COS-Tools and Andras Tantos’s Cray Simulator Question #2
Replies: 8 comments 2 replies
|
Thanks for reporting the problem with hello.sh. Code has been pushed to resolve the issue. Please provide a copy of your test program. Code produced by cal and ldr has been tested successfully in an environment where DtCyber serves as a frontend to COS 1.17. Binaries produced by ldr can be uploaded to DtCyber running NOS 2.8.7, then transmitted to COS 1.17 using the Cray Station interface. This preserves the Cray file structure created by ldr, and the binaries load and run successfully. I suspect that build_exp_disk does not preserve the file structure properly. For example, perhaps it wraps it unnecessarily in a second layer of file structure. If you provide a copy of your test program, I can try running it using the Cray Station interface. |
|
I had overlooked the attachment of your test program earlier. After seeing it, I assembled and loaded it and then uploaded the executable to host NCCMAX, a CDC Cyber 865 at the Nostalgic Computing Center (http://www.nostalgiccomputing.org). NCCMAX is connected via a Cray Station interface to a Cray X-MP running COS 1.17. The test program loaded successfully and ran, but it aborted because it does not include an operating system call to terminate gracefully. Consequently, an illegal instruction is encountered after the "S2 S2+S3" instruction at the end of the program. When I modified the program to add a call to the "ADV" macro after "S2 S2+S3", the program ran to completion and exited normally. The "ADV" macro executes an OS call to advance to the next job step, i.e., the usual way that a COS program signals normal completion. Here is the job output after adding the ADV macro: Here is the NOS procedure file used for executing the COS executable. The executable is uploaded to the NOS host in binary mode using Kermit: Here is the source code of the version of test.txt that ran to completion successfully: |
|
Hi Kevin,
Thanks once again for looking at this-delighted it worked. On my end I
have DtCyber compiled,NOS2.8.7 up and
working and a test site.cfg to connect to Tanto's simulator.
Going to take a day or two looking at the manual and docs on
bitsavers-it's quite a learning curve!
Best wishes,
Ian
…On 08/04/2023 13:59, Kevin Jordan wrote:
I had overlooked the attachment of your test program earlier. After
seeing it, I assembled and loaded it and then uploaded the executable
to host NCCMAX, a CDC Cyber 865 at the Nostalgic Computing Center
(http://www.nostalgiccomputing.org). NCCMAX is connected via a Cray
Station interface to a Cray X-MP running COS 1.17. The test program
loaded successfully and ran, but it aborted because it does not
include an operating system call to terminate gracefully.
Consequently, an illegal instruction is encountered after the "S2
S2+S3" instruction at the end of the program. When I modified the
program to add a call to the "ADV" macro after "S2 S2+S3", the program
ran to completion and exited normally. The "ADV" macro executes an OS
call to advance to the next job step, i.e., the usual way that a COS
program signals normal completion.
Here is the job output after adding the ADV macro:
|06:59:37.5975 0.0007 CSP CRAY XMP-14 SN 302 LEADING EDGE TECHNOLOGIES
04/08/83 06:59:37.6004 0.0008 CSP 06:59:37.6039 0.0009 CSP CRAY
OPERATING SYSTEM COS 1.17 ASSEMBLY DATE 02/28/89 06:59:37.6081 0.0010
CSP 06:59:37.6107 0.0011 CSP 06:59:37.6371 0.0013 CSP JOB,JN=CRAYRUN.
06:59:37.7423 0.0105 CSP ACCOUNT,AC=,APW=,UPW=. 06:59:37.7465 0.0110
CSP 06:59:37.7493 0.0111 CSP
.......................................................................
06:59:37.7522 0.0112 CSP 06:59:37.7923 0.0123 CSP ECHO,ON=ALL.
06:59:37.8026 0.0130 CSP OPTION,STAT=ON. 06:59:37.8171 0.0160 CSP
FETCH,DN=BIN,MF=FE,DF=TR,^ 06:59:37.8239 0.0190 CSP TEXT=.
