8 6. Backup and Data Recovery
9 7. Issues with Specific Versions of cryptsetup
18 This is the FAQ (Frequently Asked Questions) for cryptsetup. It
19 covers Linux disk encryption with plain dm-crypt (one passphrase,
20 no management, no descriptor on disk) and LUKS (multiple user keys
21 with one master key, anti-forensics, descriptor block at start of
22 device, ...). The latest version should usually be available at
23 http://code.google.com/p/cryptsetup/wiki/FrequentlyAskedQuestions
25 ATTENTION: If you are going to read just one thing, make it the
26 section on Backup and Data Recovery. By far the most questions on
27 the cryptsetup mailing list are from people that just managed to
28 somehow format or overwrite the start of their LUKS partitions. In
29 most cases, there is nothing that can be done to help these poor
30 souls recover their data. Make sure you understand the problem and
31 limitations imposed by the LUKS security model BEFORE you face such
37 Current FAQ maintainer is Arno Wagner <arno@wagner.name>. Other
38 contributors are listed at the end. If you want to contribute, send
39 your article, including a descriptive headline, to the maintainer,
40 or the dm-crypt mailing list with something like "FAQ ..." in the
41 subject. Please note that by contributing to this FAQ, you accept
42 the license described below.
44 This work is under the "Attribution-Share Alike 3.0 Unported"
45 license, which means distribution is unlimited, you may create
46 derived works, but attributions to original authors and this
47 license statement must be retained and the derived work must be
48 under the same license. See
49 http://creativecommons.org/licenses/by-sa/3.0/ for more details of
52 Side note: I did text license research some time ago and I think
53 this license is best suited for the purpose at hand and creates the
57 * Where is the project website?
59 There is the project website at http://code.google.com/p/cryptsetup/
60 Please do not post questions there, nobody will read them. Use
61 the mailing-list instead.
64 * Is there a mailing-list?
66 Instructions on how to subscribe to the mailing-list are at on the
67 project website. People are generally helpful and friendly on the
70 The question of how to unsubscribe from the list does crop up
71 sometimes. For this you need your list management URL, which is
72 sent to you initially and once at the start of each month. Go to
73 the URL mentioned in the email and select "unsubscribe". This page
74 also allows you to request a password reminder.
76 Alternatively, you can send an Email to dm-crypt-request@saout.de
77 with just the word "help" in the subject or message body. Make sure
78 to send it from your list address.
84 * Can I encrypt an already existing, non-empty partition to use
87 There is no converter, and it is not really needed. The way to do
88 this is to make a backup of the device in question, securely wipe
89 the device (as LUKS device initialization does not clear away old
90 data), do a luksFormat, optionally overwrite the encrypted device,
91 create a new filesystem and restore your backup on the now
92 encrypted device. Also refer to sections "Security Aspects" and
93 "Backup and Data Recovery".
95 For backup, plain GNU tar works well and backs up anything likely
96 to be in a filesystem.
99 * How do I use LUKS with a loop-device?
101 Just the same as with any block device. If you want, for example,
102 to use a 100MiB file as LUKS container, do something like this:
104 head -c 100M /dev/zero > luksfile # create empty file
105 losetup /dev/loop0 luksfile # map luksfile to /dev/loop0
106 cryptsetup luksFormat /dev/loop0 # create LUKS on the loop device
108 Afterwards just use /dev/loop0 as a you would use a LUKS partition.
109 To unmap the file when done, use "losetup -d /dev/loop0".
112 * When I add a new key-slot to LUKS, it asks for a passphrase but
113 then complains about there not being a key-slot with that
116 That is as intended. You are asked a passphrase of an existing
117 key-slot first, before you can enter the passphrase for the new
118 key-slot. Otherwise you could break the encryption by just adding a
119 new key-slot. This way, you have to know the passphrase of one of
120 the already configured key-slots in order to be able to configure a
124 * How do I read a dm-crypt key from file?
126 Note that the file will still be hashed first, just like keyboard
127 input. Use the --key-file option, like this:
129 cryptsetup create --key-file keyfile e1 /dev/loop0
132 * How do I read a LUKS slot key from file?
134 What you really do here is to read a passphrase from file, just as
135 you would with manual entry of a passphrase for a key-slot. You can
136 add a new passphrase to a free key-slot, set the passphrase of an
137 specific key-slot or put an already configured passphrase into a
138 file. In the last case make sure no trailing newline (0x0a) is
139 contained in the key file, or the passphrase will not work because
140 the whole file is used as input.
142 To add a new passphrase to a free key slot from file, use something
145 cryptsetup luksAddKey /dev/loop0 keyfile
147 To add a new passphrase to a specific key-slot, use something like
150 cryptsetup luksAddKey --key-slot 7 /dev/loop0 keyfile
152 To supply a key from file to any LUKS command, use the --key-file
153 option, e.g. like this:
155 cryptsetup luksOpen --key-file keyfile /dev/loop0 e1
158 * How do I read the LUKS master key from file?
160 The question you should ask yourself first, is why you would want
161 to do this. The only legitimate reason I can think of is if you
162 want to have two LUKS devices with the same master key. Even then,
163 I think it would be preferable to just use key-slots with the same
164 passphrase, or to use plain dm-crypt instead. If you really have a
165 good reason, please tell me. If I am convinced, I will add how to
169 * What are the security requirements for a key read from file?
171 A file-stored key or passphrase has the same security requirements
172 as one entered interactively, however you can use random bytes and
173 thereby use bytes you cannot type on the keyboard. You can use any
174 file you like as key file, for example a plain text file with a
175 human readable passphrase. To generate a file with random bytes,
176 use something like this:
178 head -c 256 /dev/random > keyfile
181 * If I map a journaled file system using dm-crypt/LUKS, does it
182 still provide its usual transactional guarantees?
184 As far as I know it does (but I may be wrong), but please note that
185 these "guarantees" are far weaker than they appear to be. For
186 example, you not not get a hard flush to disk surface even on a
187 call to fsync. In addition, the HDD itself may do independent
188 write reordering. Some other things can go wrong as well. The
189 filesystem developers are aware of these problems and typically
190 can make it work anyways. That said, dm-crypt/LUKS should not make
193 Personally, I have several instances of ext3 on dm-crypt and have
194 not noticed any specific problems.
196 Update: I did run into frequent small freezes (1-2 sec) when putting
197 a vmware image on ext3 over dm-crypt. This does indicate that the
198 transactional guarantees are in place, but at a cost. When I went
199 back to ext2, the problem went away. This also seems to have gotten
200 better with kernel 2.6.36 and the reworking of filesystem flush
201 locking. Kernel 2.6.37 is expected to improve this even further.
204 * Can I use LUKS or cryptsetup with a more secure (external) medium
205 for key storage, e.g. TPM or a smartcard?
207 Yes, see the answers on using a file-supplied key. You do have to
208 write the glue-logic yourself though. Basically you can have
209 cryptsetup read the key from STDIN and write it there with your
210 own tool that in turn gets the key from the more secure key
214 * Can I resize a dm-crypt or LUKS partition?
216 Yes, you can, as neither dm-crypt nor LUKS stores partition size.
217 Whether you should is a different question. Personally I recommend
218 backup, recreation of the encrypted partition with new size,
219 recreation of the filesystem and restore. This gets around the
220 tricky business of resizing the filesystem. The backup is really
221 non-optional here, as a lot can go wrong, resulting in partial or
222 complete data loss. Using something like gparted to resize an
223 encrypted partition is slow, but pretty safe and should be fine.
224 This will not change the size of the filesystem hidden under the
227 You also need to be aware of size-based limitations. The one
228 currently relevant is that aes-xts-plain should not be used for
229 encrypted container sizes larger than 2TiB. Use aes-xts-plain64
236 * My dm-crypt/LUKS mapping does not work! What general steps are
237 there to investigate the problem?
239 If you get a specific error message, investigate what it claims
240 first. If not, you may want to check the following things.
242 - Check that "/dev", including "/dev/mapper/control" is there. If it is
243 missing, you may have a problem with the "/dev" tree itself or you
244 may have broken udev rules.
246 - Check that you have the device mapper and the crypt target in your kernel.
247 The output of "dmsetup targets" should list a "crypt" target. If it
248 is not there or the command fails, add device mapper and
249 crypt-target to the kernel.
251 - Check that the hash-functions and ciphers you want to use are in the kernel.
252 The output of "cat /proc/crypto" needs to list them.
255 * My dm-crypt mapping suddenly stopped when upgrading cryptsetup.
257 The default cipher, hash or mode may have changed (the mode changed
258 from 1.0.x to 1.1.x). See under "Issues With Specific Versions of
262 * When I call cryptsetup from cron/CGI, I get errors about unknown
265 If you get errors about unknown parameters or the like that are not
266 present when cryptsetup is called from the shell, make sure you
267 have no older version of cryptsetup on your system that then gets
268 called by cron/CGI.For example some distributions install
269 cryptsetup into /usr/sbin, while a manual install could go to
270 /usr/local/sbin. As a debugging aid, call "cryptsetup --version"
271 from cron/CGI or the non-shell mechanism to be sure you have the
275 * Unlocking a LUKS device takes very long. Why?
