Skip to content

Latest commit

 

History

History
199 lines (135 loc) · 7.6 KB

README.md

File metadata and controls

199 lines (135 loc) · 7.6 KB

Minix QD

Introduction

This is a very quick-and-dirty distribution of the old Minix 1 and 2 operating systems by Andrew Tanenbaum. It's not official in any way.

What is Minix QD?

Minix is a pedagogical operating system designed for teaching an OS course, first developed in the 1990s. Despite being restricted in scope and suitable for understanding by students, it's actually a surprisingly powerful and thoroughly usable Unix clone; older versions will run in tiny amounts of RAM. Their entire distribution, source code and C compiler included, will fit in about 40MB of disk and it'll recompile its own kernel and userspace in 1MB of RAM. (Slowly.) They'll run entirely off floppy disk. They'll even boot on an original IBM PC with 512kB of RAM, although they're a bit cramped.

There are three versions. The current one is Minix 3, which is still under active development; look there if you're interested.

Minix QD is my quick-and-dirty distribution of the old, obsolete versions of Minix. There are two:

Minix 1 is the older version. You get: multiuser logins, all the standard Unix tools, a full Bourne shell, an ANSI C89 C compiler, a Modula-2 compiler, a vi clone, an emacs clone, a pico clone, source for nearly everything, awk, bc, games, all the man pages, and huge amounts of other stuff. It's (almost) completely self-hosted --- you can hack Minix on Minix. The maximum supported amount of RAM is 16MB.

Minix 2 gives you all of the above, plus: a better interactive shell, a richer API, TCP/IP networking, swap, and lots of other stuff I haven't discovered yet (there isn't a changelog). The maximum supported amount of RAM is 64MB.

Apart from being interesting simply as works of software art, they work so well they're actually useful on really old hardware; so, this distribution.

I don't care about instructions, I just want something I can run!

Go to the release page; download one of the img.gz files; decompress it and write it to the appropriate media; then boot it in something.

What's here?

What you will find here is the contents of the last Minix 1 and 2 releases, 1.7.5 and 2.0.4, extracted and massaged so that they'll live in a standard user directory, plus a set of scripts that will take the distributions and turn it into a bootable Minix hard drive images. Minix is self-hosting, so we need to bootstrap a Minix somehow to build it on.

You won't find all the source here, because Minix came with most source, but not all; the C compiler is ACKPACK, a special version of the Amsterdam Compiler Kit, carefully cut down and modified to run on Minix. Back when this was released, this wasn't open source.

Note: this is Minix i86. The largest binaries can get is 64kB+64kB. This is restrictive compared to Minix i386, but it'll also build and run on machines with as little as 1MB of RAM, which is the main case I'm interested in.

Instructions

You will need a Linux machine and qemu-system-i386 and rsync.

Building and booting

Do:

$ ./mkall

This will rebuild all the installation images from the files in fs for both versions. (Alternatively, you can download these from the release page.)

  • combo-1440kB.img.gz, combo-1200kB.img.gz: combined root/usr floppy disk images. These contain a single filesystem with no division into root and usr. You can boot these. By default, the disk image will be loaded into a ramdisk to free up the floppy disk drive, but if you have less than about 3MB of RAM you should mount them directly; at the boot monitor prompt. hit ESC, then type rootdev=fd0, then boot.

  • hd-64MB.img.gz: is a bootable hard drive image containing absolutely everything; compilers, games, all the source code, the lot. Don't be put off by the size. Use this if you want to run Minix in a virtual machine.

  • root-720kB.img.gz, usr-720kB.img.gz: split root and usr floppy disk images. A 720kB floppy isn't big enough for the combined disk, so you need to boot the root disk, let it load itself into a ramdisk, and then it'll prompt you to replace the floppy with the usr image. It'll ask for a device; you want /dev/fd0.

Installing onto a hard drive

Once you have a floppy based system booted, you want to copy Minix onto your computer's hard drive.

The detailed instructions are long and complicated, but the tl;dr version for Minix 2 is to log in as root, and then:

$ part

...is the partition editor; the Minix partition type is 81. If you have less than 4MB RAM, create a swap partition (of about 4MB). You'll probably want to reboot (Minix sometimes forgets to reload the partition table). I'm going to assume you have the swap in /dev/c0d0p0 and your filesystem in /dev/c0d0p1. Don't make your filesystem too big; 64MB is probably as large as you want.

To mount the swap:

# mkswap /dev/c0d0p0  # controller 0, drive 0, partition 0
# mount -s /dev/c0d0p0

To create the filesystem and copy all the data:

# mkfs /dev/c0d0p1
# mount /dev/c0d0p1 /mnt
# cpdir -vx / /mnt/
# cpdir -vx /usr/ /mnt/usr/   # only if you're using the usr-720kB floppy
# echo "root=c0d0p1" > /mnt/etc/fstab
# echo "swap=c0d0p0" >> /mnt/etc/fstab   # vi is also available
# umount /dev/c0d0p1   # note that you use a device name here, not a path

To make the hard drive bootable:

# installboot -m /dev/c0d0 /usr/mdec/masterboot
# installboot -d /dev/c0d0p1 /usr/mdec/bootblock boot
# edparams /dev/c0d0p1
c0d0p1> rootdev=c0d0p1
c0d0p1> save
c0d0p1> exit
# shutdown -h now

...then reboot. Hopefully your new system will boot.

Installing Minix 1 is exactly the same, except:

  • there's no swap, so ignore that bit

  • devices are named differently; /dev/c0d0 => /dev/hd0; /dev/c0d0p0 => /dev/hd1; /dev/c0d0p3 => /dev/hd4; /dev/c0d1 => /dev/hd5; etc.

Hacking it

If you just want to fiddle with it, or do builds, then do this:

$ ./setversion minix-2.0  # or minix-1.7
$ ./pack

...and it will create an hd.img file, similar to the hd-64MB.img.gz one above. It's also lightly massaged for use as a build box, so when it boots it'll take you straight to a shell, etc. This then allows you to then do:

$ ./runqemu

That will run QEMU and you should see Minix boot. It'll take about a second.

Once you've finished working with Minix, exit the shell with ^D. It'll shut down cleanly. Then you kill the qemu window (sorry, Minix doesn't know how to tell qemu to shut down!) and then you can do:

$ ./unpack

...and your changes will be copied out of the image into the fs directory where you and the VCS can see them. (If you unpack from the wrong version, weird things will happen. Don't do that.)

References

License

Minix 1 and 2, and all my scripts, are BSD 3-clause. See the LICENSE.