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.
80 The mailing list archive is here:
81 http://dir.gmane.org/gmane.linux.kernel.device-mapper.dm-crypt
87 * Can I encrypt an already existing, non-empty partition to use LUKS?
89 There is no converter, and it is not really needed. The way to do
90 this is to make a backup of the device in question, securely wipe
91 the device (as LUKS device initialization does not clear away old
92 data), do a luksFormat, optionally overwrite the encrypted device,
93 create a new filesystem and restore your backup on the now
94 encrypted device. Also refer to sections "Security Aspects" and
95 "Backup and Data Recovery".
97 For backup, plain GNU tar works well and backs up anything likely
98 to be in a filesystem.
101 * How do I use LUKS with a loop-device?
103 Just the same as with any block device. If you want, for example,
104 to use a 100MiB file as LUKS container, do something like this:
106 head -c 100M /dev/zero > luksfile # create empty file
107 losetup /dev/loop0 luksfile # map luksfile to /dev/loop0
108 cryptsetup luksFormat /dev/loop0 # create LUKS on loop device
110 Afterwards just use /dev/loop0 as a you would use a LUKS partition.
111 To unmap the file when done, use "losetup -d /dev/loop0".
114 * When I add a new key-slot to LUKS, it asks for a passphrase but
115 then complains about there not being a key-slot with that
118 That is as intended. You are asked a passphrase of an existing
119 key-slot first, before you can enter the passphrase for the new
120 key-slot. Otherwise you could break the encryption by just adding a
121 new key-slot. This way, you have to know the passphrase of one of
122 the already configured key-slots in order to be able to configure a
126 * How do I read a dm-crypt key from file?
128 Note that the file will still be hashed first, just like keyboard
129 input. Use the --key-file option, like this:
131 cryptsetup create --key-file keyfile e1 /dev/loop0
134 * How do I read a LUKS slot key from file?
136 What you really do here is to read a passphrase from file, just as
137 you would with manual entry of a passphrase for a key-slot. You can
138 add a new passphrase to a free key-slot, set the passphrase of an
139 specific key-slot or put an already configured passphrase into a
140 file. In the last case make sure no trailing newline (0x0a) is
141 contained in the key file, or the passphrase will not work because
142 the whole file is used as input.
144 To add a new passphrase to a free key slot from file, use something
147 cryptsetup luksAddKey /dev/loop0 keyfile
149 To add a new passphrase to a specific key-slot, use something like
152 cryptsetup luksAddKey --key-slot 7 /dev/loop0 keyfile
154 To supply a key from file to any LUKS command, use the --key-file
155 option, e.g. like this:
157 cryptsetup luksOpen --key-file keyfile /dev/loop0 e1
160 * How do I read the LUKS master key from file?
162 The question you should ask yourself first is why you would want to
163 do this. The only legitimate reason I can think of is if you want
164 to have two LUKS devices with the same master key. Even then, I
165 think it would be preferable to just use key-slots with the same
166 passphrase, or to use plain dm-crypt instead. If you really have a
167 good reason, please tell me. If I am convinced, I will add how to
171 * What are the security requirements for a key read from file?
173 A file-stored key or passphrase has the same security requirements
174 as one entered interactively, however you can use random bytes and
175 thereby use bytes you cannot type on the keyboard. You can use any
176 file you like as key file, for example a plain text file with a
177 human readable passphrase. To generate a file with random bytes,
178 use something like this:
180 head -c 256 /dev/random > keyfile
183 * If I map a journaled file system using dm-crypt/LUKS, does it still
184 provide its usual transactional guarantees?
186 As far as I know it does (but I may be wrong), but please note that
187 these "guarantees" are far weaker than they appear to be. For
188 example, you not not get a hard flush to disk surface even on a
189 call to fsync. In addition, the HDD itself may do independent
190 write reordering. Some other things can go wrong as well. The
191 filesystem developers are aware of these problems and typically
192 can make it work anyways. That said, dm-crypt/LUKS should not make
195 Personally, I have several instances of ext3 on dm-crypt and have
196 not noticed any specific problems.
198 Update: I did run into frequent small freezes (1-2 sec) when putting
199 a vmware image on ext3 over dm-crypt. This does indicate that the
200 transactional guarantees are in place, but at a cost. When I went
201 back to ext2, the problem went away. This also seems to have gotten
202 better with kernel 2.6.36 and the reworking of filesystem flush
203 locking. Kernel 2.6.38 is expected to have more improvements here.
206 * Can I use LUKS or cryptsetup with a more secure (external) medium
207 for key storage, e.g. TPM or a smartcard?
209 Yes, see the answers on using a file-supplied key. You do have to
210 write the glue-logic yourself though. Basically you can have
211 cryptsetup read the key from STDIN and write it there with your
212 own tool that in turn gets the key from the more secure key
216 * Can I resize a dm-crypt or LUKS partition?
218 Yes, you can, as neither dm-crypt nor LUKS stores partition size.
219 Whether you should is a different question. Personally I recommend
220 backup, recreation of the encrypted partition with new size,
221 recreation of the filesystem and restore. This gets around the
222 tricky business of resizing the filesystem. Resizing a dm-crypt or
223 LUKS container does not resize the filesystem in it. The backup is
224 really non-optional here, as a lot can go wrong, resulting in
225 partial or complete data loss. Using something like gparted to
226 resize an encrypted partition is slow, but typicaly works. This
227 will not change the size of the filesystem hidden under the
230 You also need to be aware of size-based limitations. The one
231 currently relevant is that aes-xts-plain should not be used for
232 encrypted container sizes larger than 2TiB. Use aes-xts-plain64
239 * My dm-crypt/LUKS mapping does not work! What general steps are
240 there to investigate the problem?
242 If you get a specific error message, investigate what it claims
243 first. If not, you may want to check the following things.
245 - Check that "/dev", including "/dev/mapper/control" is there. If it
246 is missing, you may have a problem with the "/dev" tree itself or
247 you may have broken udev rules.
249 - Check that you have the device mapper and the crypt target in your
250 kernel. The output of "dmsetup targets" should list a "crypt"
251 target. If it is not there or the command fails, add device mapper
252 and crypt-target to the kernel.
254 - Check that the hash-functions and ciphers you want to use are in
255 the kernel. The output of "cat /proc/crypto" needs to list them.
258 * My dm-crypt mapping suddenly stopped when upgrading cryptsetup.
260 The default cipher, hash or mode may have changed (the mode changed
261 from 1.0.x to 1.1.x). See under "Issues With Specific Versions of
265 * When I call cryptsetup from cron/CGI, I get errors about unknown
268 If you get errors about unknown parameters or the like that are not
269 present when cryptsetup is called from the shell, make sure you
270 have no older version of cryptsetup on your system that then gets
271 called by cron/CGI. For example some distributions install
272 cryptsetup into /usr/sbin, while a manual install could go to
273 /usr/local/sbin. As a debugging aid, call "cryptsetup --version"
274 from cron/CGI or the non-shell mechanism to be sure the right
278 * Unlocking a LUKS device takes very long. Why?
280 The iteration time for a key-slot (see Section 5 for an explanation
281 what iteration does) is calculated when setting a passphrase. By
282 default it is 1 second on the machine where the passphrase is set.
283 If you set a passphrase on a fast machine and then unlock it on a
284 slow machine, the unlocking time can be much longer. Also take into
285 account that up to 8 key-slots have to be tried in order to find the
288 If this is problem, you can add another key-slot using the slow
289 machine with the same passphrase and then remove the old key-slot.
290 The new key-slot will have an iteration count adjusted to 1 second
291 on the slow machine. Use luksKeyAdd and then luksKillSlot or
294 However, this operation will not change volume key iteration count
295 (MK iterations in output of "cryptsetup luksDump"). In order to
296 change that, you will have to backup the data in the LUKS
297 container, luksFormat on the slow machine and restore the data.
298 Note that in the original LUKS specification this value was fixed
299 to 10, but it is now derived from the PBKDF2 benchmark as well and
300 set to iterations in 0.125 sec or 1000, whichever is larger.
303 * "blkid" sees a LUKS UUID and an ext2/swap UUID on the same device.
306 Some old versions of cryptsetup have a bug where the header does
307 not get completely wiped during LUKS format and an older ext2/swap
308 signature remains on the device. This confuses blkid.
310 Fix: Wipe the unused header areas by doing a backup and restore of
311 the header with cryptsetup 1.1.x:
313 cryptsetup luksHeaderBackup --header-backup-file <file> <device>
314 cryptsetup luksHeaderRestore --header-backup-file <file> <device>
317 * cryptsetup segfaults on Gentoo amd64 hardened ...
