Subject: Re: another interesting observation...
To: Toru Nishimura <nisimura@itc.aist-nara.ac.jp>
From: Mark Abene <phiber@radicalmedia.com>
List: port-arc
Date: 02/09/2001 22:42:30
On Sat, Feb 10, 2001 at 11:08:49AM +0900, Toru Nishimura wrote:
> > I've just cross-compiled init, and the first time it actually came up,
> > prompted me for the shell, I hit enter, then it hung.
> > I entered DDB, and "ps" showed that the forked process actually execve'd
> > to "sh", and "tset" was now running but completely hung.  This is the
> > furthest I've gotten, but it was a fluke and never happened again.
> > I get the same behavior as before, init just loops forever and nothing
> > happens.  I really wish we had trace/u in DDB, but alas.
> 
> I have been thinking of EPC register value is sometimes trashed mistakenly,
> your com driver is hosed, or SR interrupt bit assignment does not match
> with hardware.
> 
> Tohru Nishimura

And where/why would EPC get trashed?  Root-device selection and even DDB work
fine, and this is over serial console, so there's nothing wrong with the com
driver.  As for the Status Register, this is an R4000.  No magic going on.
SR is as it should be.  I've verified interrupt assignments against RISC/os
header files.  Nothing odd there.

-Mark