Architectures Present in the Master Source Tree

Genetically, 4.3BSD-Quasijarus is the successor to CSRG's 4.3BSD-Tahoe release, so its kernel, toolchain, C library, etc. all include both VAX and Tahoe architectures.

Architectures Supported as Mainstream

Up to the 4.3BSD release, VAX was the only architecture supported by Berkeley UNIX, and obviously it was CSRG's primary architecture. Shortly after the 4.3BSD release the Tahoe architecture was merged in from Computer Consoles, Inc. (CCI). By 1988 CSRG shifted its attention to this architecture completely. CSRG's post-4.3 release was built and tested only on Tahoe, and it was even named 4.3BSD-Tahoe.

In retrospect, that was a very foolish move. VAX architecture is beloved and actively supported by a large user community. And where is Tahoe? It took the PUPS list a couple of days of discussion just to figure out what it was. Needless to say, my interest and support is 100% on the VAX side.

Despite the fact that CSRG's post-4.3 release was named 4.3-Tahoe and was built and tested only on Tahoe, a lot of VAX hardware support improvements were developed outside of Berkeley and merged into the main tree by CSRG from 1986 (4.3 release) to around late 1988. This is one of the reasons I have made the decision to use 4.3-Tahoe instead of 4.3 as the starting base for Quasijarus Project.

Since my interest is in VAXen and only in VAXen, VAX is the only architecture I will maintain, build, and test. The Tahoe-specific portions of the source tree will be left as they were in the 4.3-Tahoe release. However, I have SCCS-reconstructed them (see SCCS-Reconstruction), and I will gladly incorporate any contributions to these sources into the master source tree (or give people direct write access to these directories if technically possible).

You can find information about the 4.3BSD-Quasijarus VAX releases on the releases page.

You can find information about VAX hardware support, both present and future, on the VAX hardware support page.

More on the tahoe* Subdirs in the Kernel

There were two different 4.3BSD-Tahoe tapes, the original one and the one with HCX-9 support hacked in. The latter is the one on Marshall Kirk McKusick's CSRG Archives CD-ROM, and it's the one I used for most of my SCCS-reconstruction. However, when I started SCCS-reconstructing the tahoe* subdirs in the kernel, I found out that the HCX-9 hacks are shockingly dirty. Many files were hacked manually, bypassing SCCS, and even worse, in some places #include directives were replaced with the files being included smacked inline!

Clearly this is absolutely unacceptable for a system that's supposed to be open-ended and continously maintainable. However, cleaning up that mess and adding HCX-9 support in a clean and legitimate fashion requires knowing at least something about the Tahoe architecture and having a test machine. I don't know anything about the Tahoe architecture, I don't have any Tahoe machines, and I just don't really care about the Tahoe architecture, so I can't do it.

If you use and care about the Tahoe architecture, and you want to add clean and legitimate HCX-9 support to 4.3BSD-Quasijarus, feel free to take a stab. If you do it properly (not like Keith Bostic did it on the 4.3BSD-Tahoe+HCX-9 tape), I would be glad to merge your changes into the master source tree. For starters I would suggest looking at the deltas from the Reno development branch, since there the HCX-9 support was done normally, via SCCS.

@(#)arch.html 1.1 03/12/18

Michael Sokolov
msokolov@ivan.Harhan.ORG