8 6. Backup and Data Recovery
9 7. Interoperability with other Disk Encryption Tools
10 8. Issues with Specific Versions of cryptsetup
19 This is the FAQ (Frequently Asked Questions) for cryptsetup. It
20 covers Linux disk encryption with plain dm-crypt (one passphrase,
21 no management, no metadata on disk) and LUKS (multiple user keys
22 with one master key, anti-forensic features, metadata block at
23 start of device, ...). The latest version of this FAQ should
24 usually be available at
25 http://code.google.com/p/cryptsetup/wiki/FrequentlyAskedQuestions
30 ATTENTION: If you are going to read just one thing, make it the
31 section on Backup and Data Recovery. By far the most questions on
32 the cryptsetup mailing list are from people that managed to damage
33 the start of their LUKS partitions, i.e. the LUKS header. In
34 most cases, there is nothing that can be done to help these poor
35 souls recover their data. Make sure you understand the problem and
36 limitations imposed by the LUKS security model BEFORE you face
37 such a disaster! In particular, make sure you have a current header
38 backup before doing any potentially dangerous operations.
40 BACKUP: Yes, encrypted disks die, just as normal ones do. A full
41 backup is mandatory, see Section "6. Backup and Data Recovery" on
42 options for doing encrypted backup.
44 DISTRIBUTION INSTALLERS: Some distribution installers offer to
45 create LUKS containers in a way that can be mistaken as activation
46 of an existing container. Creating a new LUKS container on top of
47 an existing one leads to permanent, complete and irreversible data
48 loss. It is strongly recommended to only use distribution
49 installers after a complete backup of all LUKS containers has been
52 NO WARNING ON NON-INERACTIVE FORMAT: If you feed cryptsetup from
53 STDIN (e.g. via GnuPG) on LUKS format, it does not give you the
54 warning that you are about to format (and e.g. will lose any
55 pre-existing LUKS container on the target), as it assumes it is
56 used from a script. In this scenario, the responsibility for
57 warning the user and possibly checking for an existing LUKS header
58 is shifted to the script. This is a more general form of the
61 LUKS PASSPHRASE IS NOT THE MASTER KEY: The LUKS passphrase is not
62 used in deriving the master key. It is used in decrypting a master
63 key that is randomly selected on header creation. This means that
64 if you create a new LUKS header on top of an old one with
65 exactly the same parameters and exactly the same passphrase as the
66 old one, it will still have a different master key and your data
67 will be permanently lost.
69 PASSPHRASE CHARACTER SET: Some people have had difficulties with
70 this when upgrading distributions. It is highly advisable to only
71 use the 94 printable characters from the first 128 characters of
72 the ASCII table, as they will always have the same binary
73 representation. Other characters may have different encoding
74 depending on system configuration and your passphrase will not
75 work with a different encoding. A table of the standardized first
76 128 ASCII caracters can, e.g. be found on
77 http://en.wikipedia.org/wiki/ASCII
80 * 1.3 System Specific warnings
82 - Ubuntu as of 4/2011: It seems the installer offers to create
83 LUKS partitions in a way that several people mistook for an offer
84 to activate their existing LUKS partition. The installer gives no
85 or an inadequate warning and will destroy your old LUKS header,
86 causing permanent data loss. See also the section on Backup and
89 This issue has been acknowledged by the Ubuntu dev team, see here:
90 http://launchpad.net/bugs/420080
95 Current FAQ maintainer is Arno Wagner <arno@wagner.name>. Other
96 contributors are listed at the end. If you want to contribute, send
97 your article, including a descriptive headline, to the maintainer,
98 or the dm-crypt mailing list with something like "FAQ ..." in the
99 subject. You can also send more raw information and have me write
100 the section. Please note that by contributing to this FAQ, you
101 accept the license described below.
103 This work is under the "Attribution-Share Alike 3.0 Unported"
104 license, which means distribution is unlimited, you may create
105 derived works, but attributions to original authors and this
106 license statement must be retained and the derived work must be
107 under the same license. See
108 http://creativecommons.org/licenses/by-sa/3.0/ for more details of
111 Side note: I did text license research some time ago and I think
112 this license is best suited for the purpose at hand and creates the
116 * 1.5 Where is the project website?
118 There is the project website at http://code.google.com/p/cryptsetup/
119 Please do not post questions there, nobody will read them. Use
120 the mailing-list instead.
123 * 1.6 Is there a mailing-list?
125 Instructions on how to subscribe to the mailing-list are at on the
126 project website. People are generally helpful and friendly on the
129 The question of how to unsubscribe from the list does crop up
130 sometimes. For this you need your list management URL, which is
131 sent to you initially and once at the start of each month. Go to
132 the URL mentioned in the email and select "unsubscribe". This page
133 also allows you to request a password reminder.
135 Alternatively, you can send an Email to dm-crypt-request@saout.de
136 with just the word "help" in the subject or message body. Make sure
137 to send it from your list address.
139 The mailing list archive is here:
140 http://dir.gmane.org/gmane.linux.kernel.device-mapper.dm-crypt
146 * 2.1 What is the difference between "plain" and LUKS format?
148 Plain format is just that: It has no metadata on disk, reads all
149 paramters from the commandline (or the defaults), derives a
150 master-key from the passphrase and then uses that to de-/encrypt
151 the sectors of the device, with a direct 1:1 mapping between
152 encrypted and decrypted sectors.
154 Primary advantage is high resilience to damage, as one damaged
155 encrypted sector results in exactly one damaged decrypted sector.
156 Also, it is not readily apparent that there even is encrypted data
157 on the device, as an overwrite with crypto-grade randomness (e.g.
158 from /dev/urandom) looks exactly the same on disk.
160 Side-note: That has limited value against the authorities. In
161 civilized countries, they cannot force you to give up a crypto-key
162 anyways. In the US, the UK and dictatorships around the world,
163 they can force you to give up the keys (using imprisonment or worse
164 to pressure you), and in the worst case, they only need a
165 nebulous "suspicion" about the presence of encrypted data. My
166 advice is to either be ready to give up the keys or to not have
167 encrypted data when traveling to those countries, especially when
168 crossing the borders.
170 Disadvantages are that you do not have all the nice features that
171 the LUKS metadata offers, like multiple passphrases that can be
172 changed, the cipher being stored in the metadata, anti-forensic
173 properties like key-slot diffusion and salts, etc..
175 LUKS format uses a metadata header and 8 key-slot areas that are
176 being placed ath the begining of the disk, see below under "What
177 does the LUKS on-disk format looks like?". The passphrases are used
178 to decryt a single master key that is stored in the anti-forensic
181 Advantages are a higher usability, automatic configuration of
182 non-default crypto parameters, defenses against low-entropy
183 passphrases like salting and iterated PBKDF2 passphrase hashing,
184 the ability to change passhrases, and others.
186 Disadvantages are that it is readily obvious there is encrypted
187 data on disk (but see side note above) and that damage to the
188 header or key-slots usually results in permanent data-loss. See
189 below under "6. Backup and Data Recovery" on how to reduce that
190 risk. Also the sector numbers get shifted by the length of the
191 header and key-slots and there is a loss of that size in capacity
192 (1MB+4096B for defaults and 2MB for the most commonly used
193 non-default XTS mode).