277 The iteration time for a key-slot (see Section 5 for an explanation
278 what iteration does) is calculated when setting a passphrase. By
279 default it is 1 second on the machine where the passphrase is set.
280 If you set a passphrase on a fast machine and then unlock it on a
281 slow machine, the unlocking time can be much longer. Also take into
282 account that up to 8 key-slots have to be tried in order to find the
285 If this is problem, you can add another key-slot using the slow
286 machine with the same passphrase and then remove the old key-slot.
287 The new key-slot will have an iteration count adjusted to 1 second
288 on the slow machine. Use luksKeyAdd and then luksKillSlot or
291 However, this operation will not change volume key iteration count
292 (MK iterations in output of "cryptsetup luksDump"). In order to
293 change that, you will have to backup the data in the LUKS
294 container, luksFormat on the slow machine and restore the data.
295 Note that in the original LUKS specification this value was fixed
296 to 10, but it is now derived from the PBKDF2 benchmark as well and
297 set to iterations in 0.125 sec or 1000, whichever is larger.
300 * "blkid" sees a LUKS UUID and an ext2/swap UUID on the same device.
303 Some old versions of cryptsetup have a bug where the header does
304 not get completely wiped during LUKS format and an older ext2/swap
305 signature remains on the device. This confuses blkid.
307 Fix: Wipe the unused header areas by doing a backup and restore of
308 the header with cryptsetup 1.1.x:
310 cryptsetup luksHeaderBackup --header-backup-file <file> <device>
311 cryptsetup luksHeaderRestore --header-backup-file <file> <device>
313 If you cannot use a 1.1.x cryptsetup, you can also do a manual wipe
314 of the area in question with the command below. Be very, VERY,
315 careful and make sure to do a backup of the header before. If you
316 get this wrong, your device may become permanently inaccessible.
318 dd if=/dev/zero of=<device> bs=512 seek=2 count=6
321 * cryptsetup segfaults on Gentoo amd64 hardened ...
323 There seems to be some inteference between the hardening and and
324 the way cryptsetup benchmarks PBKDF2. The solution to this is
325 currently not quite clear for an encrypted root filesystem. For
326 other uses, you can apparently specify USE="dynamic" as compile
327 flag, see http://bugs.gentoo.org/show_bug.cgi?id=283470
333 * Can a bad RAM module cause problems?
335 LUKS and dm-crypt can give the RAM quite a workout, especially when
336 combined with software RAID. In particular the combination RAID5 +
337 LUKS + XFS seems to uncover RAM problems that never caused obvious
338 problems before. Symptoms vary, but often the problem manifest
339 itself when copying large amounts of data, typically several times
340 larger than your main memory.
342 Side note: One thing you should always do on large data movements is
343 to run a verify, for example with the "-d" option of "tar" or by
344 doing a set of MD5 checksums on the source or target with
346 find . -type f -exec md5sum \{\} \; > checksum-file
348 and then a "md5sum -c checksum-file" on the other side. If you get
349 mismatches here, RAM is the primary suspect. A lesser suspect is
350 an overclocked CPU. I have found countless hardware problems in
351 verify runs after copying or making backups. Bit errors are much
352 more common than most people think.
354 Some RAM issues are even worse and corrupt structures in one of the
355 layers. This typically results in lockups, CPU state dumps in the
356 system logs, kernel panic or other things. It is quite possible to
357 have the problem with an encrypted device, but not with an
358 otherwise the same unencrypted device. The reason for that is that
359 encryption has an error amplification property: You flip one bit
360 in an encrypted data block, and the decrypted version has half of
361 its bits flipped. This is an important security property for modern
362 ciphers. With the usual modes in cryptsetup (CBC, ESSIV, XTS), you
363 get up to a completely changed 512 byte block per bit error. A
364 corrupt block causes a lot more havoc than the occasionally
365 flipped single bit and can result various obscure errors.
367 Note however that a verify run on copying between encrypted or
368 unencrypted devices can also show you corruption when the copying
369 itself did not report any problems. If you find defect RAM, assume
370 all backups and copied data to be suspect, unless you did a verify.
