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

How can I reproduce this?
Your theory might be correct indeed, but I need to look into it. I merged the PR already since it needed to be done anyway.

from sub-iot-stack.

here is my test method:

two B-L072Z-LRWAN1 boards, one is gateway, the other is sensor_pull. use the default master code.

gateway keep sending a unicast read request to sensor_pull node every 10 second. below is my python code change from gateway.py :

# ... some other code

    def query_node(self):
        unicast_uid = "0x46323039002C0030"
        addressee_id = int(unicast_uid, 16)
        print(addressee_id)
        configuration = Configuration(qos=QoS(resp_mod=ResponseMode.RESP_MODE_ANY),
                addressee=Addressee(access_class=0x11, id_type=IdType.UID, id=addressee_id))

        alp_command = Command.create_with_read_file_action(
                file_id = 0x40,
                length = 2,
                interface_type = InterfaceType.D7ASP,
                interface_configuration = configuration)
        print(alp_command)

        self.modem.execute_command_async(alp_command)

    def run(self):
        self.log.info('Started')

        keep_running = True
        while keep_running:
            self.query_node()
            time.sleep(10)

if __name__ == '__main__':
    Gateway().run()

then sensor_pull receive the request and send reponse back to gateway.

after a while (10 minutes or less an hour), i got the assertion fail log from the sensor_pull's JLinkRTTClient terminal.

from sub-iot-stack.

How can I reproduce this?
Your theory might be correct indeed, but I need to look into it. I merged the PR already since it needed to be done anyway.

There is one place i have found could cause atomic nested problem:

1. configure_next_event() was called in a atomic context
2. in configure_next_event(), called sched_post_task_prio() which called start_atomic() and end_atomic() in it.
3. after exit from sched_post_task_prio(), the irq is enabled, but still run in configure_next_event() function.

After this pull request the problem will be solved.

from sub-iot-stack.

I revert the irq nesting counter commit in the PR on master, this seems to be causing problems with UART RX interrupt. Until the cause of this is found it's better to revert so gateway works again. TBC

from sub-iot-stack.

There have been major changes concerning the timer, we assume that this is not a problem anymore. If this is incorrect, don't hesitate to open the issue again.

from sub-iot-stack.