196 * 2.2 Can I encrypt an already existing, non-empty partition to use
199 There is no converter, and it is not really needed. The way to do
200 this is to make a backup of the device in question, securely wipe
201 the device (as LUKS device initialization does not clear away old
202 data), do a luksFormat, optionally overwrite the encrypted device,
203 create a new filesystem and restore your backup on the now
204 encrypted device. Also refer to sections "Security Aspects" and
205 "Backup and Data Recovery".
207 For backup, plain GNU tar works well and backs up anything likely
208 to be in a filesystem.
211 * 2.3 How do I use LUKS with a loop-device?
213 This can be very handy for experiments. Setup is just the same as
214 with any block device. If you want, for example, to use a 100MiB
215 file as LUKS container, do something like this:
217 head -c 100M /dev/zero > luksfile # create empty file
218 losetup /dev/loop0 luksfile # map luksfile to /dev/loop0
219 cryptsetup luksFormat /dev/loop0 # create LUKS on loop device
221 Afterwards just use /dev/loop0 as a you would use a LUKS partition.
222 To unmap the file when done, use "losetup -d /dev/loop0".
225 * 2.4 When I add a new key-slot to LUKS, it asks for a passphrase but
226 then complains about there not being a key-slot with that
229 That is as intended. You are asked a passphrase of an existing
230 key-slot first, before you can enter the passphrase for the new
231 key-slot. Otherwise you could break the encryption by just adding a
232 new key-slot. This way, you have to know the passphrase of one of
233 the already configured key-slots in order to be able to configure a
237 * 2.5 Encrytion on top of RAID or the other way round?
239 Unless you have special needs, place encryption between RAID and
240 filesystem, i.e. encryption on top of RAID. You can do it the other
241 way round, but you have to be aware that you then need to give the
242 pasphrase for each individual disk and RAID autotetection will not
243 work anymore. Therefore it is better to encrypt the RAID device,
247 * 2.6 How do I read a dm-crypt key from file?
249 Note that the file will still be hashed first, just like keyboard
250 input. Use the --key-file option, like this:
252 cryptsetup create --key-file keyfile e1 /dev/loop0
255 * 2.7 How do I read a LUKS slot key from file?
257 What you really do here is to read a passphrase from file, just as
258 you would with manual entry of a passphrase for a key-slot. You can
259 add a new passphrase to a free key-slot, set the passphrase of an
260 specific key-slot or put an already configured passphrase into a
261 file. In the last case make sure no trailing newline (0x0a) is
262 contained in the key file, or the passphrase will not work because
263 the whole file is used as input.
265 To add a new passphrase to a free key slot from file, use something
268 cryptsetup luksAddKey /dev/loop0 keyfile
270 To add a new passphrase to a specific key-slot, use something like
273 cryptsetup luksAddKey --key-slot 7 /dev/loop0 keyfile
275 To supply a key from file to any LUKS command, use the --key-file
276 option, e.g. like this:
278 cryptsetup luksOpen --key-file keyfile /dev/loop0 e1
281 * 2.8 How do I read the LUKS master key from file?
283 The question you should ask yourself first is why you would want to
284 do this. The only legitimate reason I can think of is if you want
285 to have two LUKS devices with the same master key. Even then, I
286 think it would be preferable to just use key-slots with the same
287 passphrase, or to use plain dm-crypt instead. If you really have a
288 good reason, please tell me. If I am convinced, I will add how to
292 * 2.9 What are the security requirements for a key read from file?
294 A file-stored key or passphrase has the same security requirements
295 as one entered interactively, however you can use random bytes and
296 thereby use bytes you cannot type on the keyboard. You can use any
297 file you like as key file, for example a plain text file with a
298 human readable passphrase. To generate a file with random bytes,
299 use something like this:
301 head -c 256 /dev/random > keyfile
304 * 2.10 If I map a journaled file system using dm-crypt/LUKS, does it
305 still provide its usual transactional guarantees?
307 As far as I know it does (but I may be wrong), but please note that
308 these "guarantees" are far weaker than they appear to be. For
309 example, you may not get a hard flush to disk surface even on a
310 call to fsync. In addition, the HDD itself may do independent
311 write reordering. Some other things can go wrong as well. The
312 filesystem developers are aware of these problems and typically
313 can make it work anyways. That said, dm-crypt/LUKS should not make
316 Personally, I have several instances of ext3 on dm-crypt and have
317 not noticed any specific problems.
319 Update: I did run into frequent small freezes (1-2 sec) when putting
320 a vmware image on ext3 over dm-crypt. This does indicate that the
321 transactional guarantees are in place, but at a cost. When I went
322 back to ext2, the problem went away. This also seems to have gotten
323 better with kernel 2.6.36 and the reworking of filesystem flush
324 locking. Kernel 2.6.38 is expected to have more improvements here.
327 * 2.11 Can I use LUKS or cryptsetup with a more secure (external)
328 medium for key storage, e.g. TPM or a smartcard?
330 Yes, see the answers on using a file-supplied key. You do have to
331 write the glue-logic yourself though. Basically you can have
332 cryptsetup read the key from STDIN and write it there with your
333 own tool that in turn gets the key from the more secure key
337 * 2.12 Can I resize a dm-crypt or LUKS partition?
339 Yes, you can, as neither dm-crypt nor LUKS stores partition size.
340 Whether you should is a different question. Personally I recommend
341 backup, recreation of the encrypted partition with new size,
342 recreation of the filesystem and restore. This gets around the
343 tricky business of resizing the filesystem. Resizing a dm-crypt or
344 LUKS container does not resize the filesystem in it. The backup is
345 really non-optional here, as a lot can go wrong, resulting in
346 partial or complete data loss. Using something like gparted to
347 resize an encrypted partition is slow, but typicaly works. This
348 will not change the size of the filesystem hidden under the
351 You also need to be aware of size-based limitations. The one
352 currently relevant is that aes-xts-plain should not be used for
353 encrypted container sizes larger than 2TiB. Use aes-xts-plain64
360 * 3.1 My dm-crypt/LUKS mapping does not work! What general steps are
361 there to investigate the problem?
363 If you get a specific error message, investigate what it claims
364 first. If not, you may want to check the following things.
366 - Check that "/dev", including "/dev/mapper/control" is there. If it
367 is missing, you may have a problem with the "/dev" tree itself or
368 you may have broken udev rules.
370 - Check that you have the device mapper and the crypt target in your
371 kernel. The output of "dmsetup targets" should list a "crypt"
372 target. If it is not there or the command fails, add device mapper
373 and crypt-target to the kernel.
375 - Check that the hash-functions and ciphers you want to use are in
376 the kernel. The output of "cat /proc/crypto" needs to list them.
379 * 3.2 My dm-crypt mapping suddenly stopped when upgrading cryptsetup.