06:59:42.2974 0.0193 SCP AAHR 08.36.26.USER,Z160000,. 06:59:42.2994
0.0193 SCP AAHR 08.36.26.CHARGE, * , 06:59:42.3013 0.0193 SCP AAHR
08.36.29. TRO - PF AND TAPE TRANSPARENT OUTPUT. 06:59:42.3033 0.0193
SCP AAHR 08.36.30.TRO: DATASET BIN , 63B WORDS, 0.277 06:59:42.6143
0.0193 SCP SS004 - DATASET RECEIVED FROM FRONT END 06:59:43.9493
0.0210 CSP BIN. 06:59:43.9828 0.0216 CSP END OF JOB 06:59:43.9877
0.0219 EXP SY005 - BIN 512 WRDS, 1 IOS, 1 REQ, 1 STRS, .019 SEC
06:59:43.9901 0.0219 EXP - 29-1-32A 18 STRS READ: 1 REQ, 1 STRS, .021
SEC 06:59:43.9997 0.0220 CSP 06:59:44.0022 0.0221 CSP 06:59:44.0068
0.0224 CSP CHARGES 06:59:44.0145 0.0225 CSP CS032 - DATASET NOT LOCAL:
06:59:44.0182 0.0226 ABORT AB025 - USER PROGRAM REQUESTED ABORT
06:59:44.0204 0.0226 ABORT - P=00001417B TASK-ID=0001 06:59:44.0228
0.0226 ABORT - BASE=00775000 LIMIT=01153000 CPU-NUMBER=00
06:59:44.0250 0.0226 ABORT - JOB STEP ABORTED. |
Here is the NOS procedure file used for executing the COS executable.
The executable is uploaded to the NOS host in binary mode using Kermit:
|.PROC,CRAYRUN,F=CRAYBIN. REWIND,F. COPYBF,F,ZZZCBIN.
REPLACE,ZZZCBIN=CSTAGE. CSUBMIT,CRAYJOB,TO. REVERT,NOLIST.
.DATA,CRAYJOB. JOB,JN=CRAYRUN. ACCOUNT,AC=CRAY,APW=XYZZY,UPW=QUASAR.
ECHO,ON=ALL. OPTION,STAT=ON. FETCH,DN=BIN,MF=FE,DF=TR,^
TEXT='USER,nosuser,nospass,BIN=CSTAGE.CTASK.'. BIN. |
Here is the source code of the version of test.txt that ran to
completion successfully:
|TITLE 'TEST' SUBTITLE 'TEST' IDENT TEST COMMENT 'TEST' ENTRY TEST EXT
ADV START TEST TEST S1 =* S2 S2+S3 ; CLASS=1, OR=1, FC=3 R ADV ;
advance to next job step END |
—
Reply to this email directly, view it on GitHub
<#2 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AK7IQO7WURV2PJHZCD53KSTXAFOKPANCNFSM6AAAAAAWWUAZJE>.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
|
Hi Kevin,
Thanks for the mails-I have the two emulators talking together (on the
DtCyber console on the B,A side
I see a line for crs and it is reporting cray cos 1.17 02/28/89).On the
cray side the Link command shows a connection
with the FE id.So all looks happy.
However.....
when I run /HELLO,ICF I get an application not present as per the
screenshot. Running /LIMITS shows
that ICF is a valid applicaion for the guest user.Using the INSTALL user
also produces an "invalid application"
message. When I use help to produce a list of commands ICF is not listed...
Anyway,will try a batch job as see what happens and have a look in the
docs to see if there is any clue as
to why ICF appears to be missing.
Once again I really appreciate your help and time.
Best wishes,
Ian
…On 09/04/2023 13:58, Kevin Jordan wrote:
P.S.: The Cray Station subsystem on NOS 2.8.7 includes the |ICF|
network application (Interactive Cray Facility). This allows you to
log into COS interactively from NOS. To use it, log into NOS, then
execute the command |HELLO,ICF|. This will disconnect your terminal
from the NOS interactive facility (IAF) and connect it to the Cray
Interactive Facility (ICF). You can then interact with COS in the same
way that you would use interactive mode according to Tantos' "Cray
Files" discussion.
—
Reply to this email directly, view it on GitHub
<#2 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AK7IQOZKC35ZEUTVHYBJBUDXAKW5ZANCNFSM6AAAAAAWWUAZJE>.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
|
Hi folks, Note the full stop (period) after the NOS password and the addition of GET to get the file to be local. With this, I get the following output on my 865/XMP system: Not sure why there is an abort at the end. Some program COS expects that has sadly been lost? |
|
Hi Nick,
Yes, the abort at the end is due to a missing program that would
normally perform accounting at end of job.
cheers,
Kevin
…On Tue, May 2, 2023 at 8:02 AM Nick Glazzard ***@***.***> wrote:
Hi folks,
I found I needed a slight variation of the CRAYRUN CCL for it to work.
