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 LUKS?
86 There is no converter, and it is not really needed. The way to do
87 this is to make a backup of the device in question, securely wipe
88 the device (as LUKS device initialization does not clear away old
89 data), do a luksFormat, optionally overwrite the encrypted device,
90 create a new filesystem and restore your backup on the now
91 encrypted device. Also refer to sections "Security Aspects" and
92 "Backup and Data Recovery".
94 For backup, plain GNU tar works well and backs up anything likely
95 to be in a filesystem.
98 * How do I use LUKS with a loop-device?
100 Just the same as with any block device. If you want, for example,
101 to use a 100MiB file as LUKS container, do something like this:
103 head -c 100M /dev/zero > luksfile # create empty file
104 losetup /dev/loop0 luksfile # map luksfile to /dev/loop0
105 cryptsetup luksFormat /dev/loop0 # create LUKS on loop device
107 Afterwards just use /dev/loop0 as a you would use a LUKS partition.
108 To unmap the file when done, use "losetup -d /dev/loop0".
111 * When I add a new key-slot to LUKS, it asks for a passphrase but
112 then complains about there not being a key-slot with that
115 That is as intended. You are asked a passphrase of an existing
116 key-slot first, before you can enter the passphrase for the new
117 key-slot. Otherwise you could break the encryption by just adding a
118 new key-slot. This way, you have to know the passphrase of one of
119 the already configured key-slots in order to be able to configure a
123 * How do I read a dm-crypt key from file?
125 Note that the file will still be hashed first, just like keyboard
126 input. Use the --key-file option, like this:
128 cryptsetup create --key-file keyfile e1 /dev/loop0
131 * How do I read a LUKS slot key from file?
133 What you really do here is to read a passphrase from file, just as
134 you would with manual entry of a passphrase for a key-slot. You can
135 add a new passphrase to a free key-slot, set the passphrase of an
136 specific key-slot or put an already configured passphrase into a
137 file. In the last case make sure no trailing newline (0x0a) is
138 contained in the key file, or the passphrase will not work because
139 the whole file is used as input.
141 To add a new passphrase to a free key slot from file, use something
144 cryptsetup luksAddKey /dev/loop0 keyfile
146 To add a new passphrase to a specific key-slot, use something like
149 cryptsetup luksAddKey --key-slot 7 /dev/loop0 keyfile
151 To supply a key from file to any LUKS command, use the --key-file
152 option, e.g. like this:
154 cryptsetup luksOpen --key-file keyfile /dev/loop0 e1
157 * How do I read the LUKS master key from file?
159 The question you should ask yourself first is why you would want to
160 do this. The only legitimate reason I can think of is if you want
161 to have two LUKS devices with the same master key. Even then, I
162 think it would be preferable to just use key-slots with the same
163 passphrase, or to use plain dm-crypt instead. If you really have a
164 good reason, please tell me. If I am convinced, I will add how to
168 * What are the security requirements for a key read from file?
170 A file-stored key or passphrase has the same security requirements
171 as one entered interactively, however you can use random bytes and
172 thereby use bytes you cannot type on the keyboard. You can use any
173 file you like as key file, for example a plain text file with a
174 human readable passphrase. To generate a file with random bytes,
175 use something like this:
177 head -c 256 /dev/random > keyfile
180 * If I map a journaled file system using dm-crypt/LUKS, does it still
181 provide its usual transactional guarantees?
183 As far as I know it does (but I may be wrong), but please note that
184 these "guarantees" are far weaker than they appear to be. For
185 example, you not not get a hard flush to disk surface even on a
186 call to fsync. In addition, the HDD itself may do independent
187 write reordering. Some other things can go wrong as well. The
188 filesystem developers are aware of these problems and typically
189 can make it work anyways. That said, dm-crypt/LUKS should not make
192 Personally, I have several instances of ext3 on dm-crypt and have
193 not noticed any specific problems.
