Mac Os 9 booting on: G4 Tower MDD FW800 (Detailed Posts)

[lokki]

not yet, sorry i've been busy with life 

[Jakl]

I get the same result as you Snakecoil.
Tried with the 10.4.11 first.
Then wiped HD and installed 10.2.8 in the FW800 and still the
same result with the cd drive opening and closing after the
load hd:(),Firmware and go commands.

[DieHard]

Maybe we can ship machines to Blitter and he can burn them for a small fee, LOL... The firmware God smiles upon him...

Just a suggestion, maybe you guys should compare the exact part number of Blitter's Logic Board to your own

Blitter, Can you please post your MB info. ?

[SnakeCoils]

Success! :-) GREAT SUCCESS! :-)

I have found the way to start the firmware downgrade: I were right when I said a step was missing, I found the answer reading this page on Apple site

https://support.apple.com/kb/DL1175

If you look at point 3 of the article it say:

Press and hold in the Power button until you hear a long tone or see the Power button light flashing repeatedly. Release the Power button when you hear the tone or see the Power button light flashing. The update starts automatically.

And this is exactly what is needed to do: starting the Mac keeping pressed the power button! :-)

In brief here are the steps to follow:

1) Copy the Firmware file extracted in the way Blitter has said some posts ago, no matter about the OSX installed, a blank initialized partition is good. If the partition is the first in Disk Utility list its number should be 10

2) Shut off the Mac and be prepared to do the following steps as quick as possible

3) Press and keep pressed the front Power Button until it flashes, or until you hear a continuous tone

4) Release the Power Button and be prepared to press the key combination to enter Open Firmware (Cmd + Alt + O + F)

5) When you hear the Boing sound of reboot keep pressed the Open Firmware keys until the prompt comes out

6) At the prompt enter the following line

load hd:xx,Firmware

where, of course, the "xx" is the number of the partition where the "Firmware" file is

7) At the next prompt digit

go

The screen becomes black, the progress bar appear and the CD tray is ejected

9) When progress bar has reached the end the Mac reboots, the CD tray close and you have an OS9 bootable FW800 PowerMac :-)

The Info panel reports correctly 4.4.8f2 as firmware version and the OS9 boot is available as option in Startup Disk preference panel. I used the OS9General.dmg extracted from the Install disks of my other FW400 MDD.
OS9 boot went fine, I skipped the annoying registration screen pressing CMD + Q and everithing seems ok.
For who is interested in the FW800 port I can confirm it works well under OS9 but I don't know about its speed in this environment if someone has some disk benchmark tool to suggest I will be happy to do some test. In OSX 10.4 the speed is full 80 Mb/s, I have tested it with Xbench and obtained 76 Mb/s with an old 320 Gb Seagate Freeagent powered directly from Firewire 800 port.
I don't know anything about the Bluetooth and Airport Extreme functionality since I haven't these devices.
I will perform further tests in the days to come but until now I am really really happy! :-)

About the firmware upgrade back to 4.6.0f1 I don't know... I accepted the risk of downgrade because this FW800 machine was abandoned in my Mac stuff from long time but if the MDD is a tool of your job I can't reccomend to do this step until you are absolutely sure that none of the functionality is missed or altered.

If Blitter has performed this modification long time ago maybe he can share with us his experience. Here all seems to work very well! :-)

Enjoy!

[MacTron]

Congratulations !! 

To test the Firewire speed (and others hard disks) you can use:

https://www.adrive.com/public/hpY8s3/QuickBench.sit

[SnakeCoils]

Thanks MacTron, but the most part of congratulations must go to Blitter first, without his input we would still dream of an OS9 bootable FW800 machine :-D
I have just done some benchmarks under OS9 using QuickBench and the maximum I have read is 34.42 Mb/s so I can confirm that FW800 devices runs at FW400 speed (but under OSX they run back at their full FW800 speed).
Just for fun I have tried to launch Quickbench in Classic under 10.4.11 too but in this way the volume drag'n'drop is not supported so Quickbench can't lock a drive to perform the speed tests.
In the next days I will re-install OS9 / OSX 10.4 / OSX 10.5 from scratch and if all will go as I hope I will migrate my FW400 system to the new one :-)

[Jakl]

Now we're ready for a G5 Macos9 on Steroids...

[Jakl]

Thank you blitter and SnakeCoil for the hard work to realize many of our dreams come true finally.

