I have an old debian system, which I always have been able to upgrade over the years. I had installed an old kernel, compiled by me from sources.
After my last dist-upgrade (using testing distribution), I had a kernel panic at boot. My old kernel and my new libc are not able to work together anymore (I presume).
Even device names are changed: with my latest working kernel, I had /dev/hda and /dev/hdb, now those devices are recognized as /dev/sda and /dev/sdb.
Additionaly, two partitions (/dev/hda2 and /dev/hdb2) were joined in a raid array, which name was /dev/md0 (mounted as root), and now is /dev/md127. Partition /dev/hda1 was mounted as /boot, and /dev/hdb1 was the swap partition.
I used lilo as boot manager.
Now that everything is broken, I tried to use debian rescue mode, without success. I removed lilo and installed grub2, still nothing works.
So I decided to install a new minimal debian system in the old swap partition (/dev/hdb1, now /dev/sdb1) and, from there, I tried to rescue the system.
Still nothing: I can't install a working kernel (this time a precompiled one) in the old partitions. I tried to rebuild my old raid array, mount my boot partition, chroot-ed ant apt-get installed a new kernel: grub is not able to see nothing.
I don't know what else I could try...
mount --bind /sys /chroot/sys). Still need that error message. You runapt-get install linux-image-686or whatever, and surely it gives some errors.