Slackintosh 8.1 installation for OldWorld Macs

last updated 17 July 2002

Overview

This installation procedure uses a pair of boot/root disks that allow OldWorld Macs to easily use the full Slackware installation procedure (rather than manually running installpkg commands).

These disks are based on the boot/root set Keith Keller put together (which, in turn, were based on the Debian boot floppies). These, however, have modified to boot in three stages:

  1. Loading the kernel with a miBoot floppy
  2. Loading a small initial root filesystem from the second floppy
  3. After detecting and mounting the Slackintosh CDROM, loading the final root filesystem from it
They are not useful as standalone 'rescue' type disks; they can only be used in conjunction with the Slackintosh CD-ROM. They currently don't support installation over the network, though I imagine that could be added.

I have only done a full installation on a Motorola Starmax 4000/160, so I can't be sure that these floppies will work on other machines.

Disclaimer

THE FOLLOWING INFORMATION IS PROVIDED "AS IS" WITHOUT WARRANTY OF ANY KIND. IN NO EVENT SHALL THE AUTHOR BE HELD LIABLE FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THE INFORMATION. THIS INCLUDES, BUT IS NOT LIMITED TO DIRECT DAMAGES, INDIRECT DAMAGES, INCIDENTAL DAMAGES, CONSEQUENTIAL DAMAGES, SPECIAL DAMAGES AND LOSS OF BUSINESS PROFITS.

Directions

Note: these are installation directions for a completely bare hard drive (i.e. without MacOS). I'm sure you'll need to do different things if you want to set up a dual boot system. The directions also assume you have a UNIX system available.

Installation

  1. Download the current Slackintosh tree using rsync: 
    rsync -avz rsync.slackintosh.exploits.org::slackintosh-8.1 /destination-directory
  2. Use the script supplied with the tree to generate an ISO image and burn it to CD.
  3. Download the two floppy images:
  4. Write them to two floppies: 
    dd if=slackintosh-bootdisk-1.bin of=/dev/fd0
dd if=slackintosh-rootdisk-1.bin of=/dev/fd0
  5. Boot the first floppy. The floppy masquerades as a MacOS boot disk, so it needs to be booted by the Mac ROM. Unless you've started messing with Open Firmware, it should "just work."
  6. Linux should load and prompt you for the second disk.
  7. After loading the filesystem on the second disk, it begins looking through your IDE devices for a CDROM. You should make sure that your Slackintosh CD-ROM is loaded in the drive before you get to this point.
  8. You may see some DMA-related error messages at this point as the system tries to mount the CD-ROM. It should eventually work, however.
  9. If it can't automatically mount a CDROM, it will prompt you for a device node from which it should attempt to mount the CDROM. This will probably be necessary for machines with SCSI CD-ROM drives. (The device node is of the form /dev/hdc or /dev/sda or something similar to this.)
  10. Once it has mounted your CDROM to /cdrom, it will attempt to decompress /cdrom/boot/root.gz to /dev/ram1. If this fails, it should tell you so.
  11. The system will now mount this new filesystem and use the pivot_root utility to swap it into place as the root filesystem. A couple of useful utilities will be copied to the /sbin2 directory on the 'new' root filesystem. The old root filesystem will be unmounted, the memory it used deallocated, and the initialization scripts from the new root filesystem run.
  12. You can now install as usual. (i.e., partition using fdisk and run setup) I would suggest creating a 'small' (about 100MB) partition to hold your kernels as your first partition, but please read through the whole document before you make any partitioning decisions.

Installing a bootloader

Even though all the files Slackintosh will need to run are presumably on your hard drive at this point, you'll still need to set up a bootloader. This is done with two separate entities: Open Firmware and quik.

Open Firmware plays the same role as a PC-compatible's BIOS. It's the first thing 'run' by the computer after boot, and normally hands things off to the Mac ROM when it's done doing its thing. However, you can set it to boot an arbitrary filesystem on your hard drive or floppy. (Keep in mind that with IDE it's confined to booting from master devices, and reportedly on some platforms confined to booting the master device on the primary channel.)

To access Open Firmware, press and hold down Command-Option-O-F as the machine boots. If you don't have the keys down by the time the chime ends, it will probably not go into Open Firmware. I found that the system also occasionally ignores the key combination during warm boots.

If you hold down the keys and your machine halts on a black screen, the problem is probably that your firmware is set to do its I/O through a serial port, rather than with the screen and keyboard. Your options at this point for remedying the situation are:

I chose the third option. Right now, Slackintosh proper doesn't include nvsetenv, but the root floppy will place a copy of it in /sbin2 so that you can use it during the installation process.

nvsetenv's syntax is very simple:

For many machines all you need to do is change input-device to kbd and output-device to screen. However, some might need special settings; penguinppc.org has a list of these at http://penguinppc.org/projects/quik/quirks.shtml. Additionally, you'll probably want to set auto-boot? to false ... at least while you're testing.

I strongly strongly suggest making sure that you can access Open Firmware interactively at boot before you continue. Relying on using nvsetenv or "BootVariables" exclusively may leave you in a position where your system is left unable to boot correctly. If this does happen, you can, as a last resort, use Command-Option-P-R at boot to reset the Open Firmware to factory settings. Be warned; if you do this, you may reset your input-device and output-device to your machine's serial port, leaving you in quite a predicament.

Here are some commands you might find useful inside of Open Firmware:

In any case, now that you've made sure you have a way to easily alter Open Firmware settings, let's take a look at quik.