381 The default cipher, hash or mode may have changed (the mode changed
382 from 1.0.x to 1.1.x). See under "Issues With Specific Versions of
386 * 3.3 When I call cryptsetup from cron/CGI, I get errors about
389 If you get errors about unknown parameters or the like that are not
390 present when cryptsetup is called from the shell, make sure you
391 have no older version of cryptsetup on your system that then gets
392 called by cron/CGI. For example some distributions install
393 cryptsetup into /usr/sbin, while a manual install could go to
394 /usr/local/sbin. As a debugging aid, call "cryptsetup --version"
395 from cron/CGI or the non-shell mechanism to be sure the right
399 * 3.4 Unlocking a LUKS device takes very long. Why?
401 The iteration time for a key-slot (see Section 5 for an explanation
402 what iteration does) is calculated when setting a passphrase. By
403 default it is 1 second on the machine where the passphrase is set.
404 If you set a passphrase on a fast machine and then unlock it on a
405 slow machine, the unlocking time can be much longer. Also take into
406 account that up to 8 key-slots have to be tried in order to find the
409 If this is problem, you can add another key-slot using the slow
410 machine with the same passphrase and then remove the old key-slot.
411 The new key-slot will have an iteration count adjusted to 1 second
412 on the slow machine. Use luksKeyAdd and then luksKillSlot or
415 However, this operation will not change volume key iteration count
416 (MK iterations in output of "cryptsetup luksDump"). In order to
417 change that, you will have to backup the data in the LUKS
418 container (i.e. your encrypted data), luksFormat on the slow
419 machine and restore the data. Note that in the original LUKS
420 specification this value was fixed to 10, but it is now derived
421 from the PBKDF2 benchmark as well and set to iterations in 0.125
422 sec or 1000, whichever is larger. Also note that MK iterations
423 are not very security relevant. But as each key-slot already takes
424 1 second, spending the additional 0.125 seconds really does not
428 * 3.5 "blkid" sees a LUKS UUID and an ext2/swap UUID on the same
429 device. What is wrong?
431 Some old versions of cryptsetup have a bug where the header does
432 not get completely wiped during LUKS format and an older ext2/swap
433 signature remains on the device. This confuses blkid.
435 Fix: Wipe the unused header areas by doing a backup and restore of
436 the header with cryptsetup 1.1.x:
438 cryptsetup luksHeaderBackup --header-backup-file <file> <device>
439 cryptsetup luksHeaderRestore --header-backup-file <file> <device>
442 * 3.6 cryptsetup segfaults on Gentoo amd64 hardened ...
444 There seems to be some inteference between the hardening and and
445 the way cryptsetup benchmarks PBKDF2. The solution to this is
446 currently not quite clear for an encrypted root filesystem. For
447 other uses, you can apparently specify USE="dynamic" as compile
448 flag, see http://bugs.gentoo.org/show_bug.cgi?id=283470
454 * 4.1 I get the error "LUKS keyslot x is invalid." What does that
457 This means that the given keyslot has an offset that points
458 outside the valid keyslot area. Typically, the reason is a
459 corrupted LUKS header because something was written to the start of
460 the device the LUKS contaner is on. Refer to Section "Backup and
461 Data Recovery" and ask on the mailing list if you have trouble
462 diagnosing and (if still possible) repairing this.
465 * 4.2 Can a bad RAM module cause problems?
467 LUKS and dm-crypt can give the RAM quite a workout, especially when
468 combined with software RAID. In particular the combination RAID5 +
469 LUKS + XFS seems to uncover RAM problems that never caused obvious
470 problems before. Symptoms vary, but often the problem manifest
471 itself when copying large amounts of data, typically several times
472 larger than your main memory.
474 Side note: One thing you should always do on large data
475 copy/movements is to run a verify, for example with the "-d"
476 option of "tar" or by doing a set of MD5 checksums on the source
479 find . -type f -exec md5sum \{\} \; > checksum-file
481 and then a "md5sum -c checksum-file" on the other side. If you get
482 mismatches here, RAM is the primary suspect. A lesser suspect is
483 an overclocked CPU. I have found countless hardware problems in
484 verify runs after copying or making backups. Bit errors are much
485 more common than most people think.
487 Some RAM issues are even worse and corrupt structures in one of the
488 layers. This typically results in lockups, CPU state dumps in the
489 system logs, kernel panic or other things. It is quite possible to
490 have the problem with an encrypted device, but not with an
491 otherwise the same unencrypted device. The reason for that is that
492 encryption has an error amplification property: You flip one bit
493 in an encrypted data block, and the decrypted version has half of
494 its bits flipped. This is an important security property for modern
495 ciphers. With the usual modes in cryptsetup (CBC, ESSIV, XTS), you
496 get up to a completely changed 512 byte block per bit error. A
497 corrupt block causes a lot more havoc than the occasionally
498 flipped single bit and can result in various obscure errors.
500 Note, that a verify run on copying between encrypted or
501 unencrypted devices will reliably detect corruption, even when the
502 copying itself did not report any problems. If you find defect
503 RAM, assume all backups and copied data to be suspect, unless you
507 * 4.3 How do I test RAM?
509 First you should know that overclocking often makes memory
510 problems worse. So if you overclock (which I strongly recommend
511 against in a system holding data that has some worth), run the
512 tests with the overclocking active.
514 There are two good options. One is Memtest86+ and the other is
515 "memtester" by Charles Cazabon. Memtest86+ requires a reboot and
516 then takes over the machine, while memtester runs from a
517 root-shell. Both use different testing methods and I have found
518 problems fast with each one that the other needed long to find. I
519 recommend running the following procedure until the first error is
522 - Run Memtest86+ for one cycle
524 - Run memterster for one cycle (shut down as many other applications
527 - Run Memtest86+ for 24h or more
529 - Run memtester for 24h or more
531 If all that does not produce error messages, your RAM may be sound,
532 but I have had one weak bit that Memtest86+ needed around 60 hours
533 to find. If you can reproduce the original problem reliably, a good
534 additional test may be to remove half of the RAM (if you have more
535 than one module) and try whether the problem is still there and if
536 so, try with the other half. If you just have one module, get a
537 different one and try with that. If you do overclocking, reduce
538 the settings to the most conservative ones available and try with
545 * 5.1 Is LUKS insecure? Everybody can see I have encrypted data!
547 In practice it does not really matter. In most civilized countries
548 you can just refuse to hand over the keys, no harm done. In some
549 countries they can force you to hand over the keys, if they suspect
550 encryption. However the suspicion is enough, they do not have to
551 prove anything. This is for practical reasons, as even the presence
552 of a header (like the LUKS header) is not enough to prove that you
553 have any keys. It might have been an experiment, for example. Or it
554 was used as encrypted swap with a key from /dev/random. So they
555 make you prove you do not have encrypted data. Of course that is
556 just as impossible as the other way round.
558 This means that if you have a large set of random-looking data,
559 they can already lock you up. Hidden containers (encryption hidden
560 within encryption), as possible with Truecrypt, do not help
561 either. They will just assume the hidden container is there and
562 unless you hand over the key, you will stay locked up. Don't have
563 a hidden container? Though luck. Anybody could claim that.
565 Still, if you are concerned about the LUKS header, use plain
566 dm-crypt with a good passphrase. See also Section 2, "What is the
567 difference between "plain" and LUKS format?"
