The Berkeley Software Distribution
==================================
UNIX is always litigious
by Bradford Morgan White
Feb 5, 2024

The first public presentation of UNIX was made at the Symposium on
Operating Systems Principles at the IBM Research Center in Yorktown
Heights in October of 1973. Dennis Ritchie is quoted as saying it was
beautiful day, and Ken Thompson layered his own memories with a thick
coating of modesty:

> The audience was several hundred. I was pretty nervous. The
> response was the normal, polite applause. I don't recall any
> questions.

[The IBM Research Center in Yorktown Heights, image from IBM]

<https://substackcdn.com/image/fetch/w_1456,c_limit,f_webp,
q_auto:good,fl_progressive:steep/
https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2F
images%2Fb9ace03e-5fa3-43df-875b-6989e389b156_1200x675.jpeg>

In contrast to Thompson stating that he didn't recall any questions...
the two were immediately asked for copies of the operating system,
and this wasn't a simple matter for AT&T. The American Telephone and
Telegraph company had been established as a legal, nation-wide
monopoly in the USA via the Kingsbury Commitment in late 1913. This
position was further cemented during World War I when the United
States' federal government nationalized the phone system. Following
the end of the war, the phone system went back into the hands of AT&T
and the company achieved some rather remarkable regulatory capture
with the Willis Graham Act of 1921 and the Communications Act of
1934. This complicated legal history presented a very serious
question to AT&T's legal department when people began asking for
UNIX: were computer operating systems part of the common carrier
services of the phone company and therefore required to be
distributed? If they were not, then the company needn't distribute
UNIX at all, but if they were indeed, then it was only a matter of
time before the FCC would force AT&T to distribute UNIX. In the end,
the decision was made to distribute UNIX to universities and research
centers at the cost of the media plus shipping. Somehow, quite
magically, this resulted in a nice round number of $150.00 (or around
$927 in 2024) for Katholieke Universiteit in Nijmegen, The
Netherlands in December of 1974. There are some rather important
points within the license that Katholieke was granted. Licensees were
granted source code for the operating system as computer systems of
the day weren't standardized in any meaningful way. The license then
granted free use and modification within the university, but
disallowed any spread outside. Specifically, the license mentioned
that employees and students had access.

[UNIX license from Western Electric to Katholieke]

<https://www.abortretry.fail/api/v1/file/
ea99ff77-8f06-4c00-ae95-2a1f60410a44.pdf>

At this point in the computer industry, user groups were somewhat
common. IBM had SHARE, and that had inspired similar groups around
DEC, Burroughs, Rand, and so on. It was therefore somewhat natural
that a group would form around UNIX. Thus, Mel Ferentz and Lou Katz
organized a meeting of UNIX users in New York on the 15th of May in
1974. Around twenty people were in attendance, and by this time there
were just over thirty UNIX installations outside of AT&T and its
subsidiaries. This user group grew to become USENIX over time.
Following the user group's formation, a mailing list started. From
the first list on the 30th of July in 1975, we have the following
organizations listed as installation/user sites: AT&T, Brooklyn
College, Carleton College, Case Western Reserve University, The
Children's Museum, City University of New York, Columbia University,
Duke Medical Center, East Brunswick High School, Harvard University,
Hebrew University of Jerusalem, Heriot-Watt University, Johns Hopkins
University, Knox College, Naval Postgraduate School, Oregon Museum of
Science, Polytechnic University of NY, Princeton University, Rand
Corporation, St. Olaf College, Stanford University, The Spence
School, University Catholique de Louvain, University of Alberta,
University of California (Berkeley), University of Manitoba,
University of North Carolina, University of Saskatchewan, University
of Texas (Dallas), University of Toronto, University of Utah,
University of Waterloo, and the University of Wisconsin. As we know
from the license granted to Katholieke, there were more UNIX user
sites than this, but they weren't members of the user group (or at
least not at first).

<https://www.abortretry.fail/p/the-history-of-gm-naa-io-and-share>

[Dennis Ritchie (standing), Ken Thompson at the teletype, PDP-11
1972]

<https://substackcdn.com/image/fetch/w_1456,c_limit,f_webp,
q_auto:good,fl_progressive:steep/
https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2F
images%2F3926f41f-c121-450b-8e10-e46ccc1c5448_652x522.jpeg>

In this early time period, UNIX only ran on the PDP-11, but that
changed at Princeton where UNIX was ported to the IBM 360 in 1976.
The next target was the Interdata 8/32 in 1977 which was undertaken
by Ritchie and Steve Johnson (author of yacc, lint, pcc). But,
porting efforts took off only after John Lions at the University of
New South Whales wrote a commentary on the UNIX sources and
distributed them as a book, Code and Commentary, for teaching
students about operating systems. Western Electric tried to stop
dissemination, but this was apparently impossible. Likewise,
modifications of UNIX began circulating following the release of
Lions' book, and a culture we would recognize today as “open source”
began to develop helped in no small part by AT&T's policies regarding
UNIX that essentially stated UNIX would have no advertising, no
support, no bug fixes, and payment in advance. To color the time
period more thoroughly, Bob Kahn and Vint Cerf published the first
description of TCP/IP in 1974 and by January of 1976 there were sixty
three hosts on ARPAnet, and UNIX while being used globally would run
only on hardware that cost over $9000 (around $48000 in 2024).

