The function model_timer_elapsed_ticks_get() returns the total number of ticks since the timer was started. Hence it's expected that this function to return a value > 0 when called after the timer has expired for the first time. However, this isn't the case.
This function simply returns the value p_timer->total_rtc_ticks which is updated in the function timeout_update_and_schedule(). The problem is that this function is called after the user callback for the timer is called. This can be seen in the code below.
static void model_timer_cb(void * p_context)
{
model_timer_t * p_timer = (model_timer_t *) p_context;
NRF_MESH_ASSERT(p_timer->cb != NULL);
if (p_timer->remaining_ticks == 0)
{
/* Trigger callback and repeat if required */
p_timer->cb_active = true;
p_timer->cb(p_timer->p_context);
p_timer->cb_active = false;
if (p_timer->mode == MODEL_TIMER_MODE_REPEATED)
{
p_timer->remaining_ticks = p_timer->timeout_rtc_ticks;
}
}
/* Continue if time is left */
if (p_timer->remaining_ticks > 0)
{
(void) timeout_update_and_schedule(p_timer);
}
}
The attached patch moves the update code in timeout_update_and_schedule() into model_timer_cb() where it belongs.
0001-Update-model-timer-total_rtc_ticks-before-callback.patch
The unit test for model_common should probably be updated to verify this bug.
