On Wed, 30 Oct 2002, Daniel R. Killoran,Ph.D. wrote:

> Frederick Bruckman wrote:
> >On Tue, 29 Oct 2002, Daniel R. Killoran,Ph.D. wrote:
> >>  >>
> >>  >>  and hangs. It has already done adb0 at obio0. It doesn't seem to
> >>  >>  matter whether I use a serial console or not.
> >>  >
> >>  >When you use a serial console, does that mean the mouse and keyboard
> >>  >are disconnected?
> >>
> >>  No, just that I specify it in the booter. The serial console works ok
> >>  - all the boot stuff up to the line
> >>
> >>  adb0
> >>
> >>  gets put on the serial console, just as with 1.5.3
> >
> >Well, NetBSD may not like the hardware. See what happens with no
> >keyboard or mouse attached.
>
> It USED to like it! 1.4 and 1.5 worked ok, same hardware!
>
> You are not supposed to hot-plug/unplug ADB stuff - makes it hard to
> boot unless I make the booter do it auto - I'll try that.

Please do. It's essential information to solving your problem. It's
one thing if the adb driver demands such close conformance to the
protocol that it hangs with apparently good hardware, and its another
thing entirely if the kernel doesn't work at all on 68030's.

> >  > >>  I am using a IIvx with 68 meg RAM and 2 gig disk divided into
> >>  >>  root&usr, swap, & usr.
> >>  >>  I am  using the SBC kernel, as that has always worked best before.
> >>  >
> >>  >Could you try with a custom kernel built for 68030, no FPSP?
> >>
> >>  Not until I get it to boot, which is the problem in the first place!
> >>  No way to create a custom 030 kernel for 1.6
> >
> >You don't actually need to be running NetBSD 1.6 to build a 1.6
> >kernel. The whole release is even reported to build under Cygwin. I am
> >running 1.6 (i386) and I have the sources; I'll put a MacII kernel up
> >later tonight for testing. I think it's failing too early in the boot
> >to have anything to do with floating point, though. I'd guess, rather,
> >NetBSD doesn't like something about the hardware.
>
> It is failing AFTER it does the FPU test!

So try the kernel I posted to see if it makes any difference. I guess
it won't, as the kernel is compiled with --soft-float, so there's not
much chance for floating point traps and exception vectors to cause
any problems until it gets to "init", but I wanted to rule that out.

Frederick