Yeah, that's why it's called unstable :) it's been like that since the beginning on unstable. markos's script just edits a udev script to get around that. you will want to boot from some working image and then mount the root fs of the broken image and run this sed command:
sed -i -r -e "s/2.6.3\[0-1\]/2.6.30/g" $TARGETROOT/etc/init.d/udev
where $TARGETROOT is the path to the mounted root filesystem for the broken image.
the new kernel is on it's way soon and then we won't have to worry about doing things like this anymore.
also if you don't want to mess around with sed and are comfortable with shell scripts just edit /etc/init.d/udev directly and make sure your version of the kernel doesn't cause the supported_kernel() shell function to return 1.
the sed script should work though.
Let me know if this doesn't work.
Senior Software Engineer
Genesi USA, Inc.