570 * 5.2 Should I initialize (overwrite) a new LUKS/dm-crypt partition?
572 If you just create a filesystem on it, most of the old data will
573 still be there. If the old data is sensitive, you should overwrite
574 it before encrypting. In any case, not initializing will leave the
575 old data there until the specific sector gets written. That may
576 enable an attacker to determine how much and where on the
577 partition data was written. If you think this is a risk, you can
578 prevent this by overwriting the encrypted device (here assumed to
579 be named "e1") with zeros like this:
581 dd_rescue -w /dev/zero /dev/mapper/e1
583 or alternatively with one of the following more standard commands:
585 cat /dev/zero > /dev/mapper/e1
586 dd if=/dev/zero of=/dev/mapper/e1
589 * 5.3 How do I securely erase a LUKS (or other) partition?
591 For LUKS, if you are in a desperate hurry, overwrite the LUKS
592 header and key-slot area. This means overwriting the first
593 (keyslots x stripes x keysize) + offset bytes. For the default
594 parameters, this is the 1'052'672 bytes, i.e. 1MiB + 4096 of the
595 LUKS partition. For 512 bit key length (e.g. for aes-xts-plain with
596 512 bit key) this is 2MiB. (The diferent offset stems from
597 differences in the sector alignment of the key-slots.) If in doubt,
598 just be generous and overwrite the first 10MB or so, it will likely
599 still be fast enough. A single overwrite with zeros should be
600 enough. If you anticipate being in a desperate hurry, prepare the
601 command beforehand. Example with /dev/sde1 as the LUKS partition
602 and default parameters:
604 head -c 1052672 /dev/zero > /dev/sde1; sync
606 A LUKS header backup or full backup will still grant access to
607 most or all data, so make sure that an attacker does not have
608 access to backups or destroy them as well.
610 If you have time, overwrite the whole LUKS partition with a single
611 pass of zeros. This is enough for current HDDs. For SSDs or FLASH
612 (USB sticks) you may want to overwrite the whole drive several
613 times to be sure data is not retained by wear leveling. This is
614 possibly still insecure as SSD technology is not fully understood
615 in this regard. Still, due to the anti-forensic properties of the
616 LUKS key-slots, a single overwrite of an SSD or FLASH drive could
617 be enough. If in doubt, use physical destruction in addition. Here
618 is a link to some current reseach results on erasing SSDs and FLASH
620 http://www.usenix.org/events/fast11/tech/full_papers/Wei.pdf
622 Keep in mind to also erase all backups.
624 Example for a zero-overwrite erase of partition sde1 done with
627 dd_rescue -w /dev/zero /dev/sde1
630 * 5.4 How do I securely erase a backup of a LUKS partition or header?
632 That depends on the medium it is stored on. For HDD and SSD, use
633 overwrite with zeros. For an SSD or FLASH drive (USB stick), you
634 may want to overwrite the complete SSD several times and use
635 physical destruction in addition, see last item. For re-writable
636 CD/DVD, a single overwrite should also be enough, due to the
637 anti-forensic properties of the LUKS keyslots. For write-once
638 media, use physical destruction. For low security requirements,
639 just cut the CD/DVD into several parts. For high security needs,
640 shred or burn the medium. If your backup is on magnetic tape, I
641 advise physical destruction by shredding or burning, after
642 overwriting . The problem with magnetic tape is that it has a
643 higher dynamic range than HDDs and older data may well be
644 recoverable after overwrites. Also write-head alignment issues can
645 lead to data not actually being deleted at all during overwrites.
648 * 5.5 What about backup? Does it compromise security?
650 That depends. See item 6.7.
653 * 5.6 Why is all my data permanently gone if I overwrite the LUKS
656 Overwriting the LUKS header in part or in full is the most common
657 reason why access to LUKS containers is lost permanently.
658 Overwriting can be done in a number of fashions, like creating a
659 new filesystem on the raw LUKS partition, making the raw partition
660 part of a raid array and just writing to the raw partition.
662 The LUKS header contains a 256 bit "salt" value and without that no
663 decryption is possible. While the salt is not secret, it is
664 key-grade material and cannot be reconstructed. This is a
665 cryptographically strong "cannot". From observations on the
666 cryptsetup mailing-list, people typically go though the usual
667 stages of grief (Denial, Anger, Bargaining, Depression, Acceptance)
668 when this happens to them. Observed times vary between 1 day and 2
669 weeks to complete the cycle. Seeking help on the mailing-list is
670 fine. Even if we usually cannot help with getting back your data,
671 most people found the feedback comforting.
673 If your header does not contain an intact salt, best go directly
674 to the last stage ("Acceptance") and think about what to do now.
675 There is one exception that I know of: If your LUKS container is
676 still open, then it may be possible to extract the master key from
677 the running system. See Item "How do I recover the master key from
678 a mapped LUKS container?" in Section "Backup and Data Recovery".
681 * 5.7 What is a "salt"?
683 A salt is a random key-grade value added to the passphrase before
684 it is processed. It is not kept secret. The reason for using salts
685 is as follows: If an attacker wants to crack the password for a
686 single LUKS container, then every possible passphrase has to be
687 tried. Typically an attacker will not try every binary value, but
688 will try words and sentences from a dictionary.
690 If an attacker wants to attack several LUKS containers with the
691 same dictionary, then a different approach makes sense: Compute the
692 resulting slot-key for each dictionary element and store it on
693 disk. Then the test for each entry is just the slow unlocking with
694 the slot key (say 0.00001 sec) instead of calculating the slot-key
695 first (1 sec). For a single attack, this does not help. But if you
696 have more than one container to attack, this helps tremendously,
697 also because you can prepare your table before you even have the
698 container to attack! The calculation is also very simple to
699 parallelize. You could, for example, use the night-time unused CPU
700 power of your desktop PCs for this.
702 This is where the salt comes in. If the salt is combined with the
703 passphrase (in the simplest form, just appended to it), you
704 suddenly need a separate table for each salt value. With a
705 reasonably-sized salt value (256 bit, e.g.) this is quite
709 * 5.8 Is LUKS secure with a low-entropy (bad) passphrase?
711 Note: You should only use the 94 printable characters from 7 bit
712 ASCII code to prevent your passphrase from failing when the
713 character encoding changes, e.g. because of a system upgrade, see
714 also the note at the very start of this FAQ under "WARNINGS".
716 This needs a bit of theory. The quality of your passphrase is
717 directly related to its entropy (information theoretic, not
718 thermodynamic). The entropy says how many bits of "uncertainty" or
719 "randomness" are in you passphrase. In other words, that is how
720 difficult guessing the passphrase is.
722 Example: A random English sentence has about 1 bit of entropy per
723 character. A random lowercase (or uppercase) character has about
726 Now, if n is the number of bits of entropy in your passphrase and t
727 is the time it takes to process a passphrase in order to open the
728 LUKS container, then an attacker has to spend at maximum
730 attack_time_max = 2^n * t
732 time for a successful attack and on average half that. There is no
733 way getting around that relationship. However, there is one thing
734 that does help, namely increasing t, the time it takes to use a
735 passphrase, see next FAQ item.
