I tried to run NetBSD on a SPARC 5 today.  The kernel I had was pretty
old (I think it dates from before _any_ sun4m support) and all I got
out of it was "Data Access Error" (IIRC) right after the
nnn+nnn+nnn+[nnn+nnn] line.

So, think I, I'll haul over a modern source tree and build a new
kernel.  I haul over a tree supped this morning, rebuild config and
make, and build the kernel.  The resulting kernel appears to work at
first glance: it loads and goes through autoconfig without serious
problems.  Since I booted with -as (because the disk layout is somewhat
nonstandard), it then prompted for root device, default sd0a; I typed
sd0g.  I then accept the defaults for swap device and root fs type.  It
then prints the "enter shell pathname or return for sh" line, and all
of a sudden the keyboard appears to be dead - pressing return doesn't
do anything as far as I can see.  But it's not totally dead; L1-A does
get its attention.

Everything was absolutely stock; I didn't even apply any of my private
patch tree.  I did have to build init_main.o (or whatever it is that
defines main()) by hand, because it tripped the "return type of main is
not int" warning, but upon issuing that command manually with -Werror
removed, everything else went smoothly.

Anyone have any idea what I did wrong?  I was in a bit of a rush and
didn't have time to, or I'd've taken down all the autoconfig messages.
If they're important, I can try it again tomorrow.  I didn't see
anything that jumped off the screen at me as an important error, but it
wouldn't be the first time I've missed something that turns out to be
important.

The build was done on a slightly-post-1.2 system (a SS2, the one with
the kernel that didn't work on the SS5).

					der Mouse

			       mouse@rodents.montreal.qc.ca
		     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B