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 CLONING/IMAGING: If you clone or image a LUKS container, you make a
45 copy of the LUKS header and the master key will stay the same!
46 That means that if you distribute an image to several machines, the
47 same master key will be used on all of them, regardless of whether
48 you change the passphrases. Do NOT do this! If you do, a root-user
49 on any of the machines can decrypt all other copies, breaking
50 security. See also Item 6.15.
52 DISTRIBUTION INSTALLERS: Some distribution installers offer to
53 create LUKS containers in a way that can be mistaken as activation
54 of an existing container. Creating a new LUKS container on top of
55 an existing one leads to permanent, complete and irreversible data
56 loss. It is strongly recommended to only use distribution
57 installers after a complete backup of all LUKS containers has been
60 NO WARNING ON NON-INERACTIVE FORMAT: If you feed cryptsetup from
61 STDIN (e.g. via GnuPG) on LUKS format, it does not give you the
62 warning that you are about to format (and e.g. will lose any
63 pre-existing LUKS container on the target), as it assumes it is
64 used from a script. In this scenario, the responsibility for
65 warning the user and possibly checking for an existing LUKS header
66 is shifted to the script. This is a more general form of the
69 LUKS PASSPHRASE IS NOT THE MASTER KEY: The LUKS passphrase is not
70 used in deriving the master key. It is used in decrypting a master
71 key that is randomly selected on header creation. This means that
72 if you create a new LUKS header on top of an old one with
73 exactly the same parameters and exactly the same passphrase as the
74 old one, it will still have a different master key and your data
75 will be permanently lost.
77 PASSPHRASE CHARACTER SET: Some people have had difficulties with
78 this when upgrading distributions. It is highly advisable to only
79 use the 94 printable characters from the first 128 characters of
80 the ASCII table, as they will always have the same binary
81 representation. Other characters may have different encoding
82 depending on system configuration and your passphrase will not
83 work with a different encoding. A table of the standardized first
84 128 ASCII caracters can, e.g. be found on
85 http://en.wikipedia.org/wiki/ASCII
88 * 1.3 System Specific warnings
90 - Ubuntu as of 4/2011: It seems the installer offers to create
91 LUKS partitions in a way that several people mistook for an offer
92 to activate their existing LUKS partition. The installer gives no
93 or an inadequate warning and will destroy your old LUKS header,
94 causing permanent data loss. See also the section on Backup and
97 This issue has been acknowledged by the Ubuntu dev team, see here:
98 http://launchpad.net/bugs/420080
101 * 1.4 Who wrote this?
103 Current FAQ maintainer is Arno Wagner <arno@wagner.name>. Other
104 contributors are listed at the end. If you want to contribute, send
105 your article, including a descriptive headline, to the maintainer,
106 or the dm-crypt mailing list with something like "FAQ ..." in the
107 subject. You can also send more raw information and have me write
108 the section. Please note that by contributing to this FAQ, you
109 accept the license described below.
111 This work is under the "Attribution-Share Alike 3.0 Unported"
112 license, which means distribution is unlimited, you may create
113 derived works, but attributions to original authors and this
114 license statement must be retained and the derived work must be
115 under the same license. See
116 http://creativecommons.org/licenses/by-sa/3.0/ for more details of
119 Side note: I did text license research some time ago and I think
120 this license is best suited for the purpose at hand and creates the
124 * 1.5 Where is the project website?
126 There is the project website at http://code.google.com/p/cryptsetup/
127 Please do not post questions there, nobody will read them. Use
128 the mailing-list instead.
131 * 1.6 Is there a mailing-list?
133 Instructions on how to subscribe to the mailing-list are at on the
134 project website. People are generally helpful and friendly on the
137 The question of how to unsubscribe from the list does crop up
138 sometimes. For this you need your list management URL, which is
139 sent to you initially and once at the start of each month. Go to
140 the URL mentioned in the email and select "unsubscribe". This page
141 also allows you to request a password reminder.
143 Alternatively, you can send an Email to dm-crypt-request@saout.de
144 with just the word "help" in the subject or message body. Make sure
145 to send it from your list address.
147 The mailing list archive is here:
148 http://dir.gmane.org/gmane.linux.kernel.device-mapper.dm-crypt
154 * 2.1 What is the difference between "plain" and LUKS format?
156 Plain format is just that: It has no metadata on disk, reads all
157 paramters from the commandline (or the defaults), derives a
158 master-key from the passphrase and then uses that to de-/encrypt
159 the sectors of the device, with a direct 1:1 mapping between
160 encrypted and decrypted sectors.
162 Primary advantage is high resilience to damage, as one damaged
163 encrypted sector results in exactly one damaged decrypted sector.
164 Also, it is not readily apparent that there even is encrypted data
165 on the device, as an overwrite with crypto-grade randomness (e.g.
166 from /dev/urandom) looks exactly the same on disk.
168 Side-note: That has limited value against the authorities. In
169 civilized countries, they cannot force you to give up a crypto-key
170 anyways. In the US, the UK and dictatorships around the world,
171 they can force you to give up the keys (using imprisonment or worse
172 to pressure you), and in the worst case, they only need a
173 nebulous "suspicion" about the presence of encrypted data. My
174 advice is to either be ready to give up the keys or to not have
175 encrypted data when traveling to those countries, especially when
176 crossing the borders.
178 Disadvantages are that you do not have all the nice features that
179 the LUKS metadata offers, like multiple passphrases that can be
180 changed, the cipher being stored in the metadata, anti-forensic
181 properties like key-slot diffusion and salts, etc..
183 LUKS format uses a metadata header and 8 key-slot areas that are
184 being placed ath the begining of the disk, see below under "What
185 does the LUKS on-disk format looks like?". The passphrases are used
186 to decryt a single master key that is stored in the anti-forensic
189 Advantages are a higher usability, automatic configuration of
190 non-default crypto parameters, defenses against low-entropy
191 passphrases like salting and iterated PBKDF2 passphrase hashing,
192 the ability to change passhrases, and others.
194 Disadvantages are that it is readily obvious there is encrypted
195 data on disk (but see side note above) and that damage to the
196 header or key-slots usually results in permanent data-loss. See
197 below under "6. Backup and Data Recovery" on how to reduce that
198 risk. Also the sector numbers get shifted by the length of the
199 header and key-slots and there is a loss of that size in capacity
200 (1MB+4096B for defaults and 2MB for the most commonly used
201 non-default XTS mode).
204 * 2.2 Can I encrypt an already existing, non-empty partition to use
207 There is no converter, and it is not really needed. The way to do
208 this is to make a backup of the device in question, securely wipe
209 the device (as LUKS device initialization does not clear away old
210 data), do a luksFormat, optionally overwrite the encrypted device,
211 create a new filesystem and restore your backup on the now
212 encrypted device. Also refer to sections "Security Aspects" and
213 "Backup and Data Recovery".
215 For backup, plain GNU tar works well and backs up anything likely
216 to be in a filesystem.
219 * 2.3 How do I use LUKS with a loop-device?
