Subject: problems with scsi / ncr (NetBSD 1.1)
To: None <tech-kern@NetBSD.ORG, port-i386@NetBSD.ORG>
From: Matthias Drochner <drochner@zelux6.zel.kfa-juelich.de>
List: tech-kern
Date: 12/06/1995 23:38:07
When trying to access an HP Scanjet 4c from a NetBSD/i386 box
(with NCR PCI SCSI interface) I ran into a couple of more or less
serious problems:

1. uk (the generic SCSI device) is not in arch/i386/i386/conf.c.
   There is a quick workaround: a tiny lkm creates the entry points.

2. The transfers cannot be larger then 64k (=MAXPHYS).
   This is a serious limitation for a scanner, but a workaround seems
   possible (bypass physio).

3. The "real" read command produces the following error message:

uk0(ncr0:3:0): COMMAND FAILED (6 ff) @f8735400.
ncr0:3: ERROR (80:140) (8-29-0) (0/13) @ (544:900b0000).
        script cmd = 910a0000
        reg:     da 10 80 13 47 00 03 1f 01 08 03 29 80 00 0a 00.
ncr0: handshake timeout
ncr0: restart (scsi reset).
uk0(ncr0:3:0): COMMAND FAILED (6 ff) @f8735400.

The scanner begins its work just after this read call. This means,
it cannot deliver data before a couple of seconds passed.
Who can help on this?

best regards
Matthias