app_timer_start causes a hard fault

[Sawada]

The device is connected to Keil for debugging.
When BLE send and timer start-stop is repeated, it may stop at HardFault_Handler.

Looking at the call stack at that time, 0x04 is displayed due to a LOCAL ERROR of app_timer_start.
I think this is the error cause.
However, even if I set breakpoints at the following locations, it still does not stop, but an error of 0x04 occurred in app_timer_start.
I don't know why.

( 0x04: NRF_ERROR_NO_MEM)

****

app_timer.c

  static uint32_t timer_start_op_schedule(timer_node_t * p_node,
                                                                  uint32_t timeout_initial,
                                                                  uint32_t timeout_periodic,
                                                                  void * p_context)
    {
      uint8_t last_index;
      uint32_t err_code = NRF_SUCCESS;

      CRITICAL_REGION_ENTER();
       timer_user_op_t * p_user_op = user_op_alloc(&last_index);
       if (p_user_op == NULL)
       {
         err_code = NRF_ERROR_NO_MEM;      <=Break point
       }

****

[Jørgen Holmefjord]

Hi,

The API reference for app_timer_start() indicates what will cause this issue:

NRF_ERROR_NO_MEM

If the timer operations queue was full.

There is a sdk_config.h parameter for the library that allows you to change the operations queue size: APP_TIMER_CONFIG_OP_QUEUE_SIZE

Best regards,
Jørgen

[Jørgen Holmefjord]

user_op_alloc() should only return NULL if the queue is full. Did you try to increase the size of the operation queue config?

[Sawada]

ret_code_t app_timer_start(app_timer_id_t timer_id, uint32_t timeout_ticks, void * p_context)

It seems that LOCAL_ERROR (0x4) was displayed on Keil because the reference destination of the null pointer was passed to the id of the first argument.
It was solved by taking measures so as not to pass null.
Thank you.