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 with a mapped (decrypted) container or a
50 passphrase on that machine can decrypt all other copies, breaking
51 security. See also Item 6.15.
53 DISTRIBUTION INSTALLERS: Some distribution installers offer to
54 create LUKS containers in a way that can be mistaken as activation
55 of an existing container. Creating a new LUKS container on top of
56 an existing one leads to permanent, complete and irreversible data
57 loss. It is strongly recommended to only use distribution
58 installers after a complete backup of all LUKS containers has been
61 NO WARNING ON NON-INERACTIVE FORMAT: If you feed cryptsetup from
62 STDIN (e.g. via GnuPG) on LUKS format, it does not give you the
63 warning that you are about to format (and e.g. will lose any
64 pre-existing LUKS container on the target), as it assumes it is
65 used from a script. In this scenario, the responsibility for
66 warning the user and possibly checking for an existing LUKS header
67 is shifted to the script. This is a more general form of the
70 LUKS PASSPHRASE IS NOT THE MASTER KEY: The LUKS passphrase is not
71 used in deriving the master key. It is used in decrypting a master
72 key that is randomly selected on header creation. This means that
73 if you create a new LUKS header on top of an old one with
74 exactly the same parameters and exactly the same passphrase as the
75 old one, it will still have a different master key and your data
76 will be permanently lost.
78 PASSPHRASE CHARACTER SET: Some people have had difficulties with
79 this when upgrading distributions. It is highly advisable to only
80 use the 94 printable characters from the first 128 characters of
81 the ASCII table, as they will always have the same binary
82 representation. Other characters may have different encoding
83 depending on system configuration and your passphrase will not
84 work with a different encoding. A table of the standardized first
85 128 ASCII caracters can, e.g. be found on
86 http://en.wikipedia.org/wiki/ASCII
89 * 1.3 System Specific warnings
91 - Ubuntu as of 4/2011: It seems the installer offers to create
92 LUKS partitions in a way that several people mistook for an offer
93 to activate their existing LUKS partition. The installer gives no
94 or an inadequate warning and will destroy your old LUKS header,
95 causing permanent data loss. See also the section on Backup and
98 This issue has been acknowledged by the Ubuntu dev team, see here:
99 http://launchpad.net/bugs/420080
102 * 1.4 Who wrote this?
104 Current FAQ maintainer is Arno Wagner <arno@wagner.name>. Other
105 contributors are listed at the end. If you want to contribute, send
106 your article, including a descriptive headline, to the maintainer,
107 or the dm-crypt mailing list with something like "FAQ ..." in the
108 subject. You can also send more raw information and have me write
109 the section. Please note that by contributing to this FAQ, you
110 accept the license described below.
112 This work is under the "Attribution-Share Alike 3.0 Unported"
113 license, which means distribution is unlimited, you may create
114 derived works, but attributions to original authors and this
115 license statement must be retained and the derived work must be
116 under the same license. See
117 http://creativecommons.org/licenses/by-sa/3.0/ for more details of
120 Side note: I did text license research some time ago and I think
121 this license is best suited for the purpose at hand and creates the
125 * 1.5 Where is the project website?
127 There is the project website at http://code.google.com/p/cryptsetup/
128 Please do not post questions there, nobody will read them. Use
129 the mailing-list instead.
132 * 1.6 Is there a mailing-list?
134 Instructions on how to subscribe to the mailing-list are at on the
135 project website. People are generally helpful and friendly on the
138 The question of how to unsubscribe from the list does crop up
139 sometimes. For this you need your list management URL, which is
140 sent to you initially and once at the start of each month. Go to
141 the URL mentioned in the email and select "unsubscribe". This page
142 also allows you to request a password reminder.
144 Alternatively, you can send an Email to dm-crypt-request@saout.de
145 with just the word "help" in the subject or message body. Make sure
146 to send it from your list address.
148 The mailing list archive is here:
149 http://dir.gmane.org/gmane.linux.kernel.device-mapper.dm-crypt
155 * 2.1 What is the difference between "plain" and LUKS format?
157 Plain format is just that: It has no metadata on disk, reads all
158 paramters from the commandline (or the defaults), derives a
159 master-key from the passphrase and then uses that to de-/encrypt
160 the sectors of the device, with a direct 1:1 mapping between
161 encrypted and decrypted sectors.
163 Primary advantage is high resilience to damage, as one damaged
164 encrypted sector results in exactly one damaged decrypted sector.
165 Also, it is not readily apparent that there even is encrypted data
166 on the device, as an overwrite with crypto-grade randomness (e.g.
167 from /dev/urandom) looks exactly the same on disk.
169 Side-note: That has limited value against the authorities. In
170 civilized countries, they cannot force you to give up a crypto-key
171 anyways. In the US, the UK and dictatorships around the world,
172 they can force you to give up the keys (using imprisonment or worse
173 to pressure you), and in the worst case, they only need a
174 nebulous "suspicion" about the presence of encrypted data. My
175 advice is to either be ready to give up the keys or to not have
176 encrypted data when traveling to those countries, especially when
177 crossing the borders.
179 Disadvantages are that you do not have all the nice features that
180 the LUKS metadata offers, like multiple passphrases that can be
181 changed, the cipher being stored in the metadata, anti-forensic
182 properties like key-slot diffusion and salts, etc..
184 LUKS format uses a metadata header and 8 key-slot areas that are
185 being placed ath the begining of the disk, see below under "What
186 does the LUKS on-disk format looks like?". The passphrases are used
187 to decryt a single master key that is stored in the anti-forensic
190 Advantages are a higher usability, automatic configuration of
191 non-default crypto parameters, defenses against low-entropy
192 passphrases like salting and iterated PBKDF2 passphrase hashing,
193 the ability to change passhrases, and others.
195 Disadvantages are that it is readily obvious there is encrypted
196 data on disk (but see side note above) and that damage to the
197 header or key-slots usually results in permanent data-loss. See
198 below under "6. Backup and Data Recovery" on how to reduce that
199 risk. Also the sector numbers get shifted by the length of the
200 header and key-slots and there is a loss of that size in capacity
201 (1MB+4096B for defaults and 2MB for the most commonly used
202 non-default XTS mode).
205 * 2.2 Can I encrypt an already existing, non-empty partition to use
208 There is no converter, and it is not really needed. The way to do
209 this is to make a backup of the device in question, securely wipe
210 the device (as LUKS device initialization does not clear away old
211 data), do a luksFormat, optionally overwrite the encrypted device,
212 create a new filesystem and restore your backup on the now
213 encrypted device. Also refer to sections "Security Aspects" and
214 "Backup and Data Recovery".
216 For backup, plain GNU tar works well and backs up anything likely
217 to be in a filesystem.
220 * 2.3 How do I use LUKS with a loop-device?