319 There seems to be some inteference between the hardening and and
320 the way cryptsetup benchmarks PBKDF2. The solution to this is
321 currently not quite clear for an encrypted root filesystem. For
322 other uses, you can apparently specify USE="dynamic" as compile
323 flag, see http://bugs.gentoo.org/show_bug.cgi?id=283470
329 * Can a bad RAM module cause problems?
331 LUKS and dm-crypt can give the RAM quite a workout, especially when
332 combined with software RAID. In particular the combination RAID5 +
333 LUKS + XFS seems to uncover RAM problems that never caused obvious
334 problems before. Symptoms vary, but often the problem manifest
335 itself when copying large amounts of data, typically several times
336 larger than your main memory.
338 Side note: One thing you should always do on large data
339 copy/movements is to run a verify, for example with the "-d"
340 option of "tar" or by doing a set of MD5 checksums on the source
343 find . -type f -exec md5sum \{\} \; > checksum-file
345 and then a "md5sum -c checksum-file" on the other side. If you get
346 mismatches here, RAM is the primary suspect. A lesser suspect is
347 an overclocked CPU. I have found countless hardware problems in
348 verify runs after copying or making backups. Bit errors are much
349 more common than most people think.
351 Some RAM issues are even worse and corrupt structures in one of the
352 layers. This typically results in lockups, CPU state dumps in the
353 system logs, kernel panic or other things. It is quite possible to
354 have the problem with an encrypted device, but not with an
355 otherwise the same unencrypted device. The reason for that is that
356 encryption has an error amplification property: You flip one bit
357 in an encrypted data block, and the decrypted version has half of
358 its bits flipped. This is an important security property for modern
359 ciphers. With the usual modes in cryptsetup (CBC, ESSIV, XTS), you
360 get up to a completely changed 512 byte block per bit error. A
361 corrupt block causes a lot more havoc than the occasionally
362 flipped single bit and can result various obscure errors.
364 Note however that a verify run on copying between encrypted or
365 unencrypted devices can also show you corruption when the copying
366 itself did not report any problems. If you find defect RAM, assume
367 all backups and copied data to be suspect, unless you did a verify.
372 First you should know that overclocking often makes memory
373 problems worse. So if you overclock (which I strongly recommend
374 against in a system holding data that has some worth), run the
375 tests with the overclocking active.
377 There are two good options. One is Memtest86+ and the other is
378 "memtester" by Charles Cazabon. Memtest86+ requires a reboot and
379 then takes over the machine, while memtester runs from a
380 root-shell. Both use different testing methods and I have found
381 problems fast with each one that the other needed long to find. I
382 recommend running the following procedure until the first error is
385 - Run Memtest86+ for one cycle
387 - Run memterster for one cycle (shut down as many other applications
390 - Run Memtest86+ for 24h or more
392 - Run memtester for 24h or more
394 If all that does not produce error messages, your RAM may be sound,
395 but I have had one weak bit that Memtest86+ needed around 60 hours
396 to find. If you can reproduce the original problem reliably, a good
397 additional test may be to remove half of the RAM (if you have more
398 than one module) and try whether the problem is still there and if
399 so, try with the other half. If you just have one module, get a
400 different one and try with that. If you do overclocking, reduce
401 the settings to the most conservative ones available and try with
408 * Should I initialize (overwrite) a new LUKS/dm-crypt partition?
410 If you just create a filesystem on it, most of the old data will
411 still be there. If the old data is sensitive, you should overwrite
412 it before encrypting. In any case, not initializing will leave the
413 old data there until the specific sector gets written. That may
414 enable an attacker to determine how much and where on the
415 partition data was written. If you think this is a risk, you can
416 prevent this by overwriting the encrypted device (here assumed to
417 be named "e1") with zeros like this:
419 dd_rescue -w /dev/zero /dev/mapper/e1
421 or alternatively with one of the following more standard commands:
423 cat /dev/zero > /dev/mapper/e1
424 dd if=/dev/zero of=/dev/mapper/e1
427 * How do I securely erase a LUKS (or other) partition?
