On Tue, 8 Oct 1996 18:23:12 -0700 (PDT) 
 Curt Sampson <curt@portal.ca> wrote:

 > sd1(esp0:4:0): illegal request, data = 3b bb 01 3d 54 00 00 00 00 10 00 74 00 4e 20 00 00
 > sd1: could not mode sense (4); using fictitious geometry
 > sd1(esp0:4:0): illegal request, data = 00 06 01 3d ef 00 00 00 00 10 00 74 00 4e 20 00 00
 > sd1: could not mode sense (4); using fictitious geometry
 > 
 > This seems to happen whenever I do something that seems to access the
 > disk label. It's an inconvenience, but nothing more.

This is becuase Jaz drives don't support MODE SENSE.  Pretty lame.  It's
not really a big deal.  In the post-1.2 sources (which _will_ be available
soon; you may have noticed the branch-to-mainline merge that JT and I
did last evening), we have a new quirk type that disables sending
MODE SENSE to Jaz and Zip drives.

 > However, between the beta kernel dated above (which may have been
 > from slightly earlier sources) and the 1.2 release, I think something
 > changed in the esp disconnect code for the worse, because under a 1.2
 > kernel my drive now gives messages like this when I access certain parts
 > of it after mounting it:
 
 > Oct  7 22:44:14 didactic /netbsd: esp0: RESELECT: 7 bytes in FIFO!
 > Oct  7 22:44:14 didactic /netbsd: sd1(esp0:4:0): unit attention, data = 15 d0 01 39 db 00 00 f8 00 00 f4 00 00 00 00 00 00
 > 
 > Oct  7 22:59:48 didactic /netbsd: esp0: RESELECT: 7 bytes in FIFO!
 > Oct  7 22:59:48 didactic /netbsd: sd1(esp0:4:0): unit attention, data = 28 8f 01 3d 88 00 00 f8 00 00 f4 00 00 00 00 00 00
 > 
 > Oct  7 23:01:48 didactic /netbsd: esp0: RESELECT: 7 bytes in FIFO!
 > Oct  7 23:01:49 didactic /netbsd: sd1(esp0:4:0): unit attention, data = 34 5a 01 2f 2f 00 00 00 00 00 00 00 00 00 00 00 00
 > 
 > I moved the drive from ID 4 to ID 3 (which moved it to an ID with
 > disconnect disabled) and the problem went away.

Hmm ... The esp driver has undergone some heavy work in post-1.2 (Charles
Hannum needed to fix the driver for an Alpha he was using, and I tested
the changes on the SPARC and updated the SPARC version of the driver).
Quite a lot of the changes were dealing with SCSI messages, disconnect,
and whatnot.  You may find that an updated driver does the trick.

"Stay tuned."

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