222 This can be very handy for experiments. Setup is just the same as
223 with any block device. If you want, for example, to use a 100MiB
224 file as LUKS container, do something like this:
226 head -c 100M /dev/zero > luksfile # create empty file
227 losetup /dev/loop0 luksfile # map luksfile to /dev/loop0
228 cryptsetup luksFormat /dev/loop0 # create LUKS on loop device
230 Afterwards just use /dev/loop0 as a you would use a LUKS partition.
231 To unmap the file when done, use "losetup -d /dev/loop0".
234 * 2.4 When I add a new key-slot to LUKS, it asks for a passphrase but
235 then complains about there not being a key-slot with that
238 That is as intended. You are asked a passphrase of an existing
239 key-slot first, before you can enter the passphrase for the new
240 key-slot. Otherwise you could break the encryption by just adding a
241 new key-slot. This way, you have to know the passphrase of one of
242 the already configured key-slots in order to be able to configure a
246 * 2.5 Encrytion on top of RAID or the other way round?
248 Unless you have special needs, place encryption between RAID and
249 filesystem, i.e. encryption on top of RAID. You can do it the other
250 way round, but you have to be aware that you then need to give the
251 pasphrase for each individual disk and RAID autotetection will not
252 work anymore. Therefore it is better to encrypt the RAID device,
256 * 2.6 How do I read a dm-crypt key from file?
258 Note that the file will still be hashed first, just like keyboard
259 input. Use the --key-file option, like this:
261 cryptsetup create --key-file keyfile e1 /dev/loop0
264 * 2.7 How do I read a LUKS slot key from file?
266 What you really do here is to read a passphrase from file, just as
267 you would with manual entry of a passphrase for a key-slot. You can
268 add a new passphrase to a free key-slot, set the passphrase of an
269 specific key-slot or put an already configured passphrase into a
270 file. In the last case make sure no trailing newline (0x0a) is
271 contained in the key file, or the passphrase will not work because
272 the whole file is used as input.
274 To add a new passphrase to a free key slot from file, use something
277 cryptsetup luksAddKey /dev/loop0 keyfile
279 To add a new passphrase to a specific key-slot, use something like
282 cryptsetup luksAddKey --key-slot 7 /dev/loop0 keyfile
284 To supply a key from file to any LUKS command, use the --key-file
285 option, e.g. like this:
287 cryptsetup luksOpen --key-file keyfile /dev/loop0 e1
290 * 2.8 How do I read the LUKS master key from file?
292 The question you should ask yourself first is why you would want to
293 do this. The only legitimate reason I can think of is if you want
294 to have two LUKS devices with the same master key. Even then, I
295 think it would be preferable to just use key-slots with the same
296 passphrase, or to use plain dm-crypt instead. If you really have a
297 good reason, please tell me. If I am convinced, I will add how to
301 * 2.9 What are the security requirements for a key read from file?
303 A file-stored key or passphrase has the same security requirements
304 as one entered interactively, however you can use random bytes and
305 thereby use bytes you cannot type on the keyboard. You can use any
306 file you like as key file, for example a plain text file with a
307 human readable passphrase. To generate a file with random bytes,
308 use something like this:
310 head -c 256 /dev/random > keyfile
313 * 2.10 If I map a journaled file system using dm-crypt/LUKS, does it
314 still provide its usual transactional guarantees?
316 As far as I know it does (but I may be wrong), but please note that
317 these "guarantees" are far weaker than they appear to be. For
318 example, you may not get a hard flush to disk surface even on a
319 call to fsync. In addition, the HDD itself may do independent
320 write reordering. Some other things can go wrong as well. The
321 filesystem developers are aware of these problems and typically
322 can make it work anyways. That said, dm-crypt/LUKS should not make
325 Personally, I have several instances of ext3 on dm-crypt and have
326 not noticed any specific problems.
328 Update: I did run into frequent small freezes (1-2 sec) when putting
329 a vmware image on ext3 over dm-crypt. This does indicate that the
330 transactional guarantees are in place, but at a cost. When I went
331 back to ext2, the problem went away. This also seems to have gotten
332 better with kernel 2.6.36 and the reworking of filesystem flush
333 locking. Kernel 2.6.38 is expected to have more improvements here.
336 * 2.11 Can I use LUKS or cryptsetup with a more secure (external)
337 medium for key storage, e.g. TPM or a smartcard?
339 Yes, see the answers on using a file-supplied key. You do have to
340 write the glue-logic yourself though. Basically you can have
341 cryptsetup read the key from STDIN and write it there with your
342 own tool that in turn gets the key from the more secure key
346 * 2.12 Can I resize a dm-crypt or LUKS partition?
348 Yes, you can, as neither dm-crypt nor LUKS stores partition size.
349 Whether you should is a different question. Personally I recommend
350 backup, recreation of the encrypted partition with new size,
351 recreation of the filesystem and restore. This gets around the
352 tricky business of resizing the filesystem. Resizing a dm-crypt or
353 LUKS container does not resize the filesystem in it. The backup is
354 really non-optional here, as a lot can go wrong, resulting in
355 partial or complete data loss. Using something like gparted to
356 resize an encrypted partition is slow, but typicaly works. This
357 will not change the size of the filesystem hidden under the
360 You also need to be aware of size-based limitations. The one
361 currently relevant is that aes-xts-plain should not be used for
362 encrypted container sizes larger than 2TiB. Use aes-xts-plain64
369 * 3.1 My dm-crypt/LUKS mapping does not work! What general steps are
370 there to investigate the problem?
372 If you get a specific error message, investigate what it claims
373 first. If not, you may want to check the following things.
375 - Check that "/dev", including "/dev/mapper/control" is there. If it
376 is missing, you may have a problem with the "/dev" tree itself or
377 you may have broken udev rules.
379 - Check that you have the device mapper and the crypt target in your
380 kernel. The output of "dmsetup targets" should list a "crypt"
381 target. If it is not there or the command fails, add device mapper
382 and crypt-target to the kernel.
384 - Check that the hash-functions and ciphers you want to use are in
385 the kernel. The output of "cat /proc/crypto" needs to list them.
388 * 3.2 My dm-crypt mapping suddenly stopped when upgrading cryptsetup.
390 The default cipher, hash or mode may have changed (the mode changed
391 from 1.0.x to 1.1.x). See under "Issues With Specific Versions of
395 * 3.3 When I call cryptsetup from cron/CGI, I get errors about
398 If you get errors about unknown parameters or the like that are not
399 present when cryptsetup is called from the shell, make sure you
400 have no older version of cryptsetup on your system that then gets
401 called by cron/CGI. For example some distributions install
402 cryptsetup into /usr/sbin, while a manual install could go to
403 /usr/local/sbin. As a debugging aid, call "cryptsetup --version"
404 from cron/CGI or the non-shell mechanism to be sure the right
408 * 3.4 Unlocking a LUKS device takes very long. Why?
410 The iteration time for a key-slot (see Section 5 for an explanation
411 what iteration does) is calculated when setting a passphrase. By
412 default it is 1 second on the machine where the passphrase is set.
413 If you set a passphrase on a fast machine and then unlock it on a
414 slow machine, the unlocking time can be much longer. Also take into
415 account that up to 8 key-slots have to be tried in order to find the
418 If this is problem, you can add another key-slot using the slow
419 machine with the same passphrase and then remove the old key-slot.
