Nortel DMS-100 Series Maintenance Manual page 78

1-62 Maintenance overview
System component interactions
The system components interact with the XPM exception processing system
to recover from error exceptions. The system components report and display
information about error exceptions. The system components are listed
below.
If the system detects a fatal error, the system initiates local maintenance
action to drop activity from an active unit only. The maintenance action
resets or restarts the unit. When in an INSV state, the XPM unit reports
exception errors to the CM in the form of unsolicited messages. The XPM
generates a report for each exception not reported in the trap buffer. The
CM receives the report and acknowledges the message. The CM separately
logs into PMDEBUG to extract the exception information and generates a
PM185 log report. The PMDEBUG provides the ability to view and delete
exceptions data from the trap buffer at the task level.
Exception Recovery Action
Exception recovery consists of three different operations:
When the recovery process is complete, the exception reporting system is
notified that new trap data is available. The exception reporting system is
also notified that exception handling is complete. The recovery process can
set up the trapped task for recovery and allows it to resume. The recovery
process also can initiate a maintenance action to restart or reset the unit.
If the trap is not fatal, the selected task will be restored to a point of
execution. This action allows the task to restore to a known state. The
process of recovery does not consists of a task restart. The process of
recovery forces the task to perform a multi-level exit back to the task
mainline. The task recovery model specifies that a task provide a mainline.
A mainline is a do forever loop that calls the main body of the steady state
code. After a task provides a mainline, an optional call to a recovery
procedure occurs. The task starts again following a recoverable trap. The
task is forced to return to the next instruction following the call to the main
body of code. The next instruction must call to a recovery procedure used
by the task to realize the task has taken a fault. The instruction allows the
task to take special recovery action or a branch back to the start of the loop.
297-8231-550 Standard 09.01 August 1998
CM
PMDEBUG
XPM maintenance
severity (fatal or recoverable)
process fatal trap
recover task (non-fatal trap)