It occurred to me that Wojciech Puchar wrote in gmane.os.netbsd.ports.i386:
> e) try brute-force halving. as it's about 500 .o files in kernel build 
> directory, you should in 9 tries narrow the problem to exactly one file.
> then - problem will be easy to fix.

It should be easier than that - a kernel compiled with -march=pentium will
crash almost instantly when you try to run it on a pentium. It has to be
in the very early stages of the kernel, long before most device drivers even
get a remote chance of probing/matching/attaching. IMHO this eliminates
most objectfiles.

However, this is all assuming that there is a *single* source file causing
this problem. If it's a more fundamental problem, it might occur moreoften.
That said, I've ran NetBSD on many a pentium in the past (The last one
being my old laptop, which I stopped using somewhere this spring), up to
NetBSD-2.0. I never really encountered any anomalies with compiling using
-march=pentium - just the kernel. Userland and the few applications I ran
from pkgsrc ran OK.

-- 
    Martijn van Buul - pino@dohd.org - http://www.stack.nl/~martijnb/
	 Geek code: G--  - Visit OuterSpace: mud.stack.nl 3333
 The most exciting phrase to hear in science, the one that heralds new
discoveries, is not 'Eureka!' (I found it!) but 'That's funny ...' Isaac Asimov