Open Firmware can't load your Linux kernel on its own easily (if at all). That's where quik, a LILO-like kernel bootloader, comes in. In short, you tell Open Firmware to boot a disk/partition on which you've installed quik. quik is segmented into a first stage (stored on the partition's bootblock) and a second stage loader (which is stored in the partition's filesystem). The first stage is only smart enough to load the second stage into memory. The second stage is able to read ext2 partitions (though somewhat imperfectly) and will first read the quik.conf file and then present a menu of whatever options have been configured in quik.conf. After the user has chosen a kernel (or the timeout has expired) quik will then load and execute whatever kernel is chosen (or is the default).

There's a whole host of restrictions on quik that are important to understand:

This is the setup I ended up using successfully:

dump of Open Firmware settings (from nvsetenv, my changes in red)

little-endian?  false
real-mode?      false
auto-boot?      false
diag-switch?    false
fcode-debug?    false
oem-banner?     false
oem-logo?       false
use-nvramrc?    false
real-base       0xffffffff
real-size       0x100000
virt-base       0xffffffff
virt-size       0x100000
load-base       0x4000
pci-probe-list  0xffffffff
screen-#columns 0x64
screen-#rows    0x28
selftest-#megs  0x0
boot-device     ata/ata-disk@0:0
boot-file       
diag-device     fd:diags
diag-file       
input-device    kbd
output-device   screen
oem-banner      
oem-logo        
nvramrc         
boot-command    boot

Partitioning/kernel location

/boot/quik.conf 

init-message = "Slackintosh 8.1.\n"
partition=2
root=/dev/hda3
timeout = 50
default = linux

image = /vmlinux
        label = linux

printout of mac-fdisk /dev/hda 

root@diane:~# mac-fdisk /dev/hda
/dev/hda
Command (? for help): p
/dev/hda
        #                    type name                 length   base     ( size )  system
dump: name /dev/hda len 8 
/dev/hda1     Apple_partition_map Apple                    63 @ 1        ( 31.5k)  Partition map
/dev/hda2         Apple_UNIX_SVR2 boot                 204800 @ 64       (100.0M)  Linux native
/dev/hda3         Apple_UNIX_SVR2 root                2868800 @ 204864   (  1.4G)  Linux native
/dev/hda4         Apple_UNIX_SVR2 usr                 4096000 @ 3073664  (  2.0G)  Linux native
/dev/hda5         Apple_UNIX_SVR2 usrlocal           12288000 @ 7169664  (  5.9G)  Linux native
/dev/hda6         Apple_UNIX_SVR2 swap                2048000 @ 19457664 (1000.0M)  Linux swap
/dev/hda7         Apple_UNIX_SVR2 home                3072000 @ 21505664 (  1.5G)  Linux native
/dev/hda8              Apple_Free Extra              15554838 @ 24577664 (  7.4G)  Free space

Block size=512, Number of Blocks=40132502
DeviceType=0x0, DeviceId=0x0

Command (? for help):

printout of mount 

root@diane:~# mount
/dev/hda3 on / type ext2 (rw)
/dev/hda2 on /boot type ext2 (rw)
/dev/hda4 on /usr type ext2 (rw)
/dev/hda5 on /usr/local type ext2 (rw)
/dev/hda7 on /home type ext2 (rw)
none on /dev/pts type devpts (rw,gid=5,mode=620)
none on /proc type proc (rw)
Some comments:

Wait... I thought you said you stored your kernels in /boot?

Nice catch. You noticed an apparent discrepancy between my contention that kernels, bootloaders, and quik.conf are all stored in /boot and the fact that quik.conf clearly states that vmlinux is stored in the root directory. Here's the explanation:

Since quik's second-stage loader actually reads the ext2 filesystem to find your kernel, it needs to know the file location from the filesystem's point of view (as opposed to your overall tree's point of view). So, if you have kernels and quik.conf in separate partition /dev/hda2, mounted as /boot, your quik.conf would need to look something like this: 

partition=2
image=/vmlinux
	label=Linux
... since vmlinux, while it's /boot/vmlinux in your overall tree, is /vmlinux in the filesystem on partition 2.

I strongly recommend glancing through some of the additional resources listed below before getting elbow-deep in quik and Open Firmware. The instructions I give here are neither comprehensive nor are they intended to substitute for the more comprehensive quik and Open Firmware references listed below.

And finally, an executive summary of what you should do once you've installed all your Slackintosh packages and before you reboot.

  1. Make sure that any kernels you want to use are copied to /mnt/boot (where /mnt/boot has been created as a separate filesystem).
  2. Create a /mnt/boot/quik.conf file similar to the above. Use the pathnames that they will have (that is, pathnames relative to the logical filesystem they're in, omitting the /mnt/boot part).
  3. Execute quik -v -r /mnt -C /boot/quik.conf (this tells quik to treat /mnt as the real root of your filesystem tree, to use the quik.conf in /boot and to tell you what it's doing in detail).
  4. Make the necessary changes to Open Firmware (primarily, set boot-device to the Open Firmware path for the disk where you've installed quik). You can either reboot and jump into Open Firmware to do this, or use the nvsetenv command in /sbin2.
  5. Once all these changes have been made, you should be able to boot Linux by issuing a boot command at the Open Firmware prompt (or, if you've left auto-boot? set to true, by just letting the system restart).

Additional resources

Installation overviews

Open Firmware

quik

Miscellaneous Q and As

Motorola Starmax specific

Comments?

I have successfully used these floppies and instructions to install Slackintosh on my Starmax 4000/160. They have successfully booted into the installation environment on a beige G3 as well. Other than that, I have no idea whether they work. If you use them successfully, please drop me a line at    m  a  q  (at)  n  o  c  t  r  l  (dot)  e  d  u  .

Revision history