I committed the missing functions from pmap.c and built current.  I
cross compiled GENERIC and then booted it and had it recompile its own
kernel.  I didn't run into any UVM problems, although it does panic on
an overzealous check in the the ethernet's DMA driver.  I think my
last round of DMA work introduced some new and improved race
conditions.

Darrin

Rolf Grossmann <grossman@securitas.net> writes:

> Hi all,
> 
> I've finally managed to revive my old NeXT machine and tried to catch
> up with NetBSD development. Unfortunatly, the -current kernel I've
> managed to build with my cross compilation environment traps deep inside
> UVM and I have no clue what the real reason might be. So could anyone
> out there (with a known to work setup) try to build a -current kernel
> and tell me if it's just me or if -current is indeed broken for next68k
> (then maybe we can nail down the problem). This may have to do with 
> Chuck Silvers' elimination of PMAP_NEW (you certainly have to complete
> pmap.c with some of his functions), but I doubt it as those changes were
> quite straightforward and are working on other m68k platforms. So if
> you could just confirm or deny if -current is working, that would be
> helpful.
> 
> Thanks, Rolf
> 
> PS: I love that console code. Reminds me of the times I was hacking
> NetBSD on sun4c ;)