Red Hat Bugzilla – Bug 57793
Loader problems in 7.2
Last modified: 2007-04-18 12:38:49 EDT
From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 5.01; Windows NT 5.0)
Description of problem:
Lilo gives "L" prompt and fixing it from the install CD fails.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Have a working dual-boot system running with 11 partitions on /dev/hda.
Debian on /dev/hda2 boots off a lilo using boot=/dev/hda
2. Put in a 6GB drive as /dev/hdc and install a plain-vanilla RedHat 7.2
on it. Tell RedHat to use lilo with boot=/dev/hda root=/dev/hdc1 (in point-
and-click terms, that is)
3. Realise that lilo gives the "L" prompt and then hangs and try to
reconfigure the loader with the installation CD
Actual Results: I got the "L" prompt. To mend this I inserted the CD and
booted into the installation program again. After choosing gurb as the
loader and not changing anything else, the install program finishes by
saying "no kernel packages were installed on your system. Your boot loader
configuration will not be changed." This message was repeated even when I
tried to boot the install program to reconfigure lilo.
Expected Results: I expedted lilo to work "out of the box", because it
did just that with an older debian system (I believe it is 2.0 or 2.2) on
my hardware (Athlon 1GHz, 40GB hd).
I also expected the install program to actually perform the updates I
I got my old system up and running again (with a working lilo that is) by
booting it off debian rescue discs with a 2.2.18 kernel.
My own guess is that lilo fails because the partition containing /boot
(/dev/hdc1 on the RedHat system) spans more than 1024 cylinders. BUT a
message about that should be given during installation.
Installing to a system with two distinct / filesystems (one Debian, one Red Hat)
and also two different /boot partitions (I'm assuming this is true, your email
wasn't specific on that point) will never work as long as both Debian's version
of lilo and Red Hat's version of lilo are fighting over the same master boot
record. This isn't a bug, it's simply not possible. What you have to do to get
things working is add a /boot2 partition to both the Debian and Red Hat root
filesystems and under each OS you need to mount the other OSes /boot partition
as /boot2 (or choose another name if you wish). Then, the lilo.conf file in
Debian needs to be made aware of both operating systems kernels and then you
need to rewrite the master boot record from the Debian operating system. If the
/boot partition of the Red Hat installation spans more than 1024 cylinders, then
you won't be able to update the lilo boot record from Red Hat reliably, so don't
even try, just boot into Debian and make and kernel changes in there (if you
have to, install updated kernel RPMs from Red Hat into the Red Hat operating
system without making any changes to lilo.conf and without running lilo by
passing the --no-scripts option to rpm, then boot into Debian and make the lilo
changes there and rerun lilo, then you can boot back into the new Red Hat kernel).