From owner-freebsd-scsi  Thu Feb  6 12:18:44 1997
Return-Path: <owner-freebsd-scsi>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id MAA02090
          for freebsd-scsi-outgoing; Thu, 6 Feb 1997 12:18:44 -0800 (PST)
Received: from sendero.i-connect.net ([206.190.144.100])
          by freefall.freebsd.org (8.8.5/8.8.5) with ESMTP id MAA02050
          for <freebsd-scsi@FreeBSD.ORG>; Thu, 6 Feb 1997 12:18:34 -0800 (PST)
Received: (from shimon@localhost)
          by sendero.i-connect.net (8.8.5/8.8.4)
	  id NAA09331; Thu, 6 Feb 1997 13:17:01 -0800 (PST)
Message-ID: <XFMail.970206131700.Shimon@i-Connect.Net>
X-Mailer: XFMail 1.1-alpha [p0] on FreeBSD
Content-Type: text/plain; charset=iso-8859-8
Content-Transfer-Encoding: 8bit
MIME-Version: 1.0
In-Reply-To: <199702052321.SAA01744@hda.hda.com>
Date: Thu, 06 Feb 1997 11:10:41 -0800 (PST)
Organization: iConnect Corp.
From: Simon Shapiro <Shimon@i-Connect.Net>
To: Peter Dufault <dufault@hda.com>
Subject: Re: Some SCSI Questions...
Cc: freebsd-scsi@FreeBSD.ORG
Sender: owner-freebsd-scsi@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk


Hi Peter Dufault;  On 05-Feb-97 you wrote: 

...

> >     a.  What is the /dev entry?
> 
> The .ctl entry, e.g.,  /dev/rcd0a.ctl

Ah!  Almost Plan9 :-)

> >     b.  What options in config do I need to turn on?
> None.

Good!  No mistakes possible here :-)

> >     c.  (At least) header file pathname to API
> 
> sys/scsiio.h and scsi.h for the library.  Try "man 3 scsi" and "man 8
scsi".

Aha!  This is where they went.  Thanx.

> Also, I'm sending a new test release under separate mail for your input.
> 
> I don't know how well developed the reset code is.  I'd expect
> that an xs will target a given bus.  The "scsiop" escapes were
> in there to support, e.g., resetting a specific device versus
> the bus.  See the aha code. Probably Justin has to explain it from here.

A device is not my problem.  The code i have seen appears to synonym an HBA
with its bus, and reset both.  Even the Adaptec 39xx are really a single
bus device, so resetting one resets the other.

I am dealing with an HBA that has three distinct busses on the same
controller.  When and if a reset is necessary, I need a way to know what to
reset:  The HBA itself or its SCSI busses (which one?).  These are not the
same.

consider an HBA with 3 SCSI wide busses, each having only 15 devices.
Now, a SCSI_RESET command comes from above (why?).  Say why = bus timeout
(the easiest case).  Do I reset the HBA, taking down 30 disk drives that
have no problem, along with several hundreds of queued commands?

Just trying to detangle the queued, tagged, in-driver queued, etc. commands
will make you grey in the head quickly.

Now, add to this a multi-initiator, where each SCSI bus has at least TWO
HBA's on it.  each running from a separate host, maybe not FreeBSD on the
other side.  Maybe NT.  We don't know.

This is why I am trying to understand the reset policy.  The question of
differentiating driver/HBA/bus/device reset is just an opener, to get me
somewhat familiar with the surroundings.  Once we solve this problem, we
will have a very interesting set of I/O capabilities.

Thanx for all your patience...

Simon