Subject: Re: VAX 4000-300 works
To: None <port-vax@netbsd.org>
From: Michael Kukat <michael@camaronet.de>
List: port-vax
Date: 06/04/1999 12:48:12
Hi !

> > Another port with usable results: the VAX 4000/300 works !
> >
> >Port status: Primary cache works, console works, all QBUS devices are usable.
> 
> There is a 128KB backup cache in there too. There are a bunch of control
> registers accessed through IPRs (i.e. MFPR to read, MTPR to write).
> 
> There's seven pages or so in the manual and I can't type it all in but here are
> the highlights:
> 
[...very useful info...]

> 	Flush and turn on Primary cache (as you already do?)		

Yes, stole the whole stuff from the KA43, which also uses the Rigel CPU. There,
the secondary cache seems to be different, it is addressed by memory mapped
cache control registers and cache/tag areas. I like the CPU-internal version
much better :-)

> This assumes that you start off with all caches off and enable_BTS=0,
> enable_PTS=0; if you start in some other state (e.g. if the console turns on
> the Pcache for you) then the sequence is different. If the above doesn't work,
> report the initial state (Pcache state, BCCTL state) when NetBSD starts to boot
> and I'll read some other portion of the table for you.

I'll try my luck, Pcache works fine, if it's really enabled, but i think so,
according to the performance of the 4000 for now.

> >To do: Find out some memory addresses to get backup cache running, and check
> >out how to handle the CPU-internal SGEC ethernet controller and the DSSI
> >controller. For now it only runs with the good old QBUS ethernet adapters.
> >(I use the DELQA from my MV II)
> 
> The ethernet address lives at 0x20084000: read in longwords and use the 2nd
> byte of each of the first 6 longwords.
> 
> As for the rest you reall do have to get the manual (EK-KA670-TM KA670 CPU
> Module Technical Manual): start at page 147 and stop at page 198 :-) Actually
> you don't stop since the SHAC description starts at 198 and finishes at 212 or
> so. (Programming the SHAC isn't enough to use DSSI devices - you need the MSCP
> stuff; I know that Matt Thomas tried to get this released some time ago but I
> don't think anything happened about that in the end).

ARGL, again this manual stuff... Do you know a source where those manuals are
available in online form ? Just to download a PDF, PS or whatever format would
be the greatest thing :-)

> KZQSA isn't too good at handling disks - I don't think its an OpenVMS
> restriction, it's just not what the hardware does best.

I thought so by looking onto the board. No "intelligence" in it, only a bunch
of TTLs and 128 KB buffer RAM. I think, DEC really thought this to be a tape
controller for the TLZ-04, so it's used according to the manual of the box.

Thanks again for the information, i'll now try to get this running and to pull
over the SGEC stuff from the booter to the kernel. (Yes, the booter already
knows about it...)

So long... Michael