Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!mips!daver!bungi.com!news
From: phil@cs.wwu.edu (Phil Nelson)
Newsgroups: comp.sys.nsc.32k
Subject: 1/4" SCSI Tape Drives from CSC
Message-ID: <9103062255.AA22595@strawberry.cs.wwu.edu>
Date: 6 Mar 91 22:55:41 GMT
References: <<9103061534.AA10744@lev.seri.gov>>
Sender: news@daver.bungi.com
Reply-To: phil@cs.wwu.edu
Lines: 29
Approved: news@daver.bungi.com


   From: sverre@lev.Seri.GOV (Sverre Froyen)

   a) Add a general scsi ioctl which would allow any scsi command
   to be executed by scsi_hi.c.  This would be used by an mt command
   to position the tape (and could be used by a format command to format
   disk drives).  This would avoid having to teach the kernel about
   the specifics for each individual device one might want to add.

Isn't there a standard ioctl tape command set?  I don't have the POSIX
documents, but presume other programs besides mt might want to issue
commands to the tape drive.  Or is tapes one area where it is very 
difficult to provide a standard interface?

   b) I was hoping that tape read and writes would be similar enough
   to disk read and writes to be handled by the current code.

So was I, but in comparing my scsi commands manual for both my tape
and disk, it looked to me that they require slightly different bit
configurations in the read/write command blocks.  I don't have my
manuals with me at the moment, but as I remember it, my tape command
had some bits set in the command block that would make the disk read a
large number of sectors.  The tape used 3 bytes for number of blocks
to transfer and the disk used 3 and 3/4 bytes.  The 3/4 byte was at
the high order end and the tape specified them to contain some one
bits.  Has any one else noticed this or did I get a weird duck tape
controller?

--Phil