This seems to do things for me:
.PROC,CRF,F=CRAYBIN.
REWIND,F.
COPYBF,F,ZZZCBIN.
REPLACE,ZZZCBIN=CSTAGE.
CSUBMIT,CRAYJOB,TO.
REVERT,NOLIST.
.DATA,CRAYJOB.
JOB,JN=CRAYRUN.
ACCOUNT,AC=CRAY,APW=XYZZY,UPW=QUASAR.
ECHO,ON=ALL.
OPTION,STAT=ON.
FETCH,DN=BIN,MF=FE,DF=TR,^
TEXT='USER,nosuser,nospw.GET,BIN=CSTAGE.CTASK.'.
BIN.
Note the full stop (period) after the NOS password and the addition of GET
to get the file to be local. With this, I get the following output on my
865/XMP system:
18:47:04.6753 0.0002 CSP CRAY XMP-14 SN 302 LEADING EDGE TECHNOLOGIES 05/02/83
18:47:04.6763 0.0002 CSP
18:47:04.6774 0.0003 CSP CRAY OPERATING SYSTEM COS 1.17 ASSEMBLY DATE 02/28/89
18:47:04.6784 0.0004 CSP
18:47:04.6792 0.0004 CSP
18:47:04.6885 0.0005 CSP JOB,JN=CRAYRUN.
18:47:04.7184 0.0036 CSP ACCOUNT,AC=,APW=,UPW=.
18:47:04.7197 0.0038 CSP
18:47:04.7206 0.0038 CSP .......................................................................
18:47:04.7216 0.0039 CSP
18:47:04.7312 0.0043 CSP ECHO,ON=ALL.
18:47:04.7343 0.0046 CSP OPTION,STAT=ON.
18:47:04.7388 0.0056 CSP FETCH,DN=BIN,MF=FE,DF=TR,^
18:47:04.7408 0.0065 CSP TEXT=.
18:47:09.7228 0.0066 SCP AAHZ 18.48.28.USER,NICK,.
18:47:09.7238 0.0066 SCP AAHZ 18.48.28.CHARGE, * ,
18:47:09.7248 0.0066 SCP AAHZ 18.48.31. TRO - PF AND TAPE TRANSPARENT OUTPUT.
18:47:09.7259 0.0066 SCP AAHZ 18.48.31.TRO: DATASET BIN , 71B WORDS, 0.148
18:47:09.9985 0.0066 SCP SS004 - DATASET RECEIVED FROM FRONT END
18:47:10.9748 0.0072 CSP BIN.
18:47:10.9793 0.0073 USER HELLO WORLD!
18:47:10.9870 0.0075 CSP END OF JOB
18:47:10.9886 0.0076 EXP SY005 - BIN 512 WRDS, 1 IOS, 1 REQ, 1 STRS, .007 SEC
18:47:10.9895 0.0076 EXP - 29-1-22A 18 STRS READ: 1 REQ, 1 STRS, .008 SEC
18:47:10.9923 0.0076 CSP
18:47:10.9932 0.0076 CSP
18:47:10.9947 0.0077 CSP CHARGES
18:47:10.9961 0.0077 CSP CS032 - DATASET NOT LOCAL:
18:47:10.9972 0.0078 ABORT AB025 - USER PROGRAM REQUESTED ABORT
18:47:10.9980 0.0078 ABORT - P=00001417B TASK-ID=0001
18:47:10.9988 0.0078 ABORT - BASE=00775000 LIMIT=01153000 CPU-NUMBER=00
18:47:10.9996 0.0078 ABORT - JOB STEP ABORTED.
Not sure why there is an abort at the end. Some program COS expects that
has sadly been lost?
—
Reply to this email directly, view it on GitHub
<#2 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AABALGUU27A67SBPZDZDQJ3XEDZWHANCNFSM6AAAAAAWWUAZJE>
.
You are receiving this because you commented.Message ID:
***@***.***>
|
|
... and thanks for pointing out the text (".GET,") that was somehow dropped
from the CRAYRUN procedure example.
…On Tue, May 2, 2023 at 8:34 AM Kevin Jordan ***@***.***> wrote:
Hi Nick,
Yes, the abort at the end is due to a missing program that would
normally perform accounting at end of job.
cheers,
Kevin
On Tue, May 2, 2023 at 8:02 AM Nick Glazzard ***@***.***>
wrote:
> Hi folks,
> I found I needed a slight variation of the CRAYRUN CCL for it to work.
