IBM System/370 Guide

configuration comparable to that of the current System/360 model.
(Certain BPS programs have known timing dependencies that prevent their
successful execution on a Model 165.)
Note that a control program not generated specifically for the Model
165 does not include support of the Model 165 CRT and keyboard as the
primary console device. Therefore, such a control program cannot be
used on the Model 165 unless it supports a device as a console that
is also present on the Model 165. The console device must have the
same I/O device address on the Model 165 as it has on the current
system. In addition, Model 165 RMS (machine and channel check routines)
are not included in these programming systems, and the Model 165
operates under the conditions listed below.
• Single-bit processor storage errors will be corrected by ECC
hardware, but IPL disables this interrupt (recovery bit masked).
• CPU retry will not occur, since it is disabled at IPL. Any CPU
error, including those caused by a buffer failure, will result
in a hard machine check condition and generation of CPU extended
logout data.
• External Damage and Time of Day Clock Damage are enabled and cause
a machine check condition.
• Any hard machine check error (an uncorrectable or unretryable CPU
error, a double- or multiple-bit processor storage error, or a
storage protection failure) will cause a hard machine check
condition and generation of CPU extended logout data.
A machine check control switch that determines what action is taken
when a machine check condition occurs is present on the system console.
When this swi~ch is set to the PROCESS position, machine checks cause
an interrupt and a logout if they are not disabled. This setting is
to be used when an operating system containing Model 165 RMS is in
operation. When the switch is in the STOP ON CHECK position, all
machine checks cause a hard stop without an interrupt or a logout.
If the system is not set to hard stop after a machine check when
an operating system without Model 165 RMS is used, the system takes
whatever action was planned for machine checks, as follows:
• A control program without a recovery routine included (for example,
SERO
I
,
SER1, MCH for OS" or MCRR for DOS) normally enters the wait
state after a machine check interrupt and logout. The logged data
can be obtained with a stand-alone storage dump routine or displayed
on the graphic console. The operator can re-IPL and attempt to
continue operations or the CE can perform diagnostic procedures.
• A control program that contains a recovery routine will enter the
routine and attempt execution. As stated in Section 10:05, these
routines access model-dependent data and will not operate properly.
In addition, the extended logout data stored when the interrupt
occurred will have destroyed the code at locations 512 through
1511. Results are unpredictable. Therefore, the system should
be set to hard stop if a recovery routine is present in a control
program not generated for the Model 165.
For the following reasons it is advantageous for Model 165 users
to install an operating system that includes recovery management support
designed specifically for the Model 165:
• The number of re-IPL's necessary because of machine malfunctions
can be reduced. Most hardware errors will be corrected either
by Model 165 hardware recovery procedures or by RMS routines.
95

IBM System/370 Specifications

General

Brand	IBM
Model	System/370
Category	Server
Language	English