fred@rutgers.edu (Fred Barber) (02/04/90)
I have a Kennedy 9300F (125 ips, 800/1600, Pertec [?] formatted) drive connected to a Xylogics 472 interface in a Sun 3/280 running SunOS 4.0.1. The drive and I go back a long way. The Sun and 472 were new about a year ago. The drive and interface didn't work at all well together as originally installed. I was encouraged by Shugart (who purchased Kennedy somewhere along the line and don't seem to be TOO keenly interested in THEIR older products) to upgrade the formatter to current specs, which I did. Meanwhile, I was encouraged by Xylogics to install special, non-standard firmware in the 472, which I also did. Meantime, while all the parts were spread out over the country, we did the OS installation by hand. (The install scripts didn't have a prayer of working, given the severity of the magtape problems.) Now it's all back together and the NEW OS is here, necessitating a second go-around, and I'd like to tighten up the tape/interface interaction. Everything seems to work perfectly in non-interrupt mode. When the interrupts come on (that is, when the mini-root or the OS itself is actually up and running), there tends to be a flurry of anxiety (missed interrupt, attempting to resync) that dies down after a bit. Then everything is fine except that the first file on the tape seems to disappear. That is, -- "mt fsf" at BOT works fine; -- attempt to read (e.g., "tar t") at BOT fail with unexpected EOF; -- "mt stat" at BOT causes an unrequested FSF, then works; -- all operations beyond the first filemark go without a hitch; and, -- "mt fsf" at BOT followed by a series of "mt bsr" works fine, as long as you don't hit BOT. That is, you can read what you've spaced over up to but not including (I presume) the first record on the tape. Needless to say, that first file had better not contain anything too important. And as for the hard-wired offsets in the install script, forget it! Does this ring a bell with anyone? Is there a known fix, or is it known beyond all doubt that a fix is impossible? Is there anyone out there who has this drive and interface operating properly together? Fred Barber cognex!fred@ai.mit.edu Cognex Corporation Phone: 617-449-6030 15 Crawford Street Fax: 617-449-4013 Needham, Mass 02194