Echelon Neuron C - Manual (Page 181)

Neuron C Programmer’s Guide 169
• Log an error
These actions can be combined. For example, you can log an error and then take
the application offline. Alternatively, you can disable a functional block and
change functional block status.
The Neuron firmware also logs system errors for errors detected by the firmware.
Resetting the Device
You can reset a device by calling the node_reset( ) function. This function
immediately resets all processors (application, network, media access, and for
Series 5000 devices, interrupt) on the Neuron Chip or Smart Transceiver. The
Neuron reset pin is driven low; this signal can be used to reset external
transceivers and logic. You typically take this action for catastrophic errors that
require a hardware reset. A device can also be reset by expiry of the watchdog
timer, or by a reset command received from the network.
When a device is reset, it executes the entire initialization sequence. The amount
of time required for initialization is a function of the amount of off-chip memory
as well as the size of the application program, but must be less than 18 seconds
before the application is online. See the Smart Transceivers databooks
for a
detailed formula to calculate reset time.
When you reset a device, all state information that is not kept in EEPROM is
lost. All pending incoming and outgoing messages and network variable updates
are lost. The network processor could receive duplicate packets. In addition, any
packets that have been acknowledged by the network processor but not processed
by the application are lost.
Restarting the Application
You can reset the application processor, but not the network or media-access
processors or (for Series 5000 devices) the interrupt processor, using the
application_restart( ) function. You typically take this action for application
errors that can be recovered by restarting the application, but that do not require
an external hardware reset.
When you call this function, the Neuron firmware clears all timer objects and
initializes I/O objects, non-configuration network variables, and static variables
and then executes the reset clause. Some synchronization cleanup is done before
restarting the application. Any outgoing messages in progress are terminated.
Incoming messages are unaffected. Outstanding completion events and
responses are discarded. An application restart does not lose network state
information. Because only the application processor is reset, the network and
MAC processors continue to process network traffic, and the interrupt processor
continues to process interrupts.
Typical applications that use interrupt tasks share some amount of data between
when tasks and interrupt tasks. During the application restart, some of the
assumptions made for that data sharing might not be valid; you should consider
disabling all interrupts (by calling the interrupt_control( ) function) before calling
application_restart( ). You can re-enable interrupts during the application’s
when(reset) task.