[997] in linux-scsi channel archive

home help back first fref pref prev next nref lref last post

Re: Buslogic 445S and iomega jaz drive

daemon@ATHENA.MIT.EDU (Leonard N. Zubkoff)
Fri Nov 22 15:21:59 1996

Date: 	Fri, 22 Nov 1996 12:18:53 -0800
From: "Leonard N. Zubkoff" <lnz@dandelion.com>
To: groudier@club-internet.fr
CC: dodson@redwood.mza.com, linux-scsi@vger.rutgers.edu
In-reply-to: <Pine.LNX.3.91.961122215205.113A-100000@localhost> (message from
	Gerard Roudier on Fri, 22 Nov 1996 22:06:45 +0000 (GMT))


  Date: Fri, 22 Nov 1996 22:06:45 +0000 (GMT)
  From: Gerard Roudier <groudier@club-internet.fr>


  On Thu, 21 Nov 1996, Leonard N. Zubkoff wrote:

  > Sounds like the Jaz may have a buggy implementation of tagged queuing.  You
  > can disable tagged queuing entirely by booting with the command line
  > "BusLogic=TQ:Disable", or just for target 4 with "BusLogic=TQ:YYYYN".

  It has been reported to me that Jaz devices support tagged commands but 
  send REJECT messages in some situation, for example while spinning up.
  That is a indeed a nasty implementation of tagged queuing, but it seems 
  not buggy, as we can read in scsi-2 specifications 6.8.2:

     Tagged queuing may also be temporarily disabled internal to the SCSI 
     device during certain initialization periods or to control internal 
     resource utilization.  During these periods the target may elect to 
     return QUEUE FULL status or it may respond to any queue tag message 
     with a MESSAGE REJECT message.

I would think both of those conditions should be handled correctly by the host
adapter's firmware.  It's not really clear what has actually occurred between
the BT-445S firmware and the Jaz.  It's quite possible the BT-445S firmware
gets confused in this situation.  The BT-445S is rather long in the tooth now,
shall we say.

  BusLogic reports "Abort Tag" to the driver. I donnot know the protocol 
  between driver and BusLogic controller, but the semantic of "Abort Tag" 
  seems to me correct.

The message you saw was the driver reporting that the old 3.37 firmware does
not generate the Abort Tag message, and so cannot abort a tagged command.  It
had nothing to do with the Jaz returning an Abort Tag itself.

  The ncr53c8xx driver does not a lot better than BusLogic one with Jaz 
  devices. The solution is to disable Tagged Command Queuing too.

I'm going to see if BusLogic has a Jaz around that I can test with more recent
host adapters.

		Leonard

home help back first fref pref prev next nref lref last post