A kernel from the nathanw_sa branch is claimed to work
with mips3 ISA, but it doesn't currently work on my pmax.
Everything appears to be in place, but I suspect there
is some underlying assumption about TLB operations.

mi_switch in : pid=0x0, curproc=0x802a3d00, l=0x802a3d00, new=0x0
...
pmap_asid_alloc: curproc 1 'swapper' segtab 0x809b5000 asid 1
mi_switch in : pid=0x1, curproc=0x809b6000, l=0x809b6000, new=0x0
trap: TLB miss (load or instr. fetch) in kernel mode
status=0x80004, cause=0x808, epc=0x802c0000, vaddr=0x6af0
pid=2 cmd=asc0:0 usp=0x0 ksp=0xc3977dc8
Stopped in asc0:0 at    0x802c0000:     lb      a1,27376(zero)
db> 

The problem is around mip1_cpu_switch_resume() or with
the code in locore.S marked /* XXX XXX XXX */.
Setting breakpoints seems to affect how it fails.

I guess it would be good to know if other mips1 machines
are trying to run a kernel or if r4000 pmax machines can
run a kernel.

	-- Gregory McGarry <g.mcgarry@ieee.org>