From: drew@caesar.cs.colorado.edu (Drew Eckhardt) Subject: Re: Tape Drive (help please?) Date: Thu, 1 Oct 1992 22:13:18 GMT
In article <1992Oct1.120238.11321@cs.rug.nl> mod32@cs.rug.nl ( ) writes:
>In article <BvD7r2.KB1@acsu.buffalo.edu> leet@acsu.buffalo.edu (Brian D. Leet) writes:
>>
>>In article <1992Sep29.203422.26967@let.rug.nl> H.H. Bergman writes:
>>
>>>I think Wangtek makes basically the same drives with just a different
>>>interface option, either SCSI or QIC-02. If you wait a while, there *may*
>>>[Drew?] appear an SCSI driver for Linux.
>>
Tape drives should be implemented as character devices, since
block devices can write out their data any time they geel like it.
However, under Linux, character devices do I/O directly to user
space, and without some serious kludging, the current SCSI drivers
can't do that. So, I'm waiting for version 2 of the SCSI drivers.
The Long Awaited version 2 of the SCSI drivers (Coming RSN, honest :) )
(They've been 90% complete since the beggining of summer, little things
like being homeless for a month, Compiler Construction assignments that take
our group 30-40 man hours to code, debug, document, and test,
shoelace not booting Linux .97.5 or later on my system and gcc 2.2.2d
w/jump tables not running on it, etc getting in the way) uses scatter / gather
so that SCSI command results can go directly to a non-contiguous address
space, like what you have in user space, or with multiple buffers
in buffer cache. Tape drivers, as well as raw disk devices can be
implemented cleanly using this method, so there will be an effort to
support them.
-- Microsoft is responsible for propogating the evils it calls DOS and Windows, IBM for AIX (appropriately called Aches by those having to administer it), but marketing's sins don't come close to those of legal departments. Boycott AT&T for their absurd anti-BSDI lawsuit.