375 First you should know that overclocking often makes memory problems
376 worse. So if you overclock (which I strongly recommend against in a
377 system holding data that has some worth), run the tests with the
380 There are two good options. One is Memtest86+ and the other is
381 "memtester" by Charles Cazabon. Memtest86+ requires a reboot and
382 then takes over the machine, while memtester runs from a
383 root-shell. Both use different testing methods and I have found
384 problems fast with each one that the other needed long to find. I
385 recommend running the following procedure until the first error is
388 - Run Memtest86+ for one cycle
389 - Run memterster for one cycle (shut down as many other applications as possible)
390 - Run Memtest86+ for 24h or more
391 - Run memtester for 24h or more
392 If all that does not produce error messages, your RAM may be sound,
393 but I have had one weak bit that Memtest86+ needed around 60 hours
394 to find. If you can reproduce the original problem reliably, a good
395 additional test may be to remove half of the RAM (if you have more
396 than one module) and try whether the problem is still there and if
397 so, try with the other half. If you just have one module, get a
398 different one and try with that. If you do overclocking, reduce
399 the settings to the most conservative ones available and try with
406 * Should I initialize (overwrite) a new LUKS/dm-crypt partition?
408 If you just create a filesystem on it, most of the old data will
409 still be there. If the old data is sensitive, you should overwrite
410 it before encrypting. In any case, not initializing will leave the
411 old data there until the specific sector gets written. That may
412 enable an attacker to determine how much and where on the
413 partition data was written. If you think this is a risk, you can
414 prevent this by overwriting the encrypted device (here assumed to
415 be named "e1") with zeros like this:
417 dd_rescue -w /dev/zero /dev/mapper/e1
419 or alternatively with one of the following more standard commands:
421 cat /dev/zero > /dev/mapper/e1
422 dd if=/dev/zero of=/dev/mapper/e1
425 * How do I securely erase a LUKS (or other) partition?
427 For LUKS, if you are in a desperate hurry, overwrite the first few
428 kilobytes of the LUKS partition. This erases the master key salt
429 and makes access impossible. However a LUKS header backup or full
430 backup will still grant access to most or all data, so make sure
431 that an attacker does not have access to backups or destroy them as
434 To do this right, overwrite the whole LUKS partition with a single
435 pass of zeros. This is enough for current HDDs. For SSDs you may
436 want to erase the whole drive several times to be sure data is not
437 retained by wear leveling. This is possibly still insecure as SSD
438 technology is not fully understood in this regard. Still, due to
439 the anti-forensic properties of the LUKS key-slots, a single
440 overwrite of an SSD could be enough. If in doubt, use physical
441 destruction in addition. Keep in mind to also erase all backups.
443 Example for a zero-overwrite erase of partition sda10 done with
446 dd_rescue -w /dev/zero /dev/sda10
449 * How do I securely erase a backup of a LUKS partition or header?
451 That depends on the medium it is stored on. For HDD and SSD, use
452 overwrite with zeros. For an SSD, you may want to overwrite the
453 complete SSD several times and use physical destruction in addition,
454 see last item. Treat USB flash drives the same as SSDs. For
455 re-writable CD/DVD, a single overwrite should also be enough, due
456 to the anti-forensic properties of the LUKS keyslots. For
457 write-once media, use physical destruction. For low security
458 requirements, just cut the CD/DVD into several parts. For high
459 security needs, shred or burn the medium. If your backup is on
460 magnetic tape, I advise physical destruction by shredding or
461 burning. The problem with magnetic tape is that it has a higher
462 dynamic range than HDDs and older data may well be recoverable
463 after overwrites. Also write-head alignment issues can lead to
464 data not actually being deleted at all during overwrites.
467 * What about backup? Does it compromise security?
469 That depends. See next section.
472 * Why is all my data gone if I overwrite the LUKS header?
474 Overwriting the LUKS header in part or in full is the most common
475 reason why access to LUKS containers is lost permanently.
476 Overwriting can be done in a number of fashions, like creating a
477 new filesystem on the raw LUKS partition, making the raw partition
478 part of a raid array and just writing to the raw partition.
480 The LUKS header contains a 256 bit "salt" value and without that no
481 decryption is possible. While the salt is not secret, it is
482 key-grade material and cannot be reconstructed. This is a
483 cryptographically strong "cannot". From observations on the
484 cryptsetup mailing-list, people typically go though the usual
485 stages of grief (Denial, Anger, Bargaining, Depression, Acceptance)
486 when this happens to them. Observed times vary between 1 day and 2
487 weeks to complete the cycle. Seeking help on the mailing-list is
488 fine. Even if we usually cannot help with getting back your data,
489 most people found the feedback comforting.
