Steven E Lumos <slumos@bo.Egr.UNLV.EDU> writes:

[3100 still dies]

Thanks for trying the kernel. Hm.  So much for that idea:(.


>Are there any specific things I should be trying?

No, I'm afraid not. I am currently almost out of clues.
However, I've built a third kernel, netbsd.3100 in the same directory
on ftp.netbsd.org as the previous two. Could you try that one, too?

This kernel is built with the previous version of sii.c -- the one
used from 10-Oct-1996 until 29-Apr-1997 -- and makes all the spl()
levels on the 3100 be splhigh().

If *that* still breaks, and everyone affirms that the problem only
appeared sometime around April/May, then I'm actually rather skeptical
that the problem has anything to do with changes to the disk driver,
per se.

I'm afraid the next step is to revert back to the assembler and linker
versions I was using in April, and to see if _that_ fixes this
problem.

Again, thanks to everyone who's tried out kernels over the last month.