assertion fail in configure_next_event() about sub-iot-stack HOT 5 CLOSED

sorry, this assert fail appeared again after this modification.

i am doing more test and debug on it.

from sub-iot-stack.

add a log, and reproduce the assert fail.

#ifndef NDEBUG
			//check that we didn't try to schedule a timer in the past
			//normally this shouldn't happen but it IS theoretically possible...
            fire_delay = (next_fire_time - timer_get_counter_value() - timer_info->min_delay_ticks) + 1;
			//fire_delay should be in [0,COUNTER_OVERFLOW_INCREASE]. if this is not the case, it is because timer_get_counter() is
			//now larger than next_fire_event, which means we 'missed' the event
            if (((int32_t)fire_delay) <= 0) {
                log_print_stack_string(LOG_STACK_FWK, "fire_delay %u, next_fire_time %u, counter_value %u, min_delay_ticks %u",
                                       fire_delay, next_fire_time, timer_get_counter_value(),timer_info->min_delay_ticks );
            }
			assert(((int32_t)fire_delay) > 0);
#endif

log:

[237097] [PHY] Switching to sleep mode
[237098] [PHY] resuming from sleep mode
[237099] [PHY] sync_word = f498, ch_coding 2
[237100] [PHY] Switching to sleep mode
[237101] [PHY] resuming from sleep mode
[237102] [PHY] sync_word = f498, ch_coding 2
[237103] [PHY] Switching to sleep mode
[237104] [PHY] resuming from sleep mode
[237105] [PHY] sync_word = f498, ch_coding 2
[237106] [PHY] Switching to sleep mode
[237107] [NONE] fire_delay 0, next_fire_time 6575904, counter_value 6575900, min_delay_ticks 5assertion "((int32_t)fire_delay) > 0" failed: file "/Users/sam/dash7-ap-open-source-stack/stack/framework/components/timer/timer.c", line 306, function: configure_next_event

the timer_get_counter_value() is increasing all the time cause its LPTIM hardware time. Every time we call it the result maybe increase.

So this feels that it is not very suitable to call it frequently here. I made a modification call at the top of configure_next_event() store the current_time and use it in the whold function, i am testing this modification for serveral hours to see if there is any other error.

--

update:

i have tested this modification for 15 hours by keeping sending read requset cmd to sensor_pull node and get the response, the gateway and sensor_pull works well without any error.

It seems like it works.

from sub-iot-stack.

ok thanks for the detailed information, I see the problem indeed. I did not notice it myself, probably because I'm not using debug builds for longer running tests, so this code is removed by precompiler. The modification you propose looks fine, can you create a PR please?

Related to this: we will probably re-add support for a normal timer next to LPTIMER as well, since for some internal timings the delay of setting the LPTIMER is too big

from sub-iot-stack.

Thanks for your reply. i make a PR for it. Please check if this is as your expectation.

from sub-iot-stack.

ok, merged. I will close this for now

from sub-iot-stack.