Subject: Re: Streamer problem
To: Eric Delcamp <edelcamp@easynet.fr>
From: John F. Woods <jfw@jfwhome.funhouse.com>
List: current-users
Date: 06/25/1997 13:04:01
> When trying to use my streamer on NetBSD/Amiga, I got theses messages :
> /netbsd: st0(atzsc0:2:0): illegal request, data = 00 00 00 00 00 00
> /netbsd: st0: cannot set select mode.
> This is a newly inserted tape, rewinded.
> Streamer is an ARCHIVE VIPER 150 Model 2105S Rev E.

There is a problem in the tape code that is apparently fixed by the patch
below.  However:

> I also couldn't get 150Mb tape, and I use 525Mb DC6525 (1020 feet)
> (QIC-525-DC).
> Any problems for this ?

Yes.  It won't work.  The drive will probably refuse to accept the tape
because it won't recognize the density select pattern; if it doesn't refuse,
it will probably not write reliably (higher density media isn't just "better",
it frequently has completely different characteristics that prohibit using
lower density writing techniques).

I've got a bunch of no-longer-interesting 6150 and 6250 tapes, but postage
would probably be horrendous...

Patch to get Archive 150 mode sense to work:

(Message inbox:7011)
Date: Sat, 29 Jun 1996 20:48:09 +0200
Message-Id: <96062920480874@saph1.physik.uni-bonn.de>
From: juengst@saph1.physik.uni-bonn.de (H. J|ngst, ISKP, Bonn)
To: jfw@funhouse.com
Subject: Re: ARCHIVE VIPER 150 not autosensing?

> I'm running sources from about May 27; I have found that my Archive Viper
> no longer autosenses the tape density when I put in a new tape.  If I use
> the explicit density minor devices, it works, as well as writing a tape
> and reading it without removing it.

Please, try the following patch (and look at the date of the patch:-).

Henry

*** /usr/src/sys/scsi/st.c.orig Fri Aug 12 12:11:30 1994
--- /usr/src/sys/scsi/st.c      Sun Sep  4 13:47:32 1994
***************
*** 349,354 ****
--- 349,361 ----
        st->buf_queue.b_active = 0;
        st->buf_queue.b_actf = 0;
        st->buf_queue.b_actb = &st->buf_queue.b_actf;
+ 
+         /*
+          * Forget if we've loaded the media,
+          * because sometimes things are unstable at boot time.
+          * We'll get it all again at the first open.
+          */
+         sc_link->flags &= ~SDEV_MEDIA_LOADED;
  }
  
  /*
*** st.c~	Sun May 26 07:43:14 1996
--- st.c	Sat Jun 29 18:29:54 1996
***************
*** 382,397 ****
--- 382,399 ----
  	}
  
  	/*
  	 * Set up the buf queue for this device
  	 */
  	st->buf_queue.b_active = 0;
  	st->buf_queue.b_actf = 0;
  	st->buf_queue.b_actb = &st->buf_queue.b_actf;
+ 
+ /**/	sc_link->flags &= ~SDEV_MEDIA_LOADED;
  }
  
  /*
   * Use the inquiry routine in 'scsi_base' to get drive info so we can
   * Further tailor our behaviour.
   */
  void
  st_identify_drive(st, inqbuf)