Ok, I'm working backward to this is the deal:

My uVAX 3900 which has been pretty darn stable running 1.4.1 is now giving
me random SEGV exits. Programs that were known to cause problems:
	1.) running ping eventual segv's
	2.) tarring a directory tree from an NFS mount
	3.) Compiling the kernel. (even doing a make depend) 

The first thing I checked was for bad memory. The memory appears to be good
I've run several diagnostics on it from VMS.

The second thing I thought was "aha! Its the ethernet driver." but no, that
doesn't seem to be the problem because #3 above is all on local disks.

My current theory is that it is the three MSCP disk controllers I have
configured in. (#3 is at 160340 btw) When I ran my debug kernel things were
working pretty darn well. So I'm building a GENERIC kernel with the _only_
change is the additional disk controller.

This is my question, how on earth could a third disk controller result in
segv's in *userland* ? Every process is susceptible to it if it spends any
time at all executing.

--Chuck