In message <5tpulv6jsi.fsf@highland.sibyte.com>Chris G. Demetriou writes

>Any comments?


Historically, mips OSes gave the option to request that the kernel do
a fixup of unaligned addresses: do the unaligned store or load, then
continue.  (This came in useful for porting less careful software.)
Some Ultrix binaries and some third-party sw required those fixups.


I dont think we ever finished the fixup code, but at the time, I
_think_ I was trying to be consistent with what Ultrix did; and also
thinking about the SimOS stuff the guys two doors down the hall
were doing (now better known as VMware).

If you want to change it to be more consistent with netbsd/sparc,
and netbsd/alpha does the same thing, and no-one else squeals,
that sounds fine to me.

How about creating a signal(9) manpage explaining how NetBSD ports
should be consistent with other ports in thsi area?