49

Let's say I have 1 TB of data on a partition encrypted with BitLocker, TrueCrypt or VeraCrypt.

Does changing the encryption password imply rewriting all the data (i.e., will it take hours/days)?

  • 1
    For the record: Windows Bitlocker has no procedure to explicitly "rewrite" data. You must decrypt and re-encrypt the disk – usr-local-ΕΨΗΕΛΩΝ Nov 23 '18 at 11:05
  • 1
    Related to virtually wiping an entire encrypted disk by just erasing it's key, like some encrypting hard drives can do a nearly instantaneous "wipe" of terabytes – Xen2050 Nov 23 '18 at 13:05
  • 3
    Related to this (although neither answer mentions it): user-chosen passwords will be both too short, and have terrible entropy (too easily guessable). So the drive is encrypted with a good key... and then the encryption key is protected with a terrible one (ah well). – Clockwork-Muse Nov 23 '18 at 19:13
  • 1
    @Clockwork-Muse Still better than encrypt them with the original short key. – gvgramazio Nov 23 '18 at 19:59
79

No. Your password is used to encrypt only the master key. When you change the password, the master key is reencrypted but itself does not change.

(This is how some systems, such as BitLocker or LUKS, are able to have multiple passwords for the same disk: they still use a single master key for all data, but just store multiple copies of the master key encrypted with different passwords.)

  • Thank you very much! Would you have a link with details about that? Is the master key saved (encrypted by password) at the beginning (very first bytes) of the partition? – Basj Nov 22 '18 at 14:58
  • 1
    I don't have any useful links at hand, but see Twisty's answer regarding that. – grawity Nov 22 '18 at 18:17
  • 7
    Then the next question is obvious: is it possible to change the master key? – gvgramazio Nov 23 '18 at 20:01
  • @gvgramazio: Possibly, but that should be a separate thread – and you should mention what specific fde program that you are using and on what OS. (It's possible technically, but there might not be any tools available to actually do it.) Also mention the reason why you think changing it might be necessary... – grawity Nov 24 '18 at 13:08
  • The reason could be almost the same as why one wants to change the password. Maybe the password has been discovered, thus the master key has been discovered. Of course, one should have access to encrypted master key but maybe its possible. If one has the suspect that the master key has been discovered changing only the password has no effect. – gvgramazio Nov 24 '18 at 13:13
35

Grawity's answer is correct. Because encrypting data is a relatively expensive process, it makes more sense to create a single master key that does not change during the lifetime of the encrypted data. This master key can then in turn be encrypted by one or more secondary keys, which can then be flexibly changed at will.

For example, here's how BitLocker implements this (it actually uses three "layers" of keys):

  1. Data written to a BitLocker-protected volume is encrypted with a full-volume encryption key (FVEK). This key does not change until BitLocker is completely removed from a volume.
  2. The FVEK is encrypted with the volume master key (VMK) then stored (in its encrypted form) in the volume's metadata.
  3. The VMK in turn is encrypted with one or more key protectors, such as a PIN/password.

The following picture shows the process of accessing an encrypted system disk on a machine with BitLocker full volume encryption enabled:

Scheme of disk decryption

More information about this process can be found on TechNet.

  • 9
    Note: this also means that if anyone inclined enough to get copy of decrypted FVEK while they had (perhaps legitimate) access will continue to have unrestricted access to encrypted data if they come into contact with that encrypted disk, no mater how many times you change your PIN/password/VMK. Which is rather unfortunate (IOW, most times you change your passphrase, you should instead be doing full backup/wipe/recreate with new passphrase/restore cycle manually if you want protection from such cases.) – Matija Nalis Nov 23 '18 at 22:33
  • Quite true, though for this to be the case one would need either physical access or remote access with administrative rights. If an attacker has either of these...well, enough said. – Twisty Impersonator Nov 23 '18 at 22:52
  • 2
    yes, I was thinking physical access. Full disk encryption in irrelevant from security perspective while machine is running and disk is unlocked, anyway. However, it is supposed to protect your sensitive data if machine is off and lost or stolen (think laptop in taxi or airport), tampered with (think maid paid to give access in hotel room while owner is out), or hardware-failed or about to be decommissioned - now you'll still have to do the degaussing, physical shredding and incinerating of hardware instead of just recycling it (or giving to employees or selling on ebay etc) – Matija Nalis Nov 23 '18 at 23:14
  • 3
    @TwistyImpersonator The entire purpose of encrypting a disk is to protect your data when somebody has physical access. So the scenario is not moot; it is the whole point. – Lightness Races in Orbit Nov 24 '18 at 0:14
  • 1
    @LightnessRacesinOrbit I realize that. My comment was made in the context of the suggested vulnerability of the VMK before encryption is complete. In that specific window of time, encryption does not protect against an attacker with physical access or remote admin rights. – Twisty Impersonator Nov 24 '18 at 0:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.