13. BSD Unix, Part 3 - The Beginning of BSD Unix
The beginning of BSD Unix
When Ken Thompson returned to Bell Labs in 1976, Bill Joy and Chuck Haley began exploring the Unix kernel themselves.
Hmm... we could speed things up here in the kernel.
Let's try it.
Other universities want copies of our Pascal compiler.
Let's bundle it with ex and send them a tape.
In early 1977, Bill Joy put together the Berkeley Software Distribution, or BSD, bundling the Pascal system with ex, a line editor developed from em. Early BSD was not a standalone operating system: it was a collection of software to install on an existing Unix system. About thirty copies were sent out over the following year.[1]
Thirty tapes to copy...
At the time, BSD could be distributed only to recipients with an AT&T Unix source license. Berkeley charged a nominal fee for the tape and distribution.[2]
If you already have a Unix source license, it's just the tape and postage.
Berkeley already had a tradition of sharing research software. BSD carried it on. Joy's low-cost distribution policy has also been attributed to Berkeley's Free Speech Movement, hippie culture, and antiwar activism.[2]
Universities and research institutions could obtain Unix, including its source code, for a modest licensing fee. This also helped the system spread.[2][6]
We have Unix, but the antitrust agreement bars us from the computer business... We can't sell it as a product, either.
The second distribution: 2BSD
After the first BSD distribution, Joy improved the Pascal system based on user feedback. He bundled it with vi (ex's visual mode), the C shell, and other tools to form the Second Berkeley Software Distribution, or 2BSD. Nearly seventy-five tapes were shipped over the following year.[1][5]
Joy did more than fix bugs. He handled every distribution request, copied BSD onto tapes, and mailed them out himself.
Oh, you need a copy of BSD?
Another university wants a copy.
Tune the kernel. Fix the bugs.
More copying.
A VAX arrives at Berkeley, 1978
Professor Fateman's group is getting a VAX. It's 32-bit, with a whole megabyte of memory.
Really? We should put Unix on it.
They've already got VAX Unix from Bell Labs. It's called 32/V.
Hi, I'm Ozalp Babaoglu.
My adviser asked me to add virtual memory to VAX Unix. Could you look over the code?
Sure. Let's take a look.
Babaoglu, a Berkeley graduate student, was implementing virtual memory under Professor Domenico Ferrari's supervision. Joy joined him to integrate the code into 32/V and help debug it.[1][4]
At the time, 32/V did not use the VAX hardware's virtual-memory capabilities. A program's address space was limited by physical memory, but Professor Richard Fateman's work on the computer algebra system Macsyma needed more. Virtual memory uses disk storage to give programs an address space larger than physical memory.[1]
Macsyma, developed at MIT beginning in the late 1960s, worked with symbolic expressions. It could factor expressions, compute derivatives, and perform integration.
Debugging continued through the holidays at the end of 1978.
Not going away for the holidays?
The VAX is quiet. Good time to debug virtual memory.
By January 1979, most of the bugs had been fixed.[1]
Let's try Unix with virtual memory.
Finally. It runs without crashing.
Joy ported ex, vi, the C shell, and other 2BSD tools to the VAX, while Peter Kessler and Marshall Kirk McKusick ported the Pascal system. These programs were bundled with the standard 32/V utilities and the virtual-memory kernel that Babaoglu and Joy had integrated and debugged. The result was 3BSD, which began shipping in December 1979.[1]
Finally. 32-bit BSD, ready to ship.
ARPANET had been connecting research institutions funded by the Defense Advanced Research Projects Agency, or DARPA, since 1969.[3] But the sites used different computers and operating systems, making software expensive to port or rewrite. DARPA became interested in Unix's portability, and took notice of Berkeley's successful distribution of 3BSD to research institutions.[1]
BSD on the VAX. Runs nicely, too.
The end.
The dialogue in this comic is reconstructed from historical accounts, not quoted verbatim.
References
- Marshall Kirk McKusick, Twenty Years of Berkeley Unix, in Open Sources: Voices from the Open Source Revolution, 1999. Korean edition, Hanbit Media, 2013.
- Andrew Leonard, BSD Unix: Power to the people, from the code, Salon, 2000.
- SRI International, ARPANET. Its origins and the first transmission in 1969.
- Ozalp Babaoglu, Biographical note. His Unix virtual-memory research as a Berkeley graduate student.
- The Unix Heritage Society (TUHS), Original 2BSD distribution archive. The installation document
install.msand source code in2bsd.tar.gz. - Warren Toomey, The Strange Birth and Long Life of Unix, IEEE Spectrum, 2011.