IBM System/360 2050 User Manual

MAINTENANCE PHILOSOPHY
The system provides facilities that assist the CE in:
Failure Detection
Fault Location
Deferred Maintenance
Rapid Repair
Failure Detection
Error detection is accomplished by hardware
parity checking, fault locating tests (FLT's), and
functional diagnostic programs. Not all failures
can be detected however.
Failure Detection By Hardware
The central processing unit (CPU) is parity
checked. All data transfers between registers
and main storage are checked by byte. Good
parity is always written into main storage by
design (except FLT's).
Pal'ity checking is also used on the read-only
storage (ROS) at the output.
Any multiplexor and selector channel circuitry
not used by the CPU has its own parity checking
circuitry.
Each error check is individually indicated on the
CE-operator's console. To provide a high diagnos-
tic resolution, the individual checks are OR'ed in
groups by timing and function and are also indicated
on the console.
Fault Locating Tests (FL T's)
FLT' s are semi-automatic diagnostic procedures
used in the Model 50 CPU. The procedure involves
a battery of diagnostic tests, each performed by
forcing the CPU to a predetermined state (scan-in),
running the CPU for a specified number of machine
cycles (clock advance), logging the resulting state
into storage (scan-out), and then comparing the
resulting state of the selected machine area under
test (compare) with a pre-computed result.
Functional Diagnostics
The functional diagnostic programs are the highest
level of testing a system. They are system pro-
grams that check for correct system operation.
Functional diagnostics have been written for all
areas of the Model 50.
DIAGNOSTIC TECHNIQUES
Deferred Maintenance
The logout facility allows maintenance activity to be
deferred. Certain intermittent failures will be such
that long time periods elapse between interruptions.
Troubleshooting this type of fault is time-consuming.
The failure environment is difficult to reproduce.
The customer will often prefer to continue system
operation and postpone maintenance activity under
these circumstances. The logout feature allows the
customer to continue system operation while pro-
viding the CE with information on the system state
at the time of the error.
TROUBLESHOOTING FLOWCHART
Flowcharts for troubleshooting the 2050 Processing
Unit are now in the Maintenance Diagrams Manual
for the 2050.
Figure 1 shows the status of indicators after sys-
tem reset.
Roller
Indicators
Common Chan 2
SBCR unpredictable
Common Chan
3
BUFFER 1 and BUFFER 3 reset
BUFFER 2 set to
previous contents of BUFFER 1
Mpx Chan
4
BUFFER 1 and BUFFER 2 unpredictable
Mpx Chan
5
BUS OUT unpredictable
Mpx Chan
6
POLL on, MPX 1/0 STATS unpredictable
Selector Chan
1 B-REGISTER P-bits on
Selector Chan
2 C-REGISTER P-bits on
Selector Chan
3
BYTE COUNTER P-bits on
Selector Chan
4
CYCLE COUNTER stepping
Selector Chan
5
POS REG TRF, SP-D2, and INSN SCAN on;
A-CLOCK stepping
Selector Chon
6 FIN on
CPU 1
1 L-REGISTER = PSW Backup
CPU 1
2 R-REGISTER P-bits on
CPU 1
3 M-REGISTER P-bits on
CPU 1
4
H-REGISTER P-bits on
CPU 1
5
SAR bit 17 on, BYTE ST ATS unchanged, BYTE
STORE STATS set to 1
CPU 1
6 ROSDR = Halt Loop
CPU 2
1
ROSDR = Halt Loop
CPU 2
2
ROSDR = Halt Loop, MOVER FUNCTION
unpredictable
CPU 2
3
ONE SYL OP on, REFETCH unpredictable. NEXT
ROS ADDRESS = Halt loop, ILC unpredictable
CPU 2
4
1/0 REGISTER P-bit, L BYTE CNTR, and M
BYTE CNTR on; F-REG, EDIT STATS, ond CARRY
unchanged
CPU 2
5
LSAR =Hex 17, LSFN = 1, J-REG = 7, MD= 3,
G 1 and G2 have S- and P-bits on
CPU 2
7
CURRENT ROS ADDRESS = Halt Loop
CPU 2
8
PREVIOUS ROS ADDRESS= Halt Loop
Most other indicators ore off, but for definite status refer to machine logic.
FIGURE 1. STATUS OF INDICATORS AFTER SYSTEM RESET
Diagnostic Techniquea (3/71) 11

IBM System/360 2050 Specifications

General

Brand	IBM
Model	System/360 2050
Category	Computer Hardware
Language	English