Professor Robert (Bob) Fabry was at the Symposium on Operating
Systems Principles where UNIX had first been announced and he was
very excited. Returning to UC Berkeley where he was then employed, he
assembled a group to purchase a PDP-11/45. As this was a large
purchase, he coordinated the departments of computer science, math,
and statistics. With the machine purchased, Fabry then ordered a tape
of UNIX from Thompson. The actual installation of UNIX was first
undertaken by Keith Standiford in January of 1974. In 1973/1974, it
was somewhat routine for Thompson himself to be involved in nearly
every UNIX installation for a licensee. The folks at Berkeley seemed
to be interested in doing everything themselves, but things didn't go
well. Eventually, Standiford reached out to Thompson, and Thompson
would connect to the University's 11/45 over a three hundred baud
acoustic coupler to remotely debug crash dumps from New Jersey. I
personally like to imagine that it was a Novation CAT 300, but I
haven't been able to find a model number of the modem, and I haven't
found any reference to what system Thompson was actually using. Plus,
the Novation CAT 300 wouldn't be released for several years.

Following the purchase of the PDP-11/45, the departments involved
began having issues with scheduling time on the machine. Berkeley
bought several more computers. One of these was a PDP-11/70, and its
arrival coincided with the arrival of Thompson as a visiting
professor. Thompson, Bob Kridle, and Jeff Schriebman then setup V7
UNIX on the 11/70. Shortly after the installation was completed, two
graduate students, Chuck Haley and William Nelson (Bill) Joy, arrived
on campus. They were intrigued by the computer system, and they began
hacking on Thompson's Pascal compiler. Within a few weeks (from what
I have been able to find), the teletypes attached to the 11/70 were
replaced with ADM-3 screen terminals.

[ADM-3, image by Chris Jacobs, CC BY-SA 3.0]

<https://substackcdn.com/image/fetch/f_auto,q_auto:good,
fl_progressive:steep/
https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2F
images%2F98f22ee6-2f34-4ffd-b8bd-b37fb059cbe5_1887x1860.jpeg>

For Bill Joy, using ed or em on a screen terminal wasn't really
sufficient. He took a detour from hacking on Pascal, and he created
the ex editor. Together with Pascal, the V7 UNIX at Berkeley was
notably better than other UNIX systems of the time. In early 1978,
Bill Joy began offering the Berkely Software Distribution. The first
copy we know to make it out of Berkeley was to Tom Ferrin at UCSF on
the 9th of March in 1978. The license was signed on the 13th, the
media was an 800 bpi tape, and on the tape was the “Unix Pascal
system” and the “Ex text editor.” Credits were made to W.N. Joy, S.L.
Graham, C.B. Haley, K. Thompson for Pascal, and to W.N. Joy for Ex.

[Bill Joy, Silicon Valley Visionary, on the Future of Batteries ...]

<https://substackcdn.com/image/fetch/f_auto,q_auto:good,
fl_progressive:steep/
https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2F
images%2Fd5b8a872-0970-4832-8b53-86c19f7c8543_1200x630.jpeg>

BSD (now referred to as 1BSD) shipped around thirty copies in the
first half of 1978. Somewhere around June, Pascal had been further
improved, the C shell had been written, vi had been written (by Joy),
and termcap had been written (by Joy). These new tools comprised the
bulk of the Second Berkely Software Distribution or 2BSD. Joy was the
man running the show, and he'd answer the phone, create the tapes,
incorporate feedback, package, and mail the software. Seventy five
copies of 2BSD were sent out.

The DEC VAX was first introduced in 1977. It was a 32 bit ISA with
virtual memory. In the first half of 1978, Professor Richard Fateman
at Berkeley was looking for a machine with a lot of memory for a
project of his. The VAX-11/780 met his requirements and his budget,
and he got some folks together to purchase it with a bit of help from
the National Science Foundation. The VAX did have one disadvantage in
the minds of those at UC Berkeley; the VAX ran VMS. Luckily, UNIX had
already been ported to VAX as UNIX/32V (V7 UNIX variant) by John
Reiser and Tom London at Bell Labs. This port, however, didn't take
advantage of the main VAX feature, virtual memory, which limited the
available memory to 1MB.

This had to be fixed. For Fateman, virtual memory was a requirement,
and UNIX was a requirement. He then contacted Professor Domenico
Ferrari about getting virtual memory support in UNIX. One of the
graduate students working with Ferrari, Özalp Babaoğlu, then set
about this task. Along the way, Babaoğlu reached out to Joy for
assistance. Joy helped integrate Babaoğlu's memory system into 32V
and helped him debug the resulting UNIX variant.

The new version of UNIX was good. Joy knew that the 32 bit VAX
running UNIX would render the 16 bit PDP-11 obsolete, and he started
porting 2BSD to 32V. Peter Kessler and Marshall Kirk McKusick worked
on porting Pascal, while Joy handled ex, vi, C shell, and many other
BSD utilities. The group had completed their work by the end of 1979,
and Joy shipped 3BSD in December.