737 Still, if you want good security, a high-entropy passphrase is the
738 only option. For example, a low-entropy passphrase can never be
739 considered secure against a TLA-level (Three Letter Agency level,
740 i.e. government-level) attacker, no matter what tricks are used in
741 the key-derivation function. Use at least 64 bits for secret stuff.
742 That is 64 characters of English text (but only if randomly chosen)
743 or a combination of 12 truly random letters and digits.
745 For passphrase generation, do not use lines from very well-known
746 texts (religious texts, Harry potter, etc.) as they are to easy to
747 guess. For example, the total Harry Potter has about 1'500'000
748 words (my estimation). Trying every 64 character sequence starting
749 and ending at a word boundary would take only something like 20
750 days on a single CPU and is entirely feasible. To put that into
751 perspective, using a number of Amazon EC2 High-CPU Extra Large
752 instances (each gives about 8 real cores), this test costs
753 currently about 50USD/EUR, but can be made to run arbitrarily fast.
755 On the other hand, choosing 1.5 lines from, say, the Wheel of Time
756 is in itself not more secure, but the book selection adds quite a
757 bit of entropy. (Now that I have mentioned it here, don't use tWoT
758 either!) If you add 2 or 3 typos or switch some words around, then
759 this is good passphrase material.
762 * 5.9 What is "iteration count" and why is decreasing it a bad idea?
764 Iteration count is the number of PBKDF2 iterations a passphrase is
765 put through before it is used to unlock a key-slot. Iterations are
766 done with the explicit purpose to increase the time that it takes
767 to unlock a key-slot. This provides some protection against use of
768 low-entropy passphrases.
770 The idea is that an attacker has to try all possible passphrases.
771 Even if the attacker knows the passphrase is low-entropy (see last
772 item), it is possible to make each individual try take longer. The
773 way to do this is to repeatedly hash the passphrase for a certain
774 time. The attacker then has to spend the same time (given the same
775 computing power) as the user per try. With LUKS, the default is 1
776 second of PBKDF2 hashing.
778 Example 1: Lets assume we have a really bad passphrase (e.g. a
779 girlfriends name) with 10 bits of entropy. With the same CPU, an
780 attacker would need to spend around 500 seconds on average to
781 break that passphrase. Without iteration, it would be more like
782 0.0001 seconds on a modern CPU.
784 Example 2: The user did a bit better and has 32 chars of English
785 text. That would be about 32 bits of entropy. With 1 second
786 iteration, that means an attacker on the same CPU needs around 136
787 years. That is pretty impressive for such a weak passphrase.
788 Without the iterations, it would be more like 50 days on a modern
789 CPU, and possibly far less.
791 In addition, the attacker can both parallelize and use special
792 hardware like GPUs or FPGAs to speed up the attack. The attack can
793 also happen quite some time after the luksFormat operation and CPUs
794 can have become faster and cheaper. For that reason you want a
795 bit of extra security. Anyways, in Example 1 your are screwed.
796 In example 2, not necessarily. Even if the attack is faster, it
797 still has a certain cost associated with it, say 10000 EUR/USD
798 with iteration and 1 EUR/USD without iteration. The first can be
799 prohibitively expensive, while the second is something you try
800 even without solid proof that the decryption will yield something
803 The numbers above are mostly made up, but show the idea. Of course
804 the best thing is to have a high-entropy passphrase.
806 Would a 100 sec iteration time be even better? Yes and no.
807 Cryptographically it would be a lot better, namely 100 times better.
808 However, usability is a very important factor for security
809 technology and one that gets overlooked surprisingly often. For
810 LUKS, if you have to wait 2 minutes to unlock the LUKS container,
811 most people will not bother and use less secure storage instead. It
812 is better to have less protection against low-entropy passphrases
813 and people actually use LUKS, than having them do without
814 encryption altogether.
816 Now, what about decreasing the iteration time? This is generally a
817 very bad idea, unless you know and can enforce that the users only
818 use high-entropy passphrases. If you decrease the iteration time
819 without ensuring that, then you put your users at increased risk,
820 and considering how rarely LUKS containers are unlocked in a
821 typical work-flow, you do so without a good reason. Don't do it.
822 The iteration time is already low enough that users with entropy
823 low passphrases are vulnerable. Lowering it even further increases
824 this danger significantly.
827 * 5.10 Some people say PBKDF2 is insecure?
829 There is some discussion that a hash-function should have a "large
830 memory" property, i.e. that it should require a lot of memory to be
831 computed. This serves to prevent attacks using special programmable
832 circuits, like FPGAs, and attacks using graphics cards. PBKDF2
833 does not need a lot of memory and is vulnerable to these attacks.
834 However, the publication usually refered in these discussions is
835 not very convincing in proving that the presented hash really is
836 "large memory" (that may change, email the FAQ maintainer when it
837 does) and it is of limited usefulness anyways. Attackers that use
838 clusters of normal PCs will not be affected at all by a "large
839 memory" property. For example the US Secret Service is known to
840 use the off-hour time of all the office PCs of the Treasury for
841 password breaking. The Treasury has about 110'000 employees.
842 Asuming every one has an office PC, that is significant computing
843 power, all of it with plenty of memory for computing "large
844 memory" hashes. Bot-net operators also have all the memory they
845 want. The only protection against a resouceful attacker is a
846 high-entropy passphrase, see items 5.8 and 5.9.
849 * 5.11 What about iteration count with plain dm-crypt?
851 Simple: There is none. There is also no salting. If you use plain
852 dm-crypt, the only way to be secure is to use a high entropy
853 passphrase. If in doubt, use LUKS instead.
856 * 5.12 Is LUKS with default parameters less secure on a slow CPU?
858 Unfortunately, yes. However the only aspect affected is the
859 protection for low-entropy passphrase or master-key. All other
860 security aspects are independent of CPU speed.
862 The master key is less critical, as you really have to work at it
863 to give it low entropy. One possibility is to supply the master key
864 yourself. If that key is low-entropy, then you get what you
865 deserve. The other known possibility is to use /dev/urandom for
866 key generation in an entropy-startved situation (e.g. automatic
867 installation on an embedded device without network and other entropy
870 For the passphrase, don't use a low-entropy passphrase. If your
871 passphrase is good, then a slow CPU will not matter. If you insist
872 on a low-entropy passphrase on a slow CPU, use something like
873 "--iter-time=10" or higher and wait a long time on each LUKS unlock
874 and pray that the attacker does not find out in which way exactly
875 your passphrase is low entropy. This also applies to low-entropy
876 passphrases on fast CPUs. Technology can do only so much to
877 compensate for problems in front of the keyboard.
880 * 5.13 Why was the default aes-cbc-plain replaced with aes-cbc-essiv?
882 The problem is that cbc-plain has a fingerprint vulnerability, where
883 a specially crafted file placed into the crypto-container can be
884 recognized from the outside. The issue here is that for cbc-plain
885 the initialization vector (IV) is the sector number. The IV gets
886 XORed to the first data chunk of the sector to be encrypted. If you
887 make sure that the first data block to be stored in a sector
888 contains the sector number as well, the first data block to be
889 encrypted is all zeros and always encrypted to the same ciphertext.
