gabriel rosenkoetter wrote:

 > (My understanding is that we couldn't gracefully get the hardware
 > initialized without reverse engineering it all--and there's a lot of
 > it--or letting Mac OS do the work for us.)

This isn't necessarily the case. By the time the system boots, the ROMS 
have already intialized the hardware and much of MacOS itself. In fact, 
on mac68k, much of MacOS is actually in the ROMs. I think the big 
problem isn't hardware initialization (after all, most of the drivers 
can do that). I think the big problem is more along the lines of 
hardware/machine detection and possibly some interesting video memory 
issues.

There is somebody on this list that has made initial attempts at a 
bootsector boot loader and he's gotten as far as loading a second stage 
bootloader. The last I heard he was working on getting the NetBSD 
stand-alone code to work in the bootstrap environment so that he could 
load the kernel and boot. I think he was hung up on our argument passing 
scheme or something. Check the mailing list archives archives for more info.