Silo Issues

The first thing that looks out of place is your partition table. Are you using a DOS-style table, by chance? If so, you'll need to use fdisk on the hard drive and select 's' to install a Sun disklabel. You'll lose all your data (sorry), but it's required for the OpenBoot PROM to find SILO correctly (and not drop a "Bad Magic" error). Once you finish that, make slices (partitions) similar to what you've already planned out:

hda1: 20-50 MB, boot
hda2: 64-1024 MB, swap
hda3: (Reserved. Will report size of hard drive. Don't modify.)
hda4: Remainder, system root.

The next order of business is the configuration method. I'd advise against using 'siloconfig'. It's essentially 'liloconfig' modified for use with SILO. Unfortuantely, it's far from perfect. Ditch it and write your own silo.conf by hand. Mine looks a little like this:

partition = 1
root = /dev/hda4
timeout = 100
## Kernels
# 2.4.x series
image = /2.4.31/vmlinux-2.4.31
        label = ultra31

My setup avoids a problem dealing with kernel images residing on the root with the rest of the file system on my SparcStation4. I ran into a problem with SILO not being able to read past a certain point on the disk. I dealt with it like you by putting all my boot-related stuff on a seperate small partition. See this link for more info: http://lists.suse.com/archive/suse-sparc/2000-Oct/0015.html

There's two things to note about my configuration. First, the 'partition' and 'root' parameters don't match. When SILO is installed, it will reference configuration files and images on partition 1. This means silo.conf should exist under /boot. You can then point a symlink from /etc/silo.conf to /boot/silo.conf. The 'root=' parameter is passed to your booting kernel, and will properly set your system root to hda4.

Second, the 'image=' parameter refers to something in /2.4.31 which is correct when you think of hda1 as your root. It will show up properly under /boot when your main system comes up.

Finally, to install SILO you'll need just one or two more steps. Since you're booting over NFS, you'll be either using the miniroot (ramdisk) or chrooting into a more fully-featured environment. In either case, you'll need to manually mount your /proc so SILO can figure out the layout of the machine. Do a 'mount -t proc proc /proc' and you'll have a working proc until you reboot the machine (or exit the chroot). After that, simply type 'silo' (or 'silo -C /pathtoconfig/silo.conf' if you don't have something matching silo.conf in /etc). With any luck, you'll see a "/etc/silo.conf appears to be valid" message. Otherwise, tweak your silo.conf a bit more.

You'll reboot to a working system. Viola!

Good luck,
--Jason