420 The new key-slot will have an iteration count adjusted to 1 second
421 on the slow machine. Use luksKeyAdd and then luksKillSlot or
424 However, this operation will not change volume key iteration count
425 (MK iterations in output of "cryptsetup luksDump"). In order to
426 change that, you will have to backup the data in the LUKS
427 container (i.e. your encrypted data), luksFormat on the slow
428 machine and restore the data. Note that in the original LUKS
429 specification this value was fixed to 10, but it is now derived
430 from the PBKDF2 benchmark as well and set to iterations in 0.125
431 sec or 1000, whichever is larger. Also note that MK iterations
432 are not very security relevant. But as each key-slot already takes
433 1 second, spending the additional 0.125 seconds really does not
437 * 3.5 "blkid" sees a LUKS UUID and an ext2/swap UUID on the same
438 device. What is wrong?
440 Some old versions of cryptsetup have a bug where the header does
441 not get completely wiped during LUKS format and an older ext2/swap
442 signature remains on the device. This confuses blkid.
444 Fix: Wipe the unused header areas by doing a backup and restore of
445 the header with cryptsetup 1.1.x:
447 cryptsetup luksHeaderBackup --header-backup-file <file> <device>
448 cryptsetup luksHeaderRestore --header-backup-file <file> <device>
451 * 3.6 cryptsetup segfaults on Gentoo amd64 hardened ...
453 There seems to be some inteference between the hardening and and
454 the way cryptsetup benchmarks PBKDF2. The solution to this is
455 currently not quite clear for an encrypted root filesystem. For
456 other uses, you can apparently specify USE="dynamic" as compile
457 flag, see http://bugs.gentoo.org/show_bug.cgi?id=283470
463 * 4.1 I get the error "LUKS keyslot x is invalid." What does that
466 This means that the given keyslot has an offset that points
467 outside the valid keyslot area. Typically, the reason is a
468 corrupted LUKS header because something was written to the start of
469 the device the LUKS contaner is on. Refer to Section "Backup and
470 Data Recovery" and ask on the mailing list if you have trouble
471 diagnosing and (if still possible) repairing this.
474 * 4.2 Can a bad RAM module cause problems?
476 LUKS and dm-crypt can give the RAM quite a workout, especially when
477 combined with software RAID. In particular the combination RAID5 +
478 LUKS + XFS seems to uncover RAM problems that never caused obvious
479 problems before. Symptoms vary, but often the problem manifest
480 itself when copying large amounts of data, typically several times
481 larger than your main memory.
483 Side note: One thing you should always do on large data
484 copy/movements is to run a verify, for example with the "-d"
485 option of "tar" or by doing a set of MD5 checksums on the source
488 find . -type f -exec md5sum \{\} \; > checksum-file
490 and then a "md5sum -c checksum-file" on the other side. If you get
491 mismatches here, RAM is the primary suspect. A lesser suspect is
492 an overclocked CPU. I have found countless hardware problems in
493 verify runs after copying or making backups. Bit errors are much
494 more common than most people think.
496 Some RAM issues are even worse and corrupt structures in one of the
497 layers. This typically results in lockups, CPU state dumps in the
498 system logs, kernel panic or other things. It is quite possible to
499 have the problem with an encrypted device, but not with an
500 otherwise the same unencrypted device. The reason for that is that
501 encryption has an error amplification property: You flip one bit
502 in an encrypted data block, and the decrypted version has half of
503 its bits flipped. This is an important security property for modern
504 ciphers. With the usual modes in cryptsetup (CBC, ESSIV, XTS), you
505 get up to a completely changed 512 byte block per bit error. A
506 corrupt block causes a lot more havoc than the occasionally
507 flipped single bit and can result in various obscure errors.
509 Note, that a verify run on copying between encrypted or
510 unencrypted devices will reliably detect corruption, even when the
511 copying itself did not report any problems. If you find defect
512 RAM, assume all backups and copied data to be suspect, unless you
516 * 4.3 How do I test RAM?
518 First you should know that overclocking often makes memory
519 problems worse. So if you overclock (which I strongly recommend
520 against in a system holding data that has some worth), run the
521 tests with the overclocking active.
523 There are two good options. One is Memtest86+ and the other is
524 "memtester" by Charles Cazabon. Memtest86+ requires a reboot and
525 then takes over the machine, while memtester runs from a
526 root-shell. Both use different testing methods and I have found
527 problems fast with each one that the other needed long to find. I
528 recommend running the following procedure until the first error is
531 - Run Memtest86+ for one cycle
533 - Run memterster for one cycle (shut down as many other applications
536 - Run Memtest86+ for 24h or more
538 - Run memtester for 24h or more
540 If all that does not produce error messages, your RAM may be sound,
541 but I have had one weak bit that Memtest86+ needed around 60 hours
542 to find. If you can reproduce the original problem reliably, a good
543 additional test may be to remove half of the RAM (if you have more
544 than one module) and try whether the problem is still there and if
545 so, try with the other half. If you just have one module, get a
546 different one and try with that. If you do overclocking, reduce
547 the settings to the most conservative ones available and try with
554 * 5.1 Is LUKS insecure? Everybody can see I have encrypted data!
556 In practice it does not really matter. In most civilized countries
557 you can just refuse to hand over the keys, no harm done. In some
558 countries they can force you to hand over the keys, if they suspect
559 encryption. However the suspicion is enough, they do not have to
560 prove anything. This is for practical reasons, as even the presence
561 of a header (like the LUKS header) is not enough to prove that you
562 have any keys. It might have been an experiment, for example. Or it
563 was used as encrypted swap with a key from /dev/random. So they
564 make you prove you do not have encrypted data. Of course that is
565 just as impossible as the other way round.
567 This means that if you have a large set of random-looking data,
568 they can already lock you up. Hidden containers (encryption hidden
569 within encryption), as possible with Truecrypt, do not help
570 either. They will just assume the hidden container is there and
571 unless you hand over the key, you will stay locked up. Don't have
572 a hidden container? Though luck. Anybody could claim that.
574 Still, if you are concerned about the LUKS header, use plain
575 dm-crypt with a good passphrase. See also Section 2, "What is the
576 difference between "plain" and LUKS format?"
579 * 5.2 Should I initialize (overwrite) a new LUKS/dm-crypt partition?
581 If you just create a filesystem on it, most of the old data will
582 still be there. If the old data is sensitive, you should overwrite
583 it before encrypting. In any case, not initializing will leave the
584 old data there until the specific sector gets written. That may
585 enable an attacker to determine how much and where on the
586 partition data was written. If you think this is a risk, you can
587 prevent this by overwriting the encrypted device (here assumed to
588 be named "e1") with zeros like this:
590 dd_rescue -w /dev/zero /dev/mapper/e1
592 or alternatively with one of the following more standard commands:
594 cat /dev/zero > /dev/mapper/e1
595 dd if=/dev/zero of=/dev/mapper/e1
598 * 5.3 How do I securely erase a LUKS (or other) partition?
