Subject: DILOG DQ686
To: NetBSD Port-VAX mailing list <port-vax@NetBSD.ORG>
From: Erik Manders <erik@il.ft.hse.nl>
List: port-vax
Date: 06/10/1996 19:59:04
Hello all,
	Some time ago I posted a message about a DQ686 ESDI controller my
club had received in an old uVAX II. I'm pleased te report that I've got
the thing working, at least partially. I can now access and mount disks
attached to the card. I'll try booting a disk from it tomorrow. The hack
hasn't been tested a lot yet and needs to be totally rewritten before
I'd consider sending it to ragge, but it appears to work. I unfortunately
don't have docs on the Dilog.

	The 686's problem is that it doesn't report the geometry and type
correctly, as the following illustrates:

1>udaslave: I've found the following:
 >udaslave: unit 1, size 0 sectors, state 0, type 13, media id 25644036
 >udaslave: geometry: sect/trk 17, groups 15, groups/cyl 1
 >udaslave: geometry: tracks 15, cyl 0

2>udaslave: I've found the following:
 >udaslave: unit 0, size 0 sectors, state 0, type 1, media id 25641050
 >udaslave: geometry: sect/trk 0, groups 0, groups/cyl 1
 >udaslave: geometry: tracks 0, cyl 0

The disk size gets determined later and should be 0.
1) is a ra54 attached to a RQDX3, as can be seen by the geometry and the
media id. 2) is a XT4380E attached to our Dilog. The Dilog strangely
enough reports the drive an a ra80 :-) . Note the sect/trk and groups
reported as 0. Those values are later used to compute the number of
cylinders (by dividing the number of sectors) and a `devide by zero'
occurs (in uda.c, function udaonline() ).

	If anyone is interested I can mail or post the patch I rigged.

	On a related question, does anyone know what the guse_drivetype
thing in struct mscpv_guse is? I assume it has something to do with
the physical drive type (logically), but just WHAT the thing means
isn't explained as far as I can see. It just says /* drive type
index */.  At least the guse_mediaid is explained :)

  Erik Manders                                        erik@il.ft.hse.nl
--
:evil and rude: adj.  both {evil} and {rude}, but with the additional
   connotation that the rudeness was due to malice rather than incompetence.
   Hackish evil and rude is close to the mainstream sense of `evil'.
                                        --Jargon file, version 3.3.1 [edited]