Skip to content
FreeBSDNews Published Updated 4 min readViews unavailable

FreeBSD 2.0 Ships, Finally Free of AT&T Code

FreeBSD 2.0's November 22, 1994 release rebuilt the system on 4.4BSD-Lite, cleared its inherited legal uncertainty, and reset installation and compatibility.

FreeBSD 2.0 was announced by Jordan Hubbard on November 22, 1994. The project’s own announcement described it as a full 32-bit operating system for i386, i486, and Pentium-class PCs, newly ported to the 4.4BSD-Lite code base. That rebase, rather than one isolated feature, is why 2.0 became a dividing line in FreeBSD history.

The familiar shorthand that it was “free of AT&T code” captures the release’s purpose but compresses a complicated settlement and engineering transition. The historically supportable claim is the one FreeBSD made at release time: 2.0 used the unencumbered 4.4BSD-Lite base approved after the USL, University of California, Novell, and BSDi dispute, moving the project out of the legal uncertainty surrounding the earlier Net/2 lineage.

Why the 1.x code line could not simply continue

FreeBSD’s first releases grew from 386BSD and Berkeley’s Net/2 distribution. The Unix System Laboratories litigation challenged the status of code distributed through that family. The 1994 settlement encouraged BSD descendants to move to 4.4BSD-Lite, the Berkeley distribution prepared without the disputed components.

This was not a matter of deleting one file and changing a copyright banner. FreeBSD’s historical account calls 4.4BSD-Lite “rather incomplete,” while the 2.0 announcement says the system had changed almost entirely since 1.0. The team had to make a usable PC operating system again on the new base, preserving hardware support and filling integration gaps while the project’s legal future depended on finishing.

FreeBSD 1.1.5.1, released in July 1994, closed the 1.x series. The clean break explains why the 2.0 announcement did not promise a normal in-place upgrade from 1.x: users were told to back up password and user files, perform a full installation, and restore their data.

The November announcement explicitly credited “the blessing of Novell,” then the owner of USL and Unix, and called the new base unencumbered. FreeBSD 2.0’s release notes similarly listed legal approval from Novell and the University of California as a major benefit for commercial users.

That wording is stronger evidence than later retellings that assign a precise count of disputed files without context. Counts such as three removed files, roughly seventy notices, or six missing components refer to different parts or summaries of the wider Berkeley settlement and reconstruction story. They should not be presented as if they were FreeBSD 2.0’s own change statistics.

The settlement did not make every Unix-related trademark, patent, or license question disappear worldwide. It established the code base on which the project could publish with the legal confidence expressed by the release team. That distinction preserves the significance without turning a settlement into a broader legal guarantee.

2.0 was also a substantial technical release

Legal history can obscure what users received. The release notes described 64-bit file offsets, work toward POSIX compliance, improved multicast support, and multiple filesystems inherited or developed from the 4.4 line, including union and stackable filesystems, portals, kernfs, a log-structured filesystem, and NQNFS. Some were explicitly described as needing more work; this was an ambitious base, not a claim that every component was production-polished.

The installer supported tape, CD-ROM, NFS, FTP over SLIP or Ethernet, a parallel-port path, DOS floppies, and DOS hard-disk partitions. It also aimed to make coexistence with DOS and other operating systems easier, including mounting DOS filesystems and installing a multi-OS boot manager.

Hardware scope remained firmly PC-oriented. The release targeted ISA, VLB, EISA, and PCI systems from 386sx through Pentium, with lists of supported storage and network controllers. This was years before FreeBSD became the broad multi-architecture system familiar today.

The release process still looked very different

FreeBSD 2.0 was distributed from FTP mirrors and by Walnut Creek CDROM. The announcement encouraged users with direct Internet access to download two boot floppies and let the installer fetch the distribution over FTP. It thanked Walnut Creek for infrastructure and commercial support, and named Poul-Henning Kamp as release engineer.

The announcement’s candid language also shows the project’s stage of development. It called the RELEASE less dangerous than the ALPHA “we hope,” described an ALPHA-to-RELEASE path as forthcoming, and admitted that a 1.x upgrade might be too difficult to provide. Modern expectations around signed artifacts, automated CI, and long-supported branches should not be projected backward onto 1994.

Why 2.0 became the durable branch point

FreeBSD 2.0 did not instantly settle every technical problem. The more robust 2.0.5 followed in June 1995. Its importance is structural: it replaced the inherited 1.x foundation with the 4.4BSD-Lite line from which subsequent FreeBSD development proceeded.

That makes the release relevant to later features such as jails, ZFS, and the modern release-engineering process. Those technologies arrived much later, but the project could build them on a base it could distribute, improve, and support without the same Net/2 legal cloud.

FreeBSD 2.0 therefore deserves two descriptions at once: a major re-port performed under legal pressure, and a real operating-system release with a new installer, broader 4.4BSD facilities, and a deliberately broken upgrade path. Reducing it only to lawsuit history misses the engineering; reducing it only to features misses why the engineering had to happen.

Related:

Sources:

Comments