> i'm not so sure that it is the size of the kernel here that is
> the problem -- it's the fact that there are debugging symbols in
> the program, possible pushing some size lossage somewhere..

Hmmm, yesterday I diff'd the nm output of the two kernels.  Today I
cmp'd them and they are apparently the same.

MD5 (netbsd) = 5d8de24a65b422c924cd37f1d3a83b8a
MD5 (netbsd.small) = 5d8de24a65b422c924cd37f1d3a83b8a

So in this case, strip -d was like waving a dead chicken...
I have no idea why it successfully booted after that.

Sorry for the confusion.

--sjg