491 If your header does not contain an intact salt, best go directly
492 to the last one ("Acceptance") and think about what to do now.
493 There is one exception that I know of: If your LUKS container is
494 still open, then it may be possible to extract the master key from
495 the running system. Ask on the mailing-list on how to do that and
496 make sure nobody switches off the machine.
501 A salt is a random key-grade value added to the passphrase before
502 it is processed. It is not kept secret. The reason for using salts
503 is as follows: If an attacker wants to crack the password for a
504 single LUKS container, then every possible passphrase has to be
505 tried. Typically an attacker will not try every binary value, but
506 will try words and sentences from a dictionary.
508 If an attacker wants to attack several LUKS containers with the
509 same dictionary, then a different approach makes sense: Compute the
510 resulting slot-key for each dictionary element and store it on
511 disk. Then the test for each entry is just the slow unlocking with
512 the slot key (say 0.00001 sec) instead of calculating the slot-key
513 first (1 sec). For a single attack, this does not help. But if you
514 have more than one container to attack, this helps tremendously,
515 also because you can prepare your table before you even have the
516 container to attack! The calculation is also very simple to
517 parallelize. You could, for example, use the night-time unused CPU
518 power of your desktop PCs for this.
520 This is where the salt comes in. If the salt is combined with the
521 passphrase (in the simplest form, just appended to it), you
522 suddenly need a separate table for each salt value. With a
523 reasonably-sized salt value (256 bit, e.g.) this is quite
527 * Is LUKS secure with a low-entropy (bad) passphrase?
529 This needs a bit of theory. The quality of your passphrase is
530 directly related to its entropy (information theoretic, not
531 thermodynamic). The entropy says how many bits of "uncertainty" or
532 "randomness" are in you passphrase. In other words, that is how
533 difficult guessing the passphrase is.
535 Example: A random English sentence has about 1 bit of entropy per
536 character. A random lowercase (or uppercase) character has about
539 Now, if n is the number of bits of entropy in your passphrase and t
540 is the time it takes to process a passphrase in order to open the
541 LUKS container, then an attacker has to spend at maximum
543 attack_time_max = 2^n * t
545 time for a successful attack and on average half that. There is no
546 way getting around that relationship. However, there is one thing
547 that does help, namely increasing t, the time it takes to use a
548 passphrase, see next FAQ item.
550 Still, if you want good security, a high-entropy passphrase is the
551 only option. Use at least 64 bits for secret stuff. That is 64
552 characters of English text (but only if randomly chosen) or a
553 combination of 12 truly random letters and digits.
555 For passphrase generation, do not use lines from very well-known
556 texts (religious texts, Harry potter, etc.) as they are to easy to
557 guess. For example, the total Harry Potter has about 1'500'000
558 words (my estimation). Trying every 64 character sequence starting
559 and ending at a word boundary would take only something like 20
560 days on a single CPU and is entirely feasible.
562 On the other hand, choosing 1.5 lines from, say, the Wheel of Time
563 is in itself not more secure, but the book selection adds quite a
564 bit of entropy. (Now that I have mentioned it here, don't use tWoT
565 either!) If you add 2 or 3 typos or switch some words around, then
566 this is good passphrase material.
569 * What is "iteration count" and why is decreasing it a bad idea?
571 Iteration count is the number of PBKDF2 iterations a passphrase is
572 put through before it is used to unlock a key-slot. Iterations are
573 done with the explicit purpose to increase the time that it takes
574 to unlock a key-slot. This provides some protection against use of
575 low-entropy passphrases.
577 The idea is that an attacker has to try all possible passphrases.
578 Even if the attacker knows the passphrase is low-entropy (see last
579 item), it is possible to make each individual try take longer. The
580 way to do this is to repeatedly hash the passphrase for a certain
581 time. The attacker then has to spend the same time (given the same
582 computing power) as the user per try. With LUKS, the default is 1
583 second of PBKDF2 hashing.
585 Example 1: Lets assume we have a really bad passphrase (e.g. a
586 girlfriends name) with 10 bits of entropy. With the same CPU, an
587 attacker would need to spend around 500 seconds on average to
588 break that passphrase. Without iteration, it would be more like
589 0.0001 seconds on a modern CPU.
591 Example 2: The user did a bit better and has 32 chars of English
592 text. That would give use about 32 bits of entropy. With 1 second
593 iteration, that means an attacker on the same CPU needs around 136
594 years. That is pretty impressive for such a weak passphrase.