600 For LUKS, if you are in a desperate hurry, overwrite the LUKS
601 header and key-slot area. This means overwriting the first
602 (keyslots x stripes x keysize) + offset bytes. For the default
603 parameters, this is the 1'052'672 bytes, i.e. 1MiB + 4096 of the
604 LUKS partition. For 512 bit key length (e.g. for aes-xts-plain with
605 512 bit key) this is 2MiB. (The diferent offset stems from
606 differences in the sector alignment of the key-slots.) If in doubt,
607 just be generous and overwrite the first 10MB or so, it will likely
608 still be fast enough. A single overwrite with zeros should be
609 enough. If you anticipate being in a desperate hurry, prepare the
610 command beforehand. Example with /dev/sde1 as the LUKS partition
611 and default parameters:
613 head -c 1052672 /dev/zero > /dev/sde1; sync
615 A LUKS header backup or full backup will still grant access to
616 most or all data, so make sure that an attacker does not have
617 access to backups or destroy them as well.
619 If you have time, overwrite the whole LUKS partition with a single
620 pass of zeros. This is enough for current HDDs. For SSDs or FLASH
621 (USB sticks) you may want to overwrite the whole drive several
622 times to be sure data is not retained by wear leveling. This is
623 possibly still insecure as SSD technology is not fully understood
624 in this regard. Still, due to the anti-forensic properties of the
625 LUKS key-slots, a single overwrite of an SSD or FLASH drive could
626 be enough. If in doubt, use physical destruction in addition. Here
627 is a link to some current reseach results on erasing SSDs and FLASH
629 http://www.usenix.org/events/fast11/tech/full_papers/Wei.pdf
631 Keep in mind to also erase all backups.
633 Example for a zero-overwrite erase of partition sde1 done with
636 dd_rescue -w /dev/zero /dev/sde1
639 * 5.4 How do I securely erase a backup of a LUKS partition or header?
641 That depends on the medium it is stored on. For HDD and SSD, use
642 overwrite with zeros. For an SSD or FLASH drive (USB stick), you
643 may want to overwrite the complete SSD several times and use
644 physical destruction in addition, see last item. For re-writable
645 CD/DVD, a single overwrite should also be enough, due to the
646 anti-forensic properties of the LUKS keyslots. For write-once
647 media, use physical destruction. For low security requirements,
648 just cut the CD/DVD into several parts. For high security needs,
649 shred or burn the medium. If your backup is on magnetic tape, I
650 advise physical destruction by shredding or burning, after
651 overwriting . The problem with magnetic tape is that it has a
652 higher dynamic range than HDDs and older data may well be
653 recoverable after overwrites. Also write-head alignment issues can
654 lead to data not actually being deleted at all during overwrites.
657 * 5.5 What about backup? Does it compromise security?
659 That depends. See item 6.7.
662 * 5.6 Why is all my data permanently gone if I overwrite the LUKS
665 Overwriting the LUKS header in part or in full is the most common
666 reason why access to LUKS containers is lost permanently.
667 Overwriting can be done in a number of fashions, like creating a
668 new filesystem on the raw LUKS partition, making the raw partition
669 part of a raid array and just writing to the raw partition.
671 The LUKS header contains a 256 bit "salt" value and without that no
672 decryption is possible. While the salt is not secret, it is
673 key-grade material and cannot be reconstructed. This is a
674 cryptographically strong "cannot". From observations on the
675 cryptsetup mailing-list, people typically go though the usual
676 stages of grief (Denial, Anger, Bargaining, Depression, Acceptance)
677 when this happens to them. Observed times vary between 1 day and 2
678 weeks to complete the cycle. Seeking help on the mailing-list is
679 fine. Even if we usually cannot help with getting back your data,
680 most people found the feedback comforting.
682 If your header does not contain an intact salt, best go directly
683 to the last stage ("Acceptance") and think about what to do now.
684 There is one exception that I know of: If your LUKS container is
685 still open, then it may be possible to extract the master key from
686 the running system. See Item "How do I recover the master key from
687 a mapped LUKS container?" in Section "Backup and Data Recovery".
690 * 5.7 What is a "salt"?
692 A salt is a random key-grade value added to the passphrase before
693 it is processed. It is not kept secret. The reason for using salts
694 is as follows: If an attacker wants to crack the password for a
695 single LUKS container, then every possible passphrase has to be
696 tried. Typically an attacker will not try every binary value, but
697 will try words and sentences from a dictionary.
699 If an attacker wants to attack several LUKS containers with the
700 same dictionary, then a different approach makes sense: Compute the
701 resulting slot-key for each dictionary element and store it on
702 disk. Then the test for each entry is just the slow unlocking with
703 the slot key (say 0.00001 sec) instead of calculating the slot-key
704 first (1 sec). For a single attack, this does not help. But if you
705 have more than one container to attack, this helps tremendously,
706 also because you can prepare your table before you even have the
707 container to attack! The calculation is also very simple to
708 parallelize. You could, for example, use the night-time unused CPU
709 power of your desktop PCs for this.
711 This is where the salt comes in. If the salt is combined with the
712 passphrase (in the simplest form, just appended to it), you
713 suddenly need a separate table for each salt value. With a
714 reasonably-sized salt value (256 bit, e.g.) this is quite
718 * 5.8 Is LUKS secure with a low-entropy (bad) passphrase?
720 Note: You should only use the 94 printable characters from 7 bit
721 ASCII code to prevent your passphrase from failing when the
722 character encoding changes, e.g. because of a system upgrade, see
723 also the note at the very start of this FAQ under "WARNINGS".
725 This needs a bit of theory. The quality of your passphrase is
726 directly related to its entropy (information theoretic, not
727 thermodynamic). The entropy says how many bits of "uncertainty" or
728 "randomness" are in you passphrase. In other words, that is how
729 difficult guessing the passphrase is.
731 Example: A random English sentence has about 1 bit of entropy per
732 character. A random lowercase (or uppercase) character has about
735 Now, if n is the number of bits of entropy in your passphrase and t
736 is the time it takes to process a passphrase in order to open the
737 LUKS container, then an attacker has to spend at maximum
739 attack_time_max = 2^n * t
741 time for a successful attack and on average half that. There is no
742 way getting around that relationship. However, there is one thing
743 that does help, namely increasing t, the time it takes to use a
744 passphrase, see next FAQ item.
746 Still, if you want good security, a high-entropy passphrase is the
747 only option. For example, a low-entropy passphrase can never be
748 considered secure against a TLA-level (Three Letter Agency level,
749 i.e. government-level) attacker, no matter what tricks are used in
750 the key-derivation function. Use at least 64 bits for secret stuff.
751 That is 64 characters of English text (but only if randomly chosen)
752 or a combination of 12 truly random letters and digits.
754 For passphrase generation, do not use lines from very well-known
755 texts (religious texts, Harry potter, etc.) as they are to easy to
756 guess. For example, the total Harry Potter has about 1'500'000
757 words (my estimation). Trying every 64 character sequence starting
758 and ending at a word boundary would take only something like 20
759 days on a single CPU and is entirely feasible. To put that into
760 perspective, using a number of Amazon EC2 High-CPU Extra Large
761 instances (each gives about 8 real cores), this test costs
762 currently about 50USD/EUR, but can be made to run arbitrarily fast.
