I compiled a custom kernel with Simple 53c710 SCSI suport (Compaq, NCR
machines) enabled, also enabled ISA support, ramdrive, _cramfs support_
(cramfs images system are put by RH on the install CD). I made a monolitic
kernel with no loadable modules. Make sure that you select in the
configuration tool (either make menuconfig or make xconfig) Code maturity level
options -> Prompt for development and/or incomplete code/drivers "Y",
otherwise you won't be able to choose the 53c710 SCSI driver!
This step is not necessary, but I ran out of ideas.
Note from Richard Black - It may be possible (but I haven't tested yet)
to use the Red Hat Installer - recovery.html
to perform the install - the question is if the boot.img has the sim53c710
or if it is even on the drv-disk.img floppy - if so, then once you access
the CD-Rom, you should be able to switch over to the Bash# prompt (ALT-F2)
to create the partitions, etc - you will likely have to create device nodes
while you are in this mode - see the recovery page I just listed - Answer: no,
the sim53c710 driver is not present on boot.img nor the drvdisk.img floppy.
- One other method would be to do a network install - that may make things
easier too. Of course you will have to have a supported NIC -- I fancy the
ISA ne2000, they seem to be cheap these days (about USD $15.00).
End Note from Richard Black.
Created a boot and a root disk (Bootdisk-HOWTO), using the customn
kernel, busybox, the basic system libs needed by parted, STRIPPED. Make
sure that the libc-2.2.93.so is from the .i386 rpm and not from the i686
rpm (The i686 rpm was installed by default on my Athlon).
Then boot the root/boot floppies, create partitions and filesystems with
parted, copy the contents of the rootdisk to a partition, then the
kernel, then lilo it to the mbr (perhaps not necessaryly to the mbr) --
see the
Bootdisk-HOWTO for help. Now you have a "rescue" image, that you can use
if you have any problems.
If you don't have the right devices in /dev (my case), create
them with mknod.
Copy the custom kernel to a RedHat boot disk (created from boot.img),
boot it; kernel parameters I've used (see the cpqlinux site for
explanations):
"linux sim710=addr:0x8000,irq=15 mem=exactmap mem=640K@0 mem=31M@1M expert"
RH boots ok, asks for driver disk, say "no";
When you are asked to select a partitioning method, it doesn't matter
what you select, both partitioning tools will generate an anaconda error.
Don't panic.
ALT+F2 gives you a shell. ALT+F3 gives you some driver messages.
mount one of your SCSI drives and copy the "/" to it (I've used
"#cp -dpR /* /mnt/sdb"), copy the kernel from the diskette, make it
bootable with lilo.
(this step requires patience)
Note from Richard Black: I would suggest to do some sort of for loop:
for X in *.rpm; do rpm -i package; done
Instead of *.rpm, you may want to specify certain packages.
I would try installing most of the lib packages first, that way you
have less dependencies that you need.
Also I would try to stay far away from using the "--nodeps" option
because I have found that even though you force a package in by using
"--nodeps", the package doesn't work right even if you resolve the
dependency problem afterwords. My solution on these broken packages
I forced in with "--nodeps" was to remove (sometimes forcefully remove)
and then reinstall, then the package works great. So in summary
please don't force the installation of any package with "--nodeps" unless
you know exactly what you are doing.
Probably this type of an installation can be handled with slinky or miniconda: http://www.rule-project.org/en/index.php
End of note from Richard Black.
boot RH8.0.
do "#rpm -i package" for the packages you want. You may need to use
the "--nodeps" option. On my system it helped _a lot_ to do a
"# rpmdb --rebuilddb" from time to time; speeded up the installs.
Problems that I have encountered:
- LILO not booting
Solution: see cpqlinux.com, read the LILO documentation. Great stuff.
- boot messages complaining about module dependencies.
Solution: i ignored those, nothing bad hapened so far.
- ext2 fs, no journaling (parted's fault)
Solution: Use "# tune2fs -j /dev/sdx" , where x is your ext2 partition's
number that you want journaled.
- system stops booting at some time, before entering the runlevel (that
is, during /etc/rc.d/rc.sysinit's execution)
A Solution: I had errors in fstab. Make sure fstab is ok.
- can't mount other partitions
Solution: If "# fdisk -l" gives you an error concerning
"/proc/filesystems", type "# mount -n -t proc /proc" and update fstab.
Varga, thanks for your success page.
Varga Robert ( frozenfingers at home dot ro )
Search this Site!: