Subject: Re: ROM glue
To: Olivier GALIBERT <Olivier.Galibert@mines.u-nancy.fr>
From: Allen Briggs <briggs@puma.bevd.blacksburg.va.us>
List: port-mac68k
Date: 12/14/1995 09:04:03
> I'm in the process of porting macbsd to a performa 630 (yeah, the one
> with the IDE drive...)

Fantastic!

>   - To find out vectors for the 630, I get the asm code at the ROM
>     entry points for a LCIII and look for then in the 630's one (I
>     just love Rommie ;-). A problem is that vectors for the LC
>     series are of the form 408xxxxx and 40Axxxxx. Since the ROM
>     is 1Mo and starts at 40800000, they should stop at 408FFFFF. So
>     how is the MMU mapping the rom ?

Hmm...  Good question.  The ROM itself is actually mapping 16MB.  Why?
"Because that's plenty for any mac."  That could probably be dropped
back to 1MB since I think the only macs to use more than that are the
PowerMacs.

>   - I noticed in the code that some vectors are said "verified" and
>     others not.  What should I test to know if my guesses can achieve
>     the verified status ?

I'm afraid I can't answer this other than, "if they work, they're
verified."  Many of them can be checked with MacsBug, though.  At
least, ReadXPRam is a symbol that my MacsBug can find....

>      - I hacked with a battleaxe mkfs, install and the 1.8 booter

Neat.  They were forged with a battleaxe, too...  ;-)  If the Performa
630 is an 040-based machine, you'll want to pick up the 1.9 booter--at
least the called to flush the instruction caches.  I had a number of
bizarre problems while working on the Q700 until I found that bug...

-allen

-- 
Allen Briggs - end killing - allen.briggs@bev.net ** MacBSD == NetBSD/mac68k **
   Where does all my time go?  <a href="http://www.netbsd.org/">Guess.</a>