FAQ? Why doesn't this use dmsetup/cryptsetup ? about dislocker HOT 2 OPEN

FUSE had multiple advantages at the time dislocker was written:

• I knew its existence - whereas network block devices I didn't, for example, or at least not in this context
• It was straight forward to use, it wouldn't be the problem in the implementation
• It was in userspace, so easier to test and wouldn't crash the kernel if faulting in some way
• Plus, the NTFS part was using FUSE, so it seemed logical.

Now, I've done some work to extract the core code into a library. The current dislocker implementation is a quite small program which actually uses this library. Therefore, I don't think making a network block device would be too difficult to implement.
It also makes sense to use dmcrypt/cryptsetup to implement it, BitLocker really has some resemblance in LUKS itself. It might also gain in performance. The problem would be portability however.

In essence, I don't really have a plan on this road of changing from FUSE. However, I'm open to any suggestion - like the one you cited.

As for a FAQ, I would say "why not!" I'll try to make some entries in it.

from dislocker.

Library pros:
.- Portability
.- Ability to embed in other software (data recovery e.g.)
.- No kernel panic

For optimization, that's neligible, imho, most of the performance benefits would come of optimizing the encryption code to use CPU extensions (ARM secure element, VIA Padlock, Intel AES-NI), and that can be done in user-space.

This is my suggestion.

from dislocker.