Timing Issue with app_uart_put()?

[Josh Grob]

Hello everyone,

I have very pesky bug that is leaving me in fits. I have a feeling it is a timing issue, but I am not sure at this point. Here is my issue, and I would love anyone's debugging help:

I have helper functions to write chars over UART (I am using a FTDI board to make serial connection to my mac). I am using Rev 1 NRF52 chip with UART configured as 115200baud, IRQ low priority, and flow control off.

    void uart_print(char *str)
{   
    uint32_t err_code;
    int i = 0;
    while (str[i] != '\0'){
        err_code = app_uart_put(str[i]);
        APP_ERROR_CHECK(err_code);
        i++;
    }
}

void uart_println(char *str)
{
    uint32_t err_code;

    uart_print(str);
    err_code = app_uart_put('\n');
    APP_ERROR_CHECK(err_code);
}

The above functions works perfectly on initialization methods, Advertising states, and when the device is connected. Once connected I have a timer that starts and grabs samples from my accelerometer and the failure happens in one condition branch in callback method. That method is here:

void accelgyro_timer_timeout_handler(void * p_context)
{
    float sensors[3];
    char test_str[30];
    memset(&sensors[0], 0, sizeof(sensors));
    
    read_accelgyro_data(&sensors[0]);

    fft_in_buffer[0][buffer_idx] = sensors[0];
    fft_in_buffer[1][buffer_idx] = sensors[1];

    if (buffer_idx < FFT_SIZE - 1){
        buffer_idx++;
        sprintf(test_str, "Sample %d", buffer_idx);
        uart_println(test_str);
    } else if (buffer_idx == FFT_SIZE - 1) {
        uart_println("--PSD Start--");
                   .
                   .
                 <algorithm code>
                   .
                   .
    }
}

The failure occurs when trying to write the first char of the "uart_println("--PSD Start--");" statement (the actual failure is at err_code = app_uart_put(str[i]); in the uart_print method above). It immediately fails with this error in my gdb terminal:

Program received signal SIGTRAP, Trace/breakpoint trap. 0x000008e4 in ?? ()

This is baffling me since the first condition uart_println(test_str); works fine, and if I remove the uart_println statement entirely the function behaves normally. Does anybody have any suggestions on how I should debug or try a different approach. For instance, could this be solved with flow control?

Thank you for the time and help. Josh

Below is snapshot of the Serial output

[Josh Grob]

As an update I think I am hitting a seg fault since the chip keeps resetting, and faulting at the same place each time. The SIGTrap is a red herring I think since the debugger stops at the program entry point again on reset (I think). When the firmware runs without gdb, I see the crashing and resetting behavior in a loop. Now to find where the memory is getting corrupted...

[Josh Grob]

UPDATE

I did some digging in the SDK (V11), and I came across the RTT/UART logging (NRF_LOG) apis. I switched over the code to use those APIs work great and I can log out the values I need. Thus, in short I had a bad implementation, and doesn't make sense to troubleshoot my original question anymore (Doah!). Sorry for the errant post before doing my homework.

Thank you!

P.S - I implemented the NRF_LOG_USES_RAW_UART option, and nrf_log.c failed since it couldn't find UART0_CONFIG_BAUDRATE macro. I added "nrf_drv_uart.h" and everything works well now.

[Akshay]

I guess one modification you needed in your code was this, not sure if this was your actual problem. It was skipping few character for me.

 void uart_print(char *str){   
    int i = 0;
    while (str[i] != '\0'){
        while(app_uart_put(str[i])!= NRF_SUCCESS);
        i++;
    }
 }

void uart_println(char *str){
    uart_print(str);
    while(app_uart_put('\n')!= NRF_SUCCESS);
}

The 'app_uart_put' function needs to wait before the previous characters which you have already put are sent, Hence the while loop. Hope this will help others.