764 On the other hand, choosing 1.5 lines from, say, the Wheel of Time
765 is in itself not more secure, but the book selection adds quite a
766 bit of entropy. (Now that I have mentioned it here, don't use tWoT
767 either!) If you add 2 or 3 typos or switch some words around, then
768 this is good passphrase material.
771 * 5.9 What is "iteration count" and why is decreasing it a bad idea?
773 Iteration count is the number of PBKDF2 iterations a passphrase is
774 put through before it is used to unlock a key-slot. Iterations are
775 done with the explicit purpose to increase the time that it takes
776 to unlock a key-slot. This provides some protection against use of
777 low-entropy passphrases.
779 The idea is that an attacker has to try all possible passphrases.
780 Even if the attacker knows the passphrase is low-entropy (see last
781 item), it is possible to make each individual try take longer. The
782 way to do this is to repeatedly hash the passphrase for a certain
783 time. The attacker then has to spend the same time (given the same
784 computing power) as the user per try. With LUKS, the default is 1
785 second of PBKDF2 hashing.
787 Example 1: Lets assume we have a really bad passphrase (e.g. a
788 girlfriends name) with 10 bits of entropy. With the same CPU, an
789 attacker would need to spend around 500 seconds on average to
790 break that passphrase. Without iteration, it would be more like
791 0.0001 seconds on a modern CPU.
793 Example 2: The user did a bit better and has 32 chars of English
794 text. That would be about 32 bits of entropy. With 1 second
795 iteration, that means an attacker on the same CPU needs around 136
796 years. That is pretty impressive for such a weak passphrase.
797 Without the iterations, it would be more like 50 days on a modern
798 CPU, and possibly far less.
800 In addition, the attacker can both parallelize and use special
801 hardware like GPUs or FPGAs to speed up the attack. The attack can
802 also happen quite some time after the luksFormat operation and CPUs
803 can have become faster and cheaper. For that reason you want a
804 bit of extra security. Anyways, in Example 1 your are screwed.
805 In example 2, not necessarily. Even if the attack is faster, it
806 still has a certain cost associated with it, say 10000 EUR/USD
807 with iteration and 1 EUR/USD without iteration. The first can be
808 prohibitively expensive, while the second is something you try
809 even without solid proof that the decryption will yield something
812 The numbers above are mostly made up, but show the idea. Of course
813 the best thing is to have a high-entropy passphrase.
815 Would a 100 sec iteration time be even better? Yes and no.
816 Cryptographically it would be a lot better, namely 100 times better.
817 However, usability is a very important factor for security
818 technology and one that gets overlooked surprisingly often. For
819 LUKS, if you have to wait 2 minutes to unlock the LUKS container,
820 most people will not bother and use less secure storage instead. It
821 is better to have less protection against low-entropy passphrases
822 and people actually use LUKS, than having them do without
823 encryption altogether.
825 Now, what about decreasing the iteration time? This is generally a
826 very bad idea, unless you know and can enforce that the users only
827 use high-entropy passphrases. If you decrease the iteration time
828 without ensuring that, then you put your users at increased risk,
829 and considering how rarely LUKS containers are unlocked in a
830 typical work-flow, you do so without a good reason. Don't do it.
831 The iteration time is already low enough that users with entropy
832 low passphrases are vulnerable. Lowering it even further increases
833 this danger significantly.
836 * 5.10 Some people say PBKDF2 is insecure?
838 There is some discussion that a hash-function should have a "large
839 memory" property, i.e. that it should require a lot of memory to be
840 computed. This serves to prevent attacks using special programmable
841 circuits, like FPGAs, and attacks using graphics cards. PBKDF2
842 does not need a lot of memory and is vulnerable to these attacks.
843 However, the publication usually refered in these discussions is
844 not very convincing in proving that the presented hash really is
845 "large memory" (that may change, email the FAQ maintainer when it
846 does) and it is of limited usefulness anyways. Attackers that use
847 clusters of normal PCs will not be affected at all by a "large
848 memory" property. For example the US Secret Service is known to
849 use the off-hour time of all the office PCs of the Treasury for
850 password breaking. The Treasury has about 110'000 employees.
851 Asuming every one has an office PC, that is significant computing
852 power, all of it with plenty of memory for computing "large
853 memory" hashes. Bot-net operators also have all the memory they
854 want. The only protection against a resouceful attacker is a
855 high-entropy passphrase, see items 5.8 and 5.9.
858 * 5.11 What about iteration count with plain dm-crypt?
860 Simple: There is none. There is also no salting. If you use plain
861 dm-crypt, the only way to be secure is to use a high entropy
862 passphrase. If in doubt, use LUKS instead.
865 * 5.12 Is LUKS with default parameters less secure on a slow CPU?
867 Unfortunately, yes. However the only aspect affected is the
868 protection for low-entropy passphrase or master-key. All other
869 security aspects are independent of CPU speed.
871 The master key is less critical, as you really have to work at it
872 to give it low entropy. One possibility is to supply the master key
873 yourself. If that key is low-entropy, then you get what you
874 deserve. The other known possibility is to use /dev/urandom for
875 key generation in an entropy-startved situation (e.g. automatic
876 installation on an embedded device without network and other entropy
879 For the passphrase, don't use a low-entropy passphrase. If your
880 passphrase is good, then a slow CPU will not matter. If you insist
881 on a low-entropy passphrase on a slow CPU, use something like
882 "--iter-time=10" or higher and wait a long time on each LUKS unlock
883 and pray that the attacker does not find out in which way exactly
884 your passphrase is low entropy. This also applies to low-entropy
885 passphrases on fast CPUs. Technology can do only so much to
886 compensate for problems in front of the keyboard.
889 * 5.13 Why was the default aes-cbc-plain replaced with aes-cbc-essiv?
891 The problem is that cbc-plain has a fingerprint vulnerability, where
892 a specially crafted file placed into the crypto-container can be
893 recognized from the outside. The issue here is that for cbc-plain
894 the initialization vector (IV) is the sector number. The IV gets
895 XORed to the first data chunk of the sector to be encrypted. If you
896 make sure that the first data block to be stored in a sector
897 contains the sector number as well, the first data block to be
898 encrypted is all zeros and always encrypted to the same ciphertext.
899 This also works if the first data chunk just has a constant XOR
900 with the sector number. By having several shifted patterns you can
901 take care of the case of a non-power-of-two start sector number of
904 This mechanism allows you to create a pattern of sectors that have
905 the same first ciphertext block and signal one bit per sector to the
906 outside, allowing you to e.g. mark media files that way for
907 recognition without decryption. For large files this is a
908 practical attack. For small ones, you do not have enough blocks to
909 signal and take care of different file starting offsets.
911 In order to prevent this attack, the default was changed to
912 cbc-essiv. ESSIV uses a keyed hash of the sector number, with the
913 encryption key as key. This makes the IV unpredictable without
914 knowing the encryption key and the watermarking attack fails.
917 * 5.14 Are there any problems with "plain" IV? What is "plain64"?
919 First, "plain" and "plain64" are both not secure to use with CBC,
920 see previous FAQ item.