195 Update: I did run into frequent small freezes (1-2 sec) when putting
196 a vmware image on ext3 over dm-crypt. This does indicate that the
197 transactional guarantees are in place, but at a cost. When I went
198 back to ext2, the problem went away. This also seems to have gotten
199 better with kernel 2.6.36 and the reworking of filesystem flush
200 locking. Kernel 2.6.38 is expected to have more improvements here.
203 * Can I use LUKS or cryptsetup with a more secure (external) medium
204 for key storage, e.g. TPM or a smartcard?
206 Yes, see the answers on using a file-supplied key. You do have to
207 write the glue-logic yourself though. Basically you can have
208 cryptsetup read the key from STDIN and write it there with your
209 own tool that in turn gets the key from the more secure key
213 * Can I resize a dm-crypt or LUKS partition?
215 Yes, you can, as neither dm-crypt nor LUKS stores partition size.
216 Whether you should is a different question. Personally I recommend
217 backup, recreation of the encrypted partition with new size,
218 recreation of the filesystem and restore. This gets around the
219 tricky business of resizing the filesystem. Resizing a dm-crypt or
220 LUKS container does not resize the filesystem in it. The backup is
221 really non-optional here, as a lot can go wrong, resulting in
222 partial or complete data loss. Using something like gparted to
223 resize an encrypted partition is slow, but typicaly works. This
224 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
243 is missing, you may have a problem with the "/dev" tree itself or
244 you may have broken udev rules.
246 - Check that you have the device mapper and the crypt target in your
247 kernel. The output of "dmsetup targets" should list a "crypt"
248 target. If it is not there or the command fails, add device mapper
249 and crypt-target to the kernel.
251 - Check that the hash-functions and ciphers you want to use are in
252 the kernel. 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 the right
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>
314 * cryptsetup segfaults on Gentoo amd64 hardened ...
316 There seems to be some inteference between the hardening and and
317 the way cryptsetup benchmarks PBKDF2. The solution to this is
318 currently not quite clear for an encrypted root filesystem. For
319 other uses, you can apparently specify USE="dynamic" as compile
320 flag, see http://bugs.gentoo.org/show_bug.cgi?id=283470
326 * Can a bad RAM module cause problems?
328 LUKS and dm-crypt can give the RAM quite a workout, especially when
329 combined with software RAID. In particular the combination RAID5 +
330 LUKS + XFS seems to uncover RAM problems that never caused obvious
331 problems before. Symptoms vary, but often the problem manifest
332 itself when copying large amounts of data, typically several times
333 larger than your main memory.
335 Side note: One thing you should always do on large data
336 copy/movements is to run a verify, for example with the "-d"
337 option of "tar" or by doing a set of MD5 checksums on the source
340 find . -type f -exec md5sum \{\} \; > checksum-file
342 and then a "md5sum -c checksum-file" on the other side. If you get
343 mismatches here, RAM is the primary suspect. A lesser suspect is
344 an overclocked CPU. I have found countless hardware problems in
345 verify runs after copying or making backups. Bit errors are much
346 more common than most people think.
348 Some RAM issues are even worse and corrupt structures in one of the
349 layers. This typically results in lockups, CPU state dumps in the
350 system logs, kernel panic or other things. It is quite possible to
351 have the problem with an encrypted device, but not with an
352 otherwise the same unencrypted device. The reason for that is that
353 encryption has an error amplification property: You flip one bit
354 in an encrypted data block, and the decrypted version has half of
355 its bits flipped. This is an important security property for modern
356 ciphers. With the usual modes in cryptsetup (CBC, ESSIV, XTS), you
357 get up to a completely changed 512 byte block per bit error. A
358 corrupt block causes a lot more havoc than the occasionally
359 flipped single bit and can result various obscure errors.
361 Note however that a verify run on copying between encrypted or
362 unencrypted devices can also show you corruption when the copying
363 itself did not report any problems. If you find defect RAM, assume
364 all backups and copied data to be suspect, unless you did a verify.
