Table of Contents
Advertisement
attack this type of failure effectively
by exercising the logic unti 1 a pattern
can be established ..
other diagnostic capatilities that
will also impact intermittent failures
are CE Trap (See section
1 ..
18 .. 7) and
1052 Logout (See section 1 .. 17.1 .. 3) ..
1 .. 16 ... 7 PROTECTION OF CONTROL S'IORAGE, CSL
Module 0 (first 128 control words) will be
rrctected ty the resident CSL routine on
all core loads ..
Alter control storage from the 1052 will
not be operative unless the CE key is on ..
The enable control-storage store key on
the CPU will not te operative unless the CE
key is on ..
1 ..
1~
SYSTEM FAILURE
DETEcr~oN
AND HANDLING
The objectives of failure detection and
handling procedures for the Model 25 are to
rrcvide:
a..
Assurance to the custoroer that the
systero is functionally orerational ..
b..
Sufficient indication and information
to enable the CE to locate the failure
rapidly and accurately ..
A version of the
SEREP
program is also
provided ..
1 .. 17.1 ERROR HANDLING
The M:>del 25 will corr€ to a hard stcr on
second error, a second malfunction which
occurs before the first malfuncticn has
caused a machine- check interrupt to be
taken (Figure 1-119) ..
There are no special provisions on the
Model 25 for continued operation in the
rresence of a solid malfunction ..
Error restart capatilities for such
intermittent I/O failures as data checks .•
punch checks, etc.
are provided by the
operating system used by the custcroer.
Nonhardware-caused troubles (jams,
fail ur:es, etc.)
are normally net expected
1-114
(7/69)
to result in a customer-rerorted call.
The
Model 25 Functional Characteristics Manual,
Form A24-3510, provides restart procedures
for these situations; also consult the
index in this manual for the specific area
of interest.
1 .. 17 .. 1.1
~achine-Check
and Channel Control
Check
Any of the following machine-check
conditions (if the M-bit of the PSW is on),
will cause the corresponding bit to be set
in the error register (MC) and initiate the
machine-check trap routine:
O
File Control Check
1
Storage Protect Check
2
Storage Address Check
3
Control Word Parity Check
4
Storage Data Parity Check
5
ALU Error Latch
6
A-Reg Parity Check
7
B-Reg Parity Check
The machine-check trap routine stores
the following information into program
storage starting at hex location 80.
80 - Trap priority register CMMSK)
81 - Branch condition register (BA)
8 2 - Machine- check register (MC)
83 - Error count*
84-85 - Backup address (address of the
microword)
86-87 - Unused
*Bits 1-2 of byte 83 contain the logout
code:
00 for machine check or channel
control check; or 11
f
cr interface control
check.
If none of the 0 through 6 bits of the
MMSK register is en (indication that an I/O
trap is not in process), the logout area
above is printed on the console printer.
The machine-check interruft will then take
place ..
If an I/O trap is in process (any of the
O
through 6 bits of MMSK register on) when
a machine- check trap occurs, the I/O
operation involved will be terminated and
then the I/C interrupt will take place.
The logout area will be printed on the
console printer-keyboard before any CPU
instructions are executed.
Advertisement
Table of Contents
