>Okay, Jonathon tells me he can't get a working kernel out of yesterday's
>sup, either.

Just to clarify:

A kernel I built on 27 April from freshly-supped sources (probably
committed around April 26) works fine, on a 5k/240 and (I think)
a 5k/200.

A kernel built from source supped yesterday didn't work; it died
with the same symptoms Todd reports.

I've stopped using 3100s, due largely to space reasons: they're
too slow to be worth space to set up.

--jonathan stone