> This seems to do things for me:
>
> .PROC,CRF,F=CRAYBIN.
> REWIND,F.
> COPYBF,F,ZZZCBIN.
> REPLACE,ZZZCBIN=CSTAGE.
> CSUBMIT,CRAYJOB,TO.
> REVERT,NOLIST.
> .DATA,CRAYJOB.
> JOB,JN=CRAYRUN.
> ACCOUNT,AC=CRAY,APW=XYZZY,UPW=QUASAR.
> ECHO,ON=ALL.
> OPTION,STAT=ON.
> FETCH,DN=BIN,MF=FE,DF=TR,^
> TEXT='USER,nosuser,nospw.GET,BIN=CSTAGE.CTASK.'.
> BIN.
>
> Note the full stop (period) after the NOS password and the addition of
> GET to get the file to be local. With this, I get the following output on
> my 865/XMP system:
>
> 18:47:04.6753 0.0002 CSP CRAY XMP-14 SN 302 LEADING EDGE TECHNOLOGIES 05/02/83
> 18:47:04.6763 0.0002 CSP
> 18:47:04.6774 0.0003 CSP CRAY OPERATING SYSTEM COS 1.17 ASSEMBLY DATE 02/28/89
> 18:47:04.6784 0.0004 CSP
> 18:47:04.6792 0.0004 CSP
> 18:47:04.6885 0.0005 CSP JOB,JN=CRAYRUN.
> 18:47:04.7184 0.0036 CSP ACCOUNT,AC=,APW=,UPW=.
> 18:47:04.7197 0.0038 CSP
> 18:47:04.7206 0.0038 CSP .......................................................................
> 18:47:04.7216 0.0039 CSP
> 18:47:04.7312 0.0043 CSP ECHO,ON=ALL.
> 18:47:04.7343 0.0046 CSP OPTION,STAT=ON.
> 18:47:04.7388 0.0056 CSP FETCH,DN=BIN,MF=FE,DF=TR,^
> 18:47:04.7408 0.0065 CSP TEXT=.
> 18:47:09.7228 0.0066 SCP AAHZ 18.48.28.USER,NICK,.
> 18:47:09.7238 0.0066 SCP AAHZ 18.48.28.CHARGE, * ,
> 18:47:09.7248 0.0066 SCP AAHZ 18.48.31. TRO - PF AND TAPE TRANSPARENT OUTPUT.
> 18:47:09.7259 0.0066 SCP AAHZ 18.48.31.TRO: DATASET BIN , 71B WORDS, 0.148
> 18:47:09.9985 0.0066 SCP SS004 - DATASET RECEIVED FROM FRONT END
> 18:47:10.9748 0.0072 CSP BIN.
> 18:47:10.9793 0.0073 USER HELLO WORLD!
> 18:47:10.9870 0.0075 CSP END OF JOB
> 18:47:10.9886 0.0076 EXP SY005 - BIN 512 WRDS, 1 IOS, 1 REQ, 1 STRS, .007 SEC
> 18:47:10.9895 0.0076 EXP - 29-1-22A 18 STRS READ: 1 REQ, 1 STRS, .008 SEC
> 18:47:10.9923 0.0076 CSP
> 18:47:10.9932 0.0076 CSP
> 18:47:10.9947 0.0077 CSP CHARGES
> 18:47:10.9961 0.0077 CSP CS032 - DATASET NOT LOCAL:
> 18:47:10.9972 0.0078 ABORT AB025 - USER PROGRAM REQUESTED ABORT
> 18:47:10.9980 0.0078 ABORT - P=00001417B TASK-ID=0001
> 18:47:10.9988 0.0078 ABORT - BASE=00775000 LIMIT=01153000 CPU-NUMBER=00
> 18:47:10.9996 0.0078 ABORT - JOB STEP ABORTED.
>
> Not sure why there is an abort at the end. Some program COS expects that
> has sadly been lost?
>
> —
> Reply to this email directly, view it on GitHub
> <#2 (comment)>,
> or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/AABALGUU27A67SBPZDZDQJ3XEDZWHANCNFSM6AAAAAAWWUAZJE>
> .
> You are receiving this because you commented.Message ID:
> ***@***.***>
>
|
|
Thanks to investigation by Nick Glazzard, the reason that ICF was missing
has been determined, and code has been pushed to the DtCyber repository to
resolve the problem. The problem was due to a "leftover" IF statement in
the ICF startup job (leftover from usage in another context). The IF
statement allowed ICF to be started only if it would be running on a host
with machine ID "AX" or "ED".
If you execute "git pull" against your copy of the DtCyber repo and then
reinstall NOS 2.8.7, the problem should be resolved. You can also fix the
problem in your already installed system by editing the NAMSTRT file under
username NETOPS. The very last record in the file is the ICF startup job.
Remove the "IF" statement that tests for HID being either "AX" or "ED", and
then save the file. For example, log into username INSTALL, and then
execute this sequence of commands:
GET,NAMSTRT/UN=NETOPS
SCREEN,WEBTRM
FSE,NAMSTRT
locate the IF statement near the end of the file and remove it and its
corresponding ENDIF statement
then exit from FSE
REPLACE,NAMSTRT/UN=NETOPS
Then execute the following command from DSD (i.e., the NOS console) to
start ICF:
X.NAMI(RS=IC)
cheers,
Kevin
On Sun, Apr 9, 2023 at 12:55 PM nobunagashino ***@***.***>
wrote:
… Hi Kevin,
Thanks for the mails-I have the two emulators talking together (on the
DtCyber console on the B,A side
I see a line for crs and it is reporting cray cos 1.17 02/28/89).On the
cray side the Link command shows a connection
with the FE id.So all looks happy.
However.....
when I run /HELLO,ICF I get an application not present as per the
screenshot. Running /LIMITS shows
that ICF is a valid applicaion for the guest user.Using the INSTALL user
also produces an "invalid application"
message. When I use help to produce a list of commands ICF is not listed...
Anyway,will try a batch job as see what happens and have a look in the
docs to see if there is any clue as
to why ICF appears to be missing.
Once again I really appreciate your help and time.
Best wishes,
Ian
On 09/04/2023 13:58, Kevin Jordan wrote:
>
> P.S.: The Cray Station subsystem on NOS 2.8.7 includes the |ICF|
> network application (Interactive Cray Facility). This allows you to
> log into COS interactively from NOS. To use it, log into NOS, then
> execute the command |HELLO,ICF|. This will disconnect your terminal
> from the NOS interactive facility (IAF) and connect it to the Cray
> Interactive Facility (ICF). You can then interact with COS in the same
> way that you would use interactive mode according to Tantos' "Cray
> Files" discussion.
>
> —
> Reply to this email directly, view it on GitHub
> <
#2 (reply in thread)>,
> or unsubscribe
> <
https://github.com/notifications/unsubscribe-auth/AK7IQOZKC35ZEUTVHYBJBUDXAKW5ZANCNFSM6AAAAAAWWUAZJE
>.
> You are receiving this because you authored the thread.Message ID:
> ***@***.***>
>
—
Reply to this email directly, view it on GitHub
<#2 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AABALGWL4SSPXSUVCNDZBQLXALSXHANCNFSM6AAAAAAWWUAZJE>
.
You are receiving this because you commented.Message ID:
***@***.***>
|
Uh oh!
There was an error while loading. Please reload this page.
Hi,
I have been experimenting (1) with COS-Tools and Andras Tantos’s Cray Simulator and think I have
Found a few problems with the CAL assembler and maybe the link loader.
(1) The hello.sh causes the cal program to hang immediately taking almost 100% of the cpu and producing no output.
(2) I have a small test program,test.cal which assembles and links without any errors. Once I try to run it in batch or interactive mode (2) an “invalid program load table” error is reported and the program aborts.
So to be honest I don’t know if I am doing something stupid or there really is a problem with the cal/ldr programs. Any Advice/hints/tips welcome.
To assist I will try to upload the following files.
Test.txt --the assembler source of my tiny test program.
Cray-assembly.jpg shows the assembler command line and the ldr output.
Cray-run-result.jpg shows the error output from an interactive job. A Batch job produces the same error.
(1) Feels a bit like blundering around in the dark-I have never used a mainframe style computer so quite a learning curve. Fun though..
(2) I use the build_exp_disk script to load the executable on the virtual disk which can then be fetched into the CPU.
(3) You might ask why I am bothering with this? As a child of the late 1960’s CRAY computers have held something of a special spot(4). So now that I can finally get my paws on one I really want to write a program to run 2+2 and print the result.. I would also suspect ,on a more serious note, that there are computer science students out there who find this of interest. Might be an interesting way to learn assembly code.
(4) Personally, I have only seen CRAY computers in the science museum-my late father once worked on one which was in Wythenshawe Manchester owned by SHELL.
test.txt
All reactions