429 For LUKS, if you are in a desperate hurry, overwrite the LUKS
430 header and key-slot area. This means overwriting the 1'052'672
431 bytes, i.e. at 1MiB + 4096 of the LUKS partition. A single
432 overwrite with zeros should be enough. If you anticipate being in a
433 desperate hurry, prepare the command beforehand. Example with
434 /dev/sde1 as the LUKS partition:
436 head -c 1052672 /dev/zero > /dev/sde1; sync
438 A LUKS header backup or full backup will still grant access to
439 most or all data, so make sure that an attacker does not have
440 access to backups or destroy them as well.
442 If you have time, overwrite the whole LUKS partition with a single
443 pass of zeros. This is enough for current HDDs. For SSDs or FLASH
444 (USB sticks) you may want to overwrite the whole drive several
445 times to be sure data is not retained by wear leveling. This is
446 possibly still insecure as SSD technology is not fully understood
447 in this regard. Still, due to the anti-forensic properties of the
448 LUKS key-slots, a single overwrite of an SSD or FLASH drive could
449 be enough. If in doubt, use physical destruction in addition. Keep
450 in mind to also erase all backups.
452 Example for a zero-overwrite erase of partition sde1 done with
455 dd_rescue -w /dev/zero /dev/sde1
458 * How do I securely erase a backup of a LUKS partition or header?
460 That depends on the medium it is stored on. For HDD and SSD, use
461 overwrite with zeros. For an SSD or FLASH drive (USB stick), you
462 may want to overwrite the complete SSD several times and use
463 physical destruction in addition, see last item. For re-writable
464 CD/DVD, a single overwrite should also be enough, due to the
465 anti-forensic properties of the LUKS keyslots. For write-once
466 media, use physical destruction. For low security requirements,
467 just cut the CD/DVD into several parts. For high security needs,
468 shred or burn the medium. If your backup is on magnetic tape, I
469 advise physical destruction by shredding or burning, after
470 overwriting . The problem with magnetic tape is that it has a
471 higher dynamic range than HDDs and older data may well be
472 recoverable after overwrites. Also write-head alignment issues can
473 lead to data not actually being deleted at all during overwrites.
476 * What about backup? Does it compromise security?
478 That depends. See next section.
481 * Why is all my data permanently gone if I overwrite the LUKS header?
483 Overwriting the LUKS header in part or in full is the most common
484 reason why access to LUKS containers is lost permanently.
485 Overwriting can be done in a number of fashions, like creating a
486 new filesystem on the raw LUKS partition, making the raw partition
487 part of a raid array and just writing to the raw partition.
489 The LUKS header contains a 256 bit "salt" value and without that no
490 decryption is possible. While the salt is not secret, it is
491 key-grade material and cannot be reconstructed. This is a
492 cryptographically strong "cannot". From observations on the
493 cryptsetup mailing-list, people typically go though the usual
494 stages of grief (Denial, Anger, Bargaining, Depression, Acceptance)
495 when this happens to them. Observed times vary between 1 day and 2
496 weeks to complete the cycle. Seeking help on the mailing-list is
497 fine. Even if we usually cannot help with getting back your data,
498 most people found the feedback comforting.
500 If your header does not contain an intact salt, best go directly
501 to the last stage ("Acceptance") and think about what to do now.
502 There is one exception that I know of: If your LUKS container is
503 still open, then it may be possible to extract the master key from
504 the running system. Ask on the mailing-list on how to do that and
505 make sure nobody switches off the machine.
510 A salt is a random key-grade value added to the passphrase before
511 it is processed. It is not kept secret. The reason for using salts
512 is as follows: If an attacker wants to crack the password for a
513 single LUKS container, then every possible passphrase has to be
514 tried. Typically an attacker will not try every binary value, but
515 will try words and sentences from a dictionary.
517 If an attacker wants to attack several LUKS containers with the
518 same dictionary, then a different approach makes sense: Compute the
519 resulting slot-key for each dictionary element and store it on
520 disk. Then the test for each entry is just the slow unlocking with
521 the slot key (say 0.00001 sec) instead of calculating the slot-key
522 first (1 sec). For a single attack, this does not help. But if you
523 have more than one container to attack, this helps tremendously,
524 also because you can prepare your table before you even have the
525 container to attack! The calculation is also very simple to
526 parallelize. You could, for example, use the night-time unused CPU
527 power of your desktop PCs for this.
