SDK 14.0.0 -> 14.2.0 migration: app_timer_create() thinks the timer is already running.

[Eliot Stock]

In migrating an existing application from SDK 14.0.0 to SDK 14.2.0, I am finding that the first timer I create gets NRF_ERROR_INVALID_STATE on calling app_timer_create(). This is not because the module isn't initlisalised, it's because app_timer_create() thinks the timer is already running. I am quite sure it's not - I've stepped through the code from the top of main and this is the first time this timer ID is used in the code.

I'm hitting this:

if (((timer_node_t*)*p_timer_id)->is_running)

{

  return NRF_ERROR_INVALID_STATE;

}

The timer is defined using:

APP_TIMER_DEF(m_acc_int_lag_timer_id);

and created using:

err_code = app_timer_create(&m_acc_int_lag_timer_id, APP_TIMER_MODE_SINGLE_SHOT, acc_on_int_lag_timeout);

APP_ERROR_CHECK(err_code);

Looking for m_acc_int_lag_timer in my .map file, I see:

.rodata.m_acc_int_lag_timer_id
0x0000000000048a68 0x4 _build/app/acc_bma250e.c.o

and:

.rodata.m_acc_int_lag_timer_id
0x0000000000048a68 0x4 _build/app/acc_bma250e.c.o

I see nothing in the release notes for SDK 14.1.0 or 14.2.0 that would explain this.

[Martin Lesund]

Hi Eliot,

Without the whole code it is hard for me to point out what the issue is, but here is some pointers/tips that can be helpful:

Error code 8 is NRF_ERROR_INVALID_STATE. You will get this error code if have not initialized the timer module(app_timer_init()). You could also get this error code if your timer_id is 0, or if the timeout_handler function is NULL.

Do you have BSP module in your application? 

This could also be a little interesting.

[Eliot Stock]

I must have been mistaken with my debugging because with the exact same code today, I'm finding that this timer can be created and in fact it's another SDK migration issue entirely that causing the trouble. Apologies for the false alarm.