The basic problem is with the way the Red Hat installation procedure handles the pcmcia card. It tries to load it at the ioport 0x300, which, for some reason, registers as a conflict.
I don't know enough about passing variables to the kernel at boot time, or if you can even pass pcmcia paramaters at boot time. I was also unable to twiddle the right bits in the pcmcia cpio archive on the supp disk to tell it to skip over 0x300-0x30f.
Solution.
I used tomsrtbt
This great little floppy package gives you a whole lotta-linux on a single floppy, including pcmcia support and networking!
I boot-ed the laptop from the tomsrtbt disk, used fdisk to lay out the partitions like I wanted and mke2fs-ed them.
Here is how I tend to lay out my file systems:
/ 32 meg primary partition
(I make the rest of the disk an extended partition)
swap 128 meg (I like a lot of swap, you can get away with much less, depends
on how much real ram you have, the HP4000C maxes out at 32Meg)
/tmp 128 meg (again, you can get away with much less, but what the heck, I've
got 4 gigs here, and only one OS).
/var 128 meg (you only get logging and mail on this partition, I could have used
32 or 64 here, but I was feeling generous).
/usr rest of disk. I use the rest of the disk, regardless of how big the drive is,
I hate trying to decide how much space I need to split between /usr
and /home, so I make the rest /usr and make /home a symlink to /usr/home
Next, I edited /etc/pcmcia/config.opts, after it sets its 'include port ...' I added a line that said:
exclude port 0x300-0x30f
This prevents card services from using these 'conflicting' ioports for the NIC.
Next, I insmod the pcmcia related modules:
insmod pcmcia_core insmod i82365 insmod ds insmod 3c589_cs
Edit /etc/pcmcia/network.opts and set the proper info for my network..
Then I run cardmgr, it complains about the modules already being loaded, but it does bring up the interface with IP info.
For some reason, I didn't get the routing stuff quite right, so a few route(8) commands were needed to set things straight.
Once I was able to ping my NFS server, I mounted the RH CD on it, set it to export the mountpoint and nfs mounted it on the laptop.
# mkdir /install # mount nfs.server.org:/mnt/cdrom /install
The next task was to copy the install sets to the local drive. I decided to put the install sets on the partition set aside for /usr as it was the largest partition.
# mount /dev/hda8 /mnt
The only reliable way to copy things with tomsrtbt is with cpio(1l) and it happens to have a function for doing just this task.
Cpio uses a list of files to know what to stuff into the archive (or in this case, what files to copy). The easiest way to generate this list is to use find:
# pwd /install # find RedHat | cpio -pdLmv /mnt
The 'v' just makes it list the files as it copies them, I like to know the progress of things. =)
The copying can take quite a while, depending on network speed and the speed of the CD on the host.
Once the copying is finished, unmount all the partitions (you only really need to make sure you unmount the local disk, else you'll be subject to dealing with an 'unclean' partition later) and shutdown.
Boot the box from the RH install set, and off you go! Tell it to install from local HD, and point it at the partition that has the install-sets.
For information on getting X11 to work, please refer to this page.
This document assumes that you know something about Linux and Unix in general. If you need for me to expand on any parts, please let me know, I'll be happy to get more detailed if needed.