Hello,

i'm wondering if linux is logging dmesg like logs of kernels which do not run at the moment? E.g.: let's say i have kernel 2.6.9 running at the moment. I compile a new kernel (2.6.15) and test it, but the boot-up procedure with the new kernel is not successful. Now i'd like to see the dmesg output of the 2.6.15 kernel (since the only working kernel on the machine is 2.6.9 it has to be from this one).

Maybe?

I've never tried this, but it might give you what you are trying to get.

Boot the 2.6.15 kernel, let it fail, then throw in a Live CD (knoppix comes to mind). Then, once it's booted up, mount up your root partition and chroot in:
mount /mnt/hda1
chroot /mnt/hda1

Assuming /dev/hda1 is your normal root Then run:
dmesg
And see what you get.

Your other options are to examine the /var/logs to see if any of them have info on why it's failing. Usually, why it failed is (a kernel panic is *usually* because the filesystem support wasn't added or /etc/fstab is not correct if devices change or something) more important than what dmesg is going to tell you. Try to write down what the error is when it actually fails to load.

Cool

Thank you!! Yeah, you brought me to an idea!! Why using a live CD, i can just use rescue CD or installation CD, which boots me into the single mode (like Slackware's CD No. 1) and follow the procedure you suggested!! ))) Heh, somethimes the solutions for problems are in front of you, you just need a little push into the right direction to find them! Thank you MasterC!

Writing down the error is always a good choice and i'm doing that, but sometimes i need to know about things which happened before the error and are not seen on the screen anymore.