529 This is where the salt comes in. If the salt is combined with the
530 passphrase (in the simplest form, just appended to it), you
531 suddenly need a separate table for each salt value. With a
532 reasonably-sized salt value (256 bit, e.g.) this is quite
536 * Is LUKS secure with a low-entropy (bad) passphrase?
538 This needs a bit of theory. The quality of your passphrase is
539 directly related to its entropy (information theoretic, not
540 thermodynamic). The entropy says how many bits of "uncertainty" or
541 "randomness" are in you passphrase. In other words, that is how
542 difficult guessing the passphrase is.
544 Example: A random English sentence has about 1 bit of entropy per
545 character. A random lowercase (or uppercase) character has about
548 Now, if n is the number of bits of entropy in your passphrase and t
549 is the time it takes to process a passphrase in order to open the
550 LUKS container, then an attacker has to spend at maximum
552 attack_time_max = 2^n * t
554 time for a successful attack and on average half that. There is no
555 way getting around that relationship. However, there is one thing
556 that does help, namely increasing t, the time it takes to use a
557 passphrase, see next FAQ item.
559 Still, if you want good security, a high-entropy passphrase is the
560 only option. Use at least 64 bits for secret stuff. That is 64
561 characters of English text (but only if randomly chosen) or a
562 combination of 12 truly random letters and digits.
564 For passphrase generation, do not use lines from very well-known
565 texts (religious texts, Harry potter, etc.) as they are to easy to
566 guess. For example, the total Harry Potter has about 1'500'000
567 words (my estimation). Trying every 64 character sequence starting
568 and ending at a word boundary would take only something like 20
569 days on a single CPU and is entirely feasible. To put that into
570 perspective, using a number of Amazon EC2 High-CPU Extra Large
571 instances (each gives about 8 real cores), this tests costs
572 currently about $48, but can be made to run arbitrarily fast.
574 On the other hand, choosing 1.5 lines from, say, the Wheel of Time
575 is in itself not more secure, but the book selection adds quite a
576 bit of entropy. (Now that I have mentioned it here, don't use tWoT
577 either!) If you add 2 or 3 typos or switch some words around, then
578 this is good passphrase material.
581 * What is "iteration count" and why is decreasing it a bad idea?
583 Iteration count is the number of PBKDF2 iterations a passphrase is
584 put through before it is used to unlock a key-slot. Iterations are
585 done with the explicit purpose to increase the time that it takes
586 to unlock a key-slot. This provides some protection against use of
587 low-entropy passphrases.
589 The idea is that an attacker has to try all possible passphrases.
590 Even if the attacker knows the passphrase is low-entropy (see last
591 item), it is possible to make each individual try take longer. The
592 way to do this is to repeatedly hash the passphrase for a certain
593 time. The attacker then has to spend the same time (given the same
594 computing power) as the user per try. With LUKS, the default is 1
595 second of PBKDF2 hashing.
597 Example 1: Lets assume we have a really bad passphrase (e.g. a
598 girlfriends name) with 10 bits of entropy. With the same CPU, an
599 attacker would need to spend around 500 seconds on average to
600 break that passphrase. Without iteration, it would be more like
601 0.0001 seconds on a modern CPU.
603 Example 2: The user did a bit better and has 32 chars of English
604 text. That would give use about 32 bits of entropy. With 1 second
605 iteration, that means an attacker on the same CPU needs around 136
606 years. That is pretty impressive for such a weak passphrase.
607 Without the iterations, it would be more like 50 days on a modern
608 CPU, and possibly far less.
610 In addition, the attacker can both parallelize and use special
611 hardware like GPUs to speed up the attack. The attack can also
612 happen quite some time after the luksFormat operation and CPUs can
613 have become faster and cheaper. For that reason you want a bit
614 of extra security. Anyways, in Example 1 your are screwed. In
615 example 2, not necessarily. Even if the attack is faster, it still
616 has a certain cost associated with it, say 10000 EUR/USD with
617 iteration and 1 EUR/USD without iteration. The first can be
618 prohibitively expensive, while the second is something you try
619 even without solid proof that the decryption will yield something
622 The numbers above are mostly made up, but show the idea. Of course
623 the best thing is to have a high-entropy passphrase.
625 Would a 100 sec iteration time be even better? Yes and no.
626 Cryptographically it would be a lot better, namely 100 times better.
627 However, usability is a very important factor for security
628 technology and one that gets overlooked surprisingly often. For
629 LUKS, if you have to wait 2 minutes to unlock the LUKS container,
630 most people will not bother and use less secure storage instead. It
631 is better to have less protection against low-entropy passphrases
632 and people actually use LUKS, than having them do without
633 encryption altogether.
635 Now, what about decreasing the iteration time? This is generally a
636 very bad idea, unless you know and can enforce that the users only
637 use high-entropy passphrases. If you decrease the iteration time
638 without ensuring that, then you put your users at increased risk,
639 and considering how rarely LUKS containers are unlocked in a
640 typical work-flow, you do so without a good reason. Don't do it.
641 The iteration time is already low enough that users with entropy
642 low passphrases are vulnerable. Lowering it even further increases
643 this danger significantly.
646 * Is LUKS with default parameters less secure on a slow CPU?
648 Unfortunately, yes. However the only aspect affected is the
649 protection for low-entropy passphrase or master-key. All other
650 security aspects are independent of CPU speed.
652 The master key is less critical, as you really have to work at it
653 to give it low entropy. One possibility is to supply the master key
654 yourself. If that key is low-entropy, then you get what you
655 deserve. The other known possibility is to use /dev/urandom for
656 key generation in an entropy-startved situation (e.g. automatic
657 installation on an embedded device without network and other entropy
660 For the passphrase, don't use a low-entropy passphrase. If your
661 passphrase is good, then a slow CPU will not matter. If you insist
662 on a low-entropy passphrase on a slow CPU, use something like
663 "--iter-time=10" or higher and wait a long time on each LUKS unlock
664 and pray that the attacker does not find out in which way exactly
665 your passphrase is low entropy. This also applies to low-entropy
666 passphrases on fast CPUs. Technology can do only so much to
667 compensate for problems in front of the keyboard.
670 * Why was the default aes-cbc-plain replaced with aes-cbc-essiv?
672 The problem is that cbc-plain has a fingerprint vulnerability, where
673 a specially crafted file placed into the crypto-container can be
674 recognized from the outside. The issue here is that for cbc-plain
675 the initialization vector (IV) is the sector number. The IV gets
676 XORed to the first data chunk of the sector to be encrypted. If you
677 make sure that the first data block to be stored in a sector
678 contains the sector number as well, the first data block to be
679 encrypted is all zeros and always encrypted to the same ciphertext.
680 This also works if the first data chunk just has a constant XOR
681 with the sector number. By having several shifted patterns you can
682 take care of the case of a non-power-of-two start sector number of
685 This mechanism allows you to create a pattern of sectors that have
686 the same first ciphertext block and signal one bit per sector to the
687 outside, allowing you to e.g. mark media files that way for
688 recognition without decryption. For large files this is a
689 practical attack. For small ones, you do not have enough blocks to
690 signal and take care of different file starting offsets.
692 In order to prevent this attack, the default was changed to
693 cbc-essiv. ESSIV uses a keyed hash of the sector number, with the
694 encryption key as key. This makes the IV unpredictable without
695 knowing the encryption key and the watermarking attack fails.
698 * Are there any problems with "plain" IV? What is "plain64"?
700 First, "plain" and "plain64" are both not secure to use with CBC,
701 see previous FAQ item.
703 However there are modes, like XTS, that are secure with "plain" IV.
704 The next limit is that "plain" is 64 bit, with the upper 32 bit set
705 to zero. This means that on volumes larger than 2TiB, the IV
706 repeats, creating a vulnerability that potentially leaks some
707 data. To avoid this, use "plain64", which uses the full sector
708 number up to 64 bit. Note that "plain64" requires a kernel >=
709 2.6.33. Also note that "plain64" is backwards compatible for
710 volume sizes <= 2TiB, but not for those > 2TiB. Finally, "plain64"
711 does not cause any performance penalty compared to "plain".
714 * What about XTS mode?
716 XTS mode is potentially even more secure than cbc-essiv (but only if
717 cbc-essiv is insecure in your scenario). It is a NIST standard and
718 used, e.g. in Truecrypt. At the moment, if you want to use it, you
719 have to specify it manually as "aes-xts-plain", i.e.
