On Jun 29, 10:27pm, Donn Cave wrote:
> | I have an Exabyte tape streamer here, but netBSD doen't seem to work
> | with it.
...
> I will send a patch which, applied to /usr/src/sys/scsi/st.c, will
> probably improve Exabyte support.  It's mostly the old 0.9 driver's
> Exabyte support grafted into the 1.0a distribution;  you may already
> have received something similar from the same person I got it from.
> Obviously a kernel rebuild is called for.

  I have been carrying around a partial change for Exabyte support.  The
tape driver was modified quite a bit one time and I have not completely
converted my changes to the current driver.  [I don't have an Exabyte
to test things with now.]

> to trying Olaf Seibert's modifications).  The way you can tell if it's
> not supported, is that the config blurb says nothing about "rogue".
> The <EXABYTE ...> stuff comes straight from the drive, and in most
> cases the driver does not recognize it.  If it does, it will call it
> a "rogue", and it's more likely to work.  Weird, but true.  Current
> kernel source may be different.

  The tape driver will "support" tapes if they behave in a "generic"
fashion.  If a particular tape drive mis-behaves from what the driver
expects, or if it needs some special handling in certain cases, an
entry is made in a "rogue" table.  Since the Exabyte can handle both
fixed-length and variable length records, a rogue entry is probably
required so that the record type can be specified (by using the device
minor number).  Also, if I remember correctly, the Exabyte can handle
different sizes of fixed-length records, so there needs to be some way
to be able to specify the record size.

  One other problem that I ran into the last time I tried using an
Exabyte was that the drive was disconnecting.  There was some code in
the original Amiga st.c that looked like it disabled the disconnect
on the Exabyte by sending some vendor-specific data in the mode select
command.  I added that to the common scsi st.c and was able to get the
tape working, although I didn't have the tape available long enough to
do much testing.  It was after that change that the st.c driver was
modified and I haven't tried to modify my changes to work with the
current driver yet.

Michael

-- 
Michael L. Hitch			INTERNET:  osymh@montana.edu
Computer Consultant
Office of Systems and Computing Services
Montana State University	Bozeman, MT	USA