369 First you should know that overclocking often makes memory
370 problems worse. So if you overclock (which I strongly recommend
371 against in a system holding data that has some worth), run the
372 tests with the overclocking active.
374 There are two good options. One is Memtest86+ and the other is
375 "memtester" by Charles Cazabon. Memtest86+ requires a reboot and
376 then takes over the machine, while memtester runs from a
377 root-shell. Both use different testing methods and I have found
378 problems fast with each one that the other needed long to find. I
379 recommend running the following procedure until the first error is
382 - Run Memtest86+ for one cycle
384 - Run memterster for one cycle (shut down as many other applications
387 - Run Memtest86+ for 24h or more
389 - Run memtester for 24h or more
391 If all that does not produce error messages, your RAM may be sound,
392 but I have had one weak bit that Memtest86+ needed around 60 hours
393 to find. If you can reproduce the original problem reliably, a good
394 additional test may be to remove half of the RAM (if you have more
395 than one module) and try whether the problem is still there and if
396 so, try with the other half. If you just have one module, get a
397 different one and try with that. If you do overclocking, reduce
398 the settings to the most conservative ones available and try with
405 * Should I initialize (overwrite) a new LUKS/dm-crypt partition?
407 If you just create a filesystem on it, most of the old data will
408 still be there. If the old data is sensitive, you should overwrite
409 it before encrypting. In any case, not initializing will leave the
410 old data there until the specific sector gets written. That may
411 enable an attacker to determine how much and where on the
412 partition data was written. If you think this is a risk, you can
413 prevent this by overwriting the encrypted device (here assumed to
414 be named "e1") with zeros like this:
416 dd_rescue -w /dev/zero /dev/mapper/e1
418 or alternatively with one of the following more standard commands:
420 cat /dev/zero > /dev/mapper/e1
421 dd if=/dev/zero of=/dev/mapper/e1
424 * How do I securely erase a LUKS (or other) partition?
426 For LUKS, if you are in a desperate hurry, overwrite the LUKS
427 header and key-slot area. This means overwriting the 1'052'672
428 bytes, i.e. at 1MiB + 4096 of the LUKS partition. A single
429 overwrite with zeros should be enough. If you anticipate being in a
430 desperate hurry, prepare the command beforehand. Example with
431 /dev/sde1 as the LUKS partition:
433 head -c 1052672 /dev/zero > /dev/sde1; sync
435 A LUKS header backup or full backup will still grant access to
436 most or all data, so make sure that an attacker does not have
437 access to backups or destroy them as well.
439 If you have time, overwrite the whole LUKS partition with a single
440 pass of zeros. This is enough for current HDDs. For SSDs or FLASH
441 (USB sticks) you may want to overwrite the whole drive several
442 times to be sure data is not retained by wear leveling. This is
443 possibly still insecure as SSD technology is not fully understood
444 in this regard. Still, due to the anti-forensic properties of the
445 LUKS key-slots, a single overwrite of an SSD or FLASH drive could
446 be enough. If in doubt, use physical destruction in addition. Keep
447 in mind to also erase all backups.
449 Example for a zero-overwrite erase of partition sde1 done with
452 dd_rescue -w /dev/zero /dev/sde1
455 * How do I securely erase a backup of a LUKS partition or header?
457 That depends on the medium it is stored on. For HDD and SSD, use
458 overwrite with zeros. For an SSD or FLASH drive (USB stick), you
459 may want to overwrite the complete SSD several times and use
460 physical destruction in addition, see last item. For re-writable
461 CD/DVD, a single overwrite should also be enough, due to the
462 anti-forensic properties of the LUKS keyslots. For write-once
463 media, use physical destruction. For low security requirements,
464 just cut the CD/DVD into several parts. For high security needs,
465 shred or burn the medium. If your backup is on magnetic tape, I
466 advise physical destruction by shredding or burning, after
467 overwriting . The problem with magnetic tape is that it has a
468 higher dynamic range than HDDs and older data may well be
469 recoverable after overwrites. Also write-head alignment issues can
470 lead to data not actually being deleted at all during overwrites.
