From leland@wacky.acet.org Mon Nov 22 17:48:45 1993
Received: from mwunix.mitre.org by chsun.eunet.ch (8.6.4/1.34)
id RAA09132; Mon, 22 Nov 1993 17:48:41 +0100
Received: from wacky.acet.org.acet.org ([192.188.104.18]) by mwunix.mitre.org (5.65c/SMI-2.2)
id AA13343; Mon, 22 Nov 1993 11:46:42 -0500
Received: by wacky.acet.org.acet.org (4.1/SMI-4.1)
id AA06945; Sun, 21 Nov 93 16:30:44 EST
Date: Sun, 21 Nov 93 16:30:44 EST
From: leland@wacky.acet.org (Robert Leland - PSI)
Message-Id: <9311212130.AA06945@wacky.acet.org.acet.org>
To: mw@chsun
Subject: SCSI tape driver.
Cc: leland@wacky.acet.org
Status: OR
I would like to change the way scsi tape /dev/rst maps the minor device.
This will allow for the selection of more than just one density of tape.
Also though my Archive Viper 2525 drive can Read QIC-24 Tapes generated by
a sun tape drive, it has to do this by traversing the tape several times
until it figures it is a QIC-24, I would rather not wear out my tapes!
This change would allow all type drive types to specify 3 additional densities, other than
the default.
----------------------------- The Bottom Line(first)------------------------------
People currently using
/dev/rst0, rts1 .... rst15
would remain uneffected, regardless of tape drive type.
People using the Exabyte type tape drives including Python that normally use
/dev/rst16 , /dev/rst32
would use instead:
/dev/rst32, /dev/rst64
(See below for more details)
-------------------------------------------------------------------------------------
Current
* Specific to Exabyte:
* minor bit 4 [b1bbbb] (aka /dev/rst16) selects short filemarks
* Specific to 8mm Drives.
* minor bit 5 [1bbbbb] (aka /dev/rst32) selects fix block mode, 1k blocks.
*
Minor Device: --------------------------------+
| 7 | 6 | 5 | 4 | 3 | 2 | 1 | 0 |
+---+---+---+---+---+---+---+---+
^ ^ ^ ^ ^ ^ ^ ^
| | | | | | |___|_____ Unit Number
| | | | | |_____________ NoRewind
| | | | |_________________ HiDensity
| | | |_____________________ Short FileMarks
|_________________________ Fix block mode 1K
- - - - - - - - - - - - - - - - - - - - - - -
Proposed
* Specific to Exabyte:
* minor bit 5 [bb1bbbbb] (aka /dev/rst32) selects short filemarks
* Specific to 8mm Drives.
* minor bit 6 [bb1bbbbb] (aka /dev/rst64) selects fix block mode, 1k blocks.
Minor Device: --------------------------------+
| 7 | 6 | 5 | 4 | 3 | 2 | 1 | 0 |
+---+---+---+---+---+---+---+---+
^ ^ ^ ^ ^ ^ ^ ^
| | | | | | |___|_____ Unit Number
| | | | | |_____________ NoRewind
| | | |___|_________________ HiDensity
| |_________________________ Short FileMarks
|_____________________________ Fix block mode 1K
Advantages: More similar to sun mapping, Keeps bits together.
Disadvantages: Possiblely some initial confusion.
- - - - - - - - - - - - - - - - - - - - - - -
Alternate Proposed
I would rather keep the density bits together but we could...
Split the bits used for density Use bits 3 and 6:
Minor Device: --------------------------------+
| 7 | 6 | 5 | 4 | 3 | 2 | 1 | 0 |
+---+---+---+---+---+---+---+---+
^ ^ ^ ^ ^ ^ ^ ^
| | | | | | |___|_____ Unit Number
| | | | | |_____________ NoRewind
| | | | |___________________________________ HiDensity
| | | |_____________________ Short FileMarks |
| |_________________________ Fix block mode 1K |
|_________________________________________________|
This would mean that other users would remain uneffected.
-----------------------------------------------------------------------------------------
This change will also benefit other tape drives, so that specific densities can be specified.
Could really use three bits for density, however It's nice to have a spare bit for the future.
-Rob
13a0aa4ee0