lib/modules on dual distro boot

While I was trying to get one of my distros configured, I figured maybe installing a second distro might be the best way to go to check/compare file differences. Well, somehow in the meantime, either during the install or during the comparing and changing of files, I seem to have changed the location of each modprobes default locations? Here's what I mean:

I type modprobe (whatever) and it returns an error message with:
cannot locate /lib/modules/2.4.18-6mdk/module.dep no such file or directory
or something very similar to that. This is the message I get from my Slackware install. I get just the opposite in slack:
cannot locate /lib/modules/2.4.18/module.dep no such file or directory
or similar.

Does anyone know why? Or even better, any ideas on how to fix it? Other than "Don't dual boot distros" or "Don't install Mandy (or Slackware)".

Thanks for any suggestions

Oh, looks like I forgot to mention a few system specs:

Slack=/dev/hdb1
home=/dev/hdb5 (shared)
Mandy=/dev/hdb8
Others=/dev/hdb9,10

They don't share the same /lib directories.

Cool, thanks

How's about just guessing? I can't even think of anything to try.

Ok so I bagged Mandy, I love it, but thought I would try something newer. So I now have SuSE8 installed. Here's the hd:

/dev/hdb1 = Slack
/dev/hdb2 = QNX
/dev/hdb5 = /home
/dev/hdb6 = SuSE

Ok, so anyway, I installed suse no problem, it runs fine. I booted back into Slack, and get a bunch of module errors. So I login and try modprobe emu10k1, and it returns with:
error: /lib/modules/2.4.18GB/modules.dep does not exist

Well I have /lib/modules/2.4.18/modules.dep but somehow during the suse install it messed with my Slack install. If I type uname -a I get Linux 2.4.18GB i586 now. I used to get something like Linux 2.4.18 i386 So it looks like the suse install messed with something on my slack, but I don't know what. If anyone has any ideas, or fixes, or wild suggestions, please feel free to share them.
Thanks!

Your computer seems to read from the wrong partition. the folder 2.4.18GB seems to come from SuSE, and you don't have it in Slack. In slack, the folder is named .../2.4.18/....
It just seems that the modules you want to use comes from SuSE (emu10k1).... the correct place should be without this GB.
Sorry, I don't know what to do about it, but at least I can pinpoint a problem
Hope it is helpful.
Wait... may be, you mounted wrong partitions or you share some partitions (doesn't look like it though). Just an idea

What does your lilo/grub config file look like. As far as I can see you are always booting up with the kernel which is placed on the /hdb6 partition even when you boot slackware. I assume when both slack and mandrake where getting the opposite you had slack setup to boot with mandrake's kernel and the opposite for mandrake.

The kernel always looks for the directory /lib/modules/`uname -r` to find the modules. So if you have the wrong kernel it won't find the right module directory.

boot    = /dev/hda
change-rules
reset
read-only
menu-scheme = Wg:kw:Wg:Wg
lba32
prompt
timeout = 80
message = /boot/message
default = Slackware

  image  = /boot/vmlinuz
  label  = Suse8
  root   = /dev/hdb6
  vga    = 791
  initrd = /boot/initrd
  append = " hdd=ide-scsi"

  image  = /boot/vmlinuz.suse
  label  = failsafe
  root   = /dev/hdb6
  vga    = 791
  initrd = /boot/initrd.suse
  append = "ide=nodma apm=off acpi=off  hdd=ide-scsi"
  optional

  other  = /dev/hda1
  label  = WindowsXP

  image  = /boot/memtest.bin
  label  = memtest86

  image  = /boot/vmlinuz
  root   = /dev/hdb1
  label  = Slackware
  append = "hdd=ide-scsi"
  read-only

  other  = /dev/hdb2
  label  = QNX

That's a copy of my lilo.conf To me it doesn't look like I am booting the wrong kernel, but maybe I am.