I am still in the process of reinstalling and testing but it looks good for the moment.

[DieHard]

This is like an old wound that never heals...

Finally... it heals with confirmed success

No more vague stories of a friend who knows a friend that is running 9.2.2 on FW800... real facts !

Wow, tests are just like I expected.. FW800 runs at 400 Speed in OS 9 and 800 in X... The last test is to try to boot to external FW hard drive via the 800 Port by holding "alt" at bootup...

I am guessing the hard drive won't be seen on the 800 port until OS is loaded (Since firmware is from a FW400), but I am curious.

[Syntho]

Would it be worth it to get a 1.42ghz FW800 model to have a faster CPU running OS9 now?

[DieHard]

Since you already have a FW400... you can buy a 1.42 Processor board and simply install that

[Jakl]

This is like an old wound that never heals...

Finally... it heals with confirmed success

No more vague stories of a friend who knows a friend that is running 9.2.2 on FW800... real facts !

Wow, tests are just like I expected.. FW800 runs at 400 Speed in OS 9 and 800 in X... The last test is to try to boot to external FW hard drive via the 800 Port by holding "alt" at bootup...

I am guessing the hard drive won't be seen on the 800 port until OS is loaded (Since firmware is from a FW400), but I am curious.

Well Diehard I'm happy to announce that it does boot on firewire800 from pressing 'alt' at startup. I have an old rockstor enclosure that has fw800, esata and usb 2.0 ports and it functions well. Booted into macos9 and 10.4.11 but must test further - tried to copy a file from a usb stick but crashed in 10.4.11.

[SnakeCoils]

Hello again, I have done further experiments and I discovered a really funny thing: this firmware downgrade unlocked the USB 2.0 speed of the integrated MDD ports... I noticed that because after installing TechTool 6 under OSX 10.5 the USB speed indicator reaches the 480 Mbit/s and I had no USB2 card installed yet...
Quickly connecting the same Seagate Freeagent of previous test to its USB2 side (it is a double interface model) reveal I were right: the measured speed with Xbench was about 24 Mb/s so the port were effectively running in USB 2.0 mode.
The bad news is that under OS9 seems to be a bad idea to connect USB storage devices to the integrated ports because the system tends to freeze and, when it happen to function, the maximum speed is in the USB 1.1 range, as confirmed by Quickbench test I have performed where the peak trasfer were no higher than 1 Mb/s.
So under OS9 don't connect storage USB devices (like HDD and PenDrives) to integrated ports but ONLY input devices like mouse and keyboard. Of course when you install a PCI card with dedicated USB 2.0 ports you can connect to it whatever you want, they are on a separate bus and works well under OS9 too (at USB 1.1 speed, since OS9 drivers does not support higher speed).
On OSX side the USB 2.0 mode seems much more stable but I have to made further experiments to determine if it is also enough safe.
In the meantime I found this article that explain the embedded USB 2.0 chips in some MDD models:

http://www.cnet.com/news/usb-2-0-in-some-power-mac-g4-mdds-update-apple-replies/

[DieHard]

You guys rock   I don't think any of this detailed info. is anywhere else... thanks again

[SnakeCoils]

After spending a considerable amount of time in testing I cant tell that "unlocked" USB2 mode on integrated MDD port is safe and does not freeze the machine under heavy load under OSX. For OS9 remains true what I wrote in the last post: better to not use these ports for storage devices.
First let me describe the FW800 machine I am using for testing:

PowerMac MDD "FW800" Dual G4 @ 1.25 GHz flashed with MDD FW400 firmware (version 4.4.8f2)
2 Gb of CL3 PC2700 RAM (30330 timings on System Profiler)
Radeon 9000 Pro graphic board VRAM 64 Mb (firmware version 113-99702-131 Apple OEM)
ATTO PSC Scsi Board with firmware 1.6.6f0 (latest)
D-Link USB2 PCI board with 4+1 ports (Nec chipset)
Hard Drive Seagate 120 Gb model ST3120024A (Apple OEM)
SuperDrive Pioneer DVR-105 (Apple OEM)

The hard drive is splitted in 3 partitions named as follow:

Boot_C9 (4Gb) with OS 9.2.2 from OS9General.dmg (Italian version) + Quicktime 6.0.3 + Classic extensions
Boot_X4 (12Gb) with OSX 10.4.11 to latest Apple updates
Boot_X5 (95 Gb) with OSX 10.5.8 to latest Apple updates + Quicktime 7.7

Additionally I have plugged to the PSC board a Wide SCSI LaCie disk at 10.000 RPM used as source for file testing, both under OSX 10.4 and 10.5 have been installed the latest ATTO drivers for this board. No drivers needed under OS9.