473 * What about backup? Does it compromise security?
475 That depends. See next section.
478 * Why is all my data permanently gone if I overwrite the LUKS header?
480 Overwriting the LUKS header in part or in full is the most common
481 reason why access to LUKS containers is lost permanently.
482 Overwriting can be done in a number of fashions, like creating a
483 new filesystem on the raw LUKS partition, making the raw partition
484 part of a raid array and just writing to the raw partition.
486 The LUKS header contains a 256 bit "salt" value and without that no
487 decryption is possible. While the salt is not secret, it is
488 key-grade material and cannot be reconstructed. This is a
489 cryptographically strong "cannot". From observations on the
490 cryptsetup mailing-list, people typically go though the usual
491 stages of grief (Denial, Anger, Bargaining, Depression, Acceptance)
492 when this happens to them. Observed times vary between 1 day and 2
493 weeks to complete the cycle. Seeking help on the mailing-list is
494 fine. Even if we usually cannot help with getting back your data,
495 most people found the feedback comforting.
497 If your header does not contain an intact salt, best go directly
498 to the last stage ("Acceptance") and think about what to do now.
499 There is one exception that I know of: If your LUKS container is
500 still open, then it may be possible to extract the master key from
501 the running system. Ask on the mailing-list on how to do that and
502 make sure nobody switches off the machine.
507 A salt is a random key-grade value added to the passphrase before
508 it is processed. It is not kept secret. The reason for using salts
509 is as follows: If an attacker wants to crack the password for a
510 single LUKS container, then every possible passphrase has to be
511 tried. Typically an attacker will not try every binary value, but
512 will try words and sentences from a dictionary.
514 If an attacker wants to attack several LUKS containers with the
515 same dictionary, then a different approach makes sense: Compute the
516 resulting slot-key for each dictionary element and store it on
517 disk. Then the test for each entry is just the slow unlocking with
518 the slot key (say 0.00001 sec) instead of calculating the slot-key
519 first (1 sec). For a single attack, this does not help. But if you
520 have more than one container to attack, this helps tremendously,
521 also because you can prepare your table before you even have the
522 container to attack! The calculation is also very simple to
523 parallelize. You could, for example, use the night-time unused CPU
524 power of your desktop PCs for this.
526 This is where the salt comes in. If the salt is combined with the
527 passphrase (in the simplest form, just appended to it), you
528 suddenly need a separate table for each salt value. With a
529 reasonably-sized salt value (256 bit, e.g.) this is quite
533 * Is LUKS secure with a low-entropy (bad) passphrase?
535 This needs a bit of theory. The quality of your passphrase is
536 directly related to its entropy (information theoretic, not
537 thermodynamic). The entropy says how many bits of "uncertainty" or
538 "randomness" are in you passphrase. In other words, that is how
539 difficult guessing the passphrase is.
541 Example: A random English sentence has about 1 bit of entropy per
542 character. A random lowercase (or uppercase) character has about
545 Now, if n is the number of bits of entropy in your passphrase and t
546 is the time it takes to process a passphrase in order to open the
547 LUKS container, then an attacker has to spend at maximum
549 attack_time_max = 2^n * t
551 time for a successful attack and on average half that. There is no
552 way getting around that relationship. However, there is one thing
553 that does help, namely increasing t, the time it takes to use a
554 passphrase, see next FAQ item.
556 Still, if you want good security, a high-entropy passphrase is the
557 only option. Use at least 64 bits for secret stuff. That is 64
558 characters of English text (but only if randomly chosen) or a
559 combination of 12 truly random letters and digits.
