SunOS 4.0.3 driver vs. Wren drives

[peter jensen]

	Apparent sc driver change causes timing problems with Wren drives:

We recently upgraded *one* of our machines from 4.0.1 to 4.0.3 to reassure
ourselves that this was a good thing to do.  Since the upgrade, the system
has "hung" eight times.  Here's the scoop:

	Hardware:
		Sun 3/50M-4 
		CDC Wren IV (94171-300) SCSI disk
		Exabyte 8mm cartridge tape drive
		Laserprinter

The system will boot and run fine for a few days and then hang, possibly
displaying the following message several times:

(This also happens without the Exabyte connected.  The SCSI bus is
properly terminated.)

	sd0:  sdtimer: I/O request timeout
		last phase= 0x4 (DATA IN)
		csr= 0xc407  bcr= 65535  tcr= 0x1
		cbsr= 0x45 (DATA IN)  cdr= 0x24  mr= 0x2  bsr=0x89
		target= 0, lun= 0  DMA addr= 0x3600 count= 2560 (8192)
		cdb= 8 6 7d ba 10 0

I had sent in a tape containing two crash dumps and talked to a CSC rep
several times about this problem.  This morning, I got a call from the rep
explaining that the problem was due to changes in the driver to optimize
it for faster disks.  The rep did acknolwedge that there is now a timing
problem with slower drives.

Sun's position on this is that Wrens (and Exabytes) are not supported
products, and customers should call Sun Consulting to BUY a new driver to
resolve the problem.  Personally, I don't feel someone should have to buy
a custom driver for a standard SCSI device.  Also, if Sun doesn't want to
support the Wren drives, they should remove the misleading information
from format.dat:

	# @(#)format.dat 1.20 89/03/26 SMI
	#
	# Copyright (c) 1987 by Sun Microsystems, Inc.
	#
	# Data file for the 'format' program.  This file defines the known
	# disks, disk types, and partition maps.
	#
		...
	#
	# This is the list of supported disks for embedded SCSI.
	#
	disk_type = "CDC Wren III 94161-155" ...
	disk_type = "CDC Wren IV 94171-344" ...
	disk_type = "CDC Wren V 94181-702" ...
		...

Unless Engineering starts receiving more bug reports about this problem,
they will not be doing anything to resolve the timing issue. 

FLAME ON

Aren't there enough {yp | nfs | network | compiler | your choice here}
bugs already, without needing to introduce additional problems?  We're
still waiting to get an NFS problem fixed - Is anyone else having problems
with their remote files getting corrupted (some bytes being replaced by
garbage or nulls)?

FLAME OFF

I hope other customers will place calls with CSC and show them how many
sites they're affecting with this driver BUG.  My service order number is
327515.

Peter C. Jensen     			          jensen at jarsun1.ZONE1.COM
Jordan Apostal Ritter Associates    ANSYS, ARIES, and Engineering Analysis
Admin Bldg 7, North Kingstown RI 02852    (401) 884-3014 or (401) 294-4589