Since I want to test the USB2 realiability on MDD ports I have moved keyboard and mouse to the USB port of the D-Link card leaving the internal ports free for connecting the followinge devices:

TDK 8 GB USB2 pen drive
Seagate portable HD FreeAgent GO plugged to USB port and powered by an external PSU so the connection to the Mac was in the realm of data I/O only, no current drained from the weak computer's PSU.

I have done almost the same test under OS9, 10.4 and 10.5 and performed quite well, here are my results:

OS 9.2.2
Here the USB environment is 1.1 only, maximum transfer rate of 1.2 MB/sec so the main focus were on stability on intense I/O activity.
From the SCSI disk I copied a 650 Mb single file (an MKV movie) both to pen drive and to the Seagate Hard Drive at the same moment, obtained dragging the source file from the SCSI drive to pen drive and quickly after from SCSI drive to USB drive. Two progress bar windows opens and I observed the copy operation was going on, slowly but without troubles, same speed for both streams.
In the meantime I have launched the complete demostration of Graphic Calculator app and while it was running I have done an access to my remote Qnap server to download some files I needed to install later. I also have opened a couple of time the System Profiler and then doing some surfing with Classilla browser.
Well, the experience was something like to live in a slow motion movie but the copy on both devices went fine and system does not crash until... I opened System Profiler once time more and this freezed the desktop, all opened windows and ongoing processes went locked but mouse pointer was still live.
I had to manually shut down the system and then reboot to OS9 again but looking in the destination devices I discovered the files were completed so the copy was succesful, only the desktop was freezed. For completeness I have made a further experiment: unplug the Seagate drive from USB and reconnect it to Firewire 800 port then repeat the procedure above to see if the speed were different and, yes it changed but not very good: the copy stream on the Firewire dropped to 1/3 of the speed of USB stream (no other programs was running for this test) so the copy on USB side was already finished while on the Firewire side it was still at half way. It is like the USB operation steal resources to the Firewire operation but when I tried to connect to the Firewire 400 port I discovered a terrible reality:

In OS9 the stock Firewire 400 ports does not work anymore! Only the FW800 does! :-(

A quick test under OSX prove it happens on OS9 only, the two FW400 ports works well both under 10.4 and 10.5 so this could be only a driver issue, maybe in OS9 the drivers are rigidly mapped to hardware resources and if the phisical address is changed (like flashing a machine with a different machine firmware) they do not found their way anymore. So this is the first serious problem of this firmware downgrade but if you add a combo PCI card USB/Firewire this issue can be less problematic as it could seem in first place.

OSX 10.4.11 / 10.5.8

Here the MDD's USB rear ports gain back their new speed because of direct drive support, again two streams of a single large file source are copied at the same time on two different destination, in the meantime I opened Grapher (the equivalent in OSX to Graphic Calculator in OS9) and launched all the available 3D functions demonstrations: about eight different windows each with a realtime animation inside), doing a bit of net surfing, connected again to my Qnap TS-412 and all went absolutely smooth. I also have done some copy test between the two USB devices and no problems at all.

At this point I am very satisfied about the downgrade but I really would like to know if the Airport slot and Bluetooth module are working again after this modifications.

[MacTron]

Thank You for the detailed explanation.

In OS9 the stock Firewire 400 ports does not work anymore! Only the FW800 does! :-(

A quick test under OSX prove it happens on OS9 only, the two FW400 ports works well both under 10.4 and 10.5 so this could be only a driver issue, maybe in OS9 the drivers are rigidly mapped to hardware resources and if the phisical address is changed (like flashing a machine with a different machine firmware) they do not found their way anymore. So this is the first serious problem of this firmware downgrade but if you add a combo PCI card USB/Firewire this issue can be less problematic as it could seem in first place.

May be the only explanation to this is that the FW400 are subsystems of the FW800 that the firmware don't know about it. In the MDD FW400 both ports are at the same "direct level", so the this firmware don't know nothing about FW on subsystems, once flashed on a MDD FW800 whith FW ports organized in very different way.

So the FW800 at first level works at FW400 speed, but the FW400 ports are ignored for being subsystems of the main FW800.

We have to wait to DieHard opinion, he knows about this...

[Protools5LEGuy]

After spending a considerable amount of time in testing I cant tell that "unlocked" USB2 mode on integrated MDD port is safe and does not freeze the machine under heavy load under OSX. For OS9 remains true what I wrote in the last post: better to not use these ports for storage devices.
OS 9.2.2
In OS9 the stock Firewire 400 ports does not work anymore! Only the FW800 does! :-(

A quick test under OSX prove it happens on OS9 only, the two FW400 ports works well both under 10.4 and 10.5 so this could be only a driver issue, maybe in OS9 the drivers are rigidly mapped to hardware resources and if the phisical address is changed (like flashing a machine with a different machine firmware) they do not found their way anymore. So this is the first serious problem of this firmware downgrade but if you add a combo PCI card USB/Firewire this issue can be less problematic as it could seem in first place.

At this point I am very satisfied about the downgrade but I really would like to know if the Airport slot and Bluetooth module are working again after this modifications.

Great! Good news! Maybe to loose the FW-400 ports on OS9 is not a loose when you can run MacOS9 on a FW-800 !!!!!.


I think a lot of FW-800 users will love to find this unbelieble info about  booting OS9!


MORE MACOS9 MACHINES !!!!!!!

I am with Jakl

Now we're ready for a G5 Macos9 on Steroids...

 






 

[DieHard]

I had originally suspected there would be an issue with One or more of the FW ports... put I had originally expected the 800 ports not working at all; perhaps the the physical hardware address on the MB of the FW800 ports on a FW800 G4 might be the same as the hardware address of the FW400 ports on a FW400 G4... that is why the firmware thinks the 800 ports are the 400 ports in OS9...I also suspect that that is why the FW800 ports are still bootable with a FW External drive (yet the 400s are dead). 

However... it is amazing that under OSX all ports work... I would have suspected an issue in OS X also, but the drivers must be written to communicate thru a hardware abstract layer that communicates differently with the firmware... the firmware does not even know the unit has FW800 ports, yet they work.... so there is obviously more here that we do not know about in the coding... maybe there is some mutual code in the firmware (for both models) that got left in there that is benefiting us...who knows...

At any rate, if these are the only issues, Blitter's work around is still 100% success

SnakeCoils your approach is perfect... test and log results... I am sure we have out-foxed even the designers this time.  I don't think the G4 development team would even have anticipated Blitter's approach... so all we have is Trial and Error... Definitely, this opens the door for more G4s that can run Mac OS 9... and calls for a "Holy Grail" re-write of the best Stock G4 that runs OS 9.  Prior to this discovery, the best we could hope for was "stealing" dual 1.42 and putting them in FW400 MDDs... Now we can get the whole units and re-burn firmware.   I don't think we will ever get this lucky in the G5 arena... and the hardware on the G5 is so fickle that it is better we stop here anyway.

[Jakl]

So from what you said Diehard then at the open firmware level everything is opened up at startup - then it's at the startup of the Mac Operating System that depends whether the ports are used or not?

[SnakeCoils]

Thank you very much for your feedback :-) I think it is really really important to do very quickly all the possible experiments on this hybrid Mac born FW800 flashed with MDD2003 firmware because sooner the word will be widely spread and since the process is not reversible (as far as we know) I want the people that decide for this modification will know for sure what they are going to have, just to not discover bitter surprises too much late... we can only warn about the risks, but once the risk is assumed the responsabilty is entirely on the user.
Also, the research for a "soft boot" of OS9 must not to be abandoned because there are many G4 machines that are not "blessed" by an alternative firmware (later eMacs, iMacs, Powerbooks, MacMinis) so for those computer the only chance is a modified resource to cheat the system in booting the classic OS.
I suspect that the clever programmers at MorphOS Project could have some ideas about the steps to do, if someone nowadays is still digging in the G4 and G5 secrets they are for sure the right people to ask ;-)
Alternatively one could trying to do a really crazy thing: if no other checks are performed by the flasher then could be possible, at theoretical level, to flash everything has a G4 heart with the MDD 4.4.8 firmware and seeing what happens... but this has nothing of scientific, it is more or less a blind try so be prepared to brick your Mac forever.
The right way would be to make a dump of original machine firmware, make the MINIMAL necessary modification to allow the OS9 boot (without touching anything else) and then flash it back to the Mac, it could be a good idea for a Kickstarter project :-)