721 cryptsetup -c aes-xts-plain luksFormat <device>
723 For volumes >2TiB and kernels >= 2.6.33 use "plain64" (see FAQ
724 item on "plain" and "plain64"):
726 cryptsetup -c aes-xts-plain64 luksFormat <device>
728 There is a potential security issue with XTS mode and large blocks.
729 LUKS and dm-crypt always use 512B blocks and the issue does not
733 6. Backup and Data Recovery
736 * Does a backup compromise security?
738 Depends on how you do it. First, a backup is non-optional with
739 encrypted data just the same way it is with non-encrypted data.
740 Disks do break and they do not care whether they make plain or
741 encrypted data inaccessible. As a gideline, a well-treated HDD (!)
742 breaks with about 5% probability per year. This means everybody
743 will be hit sooner or later.
745 However there are risks introduced by backups. For example if you
746 change/disable a key-slot in LUKS, a binary backup of the partition
747 will still have the old key-slot. To deal with this, you have to
748 be able to change the key-slot on the backup as well, or use a
749 different set-up. One option is to have a different passphrase on
750 the backup and to make the backup with both containers open.
751 Another one is to make a backup of the original, opened container
752 to a single file, e.g. with tar, and to encrypt that file with
753 public-key-cryptography, e.g. with GnuPG. You can then keep the
754 secret key in a safe place, because it is only used to decrypt a
755 backup. The key the backup is encrypted with can be stored without
756 special security measures, as long as an attacker cannot replace
759 If you use dm-crypt, backup is simpler: As there is no key
760 management, the main risk is that you cannot wipe the backup when
761 wiping the original. However wiping the original for dm-crypt
762 should consist of forgetting the passphrase and that you can do
763 without actual access to the backup.
765 In both cases, there is an additional (usually small) risk: An
766 attacker can see how many sectors and which ones have been changed
767 since the backup. This is not possible with the public-key method
770 My personal advice is to use one USB disk (low value date) or
771 three disks (high value data) in rotating order for backups, and
772 either use different passphrases or keep them easily accessible
773 in case you need to disable a key-slot. If you do network-backup
774 or tape-backup, I strongly recommend to go the public-key path,
775 especially as you typically cannot reliably delete data in these
776 scenarios. (Well, you can burn the tape if it is under your
780 * What happens if I overwrite the start of a LUKS partition or damage
781 the LUKS header or key-slots?
783 There are two critical components for decryption: The salt values
784 in the header itself and the key-slots. If the salt values are
785 overwritten or changed, nothing (in the cryptographically strong
786 sense) can be done to access the data, unless there is a backup
787 of the LUKS header. If a key-slot is damaged, the data can still
788 be read with a different key-slot, if there is a remaining
789 undamaged and used key-slot. Note that in order to make a key-slot
790 unrecoverable in a cryptographically strong sense, changing about
791 4-6 bits in random locations of its 128kiB size is quite enough.
794 * What happens if I (quick) format a LUKS partition?
796 I have not tried the different ways to do this, but very likely you
797 will have written a new boot-sector, which in turn overwrites the
798 LUKS header, including the salts, making your data permanently
799 irretrivable, unless you have a LUKS header backup. You may also
800 damage the key-slots in part or in full. See also last item.
803 * What does the on-disk structure of dm-crypt look like?
805 There is none. dm-crypt takes a block device and gives encrypted
806 access to each of its blocks with a key derived from the passphrase
807 given. If you use a cipher different than the default, you have to
808 specify that as a parameter to cryptsetup too. If you want to
809 change the password, you basically have to create a second
810 encrypted device with the new passphrase and copy your data over.
811 On the plus side, if you accidentally overwrite any part of a
812 dm-crypt device, the damage will be limited to the are you
816 * What does the on-disk structure of LUKS look like?
818 A LUKS partition consists of a header, followed by 8 key-slot
819 descriptors, followed by 8 key slots, followed by the encrypted
822 Header and key-slot descriptors fill the first 592 bytes. The
823 key-slot size depends on the creation parameters, namely on the
824 number of anti-forensic stripes and on key block alignment.
826 With 4000 stripes (the default), each key-slot is a bit less than
827 128kiB in size. Due to sector alignment of the key-slot start,
828 that means the key block 0 is at offset 0x1000-0x20400, key
829 block 1 at offset 0x21000-0x40400, and key block 7 at offset
830 0xc1000-0xe0400. The space to the next full sector address is
831 padded with zeros. Never used key-slots are filled with what the
832 disk originally contained there, a key-slot removed with
833 "luksRemoveKey" or "luksKillSlot" gets filled with 0xff. Start of
834 bulk data (with the default 4000 stripes and 8 key-slots) is at
835 0x101000, i.e. at 1'052'672 bytes, i.e. at 1MiB + 4096 bytes from
836 the start of the partition. This is also the value given by
837 command "luksDump" with "Payload offset: 2056", just multiply by
838 the sector size (512 bytes). Incidentally, "luksHeaderBackup"
839 dumps exactly the first 1'052'672 bytes to file and
840 "luksHeaderRestore" restores them.
842 The exact specification of the format is here:
843 http://code.google.com/p/cryptsetup/wiki/Specification
846 * How do I backup a LUKS header?
848 While you could just copy the appropriate number of bytes from the
849 start of the LUKS partition, the best way is to use command option
850 "luksHeaderBackup" of cryptsetup. This protects also against
851 errors when non-standard parameters have been used in LUKS
852 partition creation. Example:
855 cryptsetup luksHeaderBackup --header-backup-file h /dev/mapper/c1
857 To restore, use the inverse command, i.e.
859 cryptsetup luksHeaderRestore --header-backup-file h /dev/mapper/c1
862 * How do I backup a LUKS partition?
864 You do a sector-image of the whole partition. This will contain
865 the LUKS header, the keys-slots and the data ares. It can be done
866 under Linux e.g. with dd_rescue (for a direct image copy) and with
867 "cat" or "dd". Example:
869 cat /dev/sda10 > sda10.img
870 dd_rescue /dev/sda10 sda10.img
872 You can also use any other backup software that is capable of making
873 a sector image of a partition. Note that compression is
874 ineffective for encrypted data, hence it does not make sense to
878 * Do I need a backup of the full partition? Would the header and
879 key-slots not be enough?
881 Backup protects you against two things: Disk loss or corruption
882 and user error. By far the most questions on the dm-crypt mailing
883 list about how to recover a damaged LUKS partition are related
884 to user error. For example, if you create a new filesystem on a
885 LUKS partition, chances are good that all data is lost
888 For this case, a header+key-slot backup would often be enough. But
889 keep in mind that a well-treated (!) HDD has roughly a failure
890 risk of 5% per year. It is highly advisable to have a complete
891 backup to protect against this case.
894 * Are there security risks from a backup of the LUKS header or a
895 whole LUKS partition?
897 Yes. One risk is that if you remove access rights for specific
898 key-slots by deleting their contents, the data can still be
899 accessed with invalidated passphrase and the backup. The other
900 risk is that if you erase a LUKS partition, a backup could still
901 grant access, especially if you only erased the LUKS header and
902 not the whole partition.
905 * I think this is overly complicated. Is there an alternative?
907 Yes, you can use plain dm-crypt. It does not allow multiple
908 passphrases, but on the plus side, it has zero on disk description
909 and if you overwrite some part of a plain dm-crypt partition,
910 exactly the overwritten parts are lost (rounded up to sector
914 7. Issues with Specific Versions of cryptsetup
917 * When using the create command for plain dm-crypt with cryptsetup
918 1.1.x, the mapping is incompatible and my data is not accessible
921 With cryptsetup 1.1.x, the distro maintainer can define different
922 default encryption modes for LUKS and plain devices. You can check
923 these compiled-in defaults using "cryptsetup --help". Moreover, the
924 plain device default changed because the old IV mode was
925 vulnerable to a watermarking attack.
927 If you are using a plain device and you need a compatible mode, just
928 specify cipher, key size and hash algorithm explicitly. For
929 compatibility with cryptsetup 1.0.x defaults, simple use the
932 cryptsetup create -c aes-cbc-plain -s 256 -h ripemd160 <name> <dev>
934 LUKS stores cipher and mode in the metadata on disk, avoiding this
938 * cryptsetup on SLED 10 has problems...
940 SLED 10 is missing an essential kernel patch for dm-crypt, which
941 is broken in its kernel as a result. There may be a very old
942 version of cryptsetup (1.0.x) provided by SLED, which should also
943 not be used anymore as well. My advice would be to drop SLED 10.
945 A. Contributors In no particular order: