Got the HP9k system booted! was: New find & HP-UX p/w?

Christian R. Fandt cfandt at
Mon Apr 18 22:18:22 CDT 2005

Upon the date 15:09 08-04-05, Carlos Murillo said something like:
>"Christian R. Fandt" wrote:
> > Upon initial bootup there may be HP-UX of some version found running on it.
> > It may need a password. How would I get around it, if possible, and set my
> > own?
> >
> > It has that Series 300 DOS Coprocessor, 98286A, which I presume either runs
> > under HP-UX as a task or independently directly under DOS 3.x. That is if
> > indeed the 7958B HDD was not wiped during demil before the military
> > surplused the system. I have no idea what to expect as I'd never booted the
> > thing before. Any HP-UX "features" I should know about before getting too
> > far into running it? I do have a doc set for HP-UX 5.5.
>Unfortunately, the series 300/400 PROMS did not allow much interaction
>with the boot process; in particular, they did not allow loading the
>seconday boot loader in a boot-management mode like the series 700 did
>(which allowed you to pass parameters to the kernel so that it
>boots in single user mode).
>Once you choose the drive that you want it to boot from, it will just go
>through the complete OS startup process.
>One method that has been reported to work once out of five times or so
>is to let hpux boot all the way to the login prompt, power-cycle the
>computer and drives, and when it reboots and starts fsck-ing the
>disks (there should be a /etc/bcheckrc: " message on the console)
>repeatedly type ctrl-C, ctrl-\ or shift-(DEL/ESC)  (DEL/ESC is a key on the
>left side of HP-HIL keyboards).  I don't remember the exact key combination
>that was used, but it did work once for me about ten years ago with hpux 9.0
>and a model 380; it interrupts and gives you a single user prompt.
>Just finish fsck-ing stuff, mount -a and edit the /etc/passwd file.

Okay, we've got success up to a limit -and that limit is the need for a 
login name and password. There is indeed HP-UX of some heretofore unknown 
version loaded. The NEC monitor I recently found at the local Salvation 
Army was able to be brightened up by adjusting the G2 gun drive controls a 
little. Yippee! A decent $3.00 17" multi-input monitor!!!

Carlos, you evidently know HP-UX on these machines, so I'll address this 
section to you. Other folks may chime-in too if they have additional 
observations. This thread will add to our archival information on these 
machines -of which I've seen relatively little over the years up to now.

Please note that I know only a little about *nix systems and am willing and 
able to listen and learn from you. I've had to work with an SVR4 '486 box 
back in the mid-90's at work and have forgotten 90% of what I picked up 
then. Age and disuse of that brief UNIX experience, I guess. I got the box 
and software tucked away here at home as the application was obsoleted in 
1994 or '95 and I rescued the box and floppy disks from salvage.

So anyway, per your instructions above, I booted to the Console Login and 
cycled power off/on. This forced the machine to fsck. After seeing the 
"/etc/bcheckrc: " I chose the first option you offered, CTRL-C. That did 
nothing. I tried CTRL-DEL/ESC next. This stopped the fsck process. Here's 
what it said:

(in bcheckrc) # <blockcursor>"

Was that block cursor the single user prompt??

What is bcheckrc and why am I in it (other than the obvious fsck)?

I hit ^D which caused the machine to appear to finish its fsck. The Console 
Login prompt appears again. This is back to where I started. Hmmm, let's 
try your last interruption suggestion . . .

I power cycle off/on and let the machine completely finish its fsck process 
and boot to the Console Login prompt in order to get the fsck process 
completed correctly. I power cycle again to force a fsck and after the 
"/etc/bcheckrc: " appears I hit CTRL-\. The results were different. Here's 
what it said:

11 Quit - core dumped
(in bcheckrc) # <blockcursor>"

Well, an error and core dump seem to not be the method of choice. This is 
where I gave up. So, I power cycled the machine in order to correctly 
complete the fsck, shut it off and put it away until I could hear from the 
collective expertise from the list.

>Not for the faint of heart, I know.

Nothing like this bothers me if I have some kind of direction in which to 
point myself. I'm like Tony, I'll tackle anything hardware-wise, but this 
software needs a little bit of coaching.

>Now, the 375 has built-in scsi if I remember correctly... that was the
>difference wrt the 370 I think.  If you need to install hpux 9.x
>contact me off list.

On this 375 there is an HPIB connector bolted into the opening labeled 
SCSI/HS HPIB. I see a small daughter board attached to the back of the HPIB 
connector with a ribbon cable extending down to a header on the main CPU 

There is also a fifty-pin header on the main board meant, evidently, for 
connecting the SCSI option in place of the HS HPIB option. I would bet that 
the SCSI connector is the standard 50-pin Micro Ribbon connector used for 
SCSI-1 and fits in place of that HS HPIB connector. There is a second HPIB 
connector right on the main CPU meant for I/O and other slow speed 
peripherals and not mass storage like the HS HPIB is meant for.

Do you know if simply removing the HS HPIB connector assembly and inserting 
a 50-pin ribbon cable into that 50-pin header would enable use of SCSI?? Or 
is there a daughter card, á la the HS HPIB, involved? Evidently cannot use 
both (:-\

Another important question is whether this machine can be made y2k 
compliant? I had to replace the 3V lithium coin cell on the main CPU board 
and the machine prompted for a date/time upon bootup. Wouldn't take 
anything above 99 so I set it at 95, exactly 10 years in the past. Date 
entry is in the form of MMDDhhmm[yy]. When 05 is entered for yy the date 
always displays as 1970. Seems 70 to 99 only work.

The EEPROM firmware (chip location U47) on my machine is part number 
1818-5062 and *may* be the latest version. Upon startup the machine 
announces: "BOOTROM   Rev. 2.0   29 Nov 90". I make that version assumption 
by believing the firmware upgrade required on this website:  would be pretty 
much the last version released since the webpage had been made some years 
after my machine (ca 1991) was made. Looks like we're SOL, but that needs 
to be confirmed by the 9000/300 experts out there.

That upgrade to a 380 looks tempting as my machine is just the one they're 
talking about. Very simple, except for the compilation of a new kernel in 
Step 1. I'm primarily a hardware guy. But, I'd need to use the machine 
frequently to really justify messing around with the upgrade.

Thanks for the help again!

Regards,   Chris F.


>Carlos E. Murillo-Sanchez    email: carlos_murillo at
>Dean of Engineering, Universidad Autonoma de Manizales, Manizales, Colombia
>"Western civilization... thought like the greeks, organized itself like
>the romans and believed in itself like the hebrew." -- Ortega y Gasset.

Christian Fandt,    Electronic/Electrical Historian
Jamestown, NY  USA      cfandt at
         Member of Antique Wireless Association

More information about the cctalk mailing list