At this point, it's important to delineate the family tree. 1BSD and
2BSD were improvements to UNIX version 6 while 3BSD was an
improvement to 32V which was itself of a port and modification of
UNIX version 7. During the creation of 3BSD, 2BSD had continued to
see additions, fixes, and releases. One important distinction to be
made is that the filename for the kernel in 3BSD became vmunix for
virtual memory UNIX.

Computer hardware and software were highly varied in 1979 and most
often nothing was compatible with anything else. For the Defense
Advanced Research Projects Agency, this was an issue. They felt that
the best they could hope for was unity at the operating system level
for the growing network of networks, ARPAnet, and to this end they
chose to standardize on UNIX due to its proven portability resulting
from having been written in C. This same system would also be used by
DARPA for work in the VLSI Project. In the autumn of 1979, Fabry
reached out to DARPA offering 3BSD as the solution to their problem.
Initially, this wasn't well received, but the success of 3BSD that
December changed opinions. Fabry secured an eighteen month contract
with DARPA that began in April of 1980. This contract stipulated that
Berkeley's new Computer Systems Research Group would add the features
to 3BSD needed by DARPA which was left rather open. In a technical
report on the matter, the government doesn't actually list anything
too specific.

[ARPA Standard UNIX Report]

<https://www.abortretry.fail/api/v1/file/
2c12ca84-b921-4d68-97a1-c046f60372c4.pdf>

Fabry hired Laura Tong to be the project administrator, and Fabry
then set about finding a tech lead. On an evening in early March, Joy
rang Fabry at his home and expressed his desire to lead UNIX
development, and Fabry agreed. Unlike earlier BSD releases (and
on-going releases of 2BSD) a more robust system of distribution
needed to be in place to handle a higher number of orders. Tong setup
just such a system, and had Fabry coordinate with Bob Guffy at AT&T
as well as the university's lawyers to find license terms that would
satisfy all parties. The system that Joy and the software team would
create was 4BSD which was available in October of 1980. This release
brought job control to the C shell initially developed by Jim Kulp
and integrated by Joy, delivermail (predecessor of sendmail), job
control signals that worked more reliably (so if you sent SIGHUP the
job would actually hangup), the curses library, the control-z
suspend/resume functionality we know today, a filesystem that
supported block sizes up to 1K, and greater hardware support.
Notably, 4BSD supported the VAX-11/750. Like prior BSD releases, 4BSD
included the Pascal compiler and that saw still more improvements,
and it included the Franz Lisp system by Richard Fateman. 4BSD saw
nine months as the system de jour, and about one hundred fifty copies
were shipped. Licenses were per institution and not per machine, and
estimates are that the distribution was running on about five hundred
computers.

4BSD drew criticism from David Kashtan at Stanford Research
Institute. He'd run some benchmarks on VMS and BSD and he claimed
that VMS was the clear winner. This didn't sit well with Joy. He went
and made a series of performance improvements to BSD (mostly in
vmunix), and a few weeks later released a paper rebutting Kashtan's
claims and showing that BSD was every bit the match to VMS. The
improved kernel was coupled with Kevin Robert Elz's
auto-configuration and a system featuring them was released as 4.1BSD
in June of 1981, and 4.1BSD for VAX became 2.8BSD for the PDP-11..
This version was current for two years and shipped four hundred
copies. This version was good enough to win CSRG another two years of
funding from DARPA, and the CS department at Berkeley also saw
funding.

Interestingly, the reason for a point release was political. The CSRG
had wanted to call this release 5BSD, but AT&T blocked that name.
AT&T were releasing System V UNIX at the time, and they felt that a
Berkeley release also named “five” would confuse customers. Point
releases of 4.?BSD were then the agreed upon solution.

The new funding round came with some more concrete goals.
Specifically, DARPA wanted a better filesystem with higher
throughput, support for multi-gigabyte adress spaces, better IPC, and
integrated networking. Marshall Kirk McKusick outlines how these
decisions were made:

> To assist in defining the new system, Duane Adams, Berkeley's
> contract monitor at DARPA, formed a group known as the "steering
> committee" to help guide the design work and ensure that the
> research community's needs were addressed. This committee met twice
> a year between April 1981 and June 1983. It included Bob Fabry,
> Bill Joy, and Sam Leffler of the University of California at
> Berkeley; Alan Nemeth and Rob Gurwitz of Bolt, Beranek, and Newman;
> Dennis Ritchie of Bell Laboratories; Keith Lantz of Stanford
> University; Rick Rashid of Carnegie-Mellon University; Bert
> Halstead of the Massachusetts Institute of Technology; Dan Lynch of
> The Information Sciences Institute; Duane Adams and Bob Baker of
> DARPA; and Jerry Popek of the University of California at Los
> Angeles. Beginning in 1984, these meetings were supplanted by
> workshops that were expanded to include many more people.

