I'm breaking in the new snapshot this weekend; so far it has been working
well, but I still get insane readings from the clock chip when I boot netbsd
so I have to boot single-user and fix the date before going multi-user.
(If I don't the NFS mounts are really unhappy for some reason.)

I think as a result of recent fixes, I can now use rdate instead of having
to look at another machine and type fast. Previously rdate would guarantee
a panic.

When I boot NeXTStep from the internal SCSI, it can set the clock and read
it later (also in NeXTStep) correctly.

However netbsd seems to mangle the clock time as it reads it; I'm not sure if
it sets it correctly on halt or not. The mangled time is usually many decades
into the past or future; it's quite remarkable.

Has anyone else seen anything remotely like this? I'm going to start
printf'ing the clock driver since I should be able to build kernels now.
(Plus I have my own cross-compilation technology that works reasonably
well -- however it can't speed up ``make depend''.)

BTW does anyone know of online pointers to docs for the clock chip?
I have an 040 slab and the dmesg says "new style" clock chip.

Todd Whitesel
toddpw @ best.com