I recently got the message:
kernel panic - not syncing! VFS unable to mount root fs on unknown block(3,3).

I've found a number of posts of people reporting this error, but so far they mostly relate to kernel compiles or software installs gone bad, so I don't know that they apply very much. I haven't touched my kernel or installed anything new in a while.

Here's what I DID do:

1) windows installation turned funky. Not sure how or why, but it died.
2) backed up all relevant data.
3) put in windows CD for repair installation.
4) aborted windows repair installation because: installer CD did not recognize existing installation, and the installer showed me to partitions with no labels and I could not rmember which was the linux and which was the windows partition
5) booted to Techiez Toolkit and relabelled my linux partition as "LINUX"
6) attempted to reboot into linux only to discover that windows installer CD had ruined my lilo bootloader, even though I had not performed the installation. Grrr....
7) put in slackware (v 10.0, I believe) recovery disk to restore lilo.
8) noticed that my linux partition was now at /dev/hda2, when it USED to be at /dev/hda3. shrugged and then mounted /dev/hda2, chroot'd it, ran lilo, and rebooted.
9) lilo shows up and lets be choose linux, which is when I got the kernel panic message.

I probably made the biggest mistake at step 8, but I don't know exactly what I did wrong, or how to fix it.


thanks in advance!

Looks like the MBR may have been corrupted. I would still try booting /dev/hda3 if that's what it used to be. I would also use a live CD to further investigate what happened.

I edited the lilo.conf file to boot from /dev/hda2 (I wasn't sure how to force it to /dev/hda3, and I'm REALLY confused by what got changed in the first place - would it make more sense to use the label of the partition, instead of /dev/hd*?), and the boot goes... a little further. I don't get a kernel panic anymore. Instead, fsck gets an error when it tries to do something with /dev/hda3, and thinks that it's a zero-length partition, I think. It then prompts me for root password to login and perform disk maintenance. If I login and run fsck, I get the same sort of error. If I try to "proceed with normal startup" by pressing Ctrl-D, the system just reboots. :S
Well... still better than kernel panic, I suppose.

looks like the partitions may be corrupted ... not good

did you save any info an the partition sizes and stuff ? you might be able to re-create the partitions. I'd try a live CD, maybe try mounting the visible partitions ...

Use a LiveCD and take a look at the disk using cfdisk. I'm suspecting a failing drive.

actually, it was much simpler than that: I forgot to change the entry for / in fstab to be /dev/hda2. duhhh! Proof that I shouldn't do this stuff when I'm tired.
Now I've got it booted and running again.
Still I'm puzzled by the change from /dev/hda3 to /dev/hda2. Was it labelling the partition that did it or should I blame the windows installer CD? ...or something else?

yeah, that would do it

well, the window$ installer CD probably installed its own bootloader into the MBR, likely corrupting the partition /dev/hda1 ... so that it no longer exists. That shifts the others down by 1 .... right ? (I'm just guessing, BTW)

Glad YOU got it working. The live CD will show YOUR data, as suggested. IMO write info down on paper as YOU go, so YOU can see it when YA want to. The cfdisk program will get YOU repaired if the need arises. Yes, you're right about the M$Windows disc changing the the sequence of the partitions. If you're dual booting, the master boot record can be repaired by Linux. You do not need to use a separate disc, unless it is YOUR desire to use one. Another method is the old Floppy Disc Boot Loader. It is highly portable if YOU have Floppy Drive Device installed in YOUR system.