Re: Crashing 5000/240

Felix Deichmann <[email protected]> Tue, 18 Jul 2017 12:27:00 +0200
Newsgroups gmane.os.netbsd.ports.pmax
Message-ID <CALX=d=fqUfqKe6KhqSZtL8Y88pD7WvWOv1pS0QtmBwPwvpZMOA@mail.gmail.com>
Hello Bj=C3=B6rn,

2017-07-16 11:59 GMT+02:00 Bj=C3=B6rn Johannesson <[email protected]>:
> Finally got around to getting NetBSD on this machine I got several years =
ago.(Which was a pain since I lost my external scsi-cdrom during my last mo=
ve.)
> I installed on gxemul and then dd:d the diskimage to an external Sun encl=
osure.
> It does however crash during boot.See https://twitter.com/herdware/status=
/886349200585367554
> I briefly talked to tnn@ and he had some vague memory about a local patch=
 he had but couldn't find it now.
> The problem seems to be some bus_space info that is inaccurate?

I tested a daily -8 from yesterday on my Pers. DS5000/33, and it works well=
.

My machine has the same "le at ioasic at tc" your crash shows, and is
similar CPU-wise (R3000A).
The crash also shows "normal_bus_mem_read_4", which is part of
macallan's switch of pmax to common mips bus_space (some time ago,
works well on my machine and gxemul so far). It could be worth a try
switching back to old pmax bus_space and see what your machine does.
Having said that, TC (and also pmax) is quite a mess, as many drivers
are not bus_space'ified yet and e. g. rely on MIPS' linear address
space...

Felix