Though I don't think this is much possible....
Is it possible that by mistake, you renamed or over wrote some kernels?
--Sarin
PS: Also is it possible to know the kernel version without booting it? May be you can try that also. Diff kernel1 kernl2 might be last try too?

have a look in your fstab file, since you installed suse maybe what was in there is no longer relevant. Its just a guess.

try posting it if there is no obvious answers, anything at all could help with this one......

Alex

I think you may be booting the wrong kernel. If you look at these two sections:

image = /boot/vmlinuz
label = Suse8
root = /dev/hdb6
vga = 791
initrd = /boot/initrd
append = " hdd=ide-scsi"

image = /boot/vmlinuz
root = /dev/hdb1
label = Slackware
append = "hdd=ide-scsi"
read-only

You'll see they have the same path to the kernel and the same name. When you run lilo, it may only use only one of the root ' / ' partitions. Like, hdb6 to find the kernels. So, when you run either suse or slack, they'll use the same kernel.

Now, I don't know which partition lilo will look at. It could be which linux is the default. Or it could be which one comes first in the lilo.conf file.

I know of three ways that might fix it.

1. Copy the kernels to each others /boot folder. Don't forget to rename one of them. Run lilo and you're done.

2. Run suse linux. Mount the root partition of slack, hdb1, at say /mnt/slack. Then change slack's image line in suse's lilo.conf file to:

image=/mnt/slack/boot/vmlinuz

Run lilo and unmount the partition. You could run slack and mount suse.

3. Boot into slack and install lilo into it's boot partition, hdb1. Then edit suse's lilo.conf file and remove the slack section and replace it with this:

other=/dev/hdb1
label=Slackware

Run lilo. You could also boot into suse and edit slack's lilo.conf.

I hope this helps.

but the 'root' bit is different. this is the important bit since when lilo reaches a different section (such as the suse section) it will see the root section and know which partition to look at).

though i think you should try linuxcool's suggestion just in case.

I know that lilo will use the order that the sections are written in unless you specify a default (to answer linuxcool's question). So in your lilo.conf it should start at the slack section, though from there i don't know. If there was no 'default' line then it would use the top section and works its way down (this is how i have mine). Though since it is set up as a menu i don't see why this would make a difference ........

im lost now....too much drink and too little sleep hasn;t helped but i don;t see how the two distros are crossing boundries without it being mentioned in the /etc/fstab file (which would have to be done explicitly unless the mandrake/suse install is weird)

Alex

P.S.sorry if that was no help, someone super clever is going to arrive with the answer soon after all this deliberating.....

webtoe, I think I know how it works. When lilo is run, it doesn't look at the ' root= ' lines to find the root partition to use. It uses the currently mounted root partition. If suse is running and you run lilo, it will use suse's partition ( hdb6 ). If it's slackware that's runnung, it will be hdb1.

So, if MasterC is using suse's lilo.conf file, he would copy slackware's kernel into suse's /boot folder. That's if he used the no. 1 way to fix the problem.

Ok, I sort of get it...

But, then would I change this line:
root = /dev/hdb1 to
root = /dev/hdb6?

Thanks.

No, leave the root= lines alone.

You need to pick which one of the three ways to fix the problem.

No. 1 might be the easiest. If you are using suse's lilo.conf to boot, then copy slack's kernel to the /boot directory of suse. Remember to rename it. Say vmlinuz.slack and also change the name in the image=/boot/vmlinuz, in slacks section of suse's lilo.conf, to image=/boot/vmlinuz.slack. Run lilo and then reboot.

If you're using slack's lilo.conf to boot, copy suse's kernel into slack's /boot directory. Rename suse's kernel ( vmlinuz.suse8 ). Change image=/boot/vmlinuz to image=/boot/vmlinuz.suse8 in suse's section of slack's lilo.conf. Run lilo and reboot.

I hope this works.

Great, sounds good. Thanks alot for your ideas guys, I will give it a shot when I get home, and let you know the results.