Chapter 21: Daemonology

The daemon image, what Kirk McKusick calls the "beastie," dates from 1976. Created by comic artist Phil Foglio and first used by Mike O'Brien, here's the story as told by Mike:

I was a bonded locksmith. Phil's roommate had unexpectedly split town, and he was the only one who knew the combination to the wall safe in their apartment. This apartment was the only one I'd ever seen that had a wall safe, but it sure did have one, and Phil had stuff locked in there. I didn't hold out much hope, since safes were far beyond my competence, but I figured "no guts, no glory'' and told Phil that I'd give it a whack. In return, I requested T-shirt art. Phil readily agreed.
Wonder of wonders, this safe was vulnerable to the same algorithm to which Master locks used to be susceptible. I opened it after about 15 minutes of manipulation. It was my greatest moment as a locksmith and Phil was overjoyed. I went down to my lab and shot some Polaroid snaps of the PDP-11 system on which I was running UNIX, and gave them to Phil with some descriptions of the visual puns I wanted: pipes, demons with forks running along the pipes, a "bit bucket" named /dev/null, all that.
What Phil came up with is the artwork that graced the first decade's worth of "UNIX T-shirts", which were made by a Ma-and-Pa operation in a Chicago suburb. They turned out transfer art using a 3M color copier in their basement. Hence, the PDP-11 is reversed (the tape drives are backward), but since Phil left off the front panel, this error was hard to detect. His trademark signature was photo-reversed, but was recopied by the T-shirt people and "re-forwardized"-- which is why it looks a little funny compared to his real signature.

The art was used on the USENIX Tenth Anniversary shirt in 1985 (I still have one). McKusick comments: "About 1 year after Usenix produced the Portland conference T-shirts, they paid Phil for the artwork. Thus, Usenix currently holds title to the copyright." I was the individual who actually paid Phil.

As I said earlier,

NET 1 was a subset of the then-current Berkeley system. It was quite similar to 4.3-Tahoe, including source code and documentation for the networking portions of the kernel, the C library and utility programs. It was available without evidence of any prior license (AT&T or Berkeley), and was (re)distributed via anonymous FTP. The source carried a Berkeley copyright notice and a legend that allowed redistribution with attribution.
In June 1991, at the USENIX Conference in Nashville, BSD Networking Release 2 was available. NET 2 contained far more than just networking code and, like NET 1, was available with no prior license. The new features included a new virtual memory system (derived from Carnegie-Mellon's Mach system, which had been ported at the University of Utah) and a port to the Intel 386/486.

But all was not happy in Eden. While BSDI's version of the release was complete by the end of 1991, it was only released to the public on April 10, 1993 as 1.0, the long delay being the consequence of USL's filing suit to prevent BSDI from shipping its product.

BSDI had distributed pre-production releases of BSD/386 (Beta version). It now began international distribution. Full source was priced at $1000. (In the January/February 1994 ;login:, Lou Katz wrote: "It works! It works!").

On Friday, February 4, 1994, Novell and the University of California agreed to drop all relevant suits and countersuits. BSDI immediately announced the availability of a release based on "4.4BSD-Lite."

In the meantime, several groups of coders had begun work on other releases (daemons multiplying nearly as rapidly as penguins). The earliest of these was 386BSD, by Lynne and Bill Jolitz, though others have proven to be of greater importance.

William Jolitz had had considerable experience with prior BSD releases while at Berkeley (2.8BSD, 2.9BSD) and he and Lynne Jolitz contributed code to Berkeley developed at Symmetric Computer Systems during the 1980s. The public 386BSD releases beginning in 1992 were based on portions of the NET 2 release plus with additional code written by the Jolitzes.

FreeBSD derived from the 386BSD 0.1 release. It was the first free software organization founded on BSD.

Initial development of FreeBSD was started in 1993, taking its source from 386BSD. As a consequence of concerns about the legality of some of the code, NetBSD, like FreeBSD, was derived from 4.3BSD via NET 2 and 386BSD. The project began as a response to the FreeBSD unified patchkit and 386BSD, with its slow development process and focus on the i386 platform. The four founders of the project, Chris Demetriou, Theo de Raadt, Adam Glass and Charles Hannum, felt that a more BSD-like development focus would be more beneficial to the project.

de Raadt suggested the name "NetBSD" and it was readily accepted by the other founders. Although Demetriou inadvertently made a premature release of the code, NetBSD 0.8 was the first official release in May 1993.

In 1994, one of the founders, Theo de Raadt, was forced out of the project. He later founded a new project, OpenBSD, from a forked version of NET 1. The first release of OpenBSD was in October 1995.

The current release of NetBSD is version 2.1 (October 2005).

I had dinner with Poul-Henning Kamp in Copenhagen and asked him about his work on FreeBSD.

"I've been involved with FreeBSD from before it existed," he told me.

You know, the basic BSD kernel suffers from a number of Ph.D. theses. I see the split as between the academic (OpenBSD) and the practical (FreeBSD). Chris Demetriou is towards the more academic side.

I love having choice. Diversity in OSes is as important as is diversity in crops.