- TRIM for keyslots
- Do we need crypt_data_path() - path to data device (if differs)?
- Resync ETA time is not accurate, calculate it better (last minute window?).
-- Crypt benchmark cannot ttest ECB mode.
-- Log doesn't work yet in early binary start (FIPS message).
+- Extend existing LUKS header to use another KDF? (https://password-hashing.net/)
+- Fix all crazy automake warnings (or switch to Cmake).