221 This can be very handy for experiments. Setup is just the same as
222 with any block device. If you want, for example, to use a 100MiB
223 file as LUKS container, do something like this:
225 head -c 100M /dev/zero > luksfile # create empty file
226 losetup /dev/loop0 luksfile # map luksfile to /dev/loop0
227 cryptsetup luksFormat /dev/loop0 # create LUKS on loop device
229 Afterwards just use /dev/loop0 as a you would use a LUKS partition.
230 To unmap the file when done, use "losetup -d /dev/loop0".
233 * 2.4 When I add a new key-slot to LUKS, it asks for a passphrase but
234 then complains about there not being a key-slot with that
237 That is as intended. You are asked a passphrase of an existing
238 key-slot first, before you can enter the passphrase for the new
239 key-slot. Otherwise you could break the encryption by just adding a
240 new key-slot. This way, you have to know the passphrase of one of
241 the already configured key-slots in order to be able to configure a
245 * 2.5 Encrytion on top of RAID or the other way round?
247 Unless you have special needs, place encryption between RAID and
248 filesystem, i.e. encryption on top of RAID. You can do it the other
249 way round, but you have to be aware that you then need to give the
250 pasphrase for each individual disk and RAID autotetection will not
251 work anymore. Therefore it is better to encrypt the RAID device,
255 * 2.6 How do I read a dm-crypt key from file?
257 Note that the file will still be hashed first, just like keyboard
258 input. Use the --key-file option, like this:
260 cryptsetup create --key-file keyfile e1 /dev/loop0
263 * 2.7 How do I read a LUKS slot key from file?
265 What you really do here is to read a passphrase from file, just as
266 you would with manual entry of a passphrase for a key-slot. You can
267 add a new passphrase to a free key-slot, set the passphrase of an
268 specific key-slot or put an already configured passphrase into a
269 file. In the last case make sure no trailing newline (0x0a) is
270 contained in the key file, or the passphrase will not work because
271 the whole file is used as input.
273 To add a new passphrase to a free key slot from file, use something
276 cryptsetup luksAddKey /dev/loop0 keyfile
278 To add a new passphrase to a specific key-slot, use something like
281 cryptsetup luksAddKey --key-slot 7 /dev/loop0 keyfile
283 To supply a key from file to any LUKS command, use the --key-file
284 option, e.g. like this:
286 cryptsetup luksOpen --key-file keyfile /dev/loop0 e1
289 * 2.8 How do I read the LUKS master key from file?
291 The question you should ask yourself first is why you would want to
292 do this. The only legitimate reason I can think of is if you want
293 to have two LUKS devices with the same master key. Even then, I
294 think it would be preferable to just use key-slots with the same
295 passphrase, or to use plain dm-crypt instead. If you really have a
296 good reason, please tell me. If I am convinced, I will add how to
300 * 2.9 What are the security requirements for a key read from file?
302 A file-stored key or passphrase has the same security requirements
303 as one entered interactively, however you can use random bytes and
304 thereby use bytes you cannot type on the keyboard. You can use any
305 file you like as key file, for example a plain text file with a
306 human readable passphrase. To generate a file with random bytes,
307 use something like this:
309 head -c 256 /dev/random > keyfile
312 * 2.10 If I map a journaled file system using dm-crypt/LUKS, does it
313 still provide its usual transactional guarantees?
315 As far as I know it does (but I may be wrong), but please note that
316 these "guarantees" are far weaker than they appear to be. For
317 example, you may not get a hard flush to disk surface even on a
318 call to fsync. In addition, the HDD itself may do independent
319 write reordering. Some other things can go wrong as well. The
320 filesystem developers are aware of these problems and typically
321 can make it work anyways. That said, dm-crypt/LUKS should not make
324 Personally, I have several instances of ext3 on dm-crypt and have
325 not noticed any specific problems.
327 Update: I did run into frequent small freezes (1-2 sec) when putting
328 a vmware image on ext3 over dm-crypt. This does indicate that the
329 transactional guarantees are in place, but at a cost. When I went
330 back to ext2, the problem went away. This also seems to have gotten
331 better with kernel 2.6.36 and the reworking of filesystem flush
332 locking. Kernel 2.6.38 is expected to have more improvements here.
335 * 2.11 Can I use LUKS or cryptsetup with a more secure (external)
336 medium for key storage, e.g. TPM or a smartcard?
338 Yes, see the answers on using a file-supplied key. You do have to
339 write the glue-logic yourself though. Basically you can have
340 cryptsetup read the key from STDIN and write it there with your
341 own tool that in turn gets the key from the more secure key
345 * 2.12 Can I resize a dm-crypt or LUKS partition?
347 Yes, you can, as neither dm-crypt nor LUKS stores partition size.
348 Whether you should is a different question. Personally I recommend
349 backup, recreation of the encrypted partition with new size,
350 recreation of the filesystem and restore. This gets around the
351 tricky business of resizing the filesystem. Resizing a dm-crypt or
352 LUKS container does not resize the filesystem in it. The backup is
353 really non-optional here, as a lot can go wrong, resulting in
354 partial or complete data loss. Using something like gparted to
355 resize an encrypted partition is slow, but typicaly works. This
356 will not change the size of the filesystem hidden under the
359 You also need to be aware of size-based limitations. The one
360 currently relevant is that aes-xts-plain should not be used for
361 encrypted container sizes larger than 2TiB. Use aes-xts-plain64
368 * 3.1 My dm-crypt/LUKS mapping does not work! What general steps are
369 there to investigate the problem?
371 If you get a specific error message, investigate what it claims
372 first. If not, you may want to check the following things.
374 - Check that "/dev", including "/dev/mapper/control" is there. If it
375 is missing, you may have a problem with the "/dev" tree itself or
376 you may have broken udev rules.
378 - Check that you have the device mapper and the crypt target in your
379 kernel. The output of "dmsetup targets" should list a "crypt"
380 target. If it is not there or the command fails, add device mapper
381 and crypt-target to the kernel.
383 - Check that the hash-functions and ciphers you want to use are in
384 the kernel. The output of "cat /proc/crypto" needs to list them.
387 * 3.2 My dm-crypt mapping suddenly stopped when upgrading cryptsetup.
389 The default cipher, hash or mode may have changed (the mode changed
390 from 1.0.x to 1.1.x). See under "Issues With Specific Versions of
394 * 3.3 When I call cryptsetup from cron/CGI, I get errors about
397 If you get errors about unknown parameters or the like that are not
398 present when cryptsetup is called from the shell, make sure you
399 have no older version of cryptsetup on your system that then gets
400 called by cron/CGI. For example some distributions install
401 cryptsetup into /usr/sbin, while a manual install could go to
402 /usr/local/sbin. As a debugging aid, call "cryptsetup --version"
403 from cron/CGI or the non-shell mechanism to be sure the right
407 * 3.4 Unlocking a LUKS device takes very long. Why?
409 The iteration time for a key-slot (see Section 5 for an explanation
410 what iteration does) is calculated when setting a passphrase. By
411 default it is 1 second on the machine where the passphrase is set.
