> OK. Problem #1:
> panic() doesn't work until things are farther along in the
> boot process :-(. Linux has been panicing all over the place,
> but I can't see the panic message :-(.
Known problem; therefore I suggested to use the tiny console code from
Milo. Luckily I didn't have to worry about that kind of problems for
looong times. You'll soon reach that point, too - won't you?
> Problem #2:
> The config name is DESKSTATION_RPC44, but in some places it
> was _RPC (note, no 44). Fixed these.
Sorry, I just tried to provide some hooks for the rPC44 - and failed :-(
> Problem #3:
> The IRQ stuff is panicing because there is no case for RPC44.
> I'm fixing it as well.
Probably a copy of the Tyne stuff should work for you.
> Anyway, I've hacked PANIC to print stuff when things aren't too smart
> Again, another tip: If you don't get any output, then maybe Linux is
> panicing before it has enough brains to print error messages. Maybe
> we should use the BIOS routines if no video address is yet set????
The problem is that the ARC BIOS enters the "program state" when Milo
calls the launch() subroutine. This also means that calling the BIOS
has no defined result anymore. In fact at that time the kernel is where
important BIOS variables are supposed to be, eg. You'd have to hack
your own hardware level print routines. The Linux kernel once had
these but I removed them ...
I'll look through my collection of old source code snipets and send you
what I find.