use the source, Luke! again
By Warren Toomey
<wkt@henry.cs.adfa.oz.au>
So you call yourself a UNIX hacker: you know what bread() is, and the various splxx() routines don't faze you. But are you really a UNIX hacker? Let's have a look at a brief history of UNIX and the community of UNIX users and hackers that grew up around it and some recent developments for real UNIX hackers.
UNIX took the academic world by storm in 1974 with the publication of Ken Thompson's paper about its design, which was published in Communications of the ACM. Although it didn't contain many radically new ideas, UNIX had an elegance, simplicity, and flexibility that other contemporary operating systems did not have. Soon lots of people were asking Bell Laboratories if they could get copies of this wondrous new system.
This was the cause of some concern within AT&T, because of the restrictions of an antitrust decree brought against them in the 1950s. This decree effectively stopped AT&T from selling or supporting software: they could only engage in telco business. Their solution to meet the UNIX demand was to charge a nominal "license" fee to obtain UNIX and to distribute tapes or disks "as is." You'd receive your disk in the mail with just a short note: "Here's your rk05. Love, Dennis."
AT&T's stance on UNIX was often seen as an OHP slide at early conferences:
"This slide was always greeted with wild applause and laughter," says Andy Tanenbaum. This lack of support was tolerated for several reasons: Ken and Dennis did unofficially fix things if you sent them bug reports, and you also had the full source code to UNIX.
At the time, having full source code access for a useful operating system was unheard of. Source code allowed UNIX users to study how the code worked (John Lions's commentary on the sixth edition), fix bugs, write code for new devices, and add extra functionality (the Berkeley Software Releases, AUSAM from UNSW). The access to full source code, combined with AT&T's "no support" policy, engendered the strong UNIX community spirit that thrived in the late 1970s and early 1980s, and brought many UNIX users groups into existence. When in doubt as to how a program (or the kernel) worked, you could always "use the source, Luke!"
During this period, UNIX became wildly popular at universities and in many other places. In 1982, a review of the antitrust decree caused the breakup of AT&T into the various "Baby Bell" companies. This gave AT&T the freedom to start selling software. Source code licenses for UNIX became very expensive, as AT&T realized that UNIX was indeed a money spinner for them. Thus the era of UNIX source code hackers ended, except for some notable activities like the 4BSD work carried out at the University of California, Berkeley.
Those organizations lucky enough to have bought a "cheap" UNIX source license before 1982 were able to obtain the 4BSD releases from UCB and continue to hack UNIX. Everybody else had to be satisfied with a binary-only license and wait for vendors to fix bugs and add extra functionality. John Lions's commentary on how the UNIX kernel worked was no longer available for study; it was restricted to one copy per source code license, and was not to be used for educational purposes.
What were UNIX hackers going to do with no UNIX source code to hack anymore? The solution was to create UNIX clones that didn't require source code licenses. One of the first was Minix, created by Andy Tanenbaum and aimed squarely at teaching operating systems. Early versions of Minix were compatible with the seventh edition UNIX; the most recent version is POSIX compliant and can run on an AT with 2 MB of memory and 30 MB of disk space.
Many Minix users tried to convince Andy to add features such as virtual memory and networking, but Andy wanted to keep the system small for teaching purposes. Eventually, a user named Linus Torvalds got annoyed enough that he used Minix to create another UNIX clone with these extra features. And so Linux was born.
While Linux was taking off like a plague of rabbits, the BSD hackers were working on removing the last vestiges of UNIX source code from their system. They thought they had done so, and BSDI released BSD/386, a version of 4.3BSD that ran on Intel platforms. AT&T, however, wasn't so sure about the complete removal of UNIX source code and took them to court about it.
AT&T is not a good company to be sued by: it has a small army of lawyers. Eventually, the conflict was settled out of court with a few compromises, and we now have several freely available BSDs: FreeBSD, NetBSD, and OpenBSD. Of course, they all come with source code.
UNIX hackers of the late 1990s surely have an abundance of source code to hack on: Linux, Minix, OpenBSD, etc. But are they really UNIX hackers, or just UNIX clone hackers? Wouldn't it be nice if we could hack on real UNIX, for old time's sake?
