After upgrading my system from RH6.2, and electing *not* to change the partition table or re-run lilo during the install process, lilo fails with a 'LI' prompt at reboot. I was (and am) running kernel 2.2.16, custom-built.
Since the kernel is replaced that is what I would expect to occur. I hesitate to class this as notabug, its too like a trap..
Known behavior but definitely undesired. Lilo should at least arrive at LILO boot: prompt if kernel has moved, and fail only upon attempting to load the invalid image - still allowing to load different 'failsafe', 'dos' or similar custom images.
It's not the kernel change that is blowing things up, it's the upgrade to lilo itself and the fact that the old lilo second stage boot loader is gone and the new one has not been mapped out for the boot sector code. It may seem like a trap, but it's also unreasonable to expect that you can remove and reinstall lilo without at least running it once on the old config file to remap things. Not-A-Bug.