grundo

From Computers Wiki
Jump to navigationJump to search

The Motorola Net6200/166 is a machine I had been trying to find for about four years, then get running for about half a year.

Specs

system-id
00 ff ff 58 08 00
name
FirePower,Powerized_LX MP
Processor
2x PowerPC 604e @ 166 MHz
Memory
8 72-pin SIMM module slots
Hard disk
Seagate BarraCuda 4 Series 4.29 GB 7200RPM Fast Wide SCSI 1MB Cache
Display adapter
Cirrus Logic GD5446 Rev A
VRAM
2 MiB
Ethernet adapter
ZNYX NetBlaster ZX345 (SA0025) -- Windows NT detects this as a DEC PCI Fast Ethernet DECchip 21140
SCSI card
Adaptec AHA-2940UW
Expansion
4 Conventional PCI slots, 3 16-bit ISA slots
Clock battery
Duracell 3V DL1/3N (one-third N)
Serial number
MP00176
Motherboard label
MLU, LX SERIES 01-07582-03 (FIREPOWER SYSTEMS INC), serial LXM01113
Firmware version
Version 03.03
Firmware timestamp
1996-09-19,01:47:29
Capacitors
21 of 25NXA220MEFC10X12.5 CAP ALUM 220UF 20% 25V RADIAL

Quirks

  • The motherboard is labeled Firepower Systems, Inc. instead of Motorola.
  • When the system was first powered on, it booted enough to give a "no RAM" beep. Later attempts did not give a beep. There are four diagnostic LEDs on the board labeled A, B, C, D. When booting with RAM, light A stays on. When booting with no RAM, light C stays on. When booting with no processor card, no lights stay on. The beep no longer happens. Purple believes this may be due to the capacitors no longer being able to function properly and only managed to work for the initial power-on before becoming useless for further attempts.
  • After recapping, we discovered that when COMM 1 is attached to serial at 19200 baud, the system prints "AbcdefgBCD", first the first seven characters, then a pause, then the last three. Sometimes, it gets to "AbcdefgBCDE" after a second pause. Sometimes, there is a garbage character instead of this 'E'. Our hypothesis after this was that the machine doesn't like non-ECC RAM.
  • After installing two sticks of EDO ECC 60ns RAM, the machine finally boots.
  • The machine cannot be run without a keyboard. It will read garbage off the keyboard port and endlessly try to interpret invalid Forth commands, as seen from the serial console.
  • Serial output only goes past "install-console" if no display is connected.
  • Apparently the hard disk is using a beta firmware. dev /pci/scsi@3 .properties outputs v1.00 Beta 7 for version.
  • The registry files found on disk do not work with any of the available offline Windows NT password manipulation tools available. I had to manually find the encrypted LM hash in the registry, decrypt it, and then brute-force the original password for the Administrator account (which is: computer). See: Old-style Windows NT LM hash recovery
  • If you reboot the computer from within Windows NT, when the system reboots into the firmware, it will claim to not detect a keyboard (using COM1 instead) and then hang. The Num Lock key on the keyboard stays on.
  • I have not yet been able to successfully boot from the optical drive. My retail Windows NT 4.0 Workstation CD bombs out immediately in Windows Setup about not being able to load the CD-ROM device, and the device it tries to load does not change with the value of OSLOADPARTITION. Other CDs, like for old versions of PowerPC Linux distros, don't get past the Forth interpreter regardless of how I try to boot the files on them.
  • The back of the case calls the serial port "COMM 1" (2x M), while the firmware calls it "com1" (1x M).

Reconstructing the boot configuration

The hard disk is configured like this:

  • Partition 1: 10 MB FAT16 partition. This contains firmware-readable executables.
  • Partition 2: Points to an extended partition (partition 5).
  • Partition 5: NTFS partition.

Given this configuration, I was able to boot off the hard disk with the following commands:

setenv OSLOADPARTITION multi(0)scsi(0)disk(0)rdisk(0)partition(2)
boot disk:,\OS\VENEER.EXE \OS\WINNT40\OSLOADER.EXE

Speculation / History

  • The computer originally came from Marquette University. There is an FTP log on the hard disk with many entries for vmsb.csd.mu.edu. Additionally, the registration information in the System Properties dialog says Engineering, Marquette University. It originally had a:
    • hostname of eng142
    • domain of eng.mu.edu
    • static IP address of 134.48.93.142
    • subnet mask of 255.255.255.0
    • default gateway of 134.48.93.100
    • primary DNS server of 134.48.1.32
    • secondary DNS server of 134.48.1.31
  • Apparently Microsoft discontinued Windows NT for PowerPC a mere month after PowerStack II machines entered production.[1]
  • In response to there being an existing commercial third-party PowerPC Windows NT application on disk, Revexia says:

    For the PowerPC port of NT? Wtf, I mean like I knew that was possible but I struggle to imagine why anyone would actually want to use a PPC for NT. I can think of a million purpose built NT machines that would just have been easier to use. Wonder if that university just had a bunch of expensive PPCs leftover and a mandate to transition to Windows for integration with other systems. Let us know what you figure out!

