Subject: Re: retina console
To: Donn Cave <donn@u.washington.edu>
From: Michael L. Hitch <osymh@gemini.oscs.montana.edu>
List: amiga-dev
Date: 05/30/1994 13:33:54
On May 29, 10:33am, Donn Cave wrote:
> Does anyone have a Retina working as a console?

  The Retina Z3 support has only been done on a real old (relatively
speaking) kernel, I think.  Markus Wild did one that I think was
based on the sources on ftp.eunet.net (probably the #744 kernel).
I've added the Zorro III address space handling from Markus's code
to the kernel sources so that I could get the Warp Engine support
working.  That stuff should be in the current kernel sources.

> It looks to me like the source I'm building, ca. May 22, doesn't support
> that.  Briefly, the general console initial initialization probes "ite"
> out of constab, and the ite probe calls console_config() in the amiga
> autoconf.c.  There the "mainbus" and "grfcc" devices are explicitly
> semi-initialized.  When control returns to the ite probe, it finds only
> grfcc device, which eventually becomes the console more or less by default.

  I don't think the Retina console configuration has been completely
added to the current kernel.  The config_console routine appears to
have the ztwobus config disabled.  There's a comment about ztwobus
knowing that the configuration isn't "for real", but I didn't see
any special handling in ztwobus dealing with that.

> In order to configure the Retina in time for this console competition, if
> I understand what's going on here, the Zorro bus mapping would have to be
> advanced to that point as well, so that the Retina device would have a
> functional address.  (In my case it's the Retina Z3 and hence the Zorro
> III bus.)

  The Zorro II I/O address space is already mapped when the initial MMU
map is set up.  All the Zorro II devices have to do is figure out the
appropriate virtual memory address that maps to their board address.
The Zorro III I/O address space had been reserved, but nothing is
intially mapped.  The zthreebus() configuration takes care of mapping
those address spaces when each Zorro III device is configured (currently
only the Warp Engine).  I'm not certain what needs to be done in ztwobus()
and zthreebus() to handle the console initialization.  I think Chris Hopps
is probably the only person who throughly understands the new configuration
process.  [I understand it enough to convert some of the SCSI drivers
and to add the Zorro III stuff, but some of the details are still rather
hazy.]

> I'm preparing myself to take a crude stab at doing that, but it sounds a
> little dicey.

  I had been thinking about adding the Retina Z3 driver to the sources
before the config.new stuff showed up.  Now that I've gotten the 53C710
and 5380 drivers mostly converted and have a working system using the
current sources [I'm currently running with the May 26 sources right
now], I can probably take a look at adding the Retina Z3 driver, unless
you really want to tackle it.  I can probably answer some questions
about the config process and Zorro III bus stuff if you need some help.

Michael

-- 
Michael L. Hitch			INTERNET:  osymh@montana.edu
Computer Consultant			BITNET:  OSYMH@MTSUNIX1.BITNET
Office of Systems and Computing Services
Montana State University	Bozeman, MT	USA

------------------------------------------------------------------------------