> Have you tried intentionally switching z and y or whatever your keyboard > layout differences are to US keyboard, i.e. using a mangled password > instead of the correct one? I tried tried all possible combinations between US and my German layout with a script. > Then again, if unlocking with a key file (if that used to work) fails, > the issues are obviously bigger. If you have the passphrase written down > somewhere, checking again that you're remembering it correctly is a > suggestion of last resort. I have it written down, the passphrase is 100% correct. (The script was superfluous but I didn't know what else to do.) > What looks a bit inconsistent in your case is one used keyslot for the > root volume and two used keyslots for the home volume. Yes. Inside root is a keyfile to automatically unlock home but home also has a key with the same passphrase as root. Not entirely sure why I set it up that way. > Can you try booting with the old installed kernel and initrd? Usually those are kept for some time after an upgrade. None of the fallback options work: mount: /new_root: no filesystem type specified. You are now being dropped into an emergency shell. sh: can't access tty; job control turned off. > Another thing I had to debug recently for a colleague was a failing > flash medium. There were no read errors, but single bits of the data had > flipped. A unreported bit flip affecting the keyslots would be > catastrophic AFAICS. That said, bitflips affecting only the keyslots and > nothing else would be a strange beast unless this is a really crappy > SSD. How often do you reboot during normal operation? I probably update and reboot every 2-6 weeks. This is a year old Dell XPS 9370 with a Toshiba SSD. Its SMART outpus looks OK to me. I posted luksDump and hexdump of the first sectors in my original mail. Wouldn't you see such corruption there, or with chk_luks_keyslots? _______________________________________________ dm-crypt mailing list dm-crypt@xxxxxxxx https://www.saout.de/mailman/listinfo/dm-crypt