UNIX turned 25 in 1993, which makes its early versions nearly antiques. Many of the old UNIX hackers (hackers of old UNIX, that is) thought the time had come to get the old, completely antiquated UNIX systems back out for sentimental reasons. After all, ITS, CTSS, and TOPS-20 had been rescued and made publicly available, why not UNIX?
At the time, UNIX was undergoing a crisis of ownership. Did AT&T own UNIX this week, or was it Novell, Hewlett-Packard, or SCO? UNIX is a trademark of someone, but I'm not sure who. After the dust had settled, SCO had the rights to the source code, and X/Open had dibs on the name "UNIX," which is probably still an adjective.
During the ownership crisis, Peter Salus, Dennis Ritchie, and John Lions had begun to lobby Novell: they wanted John's commentary on UNIX to be made publicly available in printed form. It wasn't until the UNIX source code rights had been sold to SCO that this finally was approved. It helped to have some old UNIX hackers, Mike Tilson and Doug Michels, inside SCO to fight the battle. You can now buy John Lions's commentary on 6th Edition UNIX (with source code) from Peer to Peer Communications, ISBN 1-57398-013-7. As Ken Thompson says: "After 20 years, this is still the best exposition of a 'real' operating system."
One of the restrictions on the commentary's publication is that the UNIX source contained within cannot be entered into a computer. OK, so you can read the book, but what use is source code unless you can hack at it?!
At the time that SCO bought UNIX, I began to lobby SCO to make the old source available again, unaware of the efforts to release the Lions's commentary. SCO's initial response was "this will dilute the trade secrets we have in UNIX, and it wouldn't be economically viable." My efforts drew a blank.
To help bring greater lobbying power to bear on SCO, the PDP UNIX Preservation Society (PUPS) was formed. Its aims are to fight for the release of the old UNIX source, to preserve information and source from these old systems, and to help those people who still own PDP-11s to get UNIX up and running on them. After realizing that SCO was never going to make the old UNIX source code freely available, we explored the avenue of cheap, personal-use source licenses. The society set up a Web petition on the topic and gathered nearly 400 electronic signatures.
Inside SCO, we were very fortunate to contact Dion Johnson, who took up our cause and fought tooth and nail with the naysayers and the legal eagles at SCO. The combined efforts of the PUPS petition and Dion's hard work inside SCO has finally borne fruit.
On March 10, 1998, SCO made cheap, personal-use UNIX source code licenses available for the following versions of UNIX: first through seventh edition UNIX, 32V, and derived systems that also run on PDP-11s, such as 2.11BSD. The cost of the license is US$100, and the main restriction is that you cannot distribute the source code to people without licenses. Finally, we can be real UNIX hackers and "use the source, Luke!" again.
Acknowledgments and References
I'd like to thank Dion Johnson, Steven Schultz, the members of the PDP UNIX Preservation Society, and the people who signed the PUPS petition for their help in making cheap UNIX source licenses available again. Dion, in particular, deserves a medal for his efforts on our behalf.
You can find more about the PDP UNIX Preservation Society at <http://minnie.cs.adfa.oz.au/PUPS/> and details on how to obtain your own personal UNIX source license at <http://minnie.cs.adfa.oz.au/PUPS/getlicense.html>.
SCO won't be distributing UNIX source code as part of the license. PUPS members have volunteered to write CDs and tapes to distribute old versions of UNIX to license holders. We currently have fifth, sixth, and seventh editions, 32V, 1BSD, all 2BSDs, Mini UNIX, and Xinu. We are looking for complete versions of PWB UNIX and AUSAM. We desperately want anything before fifth edition and hope these early systems haven't gone to the bit bucket. Please contact us if you have anything from this era worth preserving.
If you are licensed and want a copy of the PUPS Archive, see the PUPS Web page above for more information. We expect to be deluged with requests for copies, so if you can volunteer to write CDs or tapes for us, please let us know.
You don't need own a PDP-11 to run these old systems. The PUPS Archive has a number of excellent PDP-11 emulators. If you have bought a copy of the Lions's commentary (and you should), now you can run real sixth edition UNIX on an emulator. And if you want, you can hack the code!
Warren Toomey is a lecturer in computer science at the Australian Defence Force Academy, where he just finished his Ph.D. in network congestion. He teaches operating systems, data networks, and system administration courses. He has been playing around on UNIX since 4.2BSD.
Editor's note: This article originally appeared in a slightly different form in the AUUG Newsletter.
Copyright 1998