- Help EVRJB
EVRJB is a VAX-11 level 3 supervisor supported diagnostic designed to
diagnose ML-11 block mode memory systems on VAX-11 central processors.
It is the result of a combined effort to convert the PDP-11 versions
ML-11 exerciser and prom maintenance programs to run under the VAX-11
diagnostic environment.
The exersicer portion of EVRJB is a bliss program which runs in stand
alone mode and exercises up to 8 ML-11 units off a single RH7xx mass
bus adaptor. ML-11 subsystem faults are detected but not isolated by
the exerciser since other diagnostics are available for trouble
shooting the exact cause of these failures.
The prom maintenance program portion is also a bliss program and is
physically part of the exerciser but when run is totally disjoint
from it. This program is executed when the program section 'mlpm'
is included in the CLI command 'start/sec:' or 'run/sec:'. The
prom maintenance program will exercise a single ML-11 unit with all
ones data, all zeros data and random data to find and mask all cell
failures within memory arrays.
An ML-11 unit is a fast random access block mode mos memory system
with ecc capabilities. It is made up of three control modules and
up to 16 array modules. It is a mass bus device and such conforms
to mass bus standards.
- Attach
Once the Diagnostic Supervisor has been loaded, the device you wish to
test has to be attached and selected. Examples of this for 11/780 and
- 1/750 shown below.
- 1/780
------
DS> ATTACH RH780
Device Link? HUB
Device Name? RHn ;n decimal 0 to 7
TR? n ;n decimal 1 to 15
BR? n ;n decimal 4 to 7
DS> ATTACH ML11
Device Link? RHn ;n from above RHn
Device Name? EMan ;a Letter A to Z
;n decimal 0 to 7
Number of Array Boards? n ;n decimal 1 to 16
CHIP Size? nk ;n 16 to 64
DS> Select EMan ;a & n from above EMan
- 1/750
------
DS> ATTACH RH750
Device Link? HUB
Device Name? RHn ;n decimal 0 to 7
BR? n ;n decimal 4 to 7
DS> ATTACH ML11
Device Link? RHn ;n from above RHn
Device Name? EMan ;a letter A to Z
;n decimal 0 to 7
Number of Array Boards? n ;n decimal 1 to 16
Chip Size? nk ;n 16 or 64
DS> Select EMan ;a & n from above EMan
Once the UUT has been attached and selected the Diagnostic must be lo-
aded and started.
Examples of this for 11/780 and 11/750 are shown below.
- 1/780
------
DS> LOAD EVRJB
DS> Start/Switches
- 1/750
------
DS> LOAD EVRJB
DS> Start/Switches
- Sections
There exist five program sections within this Diagnostic
and they are:
Probe Excon Mlex Mlpm Default
The Exericser and Prom Maitenance Program are both designed to
be started via the '/SECTION:' switch and the operator should
avoid using the '/TEST:' switch to start or control program
execution.
The necessary operator inputs are provided via the conversation
sections to adequately control either programs execution.
- Probe
This is defined as test 1 and is a default section. This
section is executed first in a test sequence and probes all
registers of attached units to insure existance. This is
done to prevent non existent drive accessing from hanging
the VAX 11/7xx.
- Excon
This is the exerciser's conversation section and is a non
default program section. If this section is selected to be
executed via the DS> 'start' or 'run' command the Exercisers
software parameters may be altered. If this section is not
selected then the Exercisers default software parameters are
choosen. This section is defined as Test 2.
- Mlex
This program section is the section which contains all the
exercisers program test options and probe and is the main
default section of EVRJB. This section is defined as being
test 1 and test 5 thru 9.
- Mlpm
This program section contains the ML-11 Prom Maintenance
program. It is a non-default section and is defined as test
1, 4 and 10. It is executed by including this program section
name into the CLI command 'START/SEC:'. When this section is
selected only the Prom Maintenance program code is executed with
the exerciser code inhibited.