561 For passphrase generation, do not use lines from very well-known
562 texts (religious texts, Harry potter, etc.) as they are to easy to
563 guess. For example, the total Harry Potter has about 1'500'000
564 words (my estimation). Trying every 64 character sequence starting
565 and ending at a word boundary would take only something like 20
566 days on a single CPU and is entirely feasible. To put that into
567 perspective, using a number of Amazon EC2 High-CPU Extra Large
568 instances (each gives about 8 real cores), this tests costs
569 currently about $48, but can be made to run arbitrarily fast.
571 On the other hand, choosing 1.5 lines from, say, the Wheel of Time
572 is in itself not more secure, but the book selection adds quite a
573 bit of entropy. (Now that I have mentioned it here, don't use tWoT
574 either!) If you add 2 or 3 typos or switch some words around, then
575 this is good passphrase material.
578 * What is "iteration count" and why is decreasing it a bad idea?
580 Iteration count is the number of PBKDF2 iterations a passphrase is
581 put through before it is used to unlock a key-slot. Iterations are
582 done with the explicit purpose to increase the time that it takes
583 to unlock a key-slot. This provides some protection against use of
584 low-entropy passphrases.
586 The idea is that an attacker has to try all possible passphrases.
587 Even if the attacker knows the passphrase is low-entropy (see last
588 item), it is possible to make each individual try take longer. The
589 way to do this is to repeatedly hash the passphrase for a certain
590 time. The attacker then has to spend the same time (given the same
591 computing power) as the user per try. With LUKS, the default is 1
592 second of PBKDF2 hashing.
594 Example 1: Lets assume we have a really bad passphrase (e.g. a
595 girlfriends name) with 10 bits of entropy. With the same CPU, an
596 attacker would need to spend around 500 seconds on average to
597 break that passphrase. Without iteration, it would be more like
598 0.0001 seconds on a modern CPU.
600 Example 2: The user did a bit better and has 32 chars of English
601 text. That would give use about 32 bits of entropy. With 1 second
602 iteration, that means an attacker on the same CPU needs around 136
603 years. That is pretty impressive for such a weak passphrase.
604 Without the iterations, it would be more like 50 days on a modern
605 CPU, and possibly far less.
607 In addition, the attacker can both parallelize and use special
608 hardware like GPUs to speed up the attack. The attack can also
609 happen quite some time after the luksFormat operation and CPUs can
610 have become faster and cheaper. For that reason you want a bit
611 of extra security. Anyways, in Example 1 your are screwed. In
612 example 2, not necessarily. Even if the attack is faster, it still
613 has a certain cost associated with it, say 10000 EUR/USD with
614 iteration and 1 EUR/USD without iteration. The first can be
615 prohibitively expensive, while the second is something you try
616 even without solid proof that the decryption will yield something
619 The numbers above are mostly made up, but show the idea. Of course
620 the best thing is to have a high-entropy passphrase.
622 Would a 100 sec iteration time be even better? Yes and no.
623 Cryptographically it would be a lot better, namely 100 times better.
624 However, usability is a very important factor for security
625 technology and one that gets overlooked surprisingly often. For
626 LUKS, if you have to wait 2 minutes to unlock the LUKS container,
627 most people will not bother and use less secure storage instead. It
628 is better to have less protection against low-entropy passphrases
629 and people actually use LUKS, than having them do without
630 encryption altogether.
632 Now, what about decreasing the iteration time? This is generally a
633 very bad idea, unless you know and can enforce that the users only
634 use high-entropy passphrases. If you decrease the iteration time
635 without ensuring that, then you put your users at increased risk,
636 and considering how rarely LUKS containers are unlocked in a
637 typical work-flow, you do so without a good reason. Don't do it.
638 The iteration time is already low enough that users with entropy
639 low passphrases are vulnerable. Lowering it even further increases
640 this danger significantly.
643 * Is LUKS with default parameters less secure on a slow CPU?
645 Unfortunately, yes. However the only aspect affected is the
646 protection for low-entropy passphrase or master-key. All other
647 security aspects are independent of CPU speed.