What was developed largely by McKusick, Joy, Sam Leffler, and Rob
Gurwitz is astounding. These developments were iterated in releases
4.1a, 4.1b, and 4.1c but ultimately culminated in 4.2 which was
released in August of 1983. The two largest innovations were the
Berkeley Sockets API and the Berkeley Fast File System. The sockets
interface allowed multiple different network protocols to be used at
any time and exposed them as files. The Berkeley FFS allowed
blocksizes from 128 bytes to larger than 4096 bytes if needed. Thus,
with Berkeley's FFS, a system operator could optimize either for disk
use or for disk performance. 4.2BSD included a full TCP/IP stack and
NFS support as well. Starting in the early iterated releases, small
tools like rcp, rsh, rlogin, and rwho appeared to demonstrate
capabilities. These were intended to be short lived, but those
familiar with UNIX will certainly recognize them. 4.2BSD also brought
disk quota facilities, a better install process, better
documentation, and some new filesystem related system calls.

Events less visible to users also occurred between 1982 and 1983. Joy
left the project in late spring of 1982 for Sun Microsystems, but
still spent a little time that summer working on IPC and reorganizing
the UNIX kernel sources to isolate machine dependent code. Once at
Sun, contributions to BSD continued with Sun sending their
modifications for running BSD on the Motorola 68000 back to Berkeley.
Leffler then took over Joy's previous duties. Joy's work, however,
would make BSD the single most easily ported operating system for
about the next twenty years. Pauline Schwartz was hired to take over
distribution duties in April of 1983. In June of 1983, Fabry went on
sabbatical and Ferrari and Susan Graham took over running the CSRG.
In 1984, Leffler went to work for Lucasfilm and Mike Karels took over
as the UNIX dev lead. He'd previously worked on the 2.?BSD series for
the PDP-11. Later that year McKusick joined CSRG full-time.

4.2BSD gained over a thousand site licenses in a year and a half, and
most UNIX system vendors actually shipped 4.2BSD (or a system based
on it) rather than AT&T's System V UNIX. Likewise, many varieties of
corporate UNIX would be based upon 4.1c or 4.2BSD such as SunOS and
DEC Ultrix. An important point at this time is that BSD's codebase
was largely divergent from that of AT&T while still mostly
maintaining compatibility. Additionally, AT&T had started shipping
UNIX without source, and those customers who bought AT&T's commercial
UNIX would often then separately obtain BSD sources if they wanted to
modify some aspect of the system. For those using microcomputers,
Microsoft's (and SCO's) XENIX was the standard.

Over on the PDP-11, a mixture of 4.1a and 4.1c formed 2.9BSD, but
4.2BSD never showed up.

In 1976, comic artist Phil Foglio drew the first versions of Beastie
as T-shirt art for Mike O'Brien as payment for unlocking a safe.

[T-shirt art of BSD by Phil Foglio]

<https://substackcdn.com/image/fetch/w_1456,c_limit,f_webp,
q_auto:good,fl_progressive:steep/
https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2F
images%2F1b00a913-e45d-44e7-b206-763ebac84cd4_688x833.jpeg>

Much of this is largely a pun on services being called daemons, and
UNIX making heavy use of pipes, with /dev/null being a bit bucket.
The daemon, Beastie, gained lasting association with BSD via a
drawing by John Lasseterf Lucasfilm being used on the cover of the
Unix System Manger's Manual that was published in 1984 by USENIX for
4.2BSD.

Cover of the manual, image from jacobelder.com
<https://substackcdn.com/image/fetch/w_1456,c_limit,f_webp,
q_auto:good,fl_progressive:steep/
https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2F
images%2F0cfe4090-7d76-4d07-acfd-0eadab7936cd_640x905.png>

While 4.2BSD was successful and mostly well received, it did get some
complaints. The majority of the complaints were centered on
performance. The team then spent two years improving performance,
refining the networking stack, and they felt that they were ready to
announce an impending release at USENIX in June of 1985. This didn't
go well. The fine folks of Bolt, Beranek, and Newman (part of the
steering commitee) noted that 4.2BSD had shipped without the final
version of the their networking code, and was instead using a heavily
modified version of their initial prototype. After some bickering
back and forth, DARPA provided both network stacks to Mike Muuss
(author of ping) of the Ballistics Research Laboratory for
testing.erkeley's code was better. 4.3BSD was released in June of
1986.

Keith Bostic joined CSRG in 1986 with the condition that he be
allowed to continue and to complete a prior project. He was working
to port 4.3BSD to the PDP-11. This meant that a 250K minimal system
on VAX would be made to fit in the 64K address space of the PDP-11...
no one thought this would work. No one. The man was clearly mad, but
despite what madness may have lurked within him, Bostic made it work
using a set of overlays and auxiliary processor states, and this
formed the 2.10BSD release. He also began attending USENIX and he'd
announce the progress of the removal of all AT&T code from BSD which
started in 1986 at thirty five percent AT&T license free, and the
announcement was met with widespread cheers and applause. This became
important due to price increases from AT&T. As of the 24th of
February in 1984, the price for a commercial license of UNIX System V
Release 2 with source for a single CPU stood at $43000 (about $126000
in 2024), and each additional CPU was a further $16000 (nearly $47000
in 2024). For educational institutions, the price was lower at $800
(or $2300 in 2024) and an additional $400 for each CPU.