922 However there are modes, like XTS, that are secure with "plain" IV.
923 The next limit is that "plain" is 64 bit, with the upper 32 bit set
924 to zero. This means that on volumes larger than 2TiB, the IV
925 repeats, creating a vulnerability that potentially leaks some
926 data. To avoid this, use "plain64", which uses the full sector
927 number up to 64 bit. Note that "plain64" requires a kernel >=
928 2.6.33. Also note that "plain64" is backwards compatible for
929 volume sizes <= 2TiB, but not for those > 2TiB. Finally, "plain64"
930 does not cause any performance penalty compared to "plain".
933 * 5.15 What about XTS mode?
935 XTS mode is potentially even more secure than cbc-essiv (but only if
936 cbc-essiv is insecure in your scenario). It is a NIST standard and
937 used, e.g. in Truecrypt. At the moment, if you want to use it, you
938 have to specify it manually as "aes-xts-plain", i.e.
940 cryptsetup -c aes-xts-plain luksFormat <device>
942 For volumes >2TiB and kernels >= 2.6.33 use "plain64" (see FAQ
943 item on "plain" and "plain64"):
945 cryptsetup -c aes-xts-plain64 luksFormat <device>
947 There is a potential security issue with XTS mode and large blocks.
948 LUKS and dm-crypt always use 512B blocks and the issue does not
952 6. Backup and Data Recovery
955 * 6.1 Why do I need Backup?
957 First, disks die. The rate for well-treated (!) disk is about 5%
958 per year, which is high enough to worry about. There is some
959 indication that this may be even worse for some SSDs. This applies
960 both to LUKS and plain dm-crypt partitions.
962 Second, for LUKS, if anything damages the LUKS header or the
963 key-stripe area then decrypting the LUKS device can become
964 impossible. This is a frequent occuurence. For example an
965 accidental format as FAT or some software overwriting the first
966 sector where it suspects a partition boot sector typically makes a
967 LUKS partition permanently inacessible. See more below on LUKS
970 So, data-backup in some form is non-optional. For LUKS, you may
971 also want to store a header backup in some secure location. This
972 only needs an update if you change passphrases.
975 * 6.2 How do I backup a LUKS header?
977 While you could just copy the appropriate number of bytes from the
978 start of the LUKS partition, the best way is to use command option
979 "luksHeaderBackup" of cryptsetup. This protects also against
980 errors when non-standard parameters have been used in LUKS
981 partition creation. Example:
984 cryptsetup luksHeaderBackup --header-backup-file h <device>
986 To restore, use the inverse command, i.e.
988 cryptsetup luksHeaderRestore --header-backup-file h <device>
991 * 6.3 How do I test a LUKS header?
995 cryptsetup -v isLuks <device>
997 on the device. Without the "-v" it just signals its result via
998 exit-status. You can alos use the more general test
1002 which will also detect other types and give some more info. Omit
1003 "-p" for old versions of blkid that do not support it.
1006 * 6.4 How do I backup a LUKS or dm-crypt partition?
1008 There are two options, a sector-image and a plain file or
1009 filesystem backup of the contents of the partition. The sector
1010 image is already encrypted, but cannot be compressed and contains
1011 all empty space. The filesystem backup can be compressed, can
1012 contain only part of the encrypted device, but needs to be
1013 encrypted separately if so desired.
1015 A sector-image will contain the whole partition in encrypted form,
1016 for LUKS the LUKS header, the keys-slots and the data area. It can
1017 be done under Linux e.g. with dd_rescue (for a direct image copy)
1018 and with "cat" or "dd". Example:
1020 cat /dev/sda10 > sda10.img
1021 dd_rescue /dev/sda10 sda10.img
1023 You can also use any other backup software that is capable of making
1024 a sector image of a partition. Note that compression is
1025 ineffective for encrypted data, hence it does not make sense to
1028 For a filesystem backup, you decrypt and mount the encrypted
1029 partition and back it up as you would a normal filesystem. In this
1030 case the backup is not encrypted, unless your encryption method
1031 does that. For example you can encrypt a backup with "tar" as
1034 tar cjf - <path> | gpg --cipher-algo AES -c - > backup.tbz2.gpg
1036 And verify the backup like this if you are at "path":
1038 cat backup.tbz2.gpg | gpg - | tar djf -
1040 Note: Allways verify backups, especially encrypted ones.
1042 In both cases GnuPG will ask you interactively for your symmetric
1043 key. The verify will only output errors. Use "tar dvjf -" to get
1044 all comparison results. To make sure no data is written to disk
1045 unencrypted, turn off swap if it is not encrypted before doing the
1048 You can of course use different or no compression and you can use
1049 an asymmetric key if you have one and have a backup of the secret
1050 key that belongs to it.
1052 A second option for a filestem-level backup that can be used when
1053 the backup is also on local disk (e.g. an external USB drive) is
1054 to use a LUKS container there and copy the files to be backed up
1055 between both mounted containers. Also see next item.
1058 * 6.5 Do I need a backup of the full partition? Would the header and
1059 key-slots not be enough?
1061 Backup protects you against two things: Disk loss or corruption
1062 and user error. By far the most questions on the dm-crypt mailing
1063 list about how to recover a damaged LUKS partition are related
1064 to user error. For example, if you create a new filesystem on a
1065 LUKS partition, chances are good that all data is lost
1068 For this case, a header+key-slot backup would often be enough. But
1069 keep in mind that a well-treated (!) HDD has roughly a failure
1070 risk of 5% per year. It is highly advisable to have a complete
1071 backup to protect against this case.
1074 * *6.6 What do I need to backup if I use "decrypt_derived"?
1076 This is a script in Debian, intended for mounting /tmp or swap with
1077 a key derived from the master key of an already decrypted device.
1078 If you use this for an device with data that should be persistent,
1079 you need to make sure you either do not lose access to that master
1080 key or have a backup of the data. If you derive from a LUKS
1081 device, a header backup of that device would cover backing up the
1082 master key. Keep in mind that this does not protect against disk
1085 Note: If you recreate the LUKS header of the device you derive from
1086 (using luksFormat), the master key changes even if you use the same
1087 passphrase(s) and you will not be able to decrypt the derived
1088 device with the new LUKS header.
1091 * 6.7 Does a backup compromise security?
1093 Depends on how you do it. However if you do not have one, you are
1094 going to eventually lose your encrypted data.
1096 There are risks introduced by backups. For example if you
1097 change/disable a key-slot in LUKS, a binary backup of the partition
1098 will still have the old key-slot. To deal with this, you have to
1099 be able to change the key-slot on the backup as well, securely
1100 erase the backup or do a filesystem-level backup instead of a binary
1103 If you use dm-crypt, backup is simpler: As there is no key
1104 management, the main risk is that you cannot wipe the backup when
1105 wiping the original. However wiping the original for dm-crypt
1106 should consist of forgetting the passphrase and that you can do
1107 without actual access to the backup.
1109 In both cases, there is an additional (usually small) risk with
1110 binary backups: An attacker can see how many sectors and which
1111 ones have been changed since the backup. To prevent this, use a
1112 filesystem level backup methid that encrypts the whole backup in
1113 one go, e.g. as described above with tar and GnuPG.
1115 My personal advice is to use one USB disk (low value data) or
1116 three disks (high value data) in rotating order for backups, and
1117 either use independent LUKS partitions on them, or use encrypted
1118 backup with tar and GnuPG.
1120 If you do network-backup or tape-backup, I strongly recommend to
1121 go the filesystem backup path with independent encryption, as you
1122 typically cannot reliably delete data in these scenarios,
1123 especially in a cloud setting. (Well, you can burn the tape if it
1124 is under your control...)
1127 * 6.8 What happens if I overwrite the start of a LUKS partition or
1128 damage the LUKS header or key-slots?
1130 There are two critical components for decryption: The salt values
1131 in the header itself and the key-slots. If the salt values are
1132 overwritten or changed, nothing (in the cryptographically strong
1133 sense) can be done to access the data, unless there is a backup
1134 of the LUKS header. If a key-slot is damaged, the data can still
1135 be read with a different key-slot, if there is a remaining
1136 undamaged and used key-slot. Note that in order to make a key-slot
1137 unrecoverable in a cryptographically strong sense, changing about
1138 4-6 bits in random locations of its 128kiB size is quite enough.
1141 * 6.9 What happens if I (quick) format a LUKS partition?
1143 I have not tried the different ways to do this, but very likely you
1144 will have written a new boot-sector, which in turn overwrites the
1145 LUKS header, including the salts, making your data permanently
1146 irretrivable, unless you have a LUKS header backup. You may also
1147 damage the key-slots in part or in full. See also last item.
1150 * 6.10 How do I recover the master key from a mapped LUKS container?
1152 This is typically only needed if you managed to damage your LUKS
1153 header, but the container is still mapped, i.e. "luksOpen"ed. It
1154 also helps if you have a mapped container that you forgot or do not
1155 know a passphrase for (e.g. on a long running server.)
1157 WARNING: Things go wrong, do a full backup before trying this!
1159 WARNING: This exposes the master key of the LUKS container. Note
1160 that both ways to recreate a LUKS header with the old master key
1161 described below will write the master key to disk. Unless you are
1162 sure you have securely erased it afterwards, e.g. by writing it to
1163 an encrypted partition, RAM disk or by erasing the filesystem you
1164 wrote it to by a complete overwrite, you should change the master
1165 key afterwards. Changing the master key requires a full data
1166 backup, luksFormat and then restore of the backup.
1168 First, there is a script by Milan that automatizes the whole
1169 process, except generating a new LUKS header with the old master
1170 key (it prints the command for that though):
1172 http://code.google.com/p/cryptsetup/source/browse/trunk/misc/luks-header-from-active
1174 You can also do this manually. Here is how:
1176 - Get the master key from the device mapper. This is done by the
1177 following command. Substitute c5 for whatever you mapped to:
1179 # dmsetup table --target crypt --showkey /dev/mapper/c5
1181 0 200704 crypt aes-cbc-essiv:sha256
1182 a1704d9715f73a1bb4db581dcacadaf405e700d591e93e2eaade13ba653d0d09
1185 The result is actually one line, wrapped here for clarity. The long
1186 hex string is the master key.
1188 - Convert the master key to a binary file representation. You can
1189 do this manually, e.g. with hexedit. You can also use the tool
1190 "xxd" from vim like this:
1192 echo "a1704d9....53d0d09" | xxd -r -p > <master-key-file>
1194 - Do a luksFormat to create a new LUKS header.
1196 NOTE: If your header is intact and you just forgot the
1197 passphrase, you can just set a new passphrase, see next subitem.
1199 Unmap the device before you do that (luksClose). Then do
1201 cryptsetup luksFormat --master-key-file=<master-key-file> <luks device>
1203 Note that if the container was created with other than the default
1204 settings of the cryptsetup version you are using, you need to give
1205 additional parameters specifying the deviations. If in doubt, try
1206 the script by Milan. It does recover the other parameters as well.
1208 Side note: This is the way the decrypt_derived script gets at the
1209 master key. It just omits the conversion and hashes the master key
1212 - If the header is intact and you just forgot the passphrase, just
1213 set a new passphrase like this:
1215 cryptsetup luksAddKey --master-key-file=<master-key-file> <luks device>
1217 You may want to disable the old one afterwards.
1220 * 6.11 What does the on-disk structure of dm-crypt look like?
1222 There is none. dm-crypt takes a block device and gives encrypted
1223 access to each of its blocks with a key derived from the passphrase
1224 given. If you use a cipher different than the default, you have to
1225 specify that as a parameter to cryptsetup too. If you want to
1226 change the password, you basically have to create a second
1227 encrypted device with the new passphrase and copy your data over.
1228 On the plus side, if you accidentally overwrite any part of a
1229 dm-crypt device, the damage will be limited to the are you
1233 * 6.12 What does the on-disk structure of LUKS look like?
1235 A LUKS partition consists of a header, followed by 8 key-slot
1236 descriptors, followed by 8 key slots, followed by the encrypted
1239 Header and key-slot descriptors fill the first 592 bytes. The
1240 key-slot size depends on the creation parameters, namely on the
1241 number of anti-forensic stripes, key material offset and master
1244 With the default parameters, each key-slot is a bit less than
1245 128kiB in size. Due to sector alignment of the key-slot start,
1246 that means the key block 0 is at offset 0x1000-0x20400, key
1247 block 1 at offset 0x21000-0x40400, and key block 7 at offset
1248 0xc1000-0xe0400. The space to the next full sector address is
1249 padded with zeros. Never used key-slots are filled with what the
1250 disk originally contained there, a key-slot removed with
1251 "luksRemoveKey" or "luksKillSlot" gets filled with 0xff. Due to
1252 2MiB default alignment, start of the data area for cryptsetup 1.3
1253 and later is at 2MiB, i.e. at 0x200000. For older versions, it is
1254 at 0x101000, i.e. at 1'052'672 bytes, i.e. at 1MiB + 4096 bytes
1255 from the start of the partition. Incidentally, "luksHeaderBackup"
1256 for a LUKS container created with default parameters dumps exactly
1257 the first 2MiB (or 1'052'672 bytes for headers created with
1258 cryptsetup versions < 1.3) to file and "luksHeaderRestore" restores
1261 For non-default parameters, you have to figure out placement
1262 yourself. "luksDump" helps. See also next item. For the most common
1263 non-default settings, namely aes-xts-plain with 512 bit key, the
1264 offsets are: 1st keyslot 0x1000-0x3f800, 2nd keyslot
1265 0x40000-0x7e000, 3rd keyslot 0x7e000-0xbd800, ..., and start of
1266 bulk data at 0x200000.
1268 The exact specification of the format is here:
1269 http://code.google.com/p/cryptsetup/wiki/Specification
1272 * 6.13 What is the smallest possible LUKS container?
1274 Note: From cryptsetup 1.3 onwards, alignment is set to 1MB. With
1275 modern Linux partitioning tools that also align to 1MB, this will
1276 result in aligmnet to 2k secors and typical Flash/SSD sectors,
1277 which is highly desirable for a number of reasons. Changing the
1278 alignment is not recomended.
1280 That said, with default parameters, the data area starts at
1281 exactly 2MB offset (at 0x101000 for cryptsetup versions before
1282 1.3). The smallest data area you can have is one sector of 512
1283 bytes. Data areas of 0 bytes can be created, but fail on mapping.
1285 While you cannot put a filesystem into something this small, it may
1286 still be used to contain, for example, key. Note that with current
1287 formatting tools, a partition for a container this size will be
1288 3MiB anyways. If you put the LUKS container into a file (via
1289 losetup and a loopback device), the file needs to be 2097664 bytes
1290 in size, i.e. 2MiB + 512B.
1292 There two ways to influence the start of the data area are key-size
1295 For alignment, you can go down to 1 on the parameter. This will
1296 still leave you with a data-area starting at 0x101000, i.e.
1297 1MiB+4096B (default parameters) as alignment will be rounded up to
1298 the next multiple of 8 (i.e. 4096 bytes) If in doubt, do a dry-run
1299 on a larger file and dump the LUKS header to get actual
1302 For key-size, you can use 128 bit (e.g. AES-128 with CBC), 256 bit
1303 (e.g. AES-256 with CBC) or 512 bit (e.g. AES-256 with XTS mode).
1304 You can do 64 bit (e.g. blowfish-64 with CBC), but anything below
1305 128 bit has to be considered insecure today.
1307 Example 1 - AES 128 bit with CBC:
1309 cryptsetup luksFormat -s 128 --align-payload=8 <device>
1311 This results in a data offset of 0x81000, i.e. 516KiB or 528384
1312 bytes. Add one 512 byte sector and the smallest LUKS container size
1313 with these parameters is 516KiB + 512B or 528896 bytes.
1315 Example 2 - Blowfish 64 bit with CBC (WARNING: insecure):
1317 cryptsetup luksFormat -c blowfish -s 64 --align-payload=8 /dev/loop0
1319 This results in a data offset of 0x41000, i.e. 260kiB or 266240
1320 bytes, with a minimal LUKS conatiner size of 260kiB + 512B or
1324 * 6.14 I think this is overly complicated. Is there an alternative?
1326 Not really. Encryption comes at a price. You can use plain
1327 dm-crypt to simplify things a bit. It does not allow multiple
1328 passphrases, but on the plus side, it has zero on disk description
1329 and if you overwrite some part of a plain dm-crypt partition,
1330 exactly the overwritten parts are lost (rounded up to sector
1334 * 6.15 Can I clone a LUKS container?
1336 You can, but it breaks security, because the cloned container has
1337 the same header and hence the same master key. You cannot change
1338 the master key on a LUKS container, even if you change the
1339 passphrase(s), the master key stays the same. That means whoever
1340 has access to one of the clones can decrypt them all, completely
1341 bypassing the passphrases.
1343 The right way to do this is to first luksFormat the target
1344 container, then to clone the contents of the source container, with
1345 both containers mapped, i.e. decrypted. You can clone the decrypted
1346 contents of a LUKS container in binary mode, although you may run
1347 into secondary issuses with GUIDs in filesystems, partition tables,
1348 RAID-components and the like. These are just the normal problems
1349 binary cloning causes.
1351 Note that if you need to ship (e.g.) cloned LUKS containers with a
1352 default passphrase, that is fine as long as each container was
1353 individually created (and hence has its own master key). In this
1354 case, changing the default passphrase will make it secure again.
1357 7. Interoperability with other Disk Encryption Tools
1360 * 7.1 What is this section about?
1362 Cryptsetup for plain dm-crypt can be used to access a number of
1363 on-disk formats created by tools like loop-aes patched into
1364 losetup. This somtimes works and sometimes does not. This section
1365 collects insights into what works, what does not and where more
1366 information is required.
1368 Additional information may be found in the mailing-list archives,
1369 mentioned at the start of this FAQ document. If you have a
1370 solution working that is not yet documented here and think a wider
1371 audience may be intertested, please email the FAQ maintainer.
1374 * 7.2 loop-aes: General observations.
1376 One problem is that there are different versions of losetup around.
1377 loop-aes is a patch for losetup. Possible problems and deviations
1378 from cryptsetup option syntax include:
1380 - Offsets specifed in bytes (cryptsetup: 512 byte sectors)
1382 - The need to specify an IV offset
1384 - Encryption mode needs specifying (e.g. "-c twofish-cbc-plain")
1386 - Key size needs specifying (e.g. "-s 128" for 128 bit keys)
1388 - Passphrase hash algorithm needs specifying
1390 Also note that because plain dm-crypt and loop-aes format does not
1391 have metadata, autodetection, while feasible in most cases, would
1392 be a lot of work that nobody really wants to do. If you still have
1393 the old set-up, using a verbosity option (-v) on mapping with the
1394 old tool or having a look into the system logs after setup could
1395 give you the information you need.
1398 * 7.3 loop-aes patched into losetup on debian 5.x, kernel 2.6.32
1400 In this case, the main problem seems to be that this variant of
1401 losetup takes the offset (-o option) in bytes, while cryptsetup
1402 takes it in sectors of 512 bytes each. Example: The losetupp
1405 losetup -e twofish -o 2560 /dev/loop0 /dev/sdb1
1406 mount /dev/loop0 mountpoint
1410 cryptsetup create -c twofish -o 5 --skip 5 e1 /dev/sdb1
1411 mount /dev/mapper/e1 mountpoint
1414 * 7.4 loop-aes with 160 bit key
1416 This seems to be sometimes used with twofish and blowfish and
1417 represents a 160 bit ripemed160 hash output padded to 196 bit key
1418 length. It seems the corresponding options for cryptsetup are
1420 --cipher twofish-cbc-null -s 192 -h ripemd160:20
1423 8. Issues with Specific Versions of cryptsetup
1426 * 8.1 When using the create command for plain dm-crypt with
1427 cryptsetup 1.1.x, the mapping is incompatible and my data is not
1430 With cryptsetup 1.1.x, the distro maintainer can define different
1431 default encryption modes for LUKS and plain devices. You can check
1432 these compiled-in defaults using "cryptsetup --help". Moreover, the
1433 plain device default changed because the old IV mode was
1434 vulnerable to a watermarking attack.
1436 If you are using a plain device and you need a compatible mode, just
1437 specify cipher, key size and hash algorithm explicitly. For
1438 compatibility with cryptsetup 1.0.x defaults, simple use the
1441 cryptsetup create -c aes-cbc-plain -s 256 -h ripemd160 <name> <dev>
1443 LUKS stores cipher and mode in the metadata on disk, avoiding this
1447 * 8.2 cryptsetup on SLED 10 has problems...
1449 SLED 10 is missing an essential kernel patch for dm-crypt, which
1450 is broken in its kernel as a result. There may be a very old
1451 version of cryptsetup (1.0.x) provided by SLED, which should also
1452 not be used anymore as well. My advice would be to drop SLED 10.
1454 A. Contributors In no particular order: