Mac Os 9 booting on: xServe G4 (Detailed Posts)

[MacTron]

There is two models of DDR xServe G4:

A) A 133 Mhz System bus and ATA 100 Hard disk Interfaces and FW400.

B) A 166 Mhz System bus and ATA 133 Hard disk Interfaces and FW800.

In the A model the ATA 100 may help, but we are pushed to try the bus overclocking using the MDD procedure, never tested in the Xserve...

Getting rid the Open Firmware barrier to a  MacRISC2 kernel (Mac Os 9) it isn't easy, but it was done before. The real challenge comes now:

Comparing the PowerMac G4 MDD Architecture Block Diagram:



With the xServe G4 (133Mhz Bus) Architecture Block Diagram:



It's easy to see that the secondary PCI bus is really different from the MDD one. There is two PCI bridge interfaces ic (integrated circuit) and two ATA 100 ic. Probably Mac Os ROM 10.2.1 don't understand this part of the motherboard, so if don't bring it to a crash, it is better to ignore it by now.

So our possibilities must be limited to Uni-North 2 direct devices: Firewire, Ethernet, AGP etc...

[MacTron]

This is the Simplified Block Diagram of the xServe G4 FW800 (slot load). The FW400 model is very similar but without FW800 ports



The light red zone is the more problematic one: the ATA ic and the PCI bridge. If Mac Os ROM 10.2.1 can't "drive" the first PCI bridge we lost the CD drive, USB and serial port etc. If Mac Os ROM 10.2.1 can't "drive" the second PCI bridge we lost The PCI slots! and the possibility of booting the machine with a Hard Disk connected to a PCI card! (SATA, FW, SCSI or USB)



In this picture, the blue color zones are the problematic ones, and the red are the VERY problematic ones...

[MacTron]

So the procedure is:

- To use a basic installation of Mac Os 9.2 (ROM 10.2.1) in to a device connected to a port that this OS can "drive", We have to try FW, Ethernet, USB, PCI cards, etc...

- To trick the Open Firmware or Mac Os ROM to avoid the MacRisc3 closed door. For this we have to Hex Edit the ROM, change the NVRAM settings or in the last chance Reflash the firmware ¿?

Probably we have to remove some xServe components to try to avoid a system crash...

The successful possibilities are very low. In the best case we can achieve that the xServe boot in to Mac Os 9, but most of the components couldn't work...

In the worst case the xServe never go further than a grey screen, but at least, we try to learn something of the "inner soul" of this lovely machines...

MacTron

[IIO]

And try to trick the Open Firmware for Mac Os ROM to avoid the MacRisc3 closed door. For this we have to Hex Edit the ROM, change the NVRAM settings or in the last chance Reflash the firmware ¿?

in theory, you´d just remove the "risc3" entry (from somewhere at the beginning) of the rom. but maybe that also has disadvantes.

[MacTron]

Here we have an intro to Macintosh boot:

http://macos9lives.com/smforum/index.php?topic=1965.msg9990#msg9990

The New World ROM also sets the "compatible" property of the root node to "MacRISC2" (machines that can boot classic Mac OS using "Mac OS ROM") or "MacRISC3" (machines that can only boot OS X or another Unix-like system).

So after the machine "basic" boot procedure (Open Firmware), we have to to enable a Mac Os 9 ROM to boot on a MacRisc3 Mac (the xServe, or any other Os X only G4). Here we have a guide on how to solve this:

http://macos9lives.com/smforum/index.php?topic=1967.msg9998#msg9998

[MacTron]

In the xServe FW400 we have to change this properties in Open Firmware:
MacRISC3 and RackMac1,1

MacRISC3 should be changed by MacRISC2.

RackMac1,1 can be left as is, at first try. Later can be changed  it to  PowerMac3,6 (PM G4 MDD) u others models ¿?

[DieHard]

Mactron... I am so interested in the results of this thread. 

Circa 2003/2004 we had 2 extra Apple Xserve G4/1.33 (Slot Load) models in our warehouse and I was going to experiment converting them into OS 9 DAW Units.

Unfortunately, we had standardized on both the 933 QS and single 1.25 MDD, so when my partner and service manager heard the fans, which get pretty loud, it was a company decision to abandon research in getting 9.2.2 and DAW apps on it; which really pissed me off.  I left the company (in New York) and moved to CA at the end of 2004 for other reasons and never got a chance to experiment with another Xserve.

[supernova777]

interesting to see the inclusion of the "promise pdc20270"
this is the same chipset off the original "fasttrack" ide raid pci cards
which i was writing about suspecting them as the origin of the
"mac bootable" seritek firmware after reading the hex code inside the firmware file

[MacTron]

interesting to see the inclusion of the "promise pdc20270"
this is the same chipset off the original "fasttrack" ide raid pci cards
which i was writing about suspecting them as the origin of the
"mac bootable" seritek firmware after reading the hex code inside the firmware file

Yes, it is Interesting...

Mactron... I am so interested in the results of this thread. 

Yes, me too... LOL

Those are just "preliminary studies" ...
An "active member" of the forum  kindly offered me a machine to test ...

So, we will see ...

I'm waiting to do the real test ...

[IIO]

when i remember right, the guy where you took these pictures of the ROM hacking from never managed to come beyond the first half of the boot screen os OS9.

i would probably be much more fun to take a DDR-MDD and put it into a proper 6x19" case, right beside the PIC expansion, midi and audio IOs, ethernet switches and all this other 19" stuff.

[supernova777]

IIO do u even bother to re-read what u write before u press enter??

[blemk]

Just getting started here (this forum), but excited to see a thread started about the G4 Xserve as well. I was so close to nabbing one (dual 1.33 Ghz/167mhz) on ebay about a month ago to go do this same line of experimenting (OS 9 on it).

[DieHard]

 

Unfortunately, Initial Tests are a Failure

Preliminary Tests on Xserve G4 1.33Ghz (Single CPU)
SN# QP334024N9A

Boot Attempt 1: FireWire

Boot attempt from External FireWire 400 Hard Drive using FW800 to FW400 Cable with Modified Mac OS 9.2.2 System Folder
(ROM 10.2.1 Mofied by iMic)

Here is the picture after Holding Option Key to scan boot devices...



After Selecting the FW Drive to Boot... nothing happens and this is displayed



Boot Attempt 2: USB Flash

Boot attempt from USB Flash with Modified Mac OS 9.2.2 System Folder
(ROM 10.2.1 Mofied by iMic)

Here is the picture after Holding Option Key to scan boot devices...



USB Flash is NOT recognized as valid Boot device...


Boot Attempt 3: CD-ROM

Boot attempt Bootable CD with Modified Mac OS 9.2.2 System Folder
(ROM 10.2.1 Mofied by iMic)

Here is the picture after Holding "C" Key to boot to CD-ROM...



After Selecting the CD Drive to Boot... nothing happens and this is displayed

[supernova777]

my 2 cents...
if it was mine..
i would keep an xserve a server.. and keep a g4 a g4..
its compact it was designed to be a server.. let it serve your mac os 9 files with osx
afp is your friend!!!! afp + gigabit ethernet <3 a match made in heaven

BUT!! first swap its cpu out to an MDD steal its brain!

[DieHard]

Way too loud for me

Yes... CPU will be going into MDD and Xserve will be parted out on eBay and scrapped after iMic and Mactron are satisfied

[Jakl]

Diehard was the CD-ROM an externally connected device? If so then a fresh install on the internal HD of the X-serve via FW400/800 has yet to be tested yet then?

[Knezzen]

Diehard was the CD-ROM an externally connected device? If so then a fresh install on the internal HD of the X-serve via FW400/800 has yet to be tested yet then?

Shouldn't matter at all. Looks like a dead end getting OS9 running on these machines.
What a pity

[MacTron]

Looks like a dead end getting OS9 running on these machines.

This is what you can read everywhere, but not here, by now

Mac Os 9 booting on a xServe is a real challenge. Not a walk on park. LOL

Yes... CPU will be going into MDD and Xserve will be parted out on eBay and scrapped after iMic and Mactron are satisfied

Thank you. Now we know a bit more about this. We know that this can't be done in the "easy way". LOL

Ok. Now seriously:
What DieHard test show us is the worst scenario. Once the Mac Os ROM has to "take the control" it disappears. Can't find even to himself: the UniNorth of the xServe isn't supported by Mac OS ROM, so it can't acces to the devices directly driven by it, like Firewire.
Or may be, only the firewire chip isn't supported...

The more weird thing is the USB booting test.
To DieHard: are you sure the USB device is in good shape ie: can boot any other G4?

[MacOS9Lives.com]

Diehard was the CD-ROM an externally connected device? If so then a fresh install on the internal HD of the X-serve via FW400/800 has yet to be tested yet then?

No, I would have mentioned that... It was the Internal Slot-Load Optical drive, so at least we know it recognizes that as bootable with modified ROM

[MacTron]

Suggestions for future test:
-To try to boot from a device connected to a PCI card, like SATA drive connected to a SATA PCI card.
-To remove some Xserve components that may disable the Mac Os 9 boot, like the Hard Drive bay board ...

We have to identify the UniNorth chip to compare it to the MDD one. May be we can trick the Mac Os 9 drivers to support it, it in the way iMic do it with Sound Blaster drivers ...