[AT&T UNIX prices 1984]

<https://www.abortretry.fail/api/v1/file/
caa106e2-deee-4f25-b938-54f15bcd1f64.pdf>

Getting well into the 1980s, it became rather obvious to CSRG members
that VAX wouldn't go on forever. Computer Consoles Inc had opened a
development center in Irvine and there they developed a proprietary
minicomputer called the Power 6/32 code named Tahoe. This new
platform was aimed to compete directly with VAX. In need of an
operating system, CCI turned to CSRG. CCI provided the team several
machines and the team set about porting 4.3BSD to the 6/32. This
became 4.3BSD-Tahoe in June of 1988. Importantly, they were able to
push forward Joy's work of separating machine dependent code from the
rest of the system, and 4.3BSD-Tahoe introduced an OSI network
protocol stack, improved the kernel's virtual memory system, and
introduced more efficient TCP/IP support. Unfortunately, 4.3BSD-Tahoe
was short lived as CCI changed their company's focus, and Sperry and
Burroughs released rebranded minicomputers based on the Power 6/32
platform. Where this effort did live on, however, was in a merger of
4.3BSD-Tahoe and 2.10BSD into 2.11BSD for the PDP-11. The Tahoe
effort also meant that much of the BSD codebase had been rewritten
furthering the aim of removing AT&T licensed code.

BSD was released in source format only. Any prospective user would be
required to compile his or her system from source entirely. Given
this, any user would first need to acquire an AT&T source license
before he/she would be able to make use of BSD. As noted previously,
UNIX license fees were ridiculous. Yet, the TCP/IP stack in 4.3BSD
was unique to Berkeley. Several software developers requested that
BSD's networking code be offered separately from UNIX, and this was
done in June of 1989 as Networking Release 1. Pricing for a tape from
Berkeley was $400 (around $983 in 2024), but the license terms
allowed for free modification, free redistribution, and free
application to any use case provided that the copyright notices on
Berkeley's code remain in place, and that products incorporating the
code mention in documentation that code from the University of
California and its contributors was included. It was also available
via anonymous FTP shortly following the initial release. This
newsgroup post was made on the 7th of December in 1988, but other
documentation states public availability was November:

> Path:
utzoo!utgpu!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!mailrus!
> ames!pasteur!ucbvax!OKEEFFE.BERKELEY.EDU!bostic
> From: bos...@OKEEFFE.BERKELEY.EDU (Keith Bostic)
> Newsgroups: comp.bugs.4bsd.ucb-fixes
> Subject: V1.73 (BSD Networking Software, Release #1)
> Message-ID: <8812070154.AA18358@okeeffe.Berkeley.EDU>
> Date: 7 Dec 88 01:54:54 GMT
> Sender: dae...@ucbvax.BERKELEY.EDU
> Organization: University of California at Berkeley
> Lines: 374
> Approved: ucb-fi...@okeeffe.berkeley.edu
>
> We are happy to announce the availability of the first release of
> the BSD networking software. It consists of the standard user level
> applications, (along with their manual pages and some related
> documentation) and some kernel and C library support. It should be
> noted that this software has only been tested for compilation and
> operation on 4.3BSD and 4.3BSD-tahoe. A complete list of files is
> attached to this message.
>
> The TCP and IP code is approximately the same as that recently made
> available via the ARPANET and Usenet. Several new algorithms are
> used in TCP, in particular Van Jacobson's slow start and dynamic
> window size selection algorithms and Phil Karn's modification to
> the roundtrip timing algorithm. These changes increase throughput
> and reduce congestion and retransmission. Several fixes were made
> in the handling of IP options and other gateway support.
>
> This software suite is copyright The Regents of the University of
> California and may be freely redistributed. No previous license,
> either AT&T or Berkeley is required. The release costs $400.00 US.
> To request an order form, please contact our distribution office by
> phone at 415-642-7780, or by email at bsd-d...@ucbarpa.berkeley.edu
> or uunet!ucbarpa!bsd-dist, or by U.S. Mail at:
>
> CSRG, Computer Science Division
> University of California
> Berkeley, CA 94720
>
> Mike Karels
> Kirk McKusick

Rather than including the file list, I will simply note that this
release was very complete. It included arp, ftp clients and servers
(yes, plural on both), route, telnet, dns tools, ifconfig, inetd,
pieces of the BSD libc, sendmail, syslog, ping, and uucp.

While the networking release was being made, 4.3BSD development
continued. A new virtual memory system was implemented from MACH at
Carnegie-Melon with the porting and merging work done by Mike Hibler.
The interface for that VM system was, however, a purely Berkeley
design adhering to the architecture descriptions found in 4.2BSD. The
NFS system was upgraded to be Sun-compatible via the use of Rick
Macklem's work at the University of Geulph. This became 4.3BSD-Reno.

The original BSD license included in parts of Tahoe and Reno, and in
all of NET/1 and NET/2 read:

> Copyright (c) <year> <copyright holder>. All rights reserved.
>
> Redistribution and use in source and binary forms are permitted
> provided that the above copyright notice and this paragraph are
> duplicated in all such forms and that any documentation,
> advertising materials, and other materials related to such
> distribution and use acknowledge that the software was developed by
> the <copyright holder>. The name of the <copyright holder> may not
> be used to endorse or promote products derived from this software
> without specific prior written permission. THIS SOFTWARE IS
> PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR IMPLIED WARRANTIES,
> INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF
> MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.

Keith Bostic was still as avid as ever to rid BSD of AT&T code and
free it from the heavy costs that came with that code. At a meeting
of CSRG, he mentioned the popularity of the networking release and he
proposed an expanded release that would include more than networking.
Discussion went on for a while, but eventually McKusick and Karels
took up the kernel work with Bostic handling the utilities and C
library. Obviously, this would be a seriously large undertaking for a
single individual, and Bostic figured that others would be willing to
help. Bostic then encouraged people to coordinate over the budding
Internet with people across the globe submitting contributions. He
also encouraged people to contribute at USENIX. In a little over a
year, most of the utilities and libraries had been rewritten with
major contributions coming from Bill Jolitz, Donn Seeley, Trent Hein,
Vadim Antonov, Mike Karels, Igor Belchinsky, Pace Willisson, Jeff
Polk, and Tony Sanders. Karels and McKusick hadn't actually expected
Bostic to succeed, but with the work completed, Bostic walked into
their office with his head held high and inquired as to their
progress on the kernel. The two then went off to work on it file by
file, removing everything originally included in the 32V release, but
they were short by six files. Rather than getting a new license and
name created by the university's lawyers, the group reused what they
had and this became Network Release 2 announced on the 3rd of July in
1991.

> Path: gmdzi!unido!fauern!ira.uka.de!sol.ctr.columbia.edu!spool.mu.edu!
> caen!zaphod.mps.ohio-state.edu!cis.ohio-state.edu!ucbvax!
> OKEEFFE.BERKELEY.EDU!bostic
> From: bos...@OKEEFFE.BERKELEY.EDU (Keith Bostic)
> Newsgroups: comp.bugs.4bsd.ucb-fixes
> Subject: V1.95 (BSD Networking Software, Release #2)
> Message-ID: <9107032314.AA06592@okeeffe.Berkeley.EDU>
> Date: 3 Jul 91 23:14:59 GMT
> Sender: dae...@ucbvax.BERKELEY.EDU
> Organization: University of California at Berkeley
> Lines: 36
> Approved: ucb-fi...@okeeffe.berkeley.edu
>
> We are happy to announce the availability of the second release of the
> BSD networking software. The distribution includes approximately 75%
of
> the utilities distributed as part of 4.3BSD-Reno and the C library
> (along with manual pages and some related documentation), and much
> of the kernel. We wish to *strongly* emphasize, however, that
> significant portions of the kernel are missing and that no binary
> support is supplied for any architecture. Please note also that
> this software has only been tested for compilation and operation on
> 4.3BSD-Reno.
>
> This release is intended for system developers and others who wish to
> preview or experiment with the most recent Berkeley system. It may
also
> be useful as an update to earlier BSD or BSD-derived systems, although
> substantial work will be required to integrate portions of this
release
> into older systems. This distribution is *not* intended to be used on
> production systems, nor is it intended for sites without the expertise
> to find and fix problems that are encountered.
>
> This software suite is Copyright (C) 1991 The Regents of the
> University of California and may be freely redistributed without
> further charge. No previous license, either from AT&T or Berkeley
> is required. The release costs $850.00 US on 6250 BPI 9-track
> magnetic tape or 8mm Exabyte cassette or $950.00 US on 1600 BPI
> 9-track magnetic tape. The distribution is approximately 90Mb in
> size. To request an order form, please contact our distribution
> office by phone at 415-642-7780, or by sending email to
> bsd-d...@ucbarpa.berkeley.edu or uunet!ucbarpa!bsd-dist, or by U.S.
> Mail at:
>
> CSRG, Computer Science Division
> University of California
> Berkeley, CA 94720
>
> Mike Karels
> Kirk McKusick
> Keith Bostic
> Keith Sklower
> Marc Teitelbaum

Less than a year later, Bill Jolitz replaced the missing kernel
pieces and compiled everything for the PC-compatible Intel 80386
platform, and he released this new distribution as 386/BSD. Only a
few short months later, a group of 386/BSD users formed NetBSD to
continue enhancements and releases of Jolitz's system. This occurred
because Jolitz still had a full-time job, and the number of bug
fixes, enhancements, and additions coming into his project were
beyond his capabilities to manage alone. NetBSD has and continues to
emphasize porting NetBSD to absolutely everything, including but not
limited to ARM, MIPS, i386, AMD64, SPARC, PowerPC, Motorola 68K, SH3,
HPPA, Itanium, RISC-V, and VAX. If your toaster has a chip in it,
NetBSD will probably run on it. NetBSD has also engaged in some
research and experimentation that has led to many security and system
hardening technologies, an amazing package management system, kernel
level hardware virtualization, and some surprising levels of backward
compatibility. As is the BSD tradition, NetBSD has also merged in
changes from systems like Solaris with ZFS.

[A toaster at the Linux World Expo in 2005, running NetBSD]

<https://substackcdn.com/image/fetch/w_1456,c_limit,f_webp,
q_auto:good,fl_progressive:steep/
https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2F
images%2F9930c00b-0270-4a9d-ae51-bfa22c05721d_499x374.jpeg>

While NET2 was being worked on and released, the last version of
2.11BSD was also being prepared and released, but this time by
USENIX. On the 14th of March in 1991, Steven M. Schultz posted the
following to comp.sys.dec.micro:

> Second Distribution of Berkeley PDP-11 Software for UNIX
> Release 2.11
> (Revised January 1991)
>
> The USENIX Association is pleased to announce the distribution of a
> new release of the "Second Berkeley Software Distribution" (2.11BSD).
>
> This release will be handled by USENIX, and is available to all V7,
> System III, System V, and 2.9BSD licensees. The Association will
> continue to maintain the non-profit price of $200. The release will
> consist of two 2400 ft. 1600 bpi tapes or one TK50 tape cartridge
> (approximately 80M) and approximately 100 pages of documentation.
>
> If you have questions about the distribution of the release, or
> require 800 bpi tapes, please contact USENIX. USENIX's address and
> phone number is as follows:
>
> 2.11BSD
> USENIX Association
> 2560 Ninth St. Suite 215
> Berkeley, CA 94710
> +1-415-528-8649
>
> USENIX may also be contacted by electronic mail at:
>
> {ucbvax,decvax}!usenix!office
>
> If you have technical questions about the release, please contact
> Steven M. Schultz at:
>
> s...@wlv.imsd.contel.com
> wlbr!wlv!sms
>
> This release is in celebration of the 20th anniversary of the
> PDP-11! Work has been ongoing since the release of 2.10.1BSD in
> January 1989. This release incorporates all fixes and changes
> posted to the USENET newsgroup comp.bugs.2bsd since 2.10.1BSD was
> released.
>
> Present in this release are several more missing pieces from the
> 4.3BSD distribution:
>
> 1) the kernel logger (/dev/klog)
> 2) the namei cache and argument encapsulation calling sequence
> 3) readv(2)/writev(2) as system calls rather than
> emulation/compatibility routines
> 4) shadow password file implementation (the May 1989 4.3BSD
> update)
> 5) a TMSCP (TK50/TU81) driver with standalone support (boot-block
> and standalone driver)
> 6) Pronet and LH/DH IMP networking support
> 7) the portable ascii archive file format (ar, ranlib)
> 8) the Unibus Mapping RegisterUMR) handling of the network was
> rewritten to avoid allocating excessive UMRs.
> 9) the necessary mods to the IP portion of the networking were made
> to allow traceroute (which is present in 2.11BSD) to run
> 10) long filenames in the file system
>
> This last addition is the reason a coldstart kit is necessary. The
> 4.3BSD on-disk directory structure has been ported (along with the
> utilities that know about on-disk directories via the raw
> filesystem: fsck, ncheck, icheck, dcheck, etc.) and is not
> compatible with previous versions of UNIX for the PDP-11.
>
> A limited amount of filesystem backward compatibility with earlier
> versions of 2BSD (2.9BSD, 2.10BSD and 2.10.1BSD) is present in a
> version of dump(8) which can read old filesystems. The disk
> partition sizes have not changed from 2.10.1BSD (the urge to
> standardize the haphazard partition sizes was suppressed in the
> interest of backwards compatibility). The restor(8) utility
> automatically converts old dump tapes to the new format on input.
>
> The constant MAXNAMLEN is now 63 instead of 14. While it is
> possible the limit could be higher, with MAXPATHLEN at 256 a
> MAXNAMLEN of 63 was judged sufficient.
>
> MANY other fixes and changes have also been made, see the "Changes
> To The Kernel" document which describes the changes made to both
> the kernel and the application programs.
>
> Steven M. Schuz
> Contel Federal Systems
> 31717 La Tienda Drive
> Westlake Village CA 91359
> s...@wlv.imsd.contel.com
> wlbr!wlv!sms

A few months after the NetBSD group formed, the FreeBSD group formed.
They chose to target the PC architecture specifically and to make
their system a bit easier to use. In contrast to NetBSD which was
solely a product offered over the internet, the FreeBSD group offered
their product on CD-ROM. FreeBSD did expand to ARM, PowerPC, and
MIPS, and like NetBSD they did continue the CSRG spirit of research
and experimentation with things like bhyve virtualization and jails.
Additionally, they have continued the legacy of borrowing the best
bits of other systems like dtrace, and ZFS, but also via their
immense collection of software in ports.

