Error Checking - IBM 5100 Maintenance Information Manual

Error Checking
The tape adapter checks parity on both bus out and the
device address bus. If an. error is sensed, the 'machine
chepk' line halts processing and turns on the PROCESS
CHECK light. To clear this error condition, press
RESTART.
Parity generated by the tape adapter is put on bus in to
the controller. for parity checking. If the controller
senses an error on bus in parity, processing halts and
the. PROCESS CHECK light is turned On. Press
RESTART to clear the error condition.
CRC (Cyclic Redundancy Check)
Cyclic redundancy checking is a mathematical method of
checking transmitted bits to see if all bits were received.
The tape is written with a two-byte CRC added to each
header and data record. This CRC is generated by the
microprogram using the record type byte and the 512
data bytes.
Read eRC Errors
When a record is read, the CRC
IS
generated by the
microprogram and compared to the CRC bytes read
from the tape. If the generated CRC bytes are equal to
the CRC bytes read,. the next operation begins.
If the CRC bytes are not equal. the tape unit
backspaces, rereads the data record, and compares the
CRC bytes again. If the retry is successful (CRC bytes
equal), the next operation is allowed. If a CRC error still
exists after 10 retries, the tape unit stops and the
controller puts an error message on the display.
444
Write CRC Errors
When a record is written, the tape unit erases the track
to provide a. clean surface to write a new record,
backspaces, and reads the new record to check the
CRC. This is done on the same pass since the erase
coil is located ahead of the read/write coil.
If the CRC generated by the microprogram is equal to
the CRC read from the tape, the next operation begins.
If the CRCs do not compare, the tape unit backspaces
to the beginning of the record and attempts to write the
record a maximum of ten times. After the tenth write, a
hex 24 (bad record) is written in the record type byte.
The tape unit then skips to the next record where it
attempts the write operation again. There is no limit to
the number of records that can be labeled bad and
skipped.
If the tape unit is unable to write a hex 24 in the record
type byte after ten attempts, the controller puts an error
message (07) on the display.
Error Reporting
When any error condition is detected by the controller, it
is reported by an error code. Many error codes (such as
end-of-file) indicate a programming problem rather than
a tape unit problem. Several codes, such as a status
error (03) and a CRC error (07), can indicate a tape unit
problem (refer to Error Codes in Section 3).
In addition to the error codes. other information is
available in read/write storage that can be examined by
using the DISPLAY REGISTER switch on the control
panel. The most recent status byte is located at address
hex DOSF.
Many programming errors are never reported because
the microprogram automatically retries the operation for
any CRC error. These retries are seen as additional tape
motion and cause extra time for customer jobs.
/
\
.
./
\