UDEV migration issue? [Solved]

kellwood · Last edited by kellwood on Wed Apr 05, 2006 8:16 pm; edited 1 time in total

Hello All,

I have made several attempts to upgrade my system to a newer kernel and switch to udev but have failed. I apologize for posting but I have read the FAQ etc and I am looked at everything many times. The error that I get when I boot is:

NeddySeagoon · Posted: Tue Apr 04, 2006 4:35 pm Post subject:

kellwood,

kellwood · Posted: Tue Apr 04, 2006 4:41 pm Post subject:

Thanks for the reply. My root file system is reiserfs, which is built into the kernel. - Kev

kellwood · Posted: Wed Apr 05, 2006 12:15 pm Post subject:

Well, I am trying to figure this one out. Does /dev exists when I get that type of kernel panic? Perhaps some udev strangeness caused the scsi drives to be named differently than expected.

NeddySeagoon · Posted: Wed Apr 05, 2006 4:01 pm Post subject:

kellwood,

At the stage your error happens, /dev/ exists but is not mounted and cannot be read. With a pure udev system it should only contain /dev/null and /dev/console, other entries are harmless but unnessacary, except for strange hardware that udev does not yet support.
The kernel does not require the /dev/sdc2 entry to mount the root filesystem, that would create a cirular dependancy.
The /dev/sdc2 entry must be there when the rootfscheck script runs or you get a different error.

You have at least 3 SCSI drives, how are they attached ?
USB drives count as SCSI too, USB drives coming and going can change the /dev/sdX letters allocated by udev.
Be sure there are no USB storage devices attached when you run /sbin/lilo and again when you boot. When your system is up, you can write udev rules so USB devices don't get in the way of real hard drives.

If /dev/sdc2 (your root) is on USB, there are other things you need on your append line.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

kellwood · Posted: Wed Apr 05, 2006 5:45 pm Post subject:

Thanks again. The are four physical disk with 3+4 set up as hardware raid 0. This raid disk is /dev/sdc. Here is the output from fdisk when I am in the devfs kernel:

kellwood · Posted: Wed Apr 05, 2006 6:03 pm Post subject:

One more interesting thing - I think. While trying to boot the udev kernel, I noticed (or didn't notice) that there were no messages about finding the scsi controller that I normally see in the non-udev kernel. Unfortunately, I can't scroll back to confirm once that kernel panic occurs. The messages go by quickly but I did boot twice to try to watch for it.

Given this, I check the kernel config and the correct controller is built in. I started with a working .config to make the new kernel.

NeddySeagoon · Posted: Wed Apr 05, 2006 6:27 pm Post subject:

kellwood,

Boot you non-devfs kernel and post the output of lspci.
It will show what hardware your have.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

kellwood · Posted: Wed Apr 05, 2006 6:35 pm Post subject:

Ok, I am in my devfs kernel. Here it is

NeddySeagoon · Posted: Wed Apr 05, 2006 7:17 pm Post subject:

kellwood,

Hers your disc controllers from lspci:-

kellwood · Posted: Wed Apr 05, 2006 8:16 pm Post subject:

Many many thanks. That solved the whole issue. What had happened is that I had copied .config from an earlier kernel version thinking that all would be set. However, things had changed with respect to how FUSION drivers where handled.

NeddySeagoon · Posted: Wed Apr 05, 2006 8:54 pm Post subject:

kellwood,

You can copy the .config like that but you must run make oldconfig in the new kernel tree.
It uses the old .config as far as it can (and throws out the rubbish) and asks you to set all the new things.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.