[comp.unix.i386] 2 hardrives of different interfaces allowed to coexist in PC UNIX?

johnk@opel.COM (John Kennedy) (08/28/90)

In article <637@slammer.UUCP> todd@slammer.UUCP (Todd Merriman) writes:
-In article <25313.26ce9f6f@kuhub.cc.ukans.edu> teoh@uicbert.eecs.uic.edu writes:
->Is it possible to allow 2 different types of hard disks in a pc UNIX
->environment (using 2 different controllers)?
-
-When I installed Interactive Unix 2.2 I attempted to use an ST-506 primary
-controller and an Adaptec 1542B SCSI controller as secondary.  After
-a *day* of re-building kernels and plugging and un-plugging controllers,
-with nary a kernel that would boot, I gave up.  Fortunately, the SCSI
-controller has floppy ports, so I decided to toss the ST-506, two MFM
-drives, and use only the SCSI.
-
-The Interactive documentation states that two different contollers is
-definitely allowed, but clearly I didn't have the magic to make it
-work!

I agree.  The documentation and menus indicate you can set up a primary ST-506
and a secondary SCSI, but what I tried was the opposite:  having an old ST-506
controller and drive available as secondary.  My attempts, too, failed.

Anyone had any luck with this?  Followups to comp.unix.i386.
-- 
John Kennedy                     johnk@opel.COM
Second Source, Inc.
Annapolis, MD

sl@van-bc.wimsey.bc.ca (Stuart Lynne) (08/29/90)

In article <25@opel.COM> johnk@opel.COM (John Kennedy) writes:
>In article <637@slammer.UUCP> todd@slammer.UUCP (Todd Merriman) writes:
>-In article <25313.26ce9f6f@kuhub.cc.ukans.edu> teoh@uicbert.eecs.uic.edu writes:
>->Is it possible to allow 2 different types of hard disks in a pc UNIX
>->environment (using 2 different controllers)?
>-
>
>I agree.  The documentation and menus indicate you can set up a primary ST-506
>and a secondary SCSI, but what I tried was the opposite:  having an old ST-506
>controller and drive available as secondary.  My attempts, too, failed.
>
>Anyone had any luck with this?  Followups to comp.unix.i386.

I have made this work under SCO Xenix (SCO also maintains that it won't).

The basic problem is that the Adaptec controller checks for a normal
controller and if it finds one will not boot. 

What I did was to install a "second" hard disk controller at the "second"
address. I didn't tell the ROM BIOS that it had any hard disks.

The Adaptec doesn't see the controller so it will boot normally.

The second problem is whether the OS will pay attention to the second
controller if the first is not in use. Xenix will. I simply had to create the 
appropriate device file's. While I had an existing disk that did not need 
preparation with the mkdev hd script I suspect just telling the script that 
you had an ST506 drive as the boot disk would work (if not it's not that hard 
to do by hand).

So the big question is whether ISC will use the controller once you have
your system setup in this fashion. And whether you will have to partition
and prepare the file systems by hand.


-- 
Stuart.Lynne@wimsey.bc.ca ubc-cs!van-bc!sl 604-937-7532(voice) 

dougp@ico.isc.com (Doug Pintar) (08/30/90)

In article <25@opel.COM> johnk@opel.COM (John Kennedy) writes:
>
>I agree.  The documentation and menus indicate you can set up a primary ST-506
>and a secondary SCSI, but what I tried was the opposite:  having an old ST-506
>controller and drive available as secondary.  My attempts, too, failed.
>
>Anyone had any luck with this?  Followups to comp.unix.i386.

There are some problems trying this.  For MFM controllers, the way you get the
geometry of the drive is usually from drive-type entries in the CMOS RAM.
Since you have to tell your system that you have NO 'AT-style' hard drives
to get a SCSI adapter to boot from one of its disks, this leaves the driver
in the dark as to what the drives on the MFM controller look like.  For ESDI
it's no problem, as you can get the geometry from the drive (unless you used
some bizarre controller translation mode when the filesystems on it were
built).  Adaptec pulled a hacko in their RLL controller to record a couple
bits of the configuration in each sector header for the first few tracks of
cylinder 0.  Must have been some INTERESTING microcode that was run when you
asked the controller for the config info...  Anyway, there's supposed to be
an IOCTL to pass both the physical and apparent geometries of drives so that
you could have some sort of init program that did this prior to opening them
the first time, but it's unclear whether it actually works.  I tried it once
with no success (on a beta 2.2 system) and haven't messed with it since.  It
may have gotten fixed since I tried it.  Good luck,
DLP

wiljo@freesid.mamnix.quest.sub.org (Wiljo Heinen) (09/03/90)

johnk@opel.COM (John Kennedy) writes:

[about the possibility of running both an ST-506 and an SCSI controller
 simultaneously on ISC 2.2]

>Anyone had any luck with this?  Followups to comp.unix.i386.

Don't know yet...

I'm gonna try next week. I already gave it a (non-succeeding) try. According
to my vendor you _have_ to jumper the 'secondary' SCSI drive to DMA chan. 6,
as ISC will use this for any secondary SCSI no matter what type the
primary controller is.

As I said - I'll try next week and then I'll inform you about the correctness
of this theory.

cheers
	Wiljo
-- 
snail-mail:	    | e-mail:			  | tel. lines:
                    |                             |
Wiljo Heinen        | wiljo@freesid.ruhr.sub.org  | voice:  +49-231-162837
Rittershausstr. 55  | wheinen@ipkama.ka.sub.org   | fax:    +49-231-140192