But its not getting very good marks...
Oh, it's doing fine. It'll need changes before being accepted, that's perfectly normal :-)
I wasn't aware that it requires OFW to be resident during the kernel boot to handle the devtree.
It depends. If all you want is a device tree, you can feed it into the kernel whatever way you want (see the current PowerPC implementation, for example). If you want real services from OF, sure you need it to be there (but not necessarily at "kernel boot" time, it can be a bit earlier, too (a kernel "bootwrapper").
So it looks like its usefulness for linuxbios is minimal.
Nah, it's fine.
The good side is that it appears that a x86 implementation unified with the sparc and powerpc code will be well received.
There is no way the "OF" part can be easily unified -- but the _filesystem_ part surely can, with some per-architecture (or even more fine-grained) set of accessor routines to the actual device tree.
I'll follow up to some of the mails on LKML tomorrow, it's bedtime now ;-)
Happy new year,
Segher