595 Without the iterations, it would be more like 50 days on a modern
596 CPU, and possibly far less.
598 In addition, the attacker can both parallelize and use special
599 hardware like GPUs to speed up the attack. The attack can also
600 happen quite some time after the luksFormat operation and CPUs can
601 have become faster and cheaper. For that reason you want a bit of
602 extra security. Anyways, in Example 1 your are screwed. In example
603 2, not necessarily. Even if the attack is faster, it still has a
604 certain cost associated with it, say 10000 EUR/USD with iteration
605 and 1 EUR/USD without iteration. The first can be prohibitively
606 expensive, while the second is something you try even without
607 solid proof that the decryption will yield something useful.
609 The numbers above are mostly made up, but show the idea. Of course
610 the best thing is to have a high-entropy passphrase.
612 Would a 100 sec iteration time be even better? Yes and no.
613 Cryptographically it would be a lot better, namely 100 times better.
614 However, usability is a very important factor for security
615 technology and one that gets overlooked surprisingly often. For
616 LUKS, if you have to wait 2 minutes to unlock the LUKS container,
617 most people will not bother and use less secure storage instead. It
618 is better to have less protection against low-entropy passphrases
619 and people actually use LUKS, than having them do without
620 encryption altogether.
622 Now, what about decreasing the iteration time? This is generally a
623 very bad idea, unless you know and can enforce that the users only
624 use high-entropy passphrases. If you decrease the iteration time
625 without ensuring that, then you put your users at increased risk,
626 and considering how often LUKS containers are unlocked in a
627 typical work-flow, you do so without a good reason. Don't do it.
628 The iteration time is already low enough that some users will
629 still chose passphrases with entropy low enough that they are
630 vulnerable. Lowering it even further increases this danger
634 * Is LUKS with default parameters less secure on a slow CPU?
636 Unfortunately, yes. However the only aspect affected is the
637 protection for low-entropy passphrase or master-key. All other
638 security aspects are independent of CPU speed.
640 The master key is less critical, as you really have to work at it
641 to give it low entropy. One possibility is to supply the master key
642 yourself. If that key is low-entropy, then you get what you
643 deserve. The other known possibility is to use /dev/urandom for
644 key generation in an entropy-startved situation (e.g. automatic
645 installation on an embedded device without network and other entropy
648 For the passphrase, don't use a low-entropy passphrase. If your
649 passphrase is good, then a slow CPU will not matter. If you insist
650 on a low-entropy passphrase on a slow CPU, use something like
651 "--iter-time=10" or higher and wait a long time on each LUKS unlock
652 and pray that the attacker does not find out in which way exactly
653 your passphrase is low entropy. This also applies to low-entropy
654 passphrases on fast CPUs. Technology can do only so much to
655 compensate for problems in front of the keyboard.
658 * Why was the default aes-cbc-plain replaced with aes-cbc-essiv?
660 The problem is that cbc-plain has a fingerprint vulnerability, where
661 a specially crafted file placed into the crypto-container can be
662 recognized from the outside. The issue here is that for cbc-plain
663 the initialization vector (IV) is the sector number. The IV gets
664 XORed to the first data chunk of the sector to be encrypted. If you
665 make sure that the first data block to be stored in a sector
666 contains the sector number as well, the first data block to be
667 encrypted is all zeros and always encrypted to the same ciphertext.
668 This also works if the first data chunk just has a constant XOR
669 with the sector number. By having several shifted patterns you can
670 take care of the case of a non-power-of-two start sector number of
673 This mechanism allows you to create a pattern of sectors that have
674 the same first ciphertext block and signal one bit per sector to the
675 outside, allowing you to e.g. mark media files that way for
676 recognition without decryption. For large files this is a
677 practical attack. For small ones, you do not have enough blocks to
678 signal and take care of different file starting offsets.
680 In order to prevent this attack, the default was changed to
681 cbc-essiv. ESSIV uses a keyed hash of the sector number, with the
682 encryption key as key. This makes the IV unpredictable without
683 knowing the encryption key and the watermarking attack fails.
686 * Are there any problems with "plain" IV? What is "plain64"?