890 This also works if the first data chunk just has a constant XOR
891 with the sector number. By having several shifted patterns you can
892 take care of the case of a non-power-of-two start sector number of
895 This mechanism allows you to create a pattern of sectors that have
896 the same first ciphertext block and signal one bit per sector to the
897 outside, allowing you to e.g. mark media files that way for
898 recognition without decryption. For large files this is a
899 practical attack. For small ones, you do not have enough blocks to
900 signal and take care of different file starting offsets.
902 In order to prevent this attack, the default was changed to
903 cbc-essiv. ESSIV uses a keyed hash of the sector number, with the
904 encryption key as key. This makes the IV unpredictable without
905 knowing the encryption key and the watermarking attack fails.
908 * 5.14 Are there any problems with "plain" IV? What is "plain64"?
910 First, "plain" and "plain64" are both not secure to use with CBC,
911 see previous FAQ item.
913 However there are modes, like XTS, that are secure with "plain" IV.
914 The next limit is that "plain" is 64 bit, with the upper 32 bit set
915 to zero. This means that on volumes larger than 2TiB, the IV
916 repeats, creating a vulnerability that potentially leaks some
917 data. To avoid this, use "plain64", which uses the full sector
918 number up to 64 bit. Note that "plain64" requires a kernel >=
919 2.6.33. Also note that "plain64" is backwards compatible for
920 volume sizes <= 2TiB, but not for those > 2TiB. Finally, "plain64"
921 does not cause any performance penalty compared to "plain".
924 * 5.15 What about XTS mode?
926 XTS mode is potentially even more secure than cbc-essiv (but only if
927 cbc-essiv is insecure in your scenario). It is a NIST standard and
928 used, e.g. in Truecrypt. At the moment, if you want to use it, you
929 have to specify it manually as "aes-xts-plain", i.e.
931 cryptsetup -c aes-xts-plain luksFormat <device>
933 For volumes >2TiB and kernels >= 2.6.33 use "plain64" (see FAQ
934 item on "plain" and "plain64"):
936 cryptsetup -c aes-xts-plain64 luksFormat <device>
938 There is a potential security issue with XTS mode and large blocks.
939 LUKS and dm-crypt always use 512B blocks and the issue does not
943 6. Backup and Data Recovery
946 * 6.1 Why do I need Backup?
948 First, disks die. The rate for well-treated (!) disk is about 5%
949 per year, which is high enough to worry about. There is some
950 indication that this may be even worse for some SSDs. This applies
951 both to LUKS and plain dm-crypt partitions.
953 Second, for LUKS, if anything damages the LUKS header or the
954 key-stripe area then decrypting the LUKS device can become
955 impossible. This is a frequent occuurence. For example an
956 accidental format as FAT or some software overwriting the first
957 sector where it suspects a partition boot sector typically makes a
958 LUKS partition permanently inacessible. See more below on LUKS
961 So, data-backup in some form is non-optional. For LUKS, you may
962 also want to store a header backup in some secure location. This
963 only needs an update if you change passphrases.
966 * 6.2 How do I backup a LUKS header?
968 While you could just copy the appropriate number of bytes from the
969 start of the LUKS partition, the best way is to use command option
970 "luksHeaderBackup" of cryptsetup. This protects also against
971 errors when non-standard parameters have been used in LUKS
972 partition creation. Example:
975 cryptsetup luksHeaderBackup --header-backup-file h <device>
977 To restore, use the inverse command, i.e.
979 cryptsetup luksHeaderRestore --header-backup-file h <device>
982 * 6.3 How do I test a LUKS header?
986 cryptsetup -v isLuks <device>
988 on the device. Without the "-v" it just signals its result via
989 exit-status. You can alos use the more general test
993 which will also detect other types and give some more info. Omit
994 "-p" for old versions of blkid that do not support it.
997 * 6.4 How do I backup a LUKS or dm-crypt partition?
999 There are two options, a sector-image and a plain file or
1000 filesystem backup of the contents of the partition. The sector
1001 image is already encrypted, but cannot be compressed and contains
1002 all empty space. The filesystem backup can be compressed, can
1003 contain only part of the encrypted device, but needs to be
1004 encrypted separately if so desired.
1006 A sector-image will contain the whole partition in encrypted form,
1007 for LUKS the LUKS header, the keys-slots and the data area. It can
1008 be done under Linux e.g. with dd_rescue (for a direct image copy)
1009 and with "cat" or "dd". Example:
1011 cat /dev/sda10 > sda10.img
1012 dd_rescue /dev/sda10 sda10.img
1014 You can also use any other backup software that is capable of making
1015 a sector image of a partition. Note that compression is
1016 ineffective for encrypted data, hence it does not make sense to
1019 For a filesystem backup, you decrypt and mount the encrypted
1020 partition and back it up as you would a normal filesystem. In this
1021 case the backup is not encrypted, unless your encryption method
1022 does that. For example you can encrypt a backup with "tar" as
1025 tar cjf - <path> | gpg --cipher-algo AES -c - > backup.tbz2.gpg
1027 And verify the backup like this if you are at "path":
1029 cat backup.tbz2.gpg | gpg - | tar djf -
1031 Note: Allways verify backups, especially encrypted ones.
1033 In both cases GnuPG will ask you interactively for your symmetric
1034 key. The verify will only output errors. Use "tar dvjf -" to get
1035 all comparison results. To make sure no data is written to disk
1036 unencrypted, turn off swap if it is not encrypted before doing the
1039 You can of course use different or no compression and you can use
1040 an asymmetric key if you have one and have a backup of the secret
1041 key that belongs to it.
1043 A second option for a filestem-level backup that can be used when
1044 the backup is also on local disk (e.g. an external USB drive) is
1045 to use a LUKS container there and copy the files to be backed up
1046 between both mounted containers. Also see next item.
1049 * 6.5 Do I need a backup of the full partition? Would the header and
1050 key-slots not be enough?
1052 Backup protects you against two things: Disk loss or corruption
1053 and user error. By far the most questions on the dm-crypt mailing
1054 list about how to recover a damaged LUKS partition are related
1055 to user error. For example, if you create a new filesystem on a
1056 LUKS partition, chances are good that all data is lost
1059 For this case, a header+key-slot backup would often be enough. But
1060 keep in mind that a well-treated (!) HDD has roughly a failure
1061 risk of 5% per year. It is highly advisable to have a complete
1062 backup to protect against this case.
1065 * *6.6 What do I need to backup if I use "decrypt_derived"?
1067 This is a script in Debian, intended for mounting /tmp or swap with
1068 a key derived from the master key of an already decrypted device.
1069 If you use this for an device with data that should be persistent,
1070 you need to make sure you either do not lose access to that master
1071 key or have a backup of the data. If you derive from a LUKS
1072 device, a header backup of that device would cover backing up the
1073 master key. Keep in mind that this does not protect against disk
1076 Note: If you recreate the LUKS header of the device you derive from
1077 (using luksFormat), the master key changes even if you use the same
1078 passphrase(s) and you will not be able to decrypt the derived
1079 device with the new LUKS header.
1082 * 6.7 Does a backup compromise security?
1084 Depends on how you do it. However if you do not have one, you are
1085 going to eventually lose your encrypted data.
1087 There are risks introduced by backups. For example if you
1088 change/disable a key-slot in LUKS, a binary backup of the partition
1089 will still have the old key-slot. To deal with this, you have to
1090 be able to change the key-slot on the backup as well, securely
1091 erase the backup or do a filesystem-level backup instead of a binary
1094 If you use dm-crypt, backup is simpler: As there is no key
1095 management, the main risk is that you cannot wipe the backup when
1096 wiping the original. However wiping the original for dm-crypt
1097 should consist of forgetting the passphrase and that you can do
1098 without actual access to the backup.
1100 In both cases, there is an additional (usually small) risk with
1101 binary backups: An attacker can see how many sectors and which
1102 ones have been changed since the backup. To prevent this, use a
1103 filesystem level backup methid that encrypts the whole backup in
1104 one go, e.g. as described above with tar and GnuPG.
1106 My personal advice is to use one USB disk (low value data) or
1107 three disks (high value data) in rotating order for backups, and
1108 either use independent LUKS partitions on them, or use encrypted
1109 backup with tar and GnuPG.
1111 If you do network-backup or tape-backup, I strongly recommend to
1112 go the filesystem backup path with independent encryption, as you
1113 typically cannot reliably delete data in these scenarios,
1114 especially in a cloud setting. (Well, you can burn the tape if it
1115 is under your control...)
1118 * 6.8 What happens if I overwrite the start of a LUKS partition or
1119 damage the LUKS header or key-slots?
1121 There are two critical components for decryption: The salt values
1122 in the header itself and the key-slots. If the salt values are
1123 overwritten or changed, nothing (in the cryptographically strong
1124 sense) can be done to access the data, unless there is a backup
1125 of the LUKS header. If a key-slot is damaged, the data can still
1126 be read with a different key-slot, if there is a remaining
1127 undamaged and used key-slot. Note that in order to make a key-slot
1128 unrecoverable in a cryptographically strong sense, changing about
1129 4-6 bits in random locations of its 128kiB size is quite enough.
1132 * 6.9 What happens if I (quick) format a LUKS partition?
1134 I have not tried the different ways to do this, but very likely you
1135 will have written a new boot-sector, which in turn overwrites the
1136 LUKS header, including the salts, making your data permanently
1137 irretrivable, unless you have a LUKS header backup. You may also
1138 damage the key-slots in part or in full. See also last item.
1141 * 6.10 How do I recover the master key from a mapped LUKS container?
1143 This is typically only needed if you managed to damage your LUKS
1144 header, but the container is still mapped, i.e. "luksOpen"ed. It
1145 also helps if you have a mapped container that you forgot or do not
1146 know a passphrase for (e.g. on a long running server.)
1148 WARNING: Things go wrong, do a full backup before trying this!
1150 WARNING: This exposes the master key of the LUKS container. Note
1151 that both ways to recreate a LUKS header with the old master key
1152 described below will write the master key to disk. Unless you are
1153 sure you have securely erased it afterwards, e.g. by writing it to
1154 an encrypted partition, RAM disk or by erasing the filesystem you
1155 wrote it to by a complete overwrite, you should change the master
1156 key afterwards. Changing the master key requires a full data
1157 backup, luksFormat and then restore of the backup.
1159 First, there is a script by Milan that automatizes the whole
1160 process, except generating a new LUKS header with the old master
1161 key (it prints the command for that though):
1163 http://code.google.com/p/cryptsetup/source/browse/trunk/misc/luks-header-from-active
1165 You can also do this manually. Here is how:
1167 - Get the master key from the device mapper. This is done by the
1168 following command. Substitute c5 for whatever you mapped to:
1170 # dmsetup table --target crypt --showkey /dev/mapper/c5
1172 0 200704 crypt aes-cbc-essiv:sha256
1173 a1704d9715f73a1bb4db581dcacadaf405e700d591e93e2eaade13ba653d0d09
1176 The result is actually one line, wrapped here for clarity. The long
1177 hex string is the master key.
1179 - Convert the master key to a binary file representation. You can
1180 do this manually, e.g. with hexedit. You can also use the tool
1181 "xxd" from vim like this:
1183 echo "a1704d9....53d0d09" | xxd -r -p > <master-key-file>
1185 - Do a luksFormat to create a new LUKS header.
1187 NOTE: If your header is intact and you just forgot the
1188 passphrase, you can just set a new passphrase, see next subitem.
1190 Unmap the device before you do that (luksClose). Then do
1192 cryptsetup luksFormat --master-key-file=<master-key-file> <luks device>
1194 Note that if the container was created with other than the default
1195 settings of the cryptsetup version you are using, you need to give
1196 additional parameters specifying the deviations. If in doubt, try
1197 the script by Milan. It does recover the other parameters as well.
1199 Side note: This is the way the decrypt_derived script gets at the
1200 master key. It just omits the conversion and hashes the master key
1203 - If the header is intact and you just forgot the passphrase, just
1204 set a new passphrase like this:
1206 cryptsetup luksAddKey --master-key-file=<master-key-file> <luks device>
1208 You may want to disable the old one afterwards.
1211 * 6.11 What does the on-disk structure of dm-crypt look like?
1213 There is none. dm-crypt takes a block device and gives encrypted
1214 access to each of its blocks with a key derived from the passphrase
1215 given. If you use a cipher different than the default, you have to
1216 specify that as a parameter to cryptsetup too. If you want to
1217 change the password, you basically have to create a second
1218 encrypted device with the new passphrase and copy your data over.
1219 On the plus side, if you accidentally overwrite any part of a
1220 dm-crypt device, the damage will be limited to the are you
1224 * 6.12 What does the on-disk structure of LUKS look like?
1226 A LUKS partition consists of a header, followed by 8 key-slot
1227 descriptors, followed by 8 key slots, followed by the encrypted
1230 Header and key-slot descriptors fill the first 592 bytes. The
1231 key-slot size depends on the creation parameters, namely on the
1232 number of anti-forensic stripes, key material offset and master
1235 With the default parameters, each key-slot is a bit less than
1236 128kiB in size. Due to sector alignment of the key-slot start,
1237 that means the key block 0 is at offset 0x1000-0x20400, key
1238 block 1 at offset 0x21000-0x40400, and key block 7 at offset
1239 0xc1000-0xe0400. The space to the next full sector address is
1240 padded with zeros. Never used key-slots are filled with what the
1241 disk originally contained there, a key-slot removed with
1242 "luksRemoveKey" or "luksKillSlot" gets filled with 0xff. Due to
1243 2MiB default alignment, start of the data area for cryptsetup 1.3
1244 and later is at 2MiB, i.e. at 0x200000. For older versions, it is
1245 at 0x101000, i.e. at 1'052'672 bytes, i.e. at 1MiB + 4096 bytes
1246 from the start of the partition. Incidentally, "luksHeaderBackup"
1247 for a LUKS container created with default parameters dumps exactly
1248 the first 2MiB (or 1'052'672 bytes for headers created with
1249 cryptsetup versions < 1.3) to file and "luksHeaderRestore" restores
1252 For non-default parameters, you have to figure out placement
1253 yourself. "luksDump" helps. See also next item. For the most common
1254 non-default settings, namely aes-xts-plain with 512 bit key, the
1255 offsets are: 1st keyslot 0x1000-0x3f800, 2nd keyslot
1256 0x40000-0x7e000, 3rd keyslot 0x7e000-0xbd800, ..., and start of
1257 bulk data at 0x200000.