649 The master key is less critical, as you really have to work at it
650 to give it low entropy. One possibility is to supply the master key
651 yourself. If that key is low-entropy, then you get what you
652 deserve. The other known possibility is to use /dev/urandom for
653 key generation in an entropy-startved situation (e.g. automatic
654 installation on an embedded device without network and other entropy
657 For the passphrase, don't use a low-entropy passphrase. If your
658 passphrase is good, then a slow CPU will not matter. If you insist
659 on a low-entropy passphrase on a slow CPU, use something like
660 "--iter-time=10" or higher and wait a long time on each LUKS unlock
661 and pray that the attacker does not find out in which way exactly
662 your passphrase is low entropy. This also applies to low-entropy
663 passphrases on fast CPUs. Technology can do only so much to
664 compensate for problems in front of the keyboard.
667 * Why was the default aes-cbc-plain replaced with aes-cbc-essiv?
669 The problem is that cbc-plain has a fingerprint vulnerability, where
670 a specially crafted file placed into the crypto-container can be
671 recognized from the outside. The issue here is that for cbc-plain
672 the initialization vector (IV) is the sector number. The IV gets
673 XORed to the first data chunk of the sector to be encrypted. If you
674 make sure that the first data block to be stored in a sector
675 contains the sector number as well, the first data block to be
676 encrypted is all zeros and always encrypted to the same ciphertext.
677 This also works if the first data chunk just has a constant XOR
678 with the sector number. By having several shifted patterns you can
679 take care of the case of a non-power-of-two start sector number of
682 This mechanism allows you to create a pattern of sectors that have
683 the same first ciphertext block and signal one bit per sector to the
684 outside, allowing you to e.g. mark media files that way for
685 recognition without decryption. For large files this is a
686 practical attack. For small ones, you do not have enough blocks to
687 signal and take care of different file starting offsets.
689 In order to prevent this attack, the default was changed to
690 cbc-essiv. ESSIV uses a keyed hash of the sector number, with the
691 encryption key as key. This makes the IV unpredictable without
692 knowing the encryption key and the watermarking attack fails.
695 * Are there any problems with "plain" IV? What is "plain64"?
697 First, "plain" and "plain64" are both not secure to use with CBC,
698 see previous FAQ item.
700 However there are modes, like XTS, that are secure with "plain" IV.
701 The next limit is that "plain" is 64 bit, with the upper 32 bit set
702 to zero. This means that on volumes larger than 2TiB, the IV
703 repeats, creating a vulnerability that potentially leaks some
704 data. To avoid this, use "plain64", which uses the full sector
705 number up to 64 bit. Note that "plain64" requires a kernel >=
706 2.6.33. Also note that "plain64" is backwards compatible for
707 volume sizes <= 2TiB, but not for those > 2TiB. Finally, "plain64"
708 does not cause any performance penalty compared to "plain".
711 * What about XTS mode?
713 XTS mode is potentially even more secure than cbc-essiv (but only if
714 cbc-essiv is insecure in your scenario). It is a NIST standard and
715 used, e.g. in Truecrypt. At the moment, if you want to use it, you
716 have to specify it manually as "aes-xts-plain", i.e.
718 cryptsetup -c aes-xts-plain luksFormat <device>
720 For volumes >2TiB and kernels >= 2.6.33 use "plain64" (see FAQ
721 item on "plain" and "plain64"):
723 cryptsetup -c aes-xts-plain64 luksFormat <device>
725 There is a potential security issue with XTS mode and large blocks.
726 LUKS and dm-crypt always use 512B blocks and the issue does not
730 6. Backup and Data Recovery
733 * Does a backup compromise security?
735 Depends on how you do it. First, a backup is non-optional with
736 encrypted data just the same way it is with non-encrypted data.
737 Disks do break and they do not care whether they make plain or
738 encrypted data inaccessible. As a gideline, a well-treated HDD (!)
739 breaks with about 5% probability per year. This means everybody
740 will be hit sooner or later.
