I've read Paul Moody's Mini Howto, but I have followed a slightly different approach to build my embedded linux.
The kernel uncompresses the initrd into memory, I think in /dev/ram0. So, if you put your whole filesystem in initrd.img.gz, and you tell lilo to mount root in /dev/ram0, you have the whole filesystem in memory.
I have tried it and it works. The advantage is that you don't need a compressed filesystem for initrd and another for the final root filesystem, you only need the last one.
I hope this helps you.
--
Carlos Mateo
[EMAIL PROTECTED]
