Hey guys. I just ran slackpkg, with my updates set to use the -current mirror. Everything downloaded and installed fine, and after it was all done it suggested that I run LILO. I ran LILO and it gave a warning about "assuming LBA32" or something like that. Now when I reboot my machine, it says

"Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(3,1)"

and thats all she wrote.

I can still boot to my old kernel and view my LILO.conf and nothing in there has changed at all.

Any suggestions?

Thanks, Paul

As the slackpkg documentation says "# Automated upgrade of kernel packages aren't a good idea ".

Brian

did you forgot to make initrd??

pretty much a lost cause then?

when I installed my 2.6 kernel initially I never made a initrd, and its been working fine all this time. I guess it just slipped my mind about the kernel packages too. If its going to be a pain to try and fix I can just re-install the system. I was just trying to see if there was a solution to what I have messed up.

I think you can try to install the huge kernel and the modules if you don't have them right now from current even when you have booted your old kernel. You have to take care that your lilo config file points to the old kernel, not only "vmlinuz" because installing the kernel package will make "vmlinuz" a symlink to the newly installed kernel. And append lilo.conf for the new kernel, before rerunning /sbin/lilo of course.

Since the "huge" kernel should have all usual file systems compiled in you can boot it. I guess the kernel you're booting with is the "generic" kernel ...

so I could probably just download the new kernel, which has been partially upgraded anyway by slackpkg and just do a normal "new kernel" install? either way, this is pretty much just a learning system I am running anyways, so I can get the ins and outs figured out.

Thanks

You have quite a few choices. You don't need to reinstall everything.

First, a comment about using -current. In Slackware -current is the development version and is not stable. If you are learning, you'd be much better off with the stable version (12.0 is the most recent).

Okay, having said that, your choices are:

1. Continue using your old kernel
2. Try to install the most recent generic kernel (and modules) from whichever tree you decide to use (-current or 12.0) and create the initrd per instructions
3. Compile your own kernel

Do not use the -huge kernels, as they are primarily for installation purposes. The recommendation is to switch to -generic once your system is installed.

Brian

What I would probably do (please check yourself if it makes sense):

• Adjust lilo.conf so that you get a new entry for your old kernel that points to the full name, not only vmlinuz, for example it could point to vmlinuz-huge-smp-2.6.21.5-smp. if you have already referenced the long name you can skip this step and the next I think
• run /sbin/lilo and reboot to see if your modified lilo.conf entry is working properly
• get one of the "huge" kernels from current, either smp or without it and also download the modules
• install these two, not with "upgradepkg" but with "installpkg", else your old kernels modules might get removed and also the kernel itself. or use tar to unpack the kernel and image files in a seperate folder and copy them over manually, as you like
• append a new entry to lilo.conf for that kernel, run /sbin/lilo, reboot and pray

IMHO it won't do any harm to use the huge kernel at least to get your system working with the new kernel. You can do a custom compile later without any problem. This way Twister512 can first forget about an initrd.

Hi Paul,

I'm brand new here (two-week old linux newbie) and I just went through the exact same thing myself. The generic kernels don't have the filesystem built into the kernel like the huge ones do so that's why you got the kernel panic.

You just need to make an initrd like the others said above. This loads your filesystem module before linux loads. Piece of cake. If I can do it anyone can.

Pat's instructions in /boot/readme.initrd are pretty good. I just copied one of his examples and substituted my correct boot partition.

As far as the lba32 warning that's actually harmless. You can simply add "lba32" in /etc/lilo.conf under global (minus the quotes) and that message will go away.

Running with the default huge kernel is fine to get up and running, but it loads SO much stuff that I had conflicts with device drivers. I switched to the generic kernel pretty quickly.

Good luck,

Dig

My understanding of his situation is that he does have a working system, he just doesn't have a (new) working kernel. So why install the huge kernel just to replace it with the generic kernel? Making an initrd is straight-forward and quick and well document.

Brian

You are right. And all roads lead to Rome, sooner or later

Personally I'm still too scared to approach slackware -current yet. LoL. I simply used the default generic kernels already installed in /boot, created initrd's for them both, and then added them to LiLo. Worked like a charm and got me a more streamlined system I could really play around with. I find 2.6.21.5 to be a nice stable kernel. Is there really any reason to upgrade? Just asking, not commenting.

That is the right question to ask, although there are many answers, depending on you, your computer, and how you use your computer. I have a server running 2.6.17 and Slack 11. It is stable, fast and I see no urgent reason to upgrade it. On the other hand, I have a laptop that is running 12.0 and my own custom 2.6.22 kernel, which I intend to upgrade to 2.6.23 soon. I like to keep that system on the kernel version used in -current, and will to upgrade to -current at some point.

For the OP, who expressed interest in learning, I would recommend that he build his own kernel, because it is a good learning experience.

My own two cents.

Brian