Program getting stuck on NRF_BREAKPOINT_COND after getting into a template function

[morpho]

So I noticed the program goes into a state where it's stuck on  NRF_BREAKPOINT_COND, which happens after getting into the following writeToLogger() template function. If I comment this function call, it seems to work fine. 

   template<typename T>
    static void writeToLogger(T *data, ...)
    {
        char tmp[40] = {0};
        va_list args;
        va_start(args, data);
        vsprintf (tmp, data, args);
        xStreamBufferSend(nrfLogTaskMsgBuffer, (void*)&tmp, 40, pdFALSE);
    }
    
int main(void)
{
  NrfLogger logger;
  NrfLogger::writeToLogger<char>("Writing to register address %x", 10);
}

[Susheel Nuguru]

before i can look into it, could you tell me which version of SDK are you using ?

have you started the debugger to see the function call stack which leads to the breakpoint?

[morpho]

Here's the screenshot of the callstack. I'm using nRF5_SDK_17.0.2_d674dde

[Susheel Nuguru]

Looks like there is a softreset in between these logs? or maybe that part of the code is being executed many times. I do not know but this should be very easy for you to check if you start the code in debug mode and use some breakpoints.

[morpho]

Well, that's what I did, and like I said, the moment NRF_LOG_FLUSH() is executed, you see this line printed a bunch of times. If I comment these log functions out, program runs fine. The issue I'm pointing has to do with the logger functions itself hence my concern

[Susheel Nuguru]

I am sorry, I still cannot compile your code

And we at Nordic cannot help you debug issues around your cpp  issues as that is what we do not support officially. I even installed STLPort package from Segger just out of curiosity but it still did not help.

[morpho]

I literally just unzipped the folder inside nRF5_SDK_17.0.2_d674dde/examples/MyProjects/ and was able to build without any problem. 

These are my compiler settings.

[morpho]

My main issue isn't about C++ but moreso related to Nrf logger itself which is nordic's library. I understand I'm using C++ but I'm not quite sure why you'd be seeing this error if I can run this without any problen

[Susheel Nuguru]

I was able to compile your project when I comment out below extern linkage in your stream_buffer.h file

//#if defined( __cplusplus )
//extern "C" {
//#endif

But if i comment this out, then uncommenting writeToLogger calls does not get the functions in stream_buffer.h linked into the executable. 

Not sure what I am missing here?

But lets focus back on the actual assert you have.

Can you give more details on the id and PC value of this assert that you should be able to see in the parameters. Maybe instead of spending our time and energy on me trying to be able to replicate it, we can change focus back to trying to debug the assert from your end itself. We can try to analyze the fault based on the parameters of app_error_fault_handler

[morpho]

Now, I'm seeing the following call stack, and the error only occurs after vTaskStartScheduler().

What exact details do you need on the ID and PC value? I'm not quite sure why is unsigned int pc=0x00 even though PC register itself matches with the call address which isn't the same as the parameter

[Susheel Nuguru]

The app_error_fault_handler says that the error is 0x4002 which is an SDK related error. You should be able to get the filename and linenumber from the info variable when casted to assert_info_t.

It most likely seems to be a manual ASSERT check in integration\nrfx\legacy\nrf_drv_clock.c: 274 (nRF5 SDK17.0.2)

    ASSERT(m_clock_cb.lfclk_requests > 0);

This happens most likely when you have not initialized the nrf_drv_clock as below before attempting to start the scheduler. Add this line before vTaskStartScheduler

    /* Initialize clock driver for better time accuracy in FREERTOS */

    err_code = nrf_drv_clock_init();

    APP_ERROR_CHECK(err_code);