1259 The exact specification of the format is here:
1260 http://code.google.com/p/cryptsetup/wiki/Specification
1263 * 6.13 What is the smallest possible LUKS container?
1265 Note: From cryptsetup 1.3 onwards, alignment is set to 1MB. With
1266 modern Linux partitioning tools that also align to 1MB, this will
1267 result in aligmnet to 2k secors and typical Flash/SSD sectors,
1268 which is highly desirable for a number of reasons. Changing the
1269 alignment is not recomended.
1271 That said, with default parameters, the data area starts at
1272 exactly 2MB offset (at 0x101000 for cryptsetup versions before
1273 1.3). The smallest data area you can have is one sector of 512
1274 bytes. Data areas of 0 bytes can be created, but fail on mapping.
1276 While you cannot put a filesystem into something this small, it may
1277 still be used to contain, for example, key. Note that with current
1278 formatting tools, a partition for a container this size will be
1279 3MiB anyways. If you put the LUKS container into a file (via
1280 losetup and a loopback device), the file needs to be 2097664 bytes
1281 in size, i.e. 2MiB + 512B.
1283 There two ways to influence the start of the data area are key-size
1286 For alignment, you can go down to 1 on the parameter. This will
1287 still leave you with a data-area starting at 0x101000, i.e.
1288 1MiB+4096B (default parameters) as alignment will be rounded up to
1289 the next multiple of 8 (i.e. 4096 bytes) If in doubt, do a dry-run
1290 on a larger file and dump the LUKS header to get actual
1293 For key-size, you can use 128 bit (e.g. AES-128 with CBC), 256 bit
1294 (e.g. AES-256 with CBC) or 512 bit (e.g. AES-256 with XTS mode).
1295 You can do 64 bit (e.g. blowfish-64 with CBC), but anything below
1296 128 bit has to be considered insecure today.
1298 Example 1 - AES 128 bit with CBC:
1300 cryptsetup luksFormat -s 128 --align-payload=8 <device>
1302 This results in a data offset of 0x81000, i.e. 516KiB or 528384
1303 bytes. Add one 512 byte sector and the smallest LUKS container size
1304 with these parameters is 516KiB + 512B or 528896 bytes.
1306 Example 2 - Blowfish 64 bit with CBC (WARNING: insecure):
1308 cryptsetup luksFormat -c blowfish -s 64 --align-payload=8 /dev/loop0
1310 This results in a data offset of 0x41000, i.e. 260kiB or 266240
1311 bytes, with a minimal LUKS conatiner size of 260kiB + 512B or
1315 * 6.14 I think this is overly complicated. Is there an alternative?
1317 Not really. Encryption comes at a price. You can use plain
1318 dm-crypt to simplify things a bit. It does not allow multiple
1319 passphrases, but on the plus side, it has zero on disk description
1320 and if you overwrite some part of a plain dm-crypt partition,
1321 exactly the overwritten parts are lost (rounded up to sector
1325 7. Interoperability with other Disk Encryption Tools
1328 * 7.1 What is this section about?
1330 Cryptsetup for plain dm-crypt can be used to access a number of
1331 on-disk formats created by tools like loop-aes patched into
1332 losetup. This somtimes works and sometimes does not. This section
1333 collects insights into what works, what does not and where more
1334 information is required.
1336 Additional information may be found in the mailing-list archives,
1337 mentioned at the start of this FAQ document. If you have a
1338 solution working that is not yet documented here and think a wider
1339 audience may be intertested, please email the FAQ maintainer.
1342 * 7.2 loop-aes: General observations.
1344 One problem is that there are different versions of losetup around.
1345 loop-aes is a patch for losetup. Possible problems and deviations
1346 from cryptsetup option syntax include:
1348 - Offsets specifed in bytes (cryptsetup: 512 byte sectors)
1350 - The need to specify an IV offset
1352 - Encryption mode needs specifying (e.g. "-c twofish-cbc-plain")
1354 - Key size needs specifying (e.g. "-s 128" for 128 bit keys)
1356 - Passphrase hash algorithm needs specifying
1358 Also note that because plain dm-crypt and loop-aes format does not
1359 have metadata, autodetection, while feasible in most cases, would
1360 be a lot of work that nobody really wants to do. If you still have
1361 the old set-up, using a verbosity option (-v) on mapping with the
1362 old tool or having a look into the system logs after setup could
1363 give you the information you need.
1366 * 7.3 loop-aes patched into losetup on debian 5.x, kernel 2.6.32
1368 In this case, the main problem seems to be that this variant of
1369 losetup takes the offset (-o option) in bytes, while cryptsetup
1370 takes it in sectors of 512 bytes each. Example: The losetupp
1373 losetup -e twofish -o 2560 /dev/loop0 /dev/sdb1
1374 mount /dev/loop0 mountpoint
1378 cryptsetup create -c twofish -o 5 --skip 5 e1 /dev/sdb1
1379 mount /dev/mapper/e1 mountpoint
1382 * 7.4 loop-aes with 160 bit key
1384 This seems to be sometimes used with twofish and blowfish and
1385 represents a 160 bit ripemed160 hash output padded to 196 bit key
1386 length. It seems the corresponding options for cryptsetup are
1388 --cipher twofish-cbc-null -s 192 -h ripemd160:20
1391 8. Issues with Specific Versions of cryptsetup
1394 * 8.1 When using the create command for plain dm-crypt with
1395 cryptsetup 1.1.x, the mapping is incompatible and my data is not
1398 With cryptsetup 1.1.x, the distro maintainer can define different
1399 default encryption modes for LUKS and plain devices. You can check
1400 these compiled-in defaults using "cryptsetup --help". Moreover, the
1401 plain device default changed because the old IV mode was
1402 vulnerable to a watermarking attack.
1404 If you are using a plain device and you need a compatible mode, just
1405 specify cipher, key size and hash algorithm explicitly. For
1406 compatibility with cryptsetup 1.0.x defaults, simple use the
1409 cryptsetup create -c aes-cbc-plain -s 256 -h ripemd160 <name> <dev>
1411 LUKS stores cipher and mode in the metadata on disk, avoiding this
1415 * 8.2 cryptsetup on SLED 10 has problems...
1417 SLED 10 is missing an essential kernel patch for dm-crypt, which
1418 is broken in its kernel as a result. There may be a very old
1419 version of cryptsetup (1.0.x) provided by SLED, which should also
1420 not be used anymore as well. My advice would be to drop SLED 10.
1422 A. Contributors In no particular order: