My project has more in common with OpenDarwin (especially with XPostFacto) than with MOL so i decided to move here.
Well, i've got some results yesterday. BootX actually started working. Currently i'm working on adapting additional Forth packages (BootX installs some) to Pegasos firmware.
In any case (success or failure) source code of modified BootX will be availible in OpenDarwin CVS, consider it as a step to porting OpenDarwin too. Currently i've subscribed to opendarwin-hackers mailing list and will discuss the project there.
Success probability is ~80% by now.

I am not sure but isnt OpenDarwin already running on Pegasos?

Current source code shows that it isn't. My BootX source code is recent one and it has no support for Pegasos. The xnu kernel itself also will work only on machines with mac-io device note present in OpenFirmware tree (if mac-io is not found the kernel panics), this is also not about Pegasos.
My BootX will include a small trick which bypasses this check in the kernel. There are no other places in the code which would directly prevent Darwin (or MacOS X) from booting up on the Pegasos so i expect it to boot after this. Next problem can be absence of appropriate hardware drivers but it's also solvable because it's already solved in a project called XPostFacto for older Macs unsupported by standard MacOS X version. Unlike Linux hardware drivers in xnu are loaded and linked dynamically at boot time by the BootX so it's possible to load a 3rd party drivers (this is what XPostFacto does).

Hm, too bad I don't have a Pegasos anymore to try to help a bit in this way.

sounds promising !

Wow, this is great! I'm anxious to see this in action.

Waiting to see this discussed on hackers@, good luck!

I think the biggest problem is the drivers for the Marvell stuff, which AFAIK has no open documentation. The linux drivers might serve as a reference, but little if any code could be reused to write the IOKit drivers.

There's nothing to drive by the OS in Marvell except Gigabit Ethernet. Well, there was some code in BSD despite it was not finished. Probably can serve as code source.

I've subscribed to hackers@opendarwin.org, it's completely silent. Does noone need opendarwin? Someone on IRC told that there are no users of opendarwin, there are only some developers. But... One thing has come to my mind. There is still no working BSD version for the Pegasos. Darwin is based on BSD, may be it could fulfill this gap?

There is already a FreeBSD PPC port, which works on Macintosh machines. Once I contacted the main developer, and he wrote me, that it should not be too difficult to port it to Pegasos. He just does not have enough time to make it.
The FreeBSD PPC mailing list is available at: http://lists.freebsd.org/mailman/listinfo/freebsd-ppc
And homepage (if we can call it that ) of lead developer is available at: http://people.freebsd.org/~grehan/
CzP

To tell the truth i personally have no interest in BSD itself. There's already Linux and i see no things which can be done on BSD and can't be done on Linux. Moreover, i've had some experience with FreeBSD5.4 on x86 and got very bad feelings. As a desktop OS it is unisable, all GUI apps crash every 20 minutes, even simple things like gedit. Linux totally rules out BSD from desktops. Besides this, there are some applications for Linux which are absent in BSD. And running BSD kernel with Linux software using Linux emulation layer looks rather stupid to me, there is Linux kernel for this.
In addition ports system makes BSD totally useless on systems without unlimited high-bandwidth connection to the Internet. And even with this building KDE or Mozilla or OpenOffice from ports is... huh... terrible idea!
The only thing i liked in BSD is its kernel. It boots much faster than Linux one. But horribly low quality userland simply doesn't allow to take any real profit of it. In contrary, Linux kernel supports much more hardware and subsystems than BSD (it even doesn't have IRdA!).

Summary is: i totally frustrated and disliked BSD.

Because you tried FreeBSD at the worst possible time. 4.X was rock solid both kernel and userland. Than they made a too large step, and none of the 5.X releases were really stable. There were problems both on the kernel and the application level. I actually skipped the 5.X series where I could and jumped from 4.X directly to 6.X where it was possible.

For any release, there are a lot of prebuilt packages, you are not required to use the port system. But it's excelent to build custom systems, this is where the whole idea of Gentoo comes from.

So, you should try again with 6.0 or with a 6.1 prerelease!

CzP

I do agree with czp, the 4.x series is my all-time favorite OS for stability.
I am using the 6.x kernel right now but I keep running my main server on 4.11.

There have been a lot of changes since 4.x (perl -> c, etc...), a lot of things dropped.

Too bad it's like that...

Due to inactivity of OpenDarwin i've joined another project called FreeDarwinhttp://freedarwin.foxybanana.com.
We've set up a GIT repository and i've imported my current BootX source code there. The IRL is:
git+ssh://git@foxybanana.com/home/alex/freedarwin/bootx-devel.git
I've also got a co-developer interested in running Darwin/OS X on IBM RS6000 machine.
You're welcome to join if you wish.

The project has passed stage 1. BootX works. It succesfully loads the kernel, drivers set and launches the kernel.
I don't know what happens next, it simply hangs. I see no output from the kernel on the screen because Darwin kernel expects a graphical framebuffer and Pegasos firmware works in some weirdly-broken text mode (looks like PC VGA text mode but one character occupues 8 bytes, i've figured out by writing 256 bytes from 0 to 255 at card's PCI space base address).
Well, the next stage is: write some helper, actually a minidriver which will detect the video card and switch it to graphics mode by itself, without help of OpenFirmware.
So bad news: it will work only with GFX cards it knows; good news: it is in progress and has some success.