Kernel Panic? That's bad right?

omegaworks · 08-14-2004, 12:26 AM

Hello everyone! Ive finally settled on a replacement linux distribution for my Windows box, Debian. Ive installed two other distros before, Mandrake and RedHat (back when it was free), but their package managment turned me off very quickly.

The problem with me is that Im an upgrade-freak. I love having the most recent version of everything on my system. I decided to go with the most recent "Stable" version of Debian that was available, Woody. I assumed that it would be a little more up-to-date then where I found my system at after the installation. I changed my apt-get sources from stable to unstable and ran it again.

This only got me up to a 2.4.x kernel (cant remember exactly what the version was atm), so I decided to apt-get the newest stable kernel, 2.6.7. After going through that process, I rebooted, and was greeted with a "Kernel Panic: " message that read:

Kernel panic: VFS: Unable to mount root fs on unknown-block(33,6)

The two messages before that might also be related:
VFS: Cannot open root device "2106" or unknown-block(33,6)
Please append a correct "root=" boot option.

I have no clue what to do. Any replies are greatly appreciated!

BTW: Love this smilie:

EDIT: Forgot to say that I have a backup boot diskette that loads a different kernel, so I am able to do everything normally, save boot without the diskette.

kevinalm · 08-14-2004, 12:53 AM

Well, yes, a kernel panic is generally "A Bad Thing" TM

In this case I suspect the cause is not compiling support for your root filesystem into the kernel. Other possible causes, an incorrect root=/dev/hd? on your grub or lilo .conf kernel append line, and a stale symlink pointing to the wrong initrd.

Tinkster · 08-14-2004, 12:56 AM

I would have assumed that the upgrade wouldn't
remove the older kernel from the grub/lilo menu.

As for "current" - I'd stick with testing for the time
being, rather than unstable. And the kernel panic
probably means that the file-system your new kernel
uses is not what you've actually been using.

Cheers,
Tink

omegaworks · 08-14-2004, 01:42 AM

Quote:

Originally posted by Tinkster
I would have assumed that the upgrade wouldn't
remove the older kernel from the grub/lilo menu.

As for "current" - I'd stick with testing for the time
being, rather than unstable. And the kernel panic
probably means that the file-system your new kernel
uses is not what you've actually been using.

Cheers,
Tink

That's odd, Ive been using ext3, and there have been no reports (at least from what Ive seen) of its exclusion in v2.6.
How would I go about compiling ext3 support into my new kernel?

If all else fails, is there a way to "revert" to the previous kernel?

Tinkster · 08-14-2004, 01:54 AM

Well ... if you haven't built a initrd with ext3
compiled in, and ext3 is a module ...

/me shrugs ...

What's the last thing it says before the kernel panic?

Cheers,
Tink

omegaworks · 08-14-2004, 02:11 AM

Quote:

Originally posted by Tinkster
What's the last thing it says before the kernel panic?

Before the two messages I put in my first post?
I cant remember anything before that exactly... Im currently browsing my /var/logs for a record of the boot.

EDIT:
Here are the two I have in my first post, they're easy to miss:

VFS: Cannot open root device "2106" or unknown-block(33,6)
Please append a correct "root=" boot option.

Then the "Kernel Panic:" occurs.

omegaworks · 08-29-2004, 05:27 PM

Update on the situation: I was able to compile EXT3 support directly into the kernel, but am experiencing a different problem.

I felt it didn't deserve a whole new thread, since the kernel emits the same panic message, just in a different place:

Code:

Net: Registered protocol family 1
Net: Registered protocol family 17
Net: Registered protocol family 15
RAMDISK: cramfs filesystem found at block 0
RAMDISK: Loading 1568 blocks [1 disk] into ram disk... done
Kernel panic: VFS: Unable to mount root fs on unknown lock (33, 6)

I think the reason for my problem is that I included drivers for my other IDE controller PCI card in the kernel when I compiled it. (It was previously undetected by the previous kernel) I think this triggered some sort of 'drive-letter-shift' that caused my hde disk drive to be labeled as hdi. I cant be sure of this since the boot messages go by so fast. (Are they logged somewhere?) Ive tried pointing the relevant portions of my lilo.conf file to hdi instead of hde, but, since I have to boot the old kernel (without the new IDE controller driver), /sbin/lilo gives me a "block device not found" error (or something similar).

Please help!