Richard L. Adams Jr (Rick) was the founder of one of the earliest
internet service providers (if not the first), UUNET, after first
creating SLIP (TCP/IP over serial). He also maintained the most
popular usenet transport in the early 1980s, B News. After the
release of 386/BSD, he got together with Keith Bostic, Marshall Kirk
McKusick, Mike Karels, and Bill Jolitz and the group founded Berkeley
Software Design, Inc otherwise known as BSDi. Their first product was
BSD/386 based off of Networking Release 2 and it first released in
January of 1992. Their system retailed for $995 (about $2160 in
2024). The target market for BSDi was the nascent internet
infrastructure market. In their advertising they touted their “99%
discount over System V” and advised interested parties to contact
1-800-its-unix.

Of course, this didn't sit well with Unix System Labs (AT&T
subsidiary). Shortly after BSDi began sales, they received a cease
and desist from USL with a particular request to stop using the phone
number that included “it's UNIX,” as the ownership of the UNIX
trademark was firmly in USL's hands. BSDi complied changing their
advertisements, the number, and explaining that BSD wasn't precisely
UNIX. This, however, wasn't quite enough. USL brought a lawsuit
against BSDi seeking an injunction against the sale of BSD/386. As
part of the suit, USL claimed that BSDi's product contained USL code
and trade secrets, and that further sales of BSDi's product would
irreparably harm USL. BSDi then claimed that they shouldn't be held
liable for any code contained in Berkeley's original source
offerings, but that they were completely willing to discuss the six
added, BSDi-original files. BSDi's argument won and USL was required
to restate their complaint or have the case dismissed. USL then filed
suit against both BSDi and the University of California with roughly
the same complaints but this time seeking anle and distribution of both
Networking Release 2 and BSD/386. The
employees of both CSRG and BSDi were deposed.

In December of 1992, US District Judge Debevoise in New Jersey took
the arguments for injunction under advisement. Six weeks later, he
dismissed all but two of the complaints and suggested that the matter
should be taken to a state court before being heard in a federal
court. The University listened and filed a lawsuit against USL the
following Monday in California. The suit by the University of
California claimed that USL had failed to provide credit to the
University for BSD code contained within System V as required by
their prior licensing agreement. The University of California wasn't
asking for financial compensation but rather that USL reprint all of
their System V documentation with proper credit and attribution for
BSD code, that USL notify licensees of their mistake, and that USL
run advertisements in print news publications informing the public of
USL's mistake.

On the 21st of December in 1992, Novell announced that it would be
acquiring Unix System Laboratories including the UNIX copyright,
trademarks, and licensing contracts. The LA Times stated that this
transaction was completed with an exchange of stock wherein all of
the shares of USL would be traded for twelve million three hundred
thousand ser which USL would be a wholly owned subsidiary of Novell.
Discussions around a settlement of legal proceedings began in the
summer of 1993, but were not resolved until January of 1994. The
result was that three files were removed from Networking Release 2,
minor changes were made to a few other files, and USL copyrights were
added to about seventy files though the license of those files would
continue to be the BSD license. On the part of USL, they agreed that
with those terms met, the company would not bring litigation against
any company, group, or user who used or distributed the resulting
open source BSD system.

That resulting system was 4.4BSD-Lite and 4.4BSD-Encumbered in March
of 1994. 4.4BSD-Encumbered included the prior USL code and required a
USL UNIX license. 4.4BSD-Lite was fully free and open source
software. This version was merged into FreeBSD 2.0, NetBSD 1.0, and a
newly named BSD/OS which was previously BSD/386. For CSRG, the
revenue received from 4.4BSD allowed the group to continue as a
part-time effort integrating bug fixes and enhancements made locally
and from the BSD community. As focus shifted into the other
distributions like BSD/OS, NetBSD, and FreeBSD the rate of
submissions to Berkeley slowed. The final release from the University
was 4.4BSD-Lite Release 2 in June of 1995. This version was merged
into FreeBSD 3.0, Rhapsody (and later Darwin), NetBSD 1.3, OpenBSD
2.3, and BSD/OS 3.0.

By 1995, BSDi's BSD/OS was the most common operating system used in
the datacenters of the early internet. This is a market that BSDi had
intentionally targeted, and this would prove to be a sword of
Damocles. The strand of hair holding the sword broke as the Dot Com
Bubble began bursting and BSDi merged with Walnut Creek which itself
was later sold to Wind River Systems.

Today, BSD is among the most common operating systems on Earth
despite people not recognizing it as such. Modern macOS descends from
4.4BSD, as do NetBSD, OpenBSD, FreeBSD, Dragonfly BSD, and a few
others. In the past, various versions of BSD formed the basis of
SunOS, DYNIX, NeXTSTEP, Ultrix, and Tru64. Various editions of BSD
were merged into AT&T UNIX, and this means that modern commercial
UNIX systems like AIX and HP-UX are also running some BSD code. BSD
has been used in many commercial products owning to its permissive
licensing, and some of the most notable of those are the Sony
Playstation 3, 4, and Vita, the Netflix Open Connect Appliance,
Juniper routers, Isilon IQ clustered storage systems, Dell Compellent
storage systems, and the Weather Channel's IntelliStar forecast
computer.

From: <https://www.abortretry.fail/p/the-berkley-software-distribution>