Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!ames!apple!agate!shelby!portia!roadman
From: roadman@portia.Stanford.EDU (arthur walker)
Newsgroups: comp.sys.amiga
Subject: Re: X11, Ethernet, Shared Hard Disks
Summary: cltd and
Message-ID: <5236@portia.Stanford.EDU>
Date: 11 Sep 89 18:53:02 GMT
References: <22987@louie.udel.EDU> <848@boing.UUCP> <277@nap1.cds.wpafb.af.mil> <297@vixen.uucp>
Sender: arthur walker <roadman@Portia.stanford.edu>
Organization: Stanford University
Lines: 59

In article <297@vixen.uucp>, joe@vixen.uucp (Joe Hitchens) writes:
> What I want to do is this ... I have an A1000 and an A2000.  The 2000
> is just a stock 2000 with 1Meg and the fatter agnus, no SCSI controller.
> The 1000 has the older CLtd. controller.  I have one ST277N drive on
> the 1000.  I want to get another controller for the 2000 (possibly a
> GVP SCSI plus 2M ram) and put it on the SCSI bus with the 1000.
> 
> My main concerns are these:
>   * Would I have to have the newer CLtd. controller to do this?
No, I am doing it with an older CLTD 1000 host and a mac II right now.
Just get the SCSIDOS 3.x upgrade.  Believe me, it's really worth the $20.
You'll get over 300k diskperf reads under FFS with a stock 68000 and the
277N.
>   * Would the two initiators on the bus conflict in any way to mess up
>     communications on the bus assuming that the device ID's for all
>     controllers and drives are unique?
When you don't autoboot (when you don't have a Kronos) the 3.x stuff reads
a configuration file to set up all three layers of code (Exec device, 
device driver/strategy, and host-hardware low-level handler code). 
As long as you stipulate in this file that arbitration be used, all will
be fine....as long as all of the hosts arbitrate.
I don't know whether the boot drivers in the KRONOS EEPROM arbitrate or
not.  If not, you might have to keep other hosts off the bus while the
boot process went on.  More likely though, the EEPROM keeps a structure
identical to the one filled by the abovementtioned setup file, so all would
be safe.  Call CLTD.  

Also, some mac II programs don't use the SCSI manager to arbitrate when 
going underneath the filesystem (image backup... or polling for DISK_REMOVED)
and they can mess up an ongoing transfer.

>   * Would the configuration work (assuming it would work with two CLtd.
>     controllers) with a GVP as one of the controllers?

From what I understand, the GVP software doesn't now allow selection of the
host ID.  Maybe this could be cut into a boot EPROM.  So, only one GVP to
a bus, along with as many CLTD as you like.

Hardframe has a jumper, but whether their current boot code checks
it I don't know.  

Since CLTD is easily programmable, no problem.  One host
can even be set up as multiple host numbers.  No target mode though, yet.

If GVP uses the hardblok.s stuff to put junk in reserved cylinders, 
you will have to use their low-format and prep facilities.  Thereafter,
you must select the maountlist entries for the CLTD side correctly to
avoid overlapping notions of partitions.  The Kronos uses the EEPROM to
store a list of sectors, LOCATED ANYWHERE ON THE DISK, in which the 
blessed files for startup are to be taken.  So no collision with any
other vendor-specific use of the reserved early cylinders.

> ==========================================================================
> Joe Hitchens -- Artist, Sculptor, Animator of Sculpture, Iconographer Adept
> joe@vixen  ...!uunet!iconsys!caeco!vixen!joe

art walker
walker@meggie.stanford.edu
roadman@portia.stanford.edu