412 If you set a passphrase on a fast machine and then unlock it on a
413 slow machine, the unlocking time can be much longer. Also take into
414 account that up to 8 key-slots have to be tried in order to find the
417 If this is problem, you can add another key-slot using the slow
418 machine with the same passphrase and then remove the old key-slot.
419 The new key-slot will have an iteration count adjusted to 1 second
420 on the slow machine. Use luksKeyAdd and then luksKillSlot or
423 However, this operation will not change volume key iteration count
424 (MK iterations in output of "cryptsetup luksDump"). In order to
425 change that, you will have to backup the data in the LUKS
426 container (i.e. your encrypted data), luksFormat on the slow
427 machine and restore the data. Note that in the original LUKS
428 specification this value was fixed to 10, but it is now derived
429 from the PBKDF2 benchmark as well and set to iterations in 0.125
430 sec or 1000, whichever is larger. Also note that MK iterations
431 are not very security relevant. But as each key-slot already takes
432 1 second, spending the additional 0.125 seconds really does not
436 * 3.5 "blkid" sees a LUKS UUID and an ext2/swap UUID on the same
437 device. What is wrong?
439 Some old versions of cryptsetup have a bug where the header does
440 not get completely wiped during LUKS format and an older ext2/swap
441 signature remains on the device. This confuses blkid.
443 Fix: Wipe the unused header areas by doing a backup and restore of
444 the header with cryptsetup 1.1.x:
446 cryptsetup luksHeaderBackup --header-backup-file <file> <device>
447 cryptsetup luksHeaderRestore --header-backup-file <file> <device>
450 * 3.6 cryptsetup segfaults on Gentoo amd64 hardened ...
452 There seems to be some inteference between the hardening and and
453 the way cryptsetup benchmarks PBKDF2. The solution to this is
454 currently not quite clear for an encrypted root filesystem. For
455 other uses, you can apparently specify USE="dynamic" as compile
456 flag, see http://bugs.gentoo.org/show_bug.cgi?id=283470
462 * 4.1 I get the error "LUKS keyslot x is invalid." What does that
465 This means that the given keyslot has an offset that points
466 outside the valid keyslot area. Typically, the reason is a
467 corrupted LUKS header because something was written to the start of
468 the device the LUKS contaner is on. Refer to Section "Backup and
469 Data Recovery" and ask on the mailing list if you have trouble
470 diagnosing and (if still possible) repairing this.
473 * 4.2 Can a bad RAM module cause problems?
475 LUKS and dm-crypt can give the RAM quite a workout, especially when
476 combined with software RAID. In particular the combination RAID5 +
477 LUKS + XFS seems to uncover RAM problems that never caused obvious
478 problems before. Symptoms vary, but often the problem manifest
479 itself when copying large amounts of data, typically several times
480 larger than your main memory.
482 Side note: One thing you should always do on large data
483 copy/movements is to run a verify, for example with the "-d"
484 option of "tar" or by doing a set of MD5 checksums on the source
487 find . -type f -exec md5sum \{\} \; > checksum-file
489 and then a "md5sum -c checksum-file" on the other side. If you get
490 mismatches here, RAM is the primary suspect. A lesser suspect is
491 an overclocked CPU. I have found countless hardware problems in
492 verify runs after copying or making backups. Bit errors are much
493 more common than most people think.
495 Some RAM issues are even worse and corrupt structures in one of the
496 layers. This typically results in lockups, CPU state dumps in the
497 system logs, kernel panic or other things. It is quite possible to
498 have the problem with an encrypted device, but not with an
499 otherwise the same unencrypted device. The reason for that is that
500 encryption has an error amplification property: You flip one bit
501 in an encrypted data block, and the decrypted version has half of
502 its bits flipped. This is an important security property for modern
503 ciphers. With the usual modes in cryptsetup (CBC, ESSIV, XTS), you
504 get up to a completely changed 512 byte block per bit error. A
505 corrupt block causes a lot more havoc than the occasionally
506 flipped single bit and can result in various obscure errors.
508 Note, that a verify run on copying between encrypted or
509 unencrypted devices will reliably detect corruption, even when the
510 copying itself did not report any problems. If you find defect
511 RAM, assume all backups and copied data to be suspect, unless you
515 * 4.3 How do I test RAM?
517 First you should know that overclocking often makes memory
518 problems worse. So if you overclock (which I strongly recommend
519 against in a system holding data that has some worth), run the
520 tests with the overclocking active.
522 There are two good options. One is Memtest86+ and the other is
523 "memtester" by Charles Cazabon. Memtest86+ requires a reboot and
524 then takes over the machine, while memtester runs from a
525 root-shell. Both use different testing methods and I have found
526 problems fast with each one that the other needed long to find. I
527 recommend running the following procedure until the first error is
530 - Run Memtest86+ for one cycle
532 - Run memterster for one cycle (shut down as many other applications
535 - Run Memtest86+ for 24h or more
537 - Run memtester for 24h or more
539 If all that does not produce error messages, your RAM may be sound,
540 but I have had one weak bit that Memtest86+ needed around 60 hours
541 to find. If you can reproduce the original problem reliably, a good
542 additional test may be to remove half of the RAM (if you have more
543 than one module) and try whether the problem is still there and if
544 so, try with the other half. If you just have one module, get a
545 different one and try with that. If you do overclocking, reduce
546 the settings to the most conservative ones available and try with
553 * 5.1 Is LUKS insecure? Everybody can see I have encrypted data!
555 In practice it does not really matter. In most civilized countries
556 you can just refuse to hand over the keys, no harm done. In some
557 countries they can force you to hand over the keys, if they suspect
558 encryption. However the suspicion is enough, they do not have to
559 prove anything. This is for practical reasons, as even the presence
560 of a header (like the LUKS header) is not enough to prove that you
561 have any keys. It might have been an experiment, for example. Or it
562 was used as encrypted swap with a key from /dev/random. So they
563 make you prove you do not have encrypted data. Of course that is
564 just as impossible as the other way round.
566 This means that if you have a large set of random-looking data,
567 they can already lock you up. Hidden containers (encryption hidden
568 within encryption), as possible with Truecrypt, do not help
569 either. They will just assume the hidden container is there and
570 unless you hand over the key, you will stay locked up. Don't have
571 a hidden container? Though luck. Anybody could claim that.
573 Still, if you are concerned about the LUKS header, use plain
574 dm-crypt with a good passphrase. See also Section 2, "What is the
575 difference between "plain" and LUKS format?"
578 * 5.2 Should I initialize (overwrite) a new LUKS/dm-crypt partition?
580 If you just create a filesystem on it, most of the old data will
581 still be there. If the old data is sensitive, you should overwrite
582 it before encrypting. In any case, not initializing will leave the
583 old data there until the specific sector gets written. That may
584 enable an attacker to determine how much and where on the
585 partition data was written. If you think this is a risk, you can
586 prevent this by overwriting the encrypted device (here assumed to
587 be named "e1") with zeros like this:
589 dd_rescue -w /dev/zero /dev/mapper/e1
591 or alternatively with one of the following more standard commands:
593 cat /dev/zero > /dev/mapper/e1
594 dd if=/dev/zero of=/dev/mapper/e1
597 * 5.3 How do I securely erase a LUKS (or other) partition?
599 For LUKS, if you are in a desperate hurry, overwrite the LUKS
600 header and key-slot area. This means overwriting the first
601 (keyslots x stripes x keysize) + offset bytes. For the default
602 parameters, this is the 1'052'672 bytes, i.e. 1MiB + 4096 of the
603 LUKS partition. For 512 bit key length (e.g. for aes-xts-plain with
604 512 bit key) this is 2MiB. (The diferent offset stems from
605 differences in the sector alignment of the key-slots.) If in doubt,
606 just be generous and overwrite the first 10MB or so, it will likely
607 still be fast enough. A single overwrite with zeros should be
608 enough. If you anticipate being in a desperate hurry, prepare the
609 command beforehand. Example with /dev/sde1 as the LUKS partition
610 and default parameters:
612 head -c 1052672 /dev/zero > /dev/sde1; sync
614 A LUKS header backup or full backup will still grant access to
615 most or all data, so make sure that an attacker does not have
616 access to backups or destroy them as well.
618 If you have time, overwrite the whole LUKS partition with a single
619 pass of zeros. This is enough for current HDDs. For SSDs or FLASH
620 (USB sticks) you may want to overwrite the whole drive several
621 times to be sure data is not retained by wear leveling. This is
622 possibly still insecure as SSD technology is not fully understood
623 in this regard. Still, due to the anti-forensic properties of the
624 LUKS key-slots, a single overwrite of an SSD or FLASH drive could
625 be enough. If in doubt, use physical destruction in addition. Here
626 is a link to some current reseach results on erasing SSDs and FLASH
628 http://www.usenix.org/events/fast11/tech/full_papers/Wei.pdf
630 Keep in mind to also erase all backups.
632 Example for a zero-overwrite erase of partition sde1 done with
635 dd_rescue -w /dev/zero /dev/sde1
638 * 5.4 How do I securely erase a backup of a LUKS partition or header?
640 That depends on the medium it is stored on. For HDD and SSD, use
641 overwrite with zeros. For an SSD or FLASH drive (USB stick), you
642 may want to overwrite the complete SSD several times and use
643 physical destruction in addition, see last item. For re-writable
644 CD/DVD, a single overwrite should also be enough, due to the
645 anti-forensic properties of the LUKS keyslots. For write-once
646 media, use physical destruction. For low security requirements,
647 just cut the CD/DVD into several parts. For high security needs,
648 shred or burn the medium. If your backup is on magnetic tape, I
649 advise physical destruction by shredding or burning, after
650 overwriting . The problem with magnetic tape is that it has a
651 higher dynamic range than HDDs and older data may well be
652 recoverable after overwrites. Also write-head alignment issues can
653 lead to data not actually being deleted at all during overwrites.
656 * 5.5 What about backup? Does it compromise security?
658 That depends. See item 6.7.
661 * 5.6 Why is all my data permanently gone if I overwrite the LUKS
664 Overwriting the LUKS header in part or in full is the most common
665 reason why access to LUKS containers is lost permanently.
666 Overwriting can be done in a number of fashions, like creating a
667 new filesystem on the raw LUKS partition, making the raw partition
668 part of a raid array and just writing to the raw partition.
670 The LUKS header contains a 256 bit "salt" value and without that no
671 decryption is possible. While the salt is not secret, it is
672 key-grade material and cannot be reconstructed. This is a
673 cryptographically strong "cannot". From observations on the
674 cryptsetup mailing-list, people typically go though the usual
675 stages of grief (Denial, Anger, Bargaining, Depression, Acceptance)
676 when this happens to them. Observed times vary between 1 day and 2
677 weeks to complete the cycle. Seeking help on the mailing-list is
678 fine. Even if we usually cannot help with getting back your data,
679 most people found the feedback comforting.
681 If your header does not contain an intact salt, best go directly
682 to the last stage ("Acceptance") and think about what to do now.
683 There is one exception that I know of: If your LUKS container is
684 still open, then it may be possible to extract the master key from
685 the running system. See Item "How do I recover the master key from
686 a mapped LUKS container?" in Section "Backup and Data Recovery".
689 * 5.7 What is a "salt"?
691 A salt is a random key-grade value added to the passphrase before
692 it is processed. It is not kept secret. The reason for using salts
693 is as follows: If an attacker wants to crack the password for a
694 single LUKS container, then every possible passphrase has to be
695 tried. Typically an attacker will not try every binary value, but
696 will try words and sentences from a dictionary.
698 If an attacker wants to attack several LUKS containers with the
699 same dictionary, then a different approach makes sense: Compute the
700 resulting slot-key for each dictionary element and store it on
701 disk. Then the test for each entry is just the slow unlocking with
702 the slot key (say 0.00001 sec) instead of calculating the slot-key
703 first (1 sec). For a single attack, this does not help. But if you
704 have more than one container to attack, this helps tremendously,
705 also because you can prepare your table before you even have the
706 container to attack! The calculation is also very simple to
707 parallelize. You could, for example, use the night-time unused CPU
708 power of your desktop PCs for this.
710 This is where the salt comes in. If the salt is combined with the
711 passphrase (in the simplest form, just appended to it), you
712 suddenly need a separate table for each salt value. With a
713 reasonably-sized salt value (256 bit, e.g.) this is quite
717 * 5.8 Is LUKS secure with a low-entropy (bad) passphrase?
719 Note: You should only use the 94 printable characters from 7 bit
720 ASCII code to prevent your passphrase from failing when the
721 character encoding changes, e.g. because of a system upgrade, see
722 also the note at the very start of this FAQ under "WARNINGS".
724 This needs a bit of theory. The quality of your passphrase is
725 directly related to its entropy (information theoretic, not
726 thermodynamic). The entropy says how many bits of "uncertainty" or
727 "randomness" are in you passphrase. In other words, that is how
728 difficult guessing the passphrase is.
730 Example: A random English sentence has about 1 bit of entropy per
731 character. A random lowercase (or uppercase) character has about
734 Now, if n is the number of bits of entropy in your passphrase and t
735 is the time it takes to process a passphrase in order to open the
736 LUKS container, then an attacker has to spend at maximum
738 attack_time_max = 2^n * t
740 time for a successful attack and on average half that. There is no
741 way getting around that relationship. However, there is one thing
742 that does help, namely increasing t, the time it takes to use a
743 passphrase, see next FAQ item.
745 Still, if you want good security, a high-entropy passphrase is the
746 only option. For example, a low-entropy passphrase can never be
747 considered secure against a TLA-level (Three Letter Agency level,
748 i.e. government-level) attacker, no matter what tricks are used in
749 the key-derivation function. Use at least 64 bits for secret stuff.
750 That is 64 characters of English text (but only if randomly chosen)
751 or a combination of 12 truly random letters and digits.
753 For passphrase generation, do not use lines from very well-known
754 texts (religious texts, Harry potter, etc.) as they are to easy to
755 guess. For example, the total Harry Potter has about 1'500'000
756 words (my estimation). Trying every 64 character sequence starting
757 and ending at a word boundary would take only something like 20
758 days on a single CPU and is entirely feasible. To put that into
759 perspective, using a number of Amazon EC2 High-CPU Extra Large
760 instances (each gives about 8 real cores), this test costs
761 currently about 50USD/EUR, but can be made to run arbitrarily fast.
763 On the other hand, choosing 1.5 lines from, say, the Wheel of Time
764 is in itself not more secure, but the book selection adds quite a
765 bit of entropy. (Now that I have mentioned it here, don't use tWoT
766 either!) If you add 2 or 3 typos or switch some words around, then
767 this is good passphrase material.
770 * 5.9 What is "iteration count" and why is decreasing it a bad idea?
772 Iteration count is the number of PBKDF2 iterations a passphrase is
773 put through before it is used to unlock a key-slot. Iterations are
774 done with the explicit purpose to increase the time that it takes
775 to unlock a key-slot. This provides some protection against use of
776 low-entropy passphrases.
778 The idea is that an attacker has to try all possible passphrases.
779 Even if the attacker knows the passphrase is low-entropy (see last
780 item), it is possible to make each individual try take longer. The
781 way to do this is to repeatedly hash the passphrase for a certain
782 time. The attacker then has to spend the same time (given the same
783 computing power) as the user per try. With LUKS, the default is 1
784 second of PBKDF2 hashing.
786 Example 1: Lets assume we have a really bad passphrase (e.g. a
787 girlfriends name) with 10 bits of entropy. With the same CPU, an
788 attacker would need to spend around 500 seconds on average to
789 break that passphrase. Without iteration, it would be more like
790 0.0001 seconds on a modern CPU.
792 Example 2: The user did a bit better and has 32 chars of English
793 text. That would be about 32 bits of entropy. With 1 second
794 iteration, that means an attacker on the same CPU needs around 136
795 years. That is pretty impressive for such a weak passphrase.
796 Without the iterations, it would be more like 50 days on a modern
797 CPU, and possibly far less.
799 In addition, the attacker can both parallelize and use special
800 hardware like GPUs or FPGAs to speed up the attack. The attack can
801 also happen quite some time after the luksFormat operation and CPUs
802 can have become faster and cheaper. For that reason you want a
803 bit of extra security. Anyways, in Example 1 your are screwed.
804 In example 2, not necessarily. Even if the attack is faster, it
805 still has a certain cost associated with it, say 10000 EUR/USD
806 with iteration and 1 EUR/USD without iteration. The first can be
807 prohibitively expensive, while the second is something you try
808 even without solid proof that the decryption will yield something
811 The numbers above are mostly made up, but show the idea. Of course
812 the best thing is to have a high-entropy passphrase.
814 Would a 100 sec iteration time be even better? Yes and no.
815 Cryptographically it would be a lot better, namely 100 times better.
816 However, usability is a very important factor for security
817 technology and one that gets overlooked surprisingly often. For
818 LUKS, if you have to wait 2 minutes to unlock the LUKS container,
819 most people will not bother and use less secure storage instead. It
820 is better to have less protection against low-entropy passphrases
821 and people actually use LUKS, than having them do without
822 encryption altogether.
824 Now, what about decreasing the iteration time? This is generally a
825 very bad idea, unless you know and can enforce that the users only
826 use high-entropy passphrases. If you decrease the iteration time
827 without ensuring that, then you put your users at increased risk,
828 and considering how rarely LUKS containers are unlocked in a
829 typical work-flow, you do so without a good reason. Don't do it.
830 The iteration time is already low enough that users with entropy
831 low passphrases are vulnerable. Lowering it even further increases
832 this danger significantly.
835 * 5.10 Some people say PBKDF2 is insecure?
837 There is some discussion that a hash-function should have a "large
838 memory" property, i.e. that it should require a lot of memory to be
839 computed. This serves to prevent attacks using special programmable
840 circuits, like FPGAs, and attacks using graphics cards. PBKDF2
841 does not need a lot of memory and is vulnerable to these attacks.
842 However, the publication usually refered in these discussions is
843 not very convincing in proving that the presented hash really is
844 "large memory" (that may change, email the FAQ maintainer when it
845 does) and it is of limited usefulness anyways. Attackers that use
846 clusters of normal PCs will not be affected at all by a "large
847 memory" property. For example the US Secret Service is known to
848 use the off-hour time of all the office PCs of the Treasury for
849 password breaking. The Treasury has about 110'000 employees.
850 Asuming every one has an office PC, that is significant computing
851 power, all of it with plenty of memory for computing "large
852 memory" hashes. Bot-net operators also have all the memory they
853 want. The only protection against a resouceful attacker is a
854 high-entropy passphrase, see items 5.8 and 5.9.
857 * 5.11 What about iteration count with plain dm-crypt?
859 Simple: There is none. There is also no salting. If you use plain
860 dm-crypt, the only way to be secure is to use a high entropy
861 passphrase. If in doubt, use LUKS instead.
864 * 5.12 Is LUKS with default parameters less secure on a slow CPU?
866 Unfortunately, yes. However the only aspect affected is the
867 protection for low-entropy passphrase or master-key. All other
868 security aspects are independent of CPU speed.
870 The master key is less critical, as you really have to work at it
871 to give it low entropy. One possibility is to supply the master key
872 yourself. If that key is low-entropy, then you get what you
873 deserve. The other known possibility is to use /dev/urandom for
874 key generation in an entropy-startved situation (e.g. automatic
875 installation on an embedded device without network and other entropy
878 For the passphrase, don't use a low-entropy passphrase. If your
879 passphrase is good, then a slow CPU will not matter. If you insist
880 on a low-entropy passphrase on a slow CPU, use something like
881 "--iter-time=10" or higher and wait a long time on each LUKS unlock
882 and pray that the attacker does not find out in which way exactly
883 your passphrase is low entropy. This also applies to low-entropy
884 passphrases on fast CPUs. Technology can do only so much to
885 compensate for problems in front of the keyboard.
888 * 5.13 Why was the default aes-cbc-plain replaced with aes-cbc-essiv?
890 The problem is that cbc-plain has a fingerprint vulnerability, where
891 a specially crafted file placed into the crypto-container can be
892 recognized from the outside. The issue here is that for cbc-plain
893 the initialization vector (IV) is the sector number. The IV gets
894 XORed to the first data chunk of the sector to be encrypted. If you
895 make sure that the first data block to be stored in a sector
896 contains the sector number as well, the first data block to be
897 encrypted is all zeros and always encrypted to the same ciphertext.
898 This also works if the first data chunk just has a constant XOR
899 with the sector number. By having several shifted patterns you can
900 take care of the case of a non-power-of-two start sector number of
903 This mechanism allows you to create a pattern of sectors that have
904 the same first ciphertext block and signal one bit per sector to the
905 outside, allowing you to e.g. mark media files that way for
906 recognition without decryption. For large files this is a
907 practical attack. For small ones, you do not have enough blocks to
908 signal and take care of different file starting offsets.
910 In order to prevent this attack, the default was changed to
911 cbc-essiv. ESSIV uses a keyed hash of the sector number, with the
912 encryption key as key. This makes the IV unpredictable without
913 knowing the encryption key and the watermarking attack fails.
916 * 5.14 Are there any problems with "plain" IV? What is "plain64"?
918 First, "plain" and "plain64" are both not secure to use with CBC,
919 see previous FAQ item.
921 However there are modes, like XTS, that are secure with "plain" IV.
922 The next limit is that "plain" is 64 bit, with the upper 32 bit set
923 to zero. This means that on volumes larger than 2TiB, the IV
924 repeats, creating a vulnerability that potentially leaks some
925 data. To avoid this, use "plain64", which uses the full sector
926 number up to 64 bit. Note that "plain64" requires a kernel >=
927 2.6.33. Also note that "plain64" is backwards compatible for
928 volume sizes <= 2TiB, but not for those > 2TiB. Finally, "plain64"
929 does not cause any performance penalty compared to "plain".
932 * 5.15 What about XTS mode?
934 XTS mode is potentially even more secure than cbc-essiv (but only if
935 cbc-essiv is insecure in your scenario). It is a NIST standard and
936 used, e.g. in Truecrypt. At the moment, if you want to use it, you
937 have to specify it manually as "aes-xts-plain", i.e.
939 cryptsetup -c aes-xts-plain luksFormat <device>
941 For volumes >2TiB and kernels >= 2.6.33 use "plain64" (see FAQ
942 item on "plain" and "plain64"):
944 cryptsetup -c aes-xts-plain64 luksFormat <device>
946 There is a potential security issue with XTS mode and large blocks.
947 LUKS and dm-crypt always use 512B blocks and the issue does not
951 6. Backup and Data Recovery
954 * 6.1 Why do I need Backup?
956 First, disks die. The rate for well-treated (!) disk is about 5%
957 per year, which is high enough to worry about. There is some
958 indication that this may be even worse for some SSDs. This applies
959 both to LUKS and plain dm-crypt partitions.
961 Second, for LUKS, if anything damages the LUKS header or the
962 key-stripe area then decrypting the LUKS device can become
963 impossible. This is a frequent occuurence. For example an
964 accidental format as FAT or some software overwriting the first
965 sector where it suspects a partition boot sector typically makes a
966 LUKS partition permanently inacessible. See more below on LUKS
969 So, data-backup in some form is non-optional. For LUKS, you may
970 also want to store a header backup in some secure location. This
971 only needs an update if you change passphrases.
974 * 6.2 How do I backup a LUKS header?
976 While you could just copy the appropriate number of bytes from the
977 start of the LUKS partition, the best way is to use command option
978 "luksHeaderBackup" of cryptsetup. This protects also against
979 errors when non-standard parameters have been used in LUKS
980 partition creation. Example:
983 cryptsetup luksHeaderBackup --header-backup-file h <device>
985 To restore, use the inverse command, i.e.
987 cryptsetup luksHeaderRestore --header-backup-file h <device>
990 * 6.3 How do I test a LUKS header?
994 cryptsetup -v isLuks <device>
996 on the device. Without the "-v" it just signals its result via
997 exit-status. You can alos use the more general test
1001 which will also detect other types and give some more info. Omit
1002 "-p" for old versions of blkid that do not support it.
1005 * 6.4 How do I backup a LUKS or dm-crypt partition?
1007 There are two options, a sector-image and a plain file or
1008 filesystem backup of the contents of the partition. The sector
1009 image is already encrypted, but cannot be compressed and contains
1010 all empty space. The filesystem backup can be compressed, can
1011 contain only part of the encrypted device, but needs to be
1012 encrypted separately if so desired.
1014 A sector-image will contain the whole partition in encrypted form,
1015 for LUKS the LUKS header, the keys-slots and the data area. It can
1016 be done under Linux e.g. with dd_rescue (for a direct image copy)
1017 and with "cat" or "dd". Example:
1019 cat /dev/sda10 > sda10.img
1020 dd_rescue /dev/sda10 sda10.img
1022 You can also use any other backup software that is capable of making
1023 a sector image of a partition. Note that compression is
1024 ineffective for encrypted data, hence it does not make sense to
1027 For a filesystem backup, you decrypt and mount the encrypted
1028 partition and back it up as you would a normal filesystem. In this
1029 case the backup is not encrypted, unless your encryption method
1030 does that. For example you can encrypt a backup with "tar" as
1033 tar cjf - <path> | gpg --cipher-algo AES -c - > backup.tbz2.gpg
1035 And verify the backup like this if you are at "path":
1037 cat backup.tbz2.gpg | gpg - | tar djf -
1039 Note: Allways verify backups, especially encrypted ones.
1041 In both cases GnuPG will ask you interactively for your symmetric
1042 key. The verify will only output errors. Use "tar dvjf -" to get
1043 all comparison results. To make sure no data is written to disk
1044 unencrypted, turn off swap if it is not encrypted before doing the
1047 You can of course use different or no compression and you can use
1048 an asymmetric key if you have one and have a backup of the secret
1049 key that belongs to it.
1051 A second option for a filestem-level backup that can be used when
1052 the backup is also on local disk (e.g. an external USB drive) is
1053 to use a LUKS container there and copy the files to be backed up
1054 between both mounted containers. Also see next item.
1057 * 6.5 Do I need a backup of the full partition? Would the header and
1058 key-slots not be enough?
1060 Backup protects you against two things: Disk loss or corruption
1061 and user error. By far the most questions on the dm-crypt mailing
1062 list about how to recover a damaged LUKS partition are related
1063 to user error. For example, if you create a new filesystem on a
1064 LUKS partition, chances are good that all data is lost
1067 For this case, a header+key-slot backup would often be enough. But
1068 keep in mind that a well-treated (!) HDD has roughly a failure
1069 risk of 5% per year. It is highly advisable to have a complete
1070 backup to protect against this case.
1073 * *6.6 What do I need to backup if I use "decrypt_derived"?
1075 This is a script in Debian, intended for mounting /tmp or swap with
1076 a key derived from the master key of an already decrypted device.
1077 If you use this for an device with data that should be persistent,
1078 you need to make sure you either do not lose access to that master
1079 key or have a backup of the data. If you derive from a LUKS
1080 device, a header backup of that device would cover backing up the
1081 master key. Keep in mind that this does not protect against disk
1084 Note: If you recreate the LUKS header of the device you derive from
1085 (using luksFormat), the master key changes even if you use the same
1086 passphrase(s) and you will not be able to decrypt the derived
1087 device with the new LUKS header.
1090 * 6.7 Does a backup compromise security?
1092 Depends on how you do it. However if you do not have one, you are
1093 going to eventually lose your encrypted data.
1095 There are risks introduced by backups. For example if you
1096 change/disable a key-slot in LUKS, a binary backup of the partition
1097 will still have the old key-slot. To deal with this, you have to
1098 be able to change the key-slot on the backup as well, securely
1099 erase the backup or do a filesystem-level backup instead of a binary
1102 If you use dm-crypt, backup is simpler: As there is no key
1103 management, the main risk is that you cannot wipe the backup when
1104 wiping the original. However wiping the original for dm-crypt
1105 should consist of forgetting the passphrase and that you can do
1106 without actual access to the backup.
1108 In both cases, there is an additional (usually small) risk with
1109 binary backups: An attacker can see how many sectors and which
1110 ones have been changed since the backup. To prevent this, use a
1111 filesystem level backup methid that encrypts the whole backup in
1112 one go, e.g. as described above with tar and GnuPG.
1114 My personal advice is to use one USB disk (low value data) or
1115 three disks (high value data) in rotating order for backups, and
1116 either use independent LUKS partitions on them, or use encrypted
1117 backup with tar and GnuPG.
1119 If you do network-backup or tape-backup, I strongly recommend to
1120 go the filesystem backup path with independent encryption, as you
1121 typically cannot reliably delete data in these scenarios,
1122 especially in a cloud setting. (Well, you can burn the tape if it
1123 is under your control...)
1126 * 6.8 What happens if I overwrite the start of a LUKS partition or
1127 damage the LUKS header or key-slots?
1129 There are two critical components for decryption: The salt values
1130 in the header itself and the key-slots. If the salt values are
1131 overwritten or changed, nothing (in the cryptographically strong
1132 sense) can be done to access the data, unless there is a backup
1133 of the LUKS header. If a key-slot is damaged, the data can still
1134 be read with a different key-slot, if there is a remaining
1135 undamaged and used key-slot. Note that in order to make a key-slot
1136 unrecoverable in a cryptographically strong sense, changing about
1137 4-6 bits in random locations of its 128kiB size is quite enough.
1140 * 6.9 What happens if I (quick) format a LUKS partition?
1142 I have not tried the different ways to do this, but very likely you
1143 will have written a new boot-sector, which in turn overwrites the
1144 LUKS header, including the salts, making your data permanently
1145 irretrivable, unless you have a LUKS header backup. You may also
1146 damage the key-slots in part or in full. See also last item.
1149 * 6.10 How do I recover the master key from a mapped LUKS container?
1151 This is typically only needed if you managed to damage your LUKS
1152 header, but the container is still mapped, i.e. "luksOpen"ed. It
1153 also helps if you have a mapped container that you forgot or do not
1154 know a passphrase for (e.g. on a long running server.)
1156 WARNING: Things go wrong, do a full backup before trying this!
1158 WARNING: This exposes the master key of the LUKS container. Note
1159 that both ways to recreate a LUKS header with the old master key
1160 described below will write the master key to disk. Unless you are
1161 sure you have securely erased it afterwards, e.g. by writing it to
1162 an encrypted partition, RAM disk or by erasing the filesystem you
1163 wrote it to by a complete overwrite, you should change the master
1164 key afterwards. Changing the master key requires a full data
1165 backup, luksFormat and then restore of the backup.
1167 First, there is a script by Milan that automatizes the whole
1168 process, except generating a new LUKS header with the old master
1169 key (it prints the command for that though):
1171 http://code.google.com/p/cryptsetup/source/browse/trunk/misc/luks-header-from-active
1173 You can also do this manually. Here is how:
1175 - Get the master key from the device mapper. This is done by the
1176 following command. Substitute c5 for whatever you mapped to:
1178 # dmsetup table --target crypt --showkey /dev/mapper/c5
1180 0 200704 crypt aes-cbc-essiv:sha256
1181 a1704d9715f73a1bb4db581dcacadaf405e700d591e93e2eaade13ba653d0d09
1184 The result is actually one line, wrapped here for clarity. The long
1185 hex string is the master key.
1187 - Convert the master key to a binary file representation. You can
1188 do this manually, e.g. with hexedit. You can also use the tool
1189 "xxd" from vim like this:
1191 echo "a1704d9....53d0d09" | xxd -r -p > <master-key-file>
1193 - Do a luksFormat to create a new LUKS header.
1195 NOTE: If your header is intact and you just forgot the
1196 passphrase, you can just set a new passphrase, see next subitem.
1198 Unmap the device before you do that (luksClose). Then do
1200 cryptsetup luksFormat --master-key-file=<master-key-file> <luks device>
1202 Note that if the container was created with other than the default
1203 settings of the cryptsetup version you are using, you need to give
1204 additional parameters specifying the deviations. If in doubt, try
1205 the script by Milan. It does recover the other parameters as well.
1207 Side note: This is the way the decrypt_derived script gets at the
1208 master key. It just omits the conversion and hashes the master key
1211 - If the header is intact and you just forgot the passphrase, just
1212 set a new passphrase like this:
1214 cryptsetup luksAddKey --master-key-file=<master-key-file> <luks device>
1216 You may want to disable the old one afterwards.
1219 * 6.11 What does the on-disk structure of dm-crypt look like?
1221 There is none. dm-crypt takes a block device and gives encrypted
1222 access to each of its blocks with a key derived from the passphrase
1223 given. If you use a cipher different than the default, you have to
1224 specify that as a parameter to cryptsetup too. If you want to
1225 change the password, you basically have to create a second
1226 encrypted device with the new passphrase and copy your data over.
1227 On the plus side, if you accidentally overwrite any part of a
1228 dm-crypt device, the damage will be limited to the are you
1232 * 6.12 What does the on-disk structure of LUKS look like?
1234 A LUKS partition consists of a header, followed by 8 key-slot
1235 descriptors, followed by 8 key slots, followed by the encrypted
1238 Header and key-slot descriptors fill the first 592 bytes. The
1239 key-slot size depends on the creation parameters, namely on the
1240 number of anti-forensic stripes, key material offset and master
1243 With the default parameters, each key-slot is a bit less than
1244 128kiB in size. Due to sector alignment of the key-slot start,
1245 that means the key block 0 is at offset 0x1000-0x20400, key
1246 block 1 at offset 0x21000-0x40400, and key block 7 at offset
1247 0xc1000-0xe0400. The space to the next full sector address is
1248 padded with zeros. Never used key-slots are filled with what the
1249 disk originally contained there, a key-slot removed with
1250 "luksRemoveKey" or "luksKillSlot" gets filled with 0xff. Due to
1251 2MiB default alignment, start of the data area for cryptsetup 1.3
1252 and later is at 2MiB, i.e. at 0x200000. For older versions, it is
1253 at 0x101000, i.e. at 1'052'672 bytes, i.e. at 1MiB + 4096 bytes
1254 from the start of the partition. Incidentally, "luksHeaderBackup"
1255 for a LUKS container created with default parameters dumps exactly
1256 the first 2MiB (or 1'052'672 bytes for headers created with
1257 cryptsetup versions < 1.3) to file and "luksHeaderRestore" restores
1260 For non-default parameters, you have to figure out placement
1261 yourself. "luksDump" helps. See also next item. For the most common
1262 non-default settings, namely aes-xts-plain with 512 bit key, the
1263 offsets are: 1st keyslot 0x1000-0x3f800, 2nd keyslot
1264 0x40000-0x7e000, 3rd keyslot 0x7e000-0xbd800, ..., and start of
1265 bulk data at 0x200000.
1267 The exact specification of the format is here:
1268 http://code.google.com/p/cryptsetup/wiki/Specification
1271 * 6.13 What is the smallest possible LUKS container?
1273 Note: From cryptsetup 1.3 onwards, alignment is set to 1MB. With
1274 modern Linux partitioning tools that also align to 1MB, this will
1275 result in aligmnet to 2k secors and typical Flash/SSD sectors,
1276 which is highly desirable for a number of reasons. Changing the
1277 alignment is not recomended.
1279 That said, with default parameters, the data area starts at
1280 exactly 2MB offset (at 0x101000 for cryptsetup versions before
1281 1.3). The smallest data area you can have is one sector of 512
1282 bytes. Data areas of 0 bytes can be created, but fail on mapping.
1284 While you cannot put a filesystem into something this small, it may
1285 still be used to contain, for example, key. Note that with current
1286 formatting tools, a partition for a container this size will be
1287 3MiB anyways. If you put the LUKS container into a file (via
1288 losetup and a loopback device), the file needs to be 2097664 bytes
1289 in size, i.e. 2MiB + 512B.
1291 There two ways to influence the start of the data area are key-size
1294 For alignment, you can go down to 1 on the parameter. This will
1295 still leave you with a data-area starting at 0x101000, i.e.
1296 1MiB+4096B (default parameters) as alignment will be rounded up to
1297 the next multiple of 8 (i.e. 4096 bytes) If in doubt, do a dry-run
1298 on a larger file and dump the LUKS header to get actual
1301 For key-size, you can use 128 bit (e.g. AES-128 with CBC), 256 bit
1302 (e.g. AES-256 with CBC) or 512 bit (e.g. AES-256 with XTS mode).
1303 You can do 64 bit (e.g. blowfish-64 with CBC), but anything below
1304 128 bit has to be considered insecure today.
1306 Example 1 - AES 128 bit with CBC:
1308 cryptsetup luksFormat -s 128 --align-payload=8 <device>
1310 This results in a data offset of 0x81000, i.e. 516KiB or 528384
1311 bytes. Add one 512 byte sector and the smallest LUKS container size
1312 with these parameters is 516KiB + 512B or 528896 bytes.
1314 Example 2 - Blowfish 64 bit with CBC (WARNING: insecure):
1316 cryptsetup luksFormat -c blowfish -s 64 --align-payload=8 /dev/loop0
1318 This results in a data offset of 0x41000, i.e. 260kiB or 266240
1319 bytes, with a minimal LUKS conatiner size of 260kiB + 512B or
1323 * 6.14 I think this is overly complicated. Is there an alternative?
1325 Not really. Encryption comes at a price. You can use plain
1326 dm-crypt to simplify things a bit. It does not allow multiple
1327 passphrases, but on the plus side, it has zero on disk description
1328 and if you overwrite some part of a plain dm-crypt partition,
1329 exactly the overwritten parts are lost (rounded up to sector
1333 * 6.15 Can I clone a LUKS container?
1335 You can, but it breaks security, because the cloned container has
1336 the same header and hence the same master key. You cannot change
1337 the master key on a LUKS container, even if you change the
1338 passphrase(s), the master key stays the same. That means whoever
1339 has access to one of the clones can decrypt them all, completely
1340 bypassing the passphrases.
1342 The right way to do this is to first luksFormat the target
1343 container, then to clone the contents of the source container, with
1344 both containers mapped, i.e. decrypted. You can clone the decrypted
1345 contents of a LUKS container in binary mode, although you may run
1346 into secondary issuses with GUIDs in filesystems, partition tables,
1347 RAID-components and the like. These are just the normal problems
1348 binary cloning causes.
1351 7. Interoperability with other Disk Encryption Tools
1354 * 7.1 What is this section about?
1356 Cryptsetup for plain dm-crypt can be used to access a number of
1357 on-disk formats created by tools like loop-aes patched into
1358 losetup. This somtimes works and sometimes does not. This section
1359 collects insights into what works, what does not and where more
1360 information is required.
1362 Additional information may be found in the mailing-list archives,
1363 mentioned at the start of this FAQ document. If you have a
1364 solution working that is not yet documented here and think a wider
1365 audience may be intertested, please email the FAQ maintainer.
1368 * 7.2 loop-aes: General observations.
1370 One problem is that there are different versions of losetup around.
1371 loop-aes is a patch for losetup. Possible problems and deviations
1372 from cryptsetup option syntax include:
1374 - Offsets specifed in bytes (cryptsetup: 512 byte sectors)
1376 - The need to specify an IV offset
1378 - Encryption mode needs specifying (e.g. "-c twofish-cbc-plain")
1380 - Key size needs specifying (e.g. "-s 128" for 128 bit keys)
1382 - Passphrase hash algorithm needs specifying
1384 Also note that because plain dm-crypt and loop-aes format does not
1385 have metadata, autodetection, while feasible in most cases, would
1386 be a lot of work that nobody really wants to do. If you still have
1387 the old set-up, using a verbosity option (-v) on mapping with the
1388 old tool or having a look into the system logs after setup could
1389 give you the information you need.
1392 * 7.3 loop-aes patched into losetup on debian 5.x, kernel 2.6.32
1394 In this case, the main problem seems to be that this variant of
1395 losetup takes the offset (-o option) in bytes, while cryptsetup
1396 takes it in sectors of 512 bytes each. Example: The losetupp
1399 losetup -e twofish -o 2560 /dev/loop0 /dev/sdb1
1400 mount /dev/loop0 mountpoint
1404 cryptsetup create -c twofish -o 5 --skip 5 e1 /dev/sdb1
1405 mount /dev/mapper/e1 mountpoint
1408 * 7.4 loop-aes with 160 bit key
1410 This seems to be sometimes used with twofish and blowfish and
1411 represents a 160 bit ripemed160 hash output padded to 196 bit key
1412 length. It seems the corresponding options for cryptsetup are
1414 --cipher twofish-cbc-null -s 192 -h ripemd160:20
1417 8. Issues with Specific Versions of cryptsetup
1420 * 8.1 When using the create command for plain dm-crypt with
1421 cryptsetup 1.1.x, the mapping is incompatible and my data is not
1424 With cryptsetup 1.1.x, the distro maintainer can define different
1425 default encryption modes for LUKS and plain devices. You can check
1426 these compiled-in defaults using "cryptsetup --help". Moreover, the
1427 plain device default changed because the old IV mode was
1428 vulnerable to a watermarking attack.
1430 If you are using a plain device and you need a compatible mode, just
1431 specify cipher, key size and hash algorithm explicitly. For
1432 compatibility with cryptsetup 1.0.x defaults, simple use the
1435 cryptsetup create -c aes-cbc-plain -s 256 -h ripemd160 <name> <dev>
1437 LUKS stores cipher and mode in the metadata on disk, avoiding this
1441 * 8.2 cryptsetup on SLED 10 has problems...
1443 SLED 10 is missing an essential kernel patch for dm-crypt, which
1444 is broken in its kernel as a result. There may be a very old
1445 version of cryptsetup (1.0.x) provided by SLED, which should also
1446 not be used anymore as well. My advice would be to drop SLED 10.
1448 A. Contributors In no particular order: