Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!wuarchive!nicsn1.monsanto.com!kncarp
From: kncarp@nicsn1.monsanto.com
Newsgroups: comp.periphs.scsi
Subject: 1542B & MD21's working - kinda.
Message-ID: <1991Jun14.094730.3133@nicsn1.monsanto.com>
Date: 14 Jun 91 09:47:30 -7
Organization: Monsanto Company, St. Louis, MO
Lines: 55

I managed to get my Adaptec 1542B to drive my two Emulex MD21 controlled disk.
The key was to disable the BIOS and strickly use the driver code.  Although this
works, it is less than perfect.  Any suggestions on how to address the described
problems would be appreciated.

1) Drive formatting and verifying:
   The 1542B BIOS contains the format and verify routines for attached drives.
   In order to be able to use these routines I needed to do the following:
   A) Address and terminate a single drive as SCSI-ID 0.
   B) Enable BIOS (default) and boot machine with cable disconnected.
   C) Connect cable at DOS prompt and enter DEBUG.
   D) Run format and verify.
   E) Goto A) for second drive.
   Comments:  It appears the SCSI RESET issued by the BIOS upon startup
   causes the 1542 to hang (Red light stays on).  By disconnecting the 
   cable until after the BIOS is done with its startup, the RESET never
   gets to the disk and the controller is happy.  Unfortunetly, the only
   thing one can do after that processing is run the DEBUG or DMA test. 
  
2) Drive partitioning:
   A) Disable BIOS via jumper and install drivers in config.sys.  Make sure
      the /NORST flag is on the primary driver (ASPI4DOS).
   B) Cable everything up and reboot machine.
   C) Run AFDISK for each drive - just like the manual says. 
  
3) Drive Use:
   A) Simply reboot after all partitions are defined and the new drives will
      be available.

Problems:

Since the BIOS is disabled, due to the RESET problem (solved with the software
drivers /NORST flag), it appears that the INT 13 hook is never installed.  I
saw a flag (/I) that would allow me to force the hook not to be installed, but
nothing that worked the reverse way.  Although the disk appear to function 
under DOS (4.01 BTW), several problems exist.

A) CHKDSK won't work.

B) SuperPCK disk caching won't work - complains about unknown physical device
   for the SCSI disk, continues to work fine on the other internal disk.
   IMHO this is a major hit.
  
C) Diagnostic routines like Spinrite and QAPLUS don't know the disk exist.
   It would be nice to have something be able to perform a detail surface
   check (like Spinrites 84 patterns).
  
D) Having Format and Verify in BIOS only make future expansion painful,
   especially for SCID-ID 2-6.

Comments please.

Kevin Carpenter
kncarp@nicsn1.monsanto.com