742 However there are risks introduced by backups. For example if you
743 change/disable a key-slot in LUKS, a binary backup of the partition
744 will still have the old key-slot. To deal with this, you have to
745 be able to change the key-slot on the backup as well, or use a
746 different set-up. One option is to have a different passphrase on
747 the backup and to make the backup with both containers open.
748 Another one is to make a backup of the original, opened container
749 to a single file, e.g. with tar, and to encrypt that file with
750 public-key-cryptography, e.g. with GnuPG. You can then keep the
751 secret key in a safe place, because it is only used to decrypt a
752 backup. The key the backup is encrypted with can be stored without
753 special security measures, as long as an attacker cannot replace
756 If you use dm-crypt, backup is simpler: As there is no key
757 management, the main risk is that you cannot wipe the backup when
758 wiping the original. However wiping the original for dm-crypt
759 should consist of forgetting the passphrase and that you can do
760 without actual access to the backup.
762 In both cases, there is an additional (usually small) risk: An
763 attacker can see how many sectors and which ones have been changed
764 since the backup. This is not possible with the public-key method
767 My personal advice is to use one USB disk (low value date) or
768 three disks (high value data) in rotating order for backups, and
769 either use different passphrases or keep them easily accessible
770 in case you need to disable a key-slot. If you do network-backup
771 or tape-backup, I strongly recommend to go the public-key path,
772 especially as you typically cannot reliably delete data in these
773 scenarios. (Well, you can burn the tape if it is under your
777 * What happens if I overwrite the start of a LUKS partition or damage
778 the LUKS header or key-slots?
780 There are two critical components for decryption: The salt values
781 in the header itself and the key-slots. If the salt values are
782 overwritten or changed, nothing (in the cryptographically strong
783 sense) can be done to access the data, unless there is a backup
784 of the LUKS header. If a key-slot is damaged, the data can still
785 be read with a different key-slot, if there is a remaining
786 undamaged and used key-slot. Note that in order to make a key-slot
787 unrecoverable in a cryptographically strong sense, changing about
788 4-6 bits in random locations of its 128kiB size is quite enough.
791 * What happens if I (quick) format a LUKS partition?
793 I have not tried the different ways to do this, but very likely you
794 will have written a new boot-sector, which in turn overwrites the
795 LUKS header, including the salts, making your data permanently
796 irretrivable, unless you have a LUKS header backup. You may also
797 damage the key-slots in part or in full. See also last item.
800 * What does the on-disk structure of dm-crypt look like?
802 There is none. dm-crypt takes a block device and gives encrypted
803 access to each of its blocks with a key derived from the passphrase
804 given. If you use a cipher different than the default, you have to
805 specify that as a parameter to cryptsetup too. If you want to
806 change the password, you basically have to create a second
807 encrypted device with the new passphrase and copy your data over.
808 On the plus side, if you accidentally overwrite any part of a
809 dm-crypt device, the damage will be limited to the are you
813 * What does the on-disk structure of LUKS look like?
815 A LUKS partition consists of a header, followed by 8 key-slot
816 descriptors, followed by 8 key slots, followed by the encrypted
819 Header and key-slot descriptors fill the first 592 bytes. The
820 key-slot size depends on the creation parameters, namely on the
821 number of anti-forensic stripes and on key block alignment.
823 With 4000 stripes (the default), each key-slot is a bit less than
824 128kiB in size. Due to sector alignment of the key-slot start,
825 that means the key block 0 is at offset 0x1000-0x20400, key
826 block 1 at offset 0x21000-0x40400, and key block 7 at offset
827 0xc1000-0xe0400. The space to the next full sector address is
828 padded with zeros. Never used key-slots are filled with what the
829 disk originally contained there, a key-slot removed with
830 "luksRemoveKey" or "luksKillSlot" gets filled with 0xff. Start of
831 bulk data (with the default 4000 stripes and 8 key-slots) is at
832 0x101000, i.e. at 1'052'672 bytes, i.e. at 1MiB + 4096 bytes from
833 the start of the partition. This is also the value given by
834 command "luksDump" with "Payload offset: 2056", just multiply by
835 the sector size (512 bytes). Incidentally, "luksHeaderBackup"
836 dumps exactly the first 1'052'672 bytes to file and
837 "luksHeaderRestore" restores them.
839 The exact specification of the format is here:
840 http://code.google.com/p/cryptsetup/wiki/Specification
843 * How do I backup a LUKS header?
845 While you could just copy the appropriate number of bytes from the
846 start of the LUKS partition, the best way is to use command option
847 "luksHeaderBackup" of cryptsetup. This protects also against
848 errors when non-standard parameters have been used in LUKS
849 partition creation. Example:
852 cryptsetup luksHeaderBackup --header-backup-file h /dev/mapper/c1
854 To restore, use the inverse command, i.e.
856 cryptsetup luksHeaderRestore --header-backup-file h /dev/mapper/c1
859 * How do I backup a LUKS partition?
861 You do a sector-image of the whole partition. This will contain
862 the LUKS header, the keys-slots and the data ares. It can be done
863 under Linux e.g. with dd_rescue (for a direct image copy) and with
864 "cat" or "dd". Example:
866 cat /dev/sda10 > sda10.img
867 dd_rescue /dev/sda10 sda10.img
869 You can also use any other backup software that is capable of making
870 a sector image of a partition. Note that compression is
871 ineffective for encrypted data, hence it does not make sense to
875 * Do I need a backup of the full partition? Would the header and
876 key-slots not be enough?
878 Backup protects you against two things: Disk loss or corruption
879 and user error. By far the most questions on the dm-crypt mailing
880 list about how to recover a damaged LUKS partition are related
881 to user error. For example, if you create a new filesystem on a
882 LUKS partition, chances are good that all data is lost
885 For this case, a header+key-slot backup would often be enough. But
886 keep in mind that a well-treated (!) HDD has roughly a failure
887 risk of 5% per year. It is highly advisable to have a complete
888 backup to protect against this case.
891 * Are there security risks from a backup of the LUKS header or a
892 whole LUKS partition?
894 Yes. One risk is that if you remove access rights for specific
895 key-slots by deleting their contents, the data can still be
896 accessed with invalidated passphrase and the backup. The other
897 risk is that if you erase a LUKS partition, a backup could still
898 grant access, especially if you only erased the LUKS header and
899 not the whole partition.
902 * I think this is overly complicated. Is there an alternative?
904 Yes, you can use plain dm-crypt. It does not allow multiple
905 passphrases, but on the plus side, it has zero on disk description
906 and if you overwrite some part of a plain dm-crypt partition,
907 exactly the overwritten parts are lost (rounded up to sector
911 7. Issues with Specific Versions of cryptsetup
914 * When using the create command for plain dm-crypt with cryptsetup
915 1.1.x, the mapping is incompatible and my data is not accessible
918 With cryptsetup 1.1.x, the distro maintainer can define different
919 default encryption modes for LUKS and plain devices. You can check
920 these compiled-in defaults using "cryptsetup --help". Moreover, the
921 plain device default changed because the old IV mode was
922 vulnerable to a watermarking attack.
924 If you are using a plain device and you need a compatible mode, just
925 specify cipher, key size and hash algorithm explicitly. For
926 compatibility with cryptsetup 1.0.x defaults, simple use the
929 cryptsetup create -c aes-cbc-plain -s 256 -h ripemd160 <name> <dev>
931 LUKS stores cipher and mode in the metadata on disk, avoiding this
935 * cryptsetup on SLED 10 has problems...
937 SLED 10 is missing an essential kernel patch for dm-crypt, which
938 is broken in its kernel as a result. There may be a very old
939 version of cryptsetup (1.0.x) provided by SLED, which should also
940 not be used anymore as well. My advice would be to drop SLED 10.
942 A. Contributors In no particular order: