3. Enter SYS mode on the MCU

4. Write a function to force a watchdog reset of the MCU

5. MCU runtime error detection (If an interrupt from ESM Group 2 is triggered, immediately perform a reset and restart operation)

--------------------------------------------------------------------------------------------------

I am currently validating the forced watchdog restart functionality. Upon execution, it causes an exception in dataEntry. Is there an issue with the functionality I have implemented?

Below is the serial port output during my program's execution:

Additionally, I would like to inquire whether error detection is required during MCU operation. If so, should all channels in both ESM Group 1 and Group 2 be monitored simultaneously? If interrupts are enabled for error detection, will this affect the execution of the F021 API function in RAM？

4 months ago

0 jagadish gundavarapu 4 months ago

TI__Guru* 77776 points

Hi Shuo,

I don't find any issues in your screenshots, and they seem to be fine for me. Is it possible to provide complete project once?

--
Thanks & regards,
Jagadish.

0 shuo liu 4 months ago in reply to jagadish gundavarapu

Prodigy 80 points

Hi, gundavarapu

The issue I encountered is that executing the Force_Watchdog_Reset function causes a dataEntry exception.

As seen in the serial port output, the watchdog was being fed during the RTI interrupt. After executing the Force_Watchdog_Reset function, an exception occurred, so the watchdog feeding information was lost.

0 jagadish gundavarapu 4 months ago in reply to shuo liu

TI__Guru* 77776 points

Hi Shuo,

jagadish gundavarapu said:
I don't find any issues in your screenshots, and they seem to be fine for me. Is it possible to provide complete project once?

I mean i verified your code in screenshots and i don't see any issues there and code seems fine to me. Is it possible for you to share your complete project?

--
Thanks & regards,
Jagadish.

0 shuo liu 4 months ago in reply to jagadish gundavarapu

Prodigy 80 points

Hi, gundavarapu

Sorry for the late reply.

I will attach my project as follows:

TMS_95_Watch_Dog_Reset_Test.rar

0 jagadish gundavarapu 4 months ago in reply to shuo liu

TI__Guru* 77776 points

Hi Shuo,

I understood the root cause for the issue.

It is because of SCI, i mean after watchdog reset the peripherals are getting disable again.

And the below API is not getting called because of reset source is watchdog:

If the reset source is watchdog then the execution will directly jump to the red color highlighted line and all the blue color highlighted lines will not get executed.

That means peripherals will not get enabled by watchdog reset, however we are trying to directly call SCI initializations without enabling the clock to that corresponding peripheral and that producing data_abort.

The solution is simple, just callly system initializations, VIM initializations etc in watchdog reset case also just like below: