Actually, I raised the same issues (the GCC warnings or lack thereof) with
Lennart personally earlier. I'm not 100% sure this could be the problem
though; it could well be any interrupt-driven code since it's dying up in
that level. I'll try stripping my backup system back down to bare basics and
let it sit running make build for a while to see if that causes anything to
happen.

I certainly have the FTP space and bandwidth (nice perk for being at MIT =).
I'll send the details to you by personal mail.

The EBSA285 _is_ a PCI card though, is it not? At least, the one on my desk
looks like it could fit into a PCI slot. It doesn't have built-in ethernet
though. Actually, the board I've been designing has USB and PCMCIA
controllers, so there's certainly some expandability here. If I could just
find some spare cycles to get PCI->PCMCIA bridges working under NetBSD... I
looked into this earlier, but found that all we have in the source tree
right now is code for ISA->PCMCIA bridges, to the best of my understanding.

Personally, I'd still like to see real working SMP and clustering in NetBSD
-- I'm a little jealous after seeing OpenVMS and Digital UNIX in action here
at DEC. =) It's neat being able to watch systems staff take down one CPU
board and physically unplug it from the machine to replace it without taking
the system down..

-Edwin