On Sun, 14 Jan 1996 11:56:20 -0800 (PST) 
 Curt Sampson <curt@portal.ca> wrote:

 > Are these known to be a bit dodgey? I must admit mine doesn't work
 > so well, but I've not had a chance to try another SCSI drive yet.

Only in that the MD21 is a horribly ancient device.  There are at least a 
couple semi-kludges in the tree to deal with them.  First of all, it 
reports itself as pre-SCSI I (which is probably true :-), so the scsibus 
code doesn't trust it to deal with LUNs correctly (which is a valid 
assumption for 99% of pre-SCSI I "scsi" devices).  However, as we all 
know, the MD21 _does_ deal with LUNs ... So that required a new quirk 
type: SDEV_FORCELUNS.

Now, we've also noticed recently some SCSI protocol disagreement between 
the new NCR5380 driver and the MD21.  Quite a lot of the problem was 
linked to the use of 10 byte read and write cdbs.  The MD21, it seems, 
would really rather see 6 byte read and write cdbs.  In fact, given a 
choice, it would proably rather use the small version of any multi-format
command in SCSI, but that's just a guess.  Anyhow, that required 
workarounds in sd.c and cd.c, though it's arguable that we should have 
been using the "appropriate" command format in those drivers, anyhow.

I wouldn't say they're dodgy.  I'd rather say they have a personality 
uniquely their own.  Mine work fine, and I'm glad to put the closet full 
of Micropolis ESDI disks to use.  (I have MD21s and the Adaptec bridges 
on my Integrated Solutions systems, too...no problems to speak of.)

--------------------------------------------------------------------------
Jason R. Thorpe                                       thorpej@nas.nasa.gov
NASA Ames Research Center                               Home: 408.866.1912
NAS: M/S 258-6                                          Work: 415.604.0935
Moffett Field, CA 94035                                Pager: 415.428.6939