I've been looking into why the Microtek backend has suddenly started dying
with sane-0.71....
Ugh -- I'm not sure what all has caused this problem to rear its head now,
but part of the problem is that, apparently, the Microtek scanners do *not*
follow the SCSI-II standard for the REQUEST_SENSE command reply --- and the
Linux mid-level scsi driver (scsi.c) is coded to parse standard sense data,
not the weird stuff pumped out by these scanners.
Thus, during error conditions, the scsi driver is certainly confused by what
the scanner tells it.
Perhaps one solution is to try to ignore the scsi driver's complaints, and
have the backend send its own REQUEST_SENSE under suspicious circumstances.
I'm playing with it.
In any case, does any one have any insights into how other OS's handle scsi
CHECK_CONDITION conditions and REQUEST_SENSE requests? No one else has
complained yet (well, except for a similar unresolved problem in OS/2, and
that thing with Solaris....).
-matt m.
-- Source code, list archive, and docs: http://www.mostang.com/sane/ To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com