> Is it even remotely plausible that a method for inter-working with
> a STunOS loadable kernel module could be developed?

"AAAAAAAAAAAAAAAAAAAAAIIIIIIIIIIIIIIIGGGGGGGGGGGHHHHHH!!!!"

No.

> I imagine not for the following putative reasons:
> 
> 	(1)	they dabble into hardwired physmem in evil ways.

could be.

> 	(2)	they require more glue layer than is desirable in NetBSD.

* 1000000000

> 	(3)	even if they do use sane address/offset, structure walks
> 		down stuff like mbuf chains is too different.

yeah.

for this purpose, the entire kernel is effectively different.
enough structures have changed in different ways since the code
was common (4.2BSD) that even with lots and *lots* of work, it
probably wouldn't work...

> That said, there is a class of hardware (Sun VideoPix) which is driven by
> a StunOS lkm, and I would *LOVE* to be able to talk to it...

Sorry, i just don't see this as usable unless you get the docs
somehow, etc...

> [ ... ]
>
> 	xterm/csh dumps core at random at sub-command termination.

umm, that was a bug in libc which was introduced just before the
binaries were made, and fixed soon afterwards...  8-)


On another note:
I hope a new binary set for the sparc is done soon -- there are
several known bugs in the current snapshot that make it difficult
to use, but there's no 'well-connected' place where it can be made...



cgd

------------------------------------------------------------------------------