688 First, "plain" and "plain64" are both not safe to use with CBC, see
691 However there are modes, like XTS, that are secure with "plain" IV.
692 The next limit is that "plain" is 64 bit, with the upper 32 bit set
693 to zero. This means that on volumes larger than 2TiB, the IV
694 repeats, creating a vulnerability that potentially leaks some
695 data. To avoid this, use "plain64", which uses the full sector
696 number up to 64 bit. Note that "plain64" requires a kernel >=
697 2.6.33. Also note that "plain64" is backwards compatible for
698 volume sizes <= 2TiB, but not for those > 2TiB. Finally, "plain64"
699 does not cause any performance penalty compared to "plain".
702 * What about XTS mode?
704 XTS mode is potentially even more secure than cbc-essiv (but only if
705 cbc-essiv is insecure in your scenario). It is a NIST standard and
706 used, e.g. in Truecrypt. At the moment, if you want to use it, you
707 have to specify it manually as "aes-xts-plain", i.e.
709 cryptsetup -c aes-xts-plain luksFormat <device>
711 For volumes >2TiB and kernels >= 2.6.33 use "plain64" (see FAQ
712 item on "plain" and "plain64"):
714 cryptsetup -c aes-xts-plain64 luksFormat <device>
716 There is a potential security issue with XTS mode and large blocks.
717 LUKS and dm-crypt always use 512B blocks and the issue does not
721 6. Backup and Data Recovery
724 * Does a backup compromise security?
726 Depends on how you do it. First, a backup is non-optional with
727 encrypted data just the same way it is with non-encrypted data.
728 Disks do break and they do not care whether they make plain or
729 encrypted data inaccessible.
731 However there are risks introduced by backups. For example if you
732 change/disable a key-slot in LUKS, a binary backup of the partition
733 will still have the old key-slot. To deal with this, you have to
734 be able to change the key-slot on the backup as well, or use a
735 different set-up. One option is to have a different passphrase on
736 the backup and to make the backup with both containers open.
737 Another one is to make a backup of the original, opened container to
738 a single file, e.g. with tar, and to encrypt that file with
739 public-key-cryptography, e.g. with GnuPG. You can then keep the
740 secret key in a safe place, because it is only used to decrypt a
741 backup. The key the backup is encrypted with can be stored without
742 special security measures, as long as an attacker cannot replace
745 If you use dm-crypt, backup is simpler: As there is no key
746 management, the main risk is that you cannot wipe the backup when
747 wiping the original. However wiping the original for dm-crypt
748 should consist of forgetting the passphrase and that you can do
749 without actual access to the backup.
751 In both cases, there is an additional (usually small) risk: An
752 attacker can see how many sectors and which ones have been changed
753 since the backup. This is not possible with the public-key method
756 My personal advice is to use one USB disk (low value date) or three
757 disks (high value data) in rotating order for backups, and either
758 use different passphrases or keep them easily accessible in case
759 you need to disable a key-slot. If you do network-backup or
760 tape-backup, I strongly recommend to go the public-key path,
761 especially as you typically cannot reliably delete data in these
762 scenarios. (Well, you can burn the tape if it is under your
766 * What happens if I overwrite the start of a LUKS partition or
767 damage the LUKS header or key-slots?
769 There are two critical components for decryption: The salt values
770 in the header itself and the key-slots. If the salt values are
771 overwritten or changed, nothing (in the cryptographically strong
772 sense) can be done to access the data, unless there is a backup of
773 the LUKS header. If a key-slot is damaged, the data can still be
774 read with a different key-slot, if there is a remaining undamaged
775 and used key-slot. Note that in order to make a key-slot
776 unrecoverable in a cryptographically strong sense, changing about
777 4-6 bits in random locations of its 128kiB size is quite enough.
780 * What happens if I (quick) format a LUKS partition?
782 I have not tried the different ways to do this, but very likely you
783 will have written a new boot-sector, which in turn overwrites the
784 LUKS header, including the salts. You may also damage the key-slots
785 in part or in full. See also last item.
788 * What does the on-disk structure of dm-crypt look like?
790 There is none. dm-crypt takes a block device and gives encrypted
791 access to each of its blocks with a key derived from the passphrase
792 given. If you use a cipher different than the default, you have to
793 specify that as a parameter to cryptsetup too. If you want to
794 change the password, you basically have to create a second
795 encrypted device with the new passphrase and copy your data over.
796 On the plus side, if you accidentally overwrite any part of a
797 dm-crypt device, the damage will be limited to the are you
801 * What does the on-disk structure of LUKS look like?
803 A LUKS partition consists of a header, followed by 8 key-slot
804 descriptors, followed by 8 key slots, followed by the encrypted
807 Header and key-slot descriptors fill the first 592 bytes. The
808 key-slot size depends on the creation parameters, namely on the
809 number of anti-forensic stripes and on key block alignment.
811 With 4000 stripes (the default), each key-slot is a bit less than
812 128kiB in size. Due to sector alignment of the key-slot start,
813 that means the key block 0 is at offset 0x1000-0x20400, key block
814 1 at offset 0x21000-0x40400, and key block 7 at offset
815 0xc1000-0xe0400. The space to the next full sector address is
816 padded with zeros. Never used key-slots are filled with what the
817 disk originally contained there, a key-slot removed with
818 "luksRemoveKey" or "luksKillSlot" gets filled with 0xff. Start of
819 bulk data (with the default 4000 stripes and 8 key-slots) is at
820 0x101000, i.e. at 1'052'672 bytes, i.e. at 1MiB + 4096 bytes from
821 the start of the partition. This is also the value given by command
822 "luksDump" with "Payload offset: 2056", just multiply by the sector
823 size (512 bytes). Incidentally, "luksHeaderBackup" dumps exactly
824 the first 1'052'672 bytes to file and "luksHeaderRestore" restores
827 The exact specification of the format is here:
828 http://code.google.com/p/cryptsetup/wiki/Specification
831 * How do I backup a LUKS header?
833 While you could just copy the appropriate number of bytes from the
834 start of the LUKS partition, the best way is to use command option
835 "luksHeaderBackup" of cryptsetup. This protects also against errors
836 when non-standard parameters have been used in LUKS partition
840 cryptsetup luksHeaderBackup --header-backup-file h_bak /dev/mapper/c1
843 * How do I backup a LUKS partition?
845 You do a sector-image of the whole partition. This will contain the
846 LUKS header, the keys-slots and the data ares. It can be done
847 under Linux e.g. with dd_rescue (for a direct image copy) and with
848 "cat" or "dd". Example:
850 cat /dev/sda10 > sda10.img
851 dd_rescue /dev/sda10 sda10.img
853 You can also use any other backup software that is capable of making
854 a sector image of a partition. Note that compression is
855 ineffective for encrypted data, hence it does not sense to use it.
858 * Do I need a backup of the full partition? Would the header and
859 key-slots not be enough?
861 Backup protects you against two things: Disk loss or corruption and
862 user error. By far the most questions on the dm-crypt mailing list
863 about how to recover a damaged LUKS partition are related to user
864 error. For example, if you create a new filesystem on a LUKS
865 partition, chances are good that all data is lost permanently.
867 For this case, a header+key-slot backup would often be enough. But
868 keep in mind that a HDD has roughly a failure risk of 5% per year.
869 It is highly advisable to have a complete backup to protect against
873 * Are there security risks from a backup of the LUKS header or a
874 whole LUKS partition?
876 Yes. One risk is that if you remove access rights for specific
877 key-slots by deleting their contents, the data can still be
878 accessed with invalidated passphrase and the backup. The other risk
879 is that if you erase a LUKS partition, a backup could still grant
880 access, especially if you only erased the LUKS header and not the
884 * I think this is overly complicated. Is there an alternative?
886 Yes, you can use plain dm-crypt. It does not allow multiple
887 passphrases, but on the plus side, it has zero on disk description
888 and if you overwrite some part of a plain dm-crypt partition,
889 exactly the overwritten parts are lost (rounded up to sector
893 7. Issues with Specific Versions of cryptsetup
896 * When using the create command for plain dm-crypt with cryptsetup
897 1.1.x, the mapping is incompatible and my data is not accessible
900 With cryptsetup 1.1.x, the distro maintainer can define different
901 default encryption modes for LUKS and plain devices. You can check
902 these compiled-in defaults using "cryptsetup --help". Moreover, the
903 plain device default changed because the old IV mode was
904 vulnerable to a watermarking attack.
906 If you are using a plain device and you need a compatible mode, just
907 specify cipher, key size and hash algorithm explicitly. For
908 compatibility with cryptsetup 1.0.x defaults, simple use the
911 cryptsetup create -c aes-cbc-plain -s 256 -h ripemd160 <name> <device>
913 LUKS stores cipher and mode in the metadata on disk, avoiding this
917 * cryptsetup on SLED 10 has problems...
919 SLED 10 is missing an essential kernel patch for dm-crypt, which
920 is broken in its kernel as a result. There may be a very old
921 version of cryptsetup (1.0.x) provided by SLED, which should also
922 not be used anymore as well. My advice would be to drop SLED 10.
924 A. Contributors In no particular order: