Monthly Columns
 

Kirk McKusick Speaks at Usenix

Copyright © 1999 Pat Lynch

From hot, humid New Jersey, we fled to California, Monterey to be exact, to witness 1500 computer professionals under one roof, of which we were two. We spent 10 days there, and most was uneventful (of course the FreeBSD dinner was fun), until Thursday night. About an hour after the FreeBSD BoF (Birds of-a Feather) session, in a larger room in the Conference center, many people gathered to hear Marshall Kirk McKusick speak of the history of the BSD distributions.

It started very quiet, Kirk was talking up on stage about things before he became involved directly, of Bill Joy's adventured with the BSD distribution, of Kirk witnessing this and ultimately being drawn in. He talked of sharing the PDP 11/45 with the Math Department and the Statistics department. He also spoke of Ken Thompson's sabbatical from AT&T Labs to UCB.

As he went up through history about how originally BSD was a set of tools, and later became kernel enhancements and therefore a full OS based on UNIX, he took us up to the history of the DARPA contract and their "competition" with Bolt, Beranek, and Newman (referred to as BBN from here on), and ultimately how their networking implementation had become the choice of DARPA.

About halfway through Kirk stopped, offered a break, sold "May the Source Be With You" T-shirts and polos, and I happened to be standing discussing some interesting things with a colleague when Dennis Ritchie (yes that Dennis Ritchie) saunters into the room. I had actually met DMR before, but never actually spoke with him directly, and he stood there telling us of his side of the DARPA meetings (of which DMR was a member of the advisory board).

He said (paraphrased) that every DARPA meeting ended up the same, with the Military coming in and giving CSRG (at UCB, the group that worked on BSD) a stern warning that they were to work on the Operating System, and that BBN will work on the networking. Every time, Bob Fabry, then the advisor of CSRG, would "Yes: them to death and they'd go off and just continue the way they were going. Much to the frustration of the DARPA advisory board.

Shortly after DMR told us this story, we all sat down again to hear Kirk talk more about BSD, and Mr. Ritchie sat down as well. Kirk then talked of personnel changes within CSRG and Bill Joy's resignation to go work at Sun Microsystems, Bob Fabry leaving, and Sam Leffler, another major part in CSRG, leaving as well for Lucasfilm. Shortly after this, Kirk joined them full-time. He also took us to Keith Bostic's arrival to CSRG and his backport of 4.3BSD to the PDP-11.

The next Part in the Saga of CSRG is very important, as it led up to the lawsuit and the creation of the "future" BSD Lite. The release of the Networking Release 1, under what is now known as the Berkeley license, because of the need to separate the networking code from AT&T owned code. After that, the base system was still developed, The VM system was integrated, taken from MACH, written at Carnegie-Mellon. Mike Hibler did the work on the VM merge. As Kirk said, "Why write something from scratch, when you can take it from somewhere else." (Paraphrased). They also were able to take a Sun compatible port of NFS written by Rick Macklem at the University of Guelph in Canada. They released 4.3BSD-Reno shortly after, and Kirk said that the release was bit of a gamble due to all the new and relatively untested code. It did enable them to work the bugs out though. This occurred in early 1990.

Through much of Kirk's talk he didn't mention dates (these dates are mentioned in the article he wrote for the O'Reilly book, _Open Sources_, ISBN 1-56592-582-3), and moved through the history of the CSRG. Sometime during this period, the history of the BSD Daemon, "Beastie", Kirk calls it. Is brought up. He explains where it came from, and mentions a USENIX shirt from early on, which Dennis Ritchie comes up and models for us. Truly a treat.

Shortly after, he got into the history of the lawsuit against UCB from AT&T. He described how it was to be deposed. I think his exact words were "It was like getting an exam from a proctologist" or something to that effect. He explained that up until this point it wasn't really a problem. But Jolitz' 386BSD, BSDI's BSD/OS (who was advertising with "call 1-800-ITS-UNIX"), and a few other catalysts caused AT&T to file suit. Right afterwards through some legal maneuvering of their own, the University of California file suit of their own. Essentially quoting the "due credit" type clause from the Berkeley licensing, because AT&T was using their networking code.

He then said that it was Ray Noorda, the CEO of Novell, who just recently acquired the UNIX trademark and code from AT&T, who was the catalyst in settling out of court. The files that were disputed were removed, and each project (at that point FreeBSD, NetBSD, BSDI) were forced to write new code to replace those files.

Most all the rest of the history we know pretty well. Kirk had a question and answer session, most, of which was questions about the Beastie, and the NetBSD project then presented an unmentionable "NetBSDality" T-shirt, and the night wrapped up.

I have to thank Mr. McKusick for a great BoF, and same with Mr Ritchie and all the other "old-timers" that were there. It was truly educational for us newbies.

(Sources after the fact were from Mr. McKusick's article in "Open Sources")

Pat Lynch, lynch@bsdunix.net