Dr. Dennis Brylow's email

Dear Hunter,

This is fascinating. The machine you speak of would likely predate me; When I got to Marquette in 2005, there were no such unusual machines in service in the then Math/Stats/Computer Science Department. To my knowledge, I was the only one on campus running Yellow Dog Linux on PowerPC-based machines in the 2005-2010 time range. Engineering is a separate college, and has their own system administration team, so they might know more. But we'd definitely need to seek out folks who were involved in lab infrastructure at the time.

I can suggest two leads, if you haven't contacted them already. Mr. Steve Goodman <steven.goodman@marquette.edu>, our computer science department director of technology, did a prior stint here at Marquette the decade before my arrival, and may have had contact with this Net6200 machinery. I know we had a ton of PowerPC-based "Java Stations" in some of the computer labs in Cudahy Hall back in the day; they were all relegated to storage in the back room by the time I got here, but it sounds similar to what you're asking about. There were dozens of them.

I had briefly investigated them as backend machines for our RISC-based embedded operating systems infrastructure, but there was very little documentation available anywhere on them, and the work wasn't going to be replicable at other schools if they couldn't get their hands on those things. It would be somewhat unusual for Engineering to have such equipment lurking in their labs if they weren't already being used at scale elsewhere on campus. Your theory that Eng142 probably had a prior life rings true. I can't imagine Engineering investing in a PowerPC-based WinNT workstation, but it definitely could have been a repurposing of a relatively expensive piece of hardware after one of the special-purpose computer labs had moved on to new workstations. If so, its original home would probably have been in the MSCS Department in Cudahy Hall, or perhaps even earlier before the move over from the Physics building in the 1990's.

My second lead would be to reach out to Brad Bonczkiewicz <brad.bonczkiewicz@marquette.edu>, director of technology over in the College of Engineering. He's been around here a long time, and might know something about how this machine came to be in the eng.mu.edu domain.

Let me know if you'd like me to make electronic introductions to either or both of these two gentlemen. I'm certainly curious to hear what you learn.

Best Regards, Dennis

Brad Bonczkiewicz's email

Hi Hunter, sorry I do not have any additional information about the computer. Our records go back around 20 years which wouldn't include this system. While I've been in my position 15+ years, I have made every effort to ensure the hard drives are wiped before being disposed. If this was university property, it is critical we delete all user data, configuration files/accounts/passwords, and Marquette-licensed software for both legal and liability purposes. I would greatly appreciate if you would not connect the computer to the Internet and erase the drive when you are completed. While I do remember Windows NT 4 as a student in college, I don't remember the Motorola hardware and its unusual to find a PowerPC that is not inside of a Mac. Very interesting find and project to restore. Good luck

Things to archive

The disk image is "motorola.img" on the NAS.

  • There is a surviving third-party commercial PowerPC Windows NT application! It is a version of something called ExecSoft Diskeeper.
  • Installer for Microsoft Internet Explorer 3.01 for PowerPC
  • Various hotfixes specifically for PowerPC (checked with file on Linux)
  • The copies of Veneer, OSLoader, HAL, and the drivers directory (requested by Rairii at: https://haqueers.com/@Rairii/111851384631484546)

Documentation

Potential documentation

What a successful boot looks like over the serial port

AbcdefgBCDElxhHjklmBCDEoFGIJKLMN$1234#noFGIJKLMN$1234#stand-init

/pci/isa/interrupt-controller init

init-slaves

master-cache-on

/pci/isa/rtc

/pci/isa/nvram

init-nvram-buffer

init-options

init-security

/ clock frequency

more-memory

?bailout

scrub-memory

hardware-kludge

copy-reboot-info

nvramrc

probe-all

probe-isa

probe-pci

get-mac-address

probe-onboard

probe-ide

init-scsi

init-ide

install-console

History

  • 2023-09-09: The computer is purchased at VCF Midwest. The seller states that the computer had been stored in a barn near Racine.
  • 2023-09-10: The computer is temporarily transported to my parents' place.
  • 2023-09-28: The computer is transported to the new home.
  • 2023-12-16: 📈 The clock battery is replaced. Before replacing, it was measuring 2.9 V.
  • 2023-12-16: The capacitors on multiple components are identified as bad.
  • 2024-01-14: 📈 21 capacitors are replaced (8 on processor card, 13 on motherboard.
  • 2024-01-24: 📈 The correct RAM is installed.
  • 2024-02-04: The original Administrator password is recovered.

References