IBM System/370 Guide

written in a read-only manner that will allow reloading of the module
at any time without altering execution results). Only refreshable
modules residing in the control program nucleus, the link pack area,
or an SVC transient area that are damaged will be repaired, if possible.
Processing program modules will not be refreshed.
If PDAR cannot correct the error or if the error is an uncorrectable
type, PDAR attempts to identify the task associated with the error
so that the task can be terminated abnormally. A damage report record,
which contains both the fixed and CPU extended logout area data, the
recovery action taken" the program identification, the date, and the
time of day, is prepared and logged.
System operation continues if the error was corrected or if the
error task associated with an uncorrectable error can be identified
and terminated. System operation halts. and a re-IPL is required if
an uncorrectable error damages a portion of the control program or
if the error cannot be associated with a specific task. (A special
emergency recording routine attempts to record any existing machine
check records in SYS1.LOGREC.) The operator is informed of whatever
action is taken.
MeH for the Model 165 contains model-dependent routines and will
not execute correctly on System/360 models or on another System/370
model. (See section 60:15 for a discussion of operating system
portability.)
CHANNEL CHECK HANDLER: CCH receives control after a channel error
causes an I/O interrupt. CCH formats both an error information block
(containing channel logout data) for use by an ERP routine and a CCH
error record for recording in SYS1.LOGREC. The latter contains status
information from the logout area. program identification. date, and
time of day.
If CCH determines that operating system integrity has been impaired
by the channel error, control is given to the emergency MCH routine
for error recording., and system operations are terminated. otherwise,
the error information block and error record are passed to the
appropriate device-dependent error recovery procedure (ERP), which
logs the error record and retries the failing I/O operation, if
possible, using status information from the error information block.
If a successful retry occurs" system operation continues. If the error
is deemed permanent (uncorrectable), another error record is prepared
and recorded by the outboard recorder routine (OBR). and the task
involved is abnormally terminated (unless I/O RMS or a user-written
permanent error-handling routine is present). The operator is informed
of the abnormal termination and system operation continues.
The CCH routine is structured in a manner that makes it model
independent. A channel/model-independent module resides in the
operating system nucleus. The required channel-dependent modules
included in the operating system at system generation time are loaded
during the IPL procedure by NIP (nucleus initialization program), which
determines the types of channels that are present in the system, using
the new STORE CHANNEL ID instruction and channel configuration ·lata
specified by the user at system generation time. CCH routines are
therefore compatible for System/370 Models 155 and 165 and for
System/360 Models 65, 75, 85. 91, 195, and MP/65 systems.
ERROR RECOVERY PROCEDURES (ERP's)
These device-dependent error routines are a standard part of the
control program generated for any OS environment. They will accept
and use the Model 165 channel logout data formatted by the CCH routine.
81

IBM System/370 Specifications

General

Brand	IBM
Model	System/370
Category	Server
Language	English