- Default
The exerciser code as describe in program sections 'MLEX' and 'PROBE'
is defined as EVRJB default program sections. However when the Prom
Maintenance program is called 'MLEX' default program section will be
inhibited from executing and only Prom Maintenance is performed.
- Device
The devices supported by this diagnostic are solely ML-11 units.
- Event
Defined within the Exerciser are two event flags. The following is a brief
description of thier usage.
Event flag 21
Default is false (0)
Purpose is to control the testing of soft errors within a units
sector space.
When this flag is set and an option has completed testing all
sectors within a unit the sectors are read back and ECC logic
is utilized to detect memory cells which might have lost their
charge. When a cell looses it charge in this fashion it is defined
to be a soft error.
This function was primarily designed for DMT testing of ML-11 units.
Program flow can be interupted at any time and this event flag state
can be changed to alter further program execution.
Event flag 22
Default true (1)
Purpose is to control the inhibiting of single bit errors (sbe)
printouts to the console terminal during program run time.
In any event all detected SBE's will be logged into the report
summary code section for operator review.
Program flow can be interupted at any time and this event flag state
can be changed to alter further program execution.
- Quick
A quick verify pass is included into the exercisers functionally
via the DS> flag 'quick'.
If the quick flag is set via 'set flags quick' only the following
test options will be run with single bit error logging and optional
sbe reporting.
1. A check that the basic commands work for each unit via the
test 3 'integrity test code'.
The Write, Write Check and Read commands will be issued to
each available ML-11 unit. A one sector transfer is suff-
icient for this and no extensive data checking is done.
Failure at this level of testing will point to the need to
run EVRJA ML-11 Logic test.
2. Test 5 option 1
3. Test 6 option 2
Only a shortened version of option 2 will be used during
quick verify. Just one of the patterns will be run and no
loop reading of sectors will be done.
4. Test 7 option 3
The longer options, March and Random patterns, are not part of the
quick verify pass.
Program control will be returned back to DS> once this quick verify
pass has been completed.
- Summary
Report Summary Code is performed at the end of a pass of a test
sequence or when the operator interupts program execution and does
a DS> summary command. Program execution may then be resumed by
doing a DS> continue command.
The Exercisers Summary Report Code reports the number of bytes
written, read, write checked, drive drop status, hard and soft data
failure, array modules requiring prom maintenance and locations of
failing single bit errors.
The Prom Maintenance Program Summary Report Code reports where the
program has found failing mos ram chips and the number of bad rows
and columns found in error within each.
- Manual
There presently does not exist any operator intervention tests
implemented within this diagnostic.
- Condition
Condition codes are the means by which the Prom
Maintenance Program reports to the operator
PM'ing errors detected during program run time.
- A
This indicates that one all bad chip (greater than
10 all bad rows and or columns) has been detected
in a chip at a given bank.
This chip is not Prom Maintenanced and the error
correction is expected to correct the failing
data from this chip until field service can replace
the array module from which this chip resides.
- B
This indicates that a second all bad chip has been
detected in a given bank.
The bad chip is not Prom Maintenanced and further
testing of this array is aborted.
Field service should replace this array module
before leaving the site.
- C
Indicates that bad nibble offsets have exceeded 14
resulting in possible unsafe errors.
The arrays nibble offsets for this bank are masked
up to 14 offsets. The error correction is expected
to correct the unmasked errors which were left
behind.
Field service should replace this array module at
the earliest possible date.
- D
This indicates that Prom blast errors were detected.
This means that the selected new prom data was not
written into the proms correctly. Either the
selected prom locations were not written correctly
or other prom locations were inadvertently written
into.
This could result in either bad locations not being
masked or good locations being masked.
Later routines will determine if this situation
warrents the array to be replaced.
If no further error messages occur after a condition
D message then the operator can assume that condition
D occurance did not result in any drive unsafe errors.