> I'm not sure if that hme or not but the following patch fixed
> 'signal 11' during NFS build for me. I have some theory about
> aliased shared pages, but it's just guess with no prove at this 
> point.

One of the sparc cpus (unfortunately I can't remember which, and
wouldn't know the sun name for it anyway) will only snoop virtual
addresses if they the addresses hit the same (set of) cache line.

So if a piece of physical memory is mapped at virtual addresses
v1 and v2 then v1 == v2 module (cache_size/number of sets).
Causes real fun if/when anything is mapped twice.

This may, or may not, already be allowed for on the appropriate
architecture - or netbsd may not support those cpus.
(They would be lurking on sBus/mBus systems.)

	David

-- 
David Laight: david@l8s.co.uk