8 6. Backup and Data Recovery
9 7. Interoperability with other Disk Encryption Tools
10 8. Issues with Specific Versions of cryptsetup
19 This is the FAQ (Frequently Asked Questions) for cryptsetup. It
20 covers Linux disk encryption with plain dm-crypt (one passphrase,
21 no management, no metadata on disk) and LUKS (multiple user keys
22 with one master key, anti-forensic features, metadata block at
23 start of device, ...). The latest version of this FAQ should
24 usually be available at
25 http://code.google.com/p/cryptsetup/wiki/FrequentlyAskedQuestions
30 ATTENTION: If you are going to read just one thing, make it the
31 section on Backup and Data Recovery. By far the most questions on
32 the cryptsetup mailing list are from people that managed to damage
33 the start of their LUKS partitions, i.e. the LUKS header. In
34 most cases, there is nothing that can be done to help these poor
35 souls recover their data. Make sure you understand the problem and
36 limitations imposed by the LUKS security model BEFORE you face
37 such a disaster! In particular, make sure you have a current header
38 backup before doing any potentially dangerous operations.
40 BACKUP: Yes, encrypted disks die, just as normal ones do. A full
41 backup is mandatory, see Section "6. Backup and Data Recovery" on
42 options for doing encrypted backup.
44 CLONING/IMAGING: If you clone or image a LUKS container, you make a
45 copy of the LUKS header and the master key will stay the same!
46 That means that if you distribute an image to several machines, the
47 same master key will be used on all of them, regardless of whether
48 you change the passphrases. Do NOT do this! If you do, a root-user
49 on any of the machines with a mapped (decrypted) container or a
50 passphrase on that machine can decrypt all other copies, breaking
51 security. See also Item 6.15.
53 DISTRIBUTION INSTALLERS: Some distribution installers offer to
54 create LUKS containers in a way that can be mistaken as activation
55 of an existing container. Creating a new LUKS container on top of
56 an existing one leads to permanent, complete and irreversible data
57 loss. It is strongly recommended to only use distribution
58 installers after a complete backup of all LUKS containers has been
61 NO WARNING ON NON-INTERACTIVE FORMAT: If you feed cryptsetup from
62 STDIN (e.g. via GnuPG) on LUKS format, it does not give you the
63 warning that you are about to format (and e.g. will lose any
64 pre-existing LUKS container on the target), as it assumes it is
65 used from a script. In this scenario, the responsibility for
66 warning the user and possibly checking for an existing LUKS header
67 is shifted to the script. This is a more general form of the
70 LUKS PASSPHRASE IS NOT THE MASTER KEY: The LUKS passphrase is not
71 used in deriving the master key. It is used in decrypting a master
72 key that is randomly selected on header creation. This means that
73 if you create a new LUKS header on top of an old one with
74 exactly the same parameters and exactly the same passphrase as the
75 old one, it will still have a different master key and your data
76 will be permanently lost.
78 PASSPHRASE CHARACTER SET: Some people have had difficulties with
79 this when upgrading distributions. It is highly advisable to only
80 use the 94 printable characters from the first 128 characters of
81 the ASCII table, as they will always have the same binary
82 representation. Other characters may have different encoding
83 depending on system configuration and your passphrase will not
84 work with a different encoding. A table of the standardized first
85 128 ASCII characters can, e.g. be found on
86 http://en.wikipedia.org/wiki/ASCII
89 * 1.3 System specific warnings
91 - Ubuntu as of 4/2011: It seems the installer offers to create
92 LUKS partitions in a way that several people mistook for an offer
93 to activate their existing LUKS partition. The installer gives no
94 or an inadequate warning and will destroy your old LUKS header,
95 causing permanent data loss. See also the section on Backup and
98 This issue has been acknowledged by the Ubuntu dev team, see here:
99 http://launchpad.net/bugs/420080
101 Update 7/2012: I am unsure whether this has been fixed bu now, best
105 * 1.4 My LUKS-device is broken! Help!
107 First: Do not panic! In many cases the data is still recoverable.
108 Do not do anything hasty! Steps:
110 - Take some deep breaths. Maybe add some relaxing music. This may
111 sound funny, but I am completely serious. Often, critical damage is
112 done only after the initial problem.
114 - Do not reboot. The keys mays still be in the kernel if the device
117 - Make sure others do not reboot the system.
119 - Do not write to your disk without a clear understanding why this
120 will not make matters worse. Do a sector-level backup before any
121 writes. Often you do not need to write at all to get enough access
122 to make a backup of the data.
126 - Read section 6 of this FAQ.
128 - Ask on the mailing-list if you need more help.
131 * 1.5 Who wrote this?
133 Current FAQ maintainer is Arno Wagner <arno@wagner.name>. Other
134 contributors are listed at the end. If you want to contribute, send
135 your article, including a descriptive headline, to the maintainer,
136 or the dm-crypt mailing list with something like "FAQ ..." in the
137 subject. You can also send more raw information and have me write
138 the section. Please note that by contributing to this FAQ, you
139 accept the license described below.
141 This work is under the "Attribution-Share Alike 3.0 Unported"
142 license, which means distribution is unlimited, you may create
143 derived works, but attributions to original authors and this
144 license statement must be retained and the derived work must be
145 under the same license. See
146 http://creativecommons.org/licenses/by-sa/3.0/ for more details of
149 Side note: I did text license research some time ago and I think
150 this license is best suited for the purpose at hand and creates the
154 * 1.5 Where is the project website?
156 There is the project website at http://code.google.com/p/cryptsetup/
157 Please do not post questions there, nobody will read them. Use
158 the mailing-list instead.
161 * 1.6 Is there a mailing-list?
163 Instructions on how to subscribe to the mailing-list are at on the
164 project website. People are generally helpful and friendly on the
167 The question of how to unsubscribe from the list does crop up
168 sometimes. For this you need your list management URL, which is
169 sent to you initially and once at the start of each month. Go to
170 the URL mentioned in the email and select "unsubscribe". This page
171 also allows you to request a password reminder.
173 Alternatively, you can send an Email to dm-crypt-request@saout.de
174 with just the word "help" in the subject or message body. Make sure
175 to send it from your list address.
177 The mailing list archive is here:
178 http://dir.gmane.org/gmane.linux.kernel.device-mapper.dm-crypt
184 * 2.1 What is the difference between "plain" and LUKS format?
186 Plain format is just that: It has no metadata on disk, reads all
187 parameters from the commandline (or the defaults), derives a
188 master-key from the passphrase and then uses that to de-/encrypt
189 the sectors of the device, with a direct 1:1 mapping between
190 encrypted and decrypted sectors.
192 Primary advantage is high resilience to damage, as one damaged
193 encrypted sector results in exactly one damaged decrypted sector.
194 Also, it is not readily apparent that there even is encrypted data
195 on the device, as an overwrite with crypto-grade randomness (e.g.
196 from /dev/urandom) looks exactly the same on disk.
198 Side-note: That has limited value against the authorities. In
199 civilized countries, they cannot force you to give up a crypto-key
200 anyways. In the US, the UK and dictatorships around the world,
201 they can force you to give up the keys (using imprisonment or worse
202 to pressure you), and in the worst case, they only need a
203 nebulous "suspicion" about the presence of encrypted data. My
204 advice is to either be ready to give up the keys or to not have
205 encrypted data when traveling to those countries, especially when
206 crossing the borders.
208 Disadvantages are that you do not have all the nice features that
209 the LUKS metadata offers, like multiple passphrases that can be
210 changed, the cipher being stored in the metadata, anti-forensic
211 properties like key-slot diffusion and salts, etc..
213 LUKS format uses a metadata header and 8 key-slot areas that are
214 being placed at the beginning of the disk, see below under "What
215 does the LUKS on-disk format looks like?". The passphrases are used
216 to decrypt a single master key that is stored in the anti-forensic
219 Advantages are a higher usability, automatic configuration of
220 non-default crypto parameters, defenses against low-entropy
221 passphrases like salting and iterated PBKDF2 passphrase hashing,
222 the ability to change passphrases, and others.
224 Disadvantages are that it is readily obvious there is encrypted
225 data on disk (but see side note above) and that damage to the
226 header or key-slots usually results in permanent data-loss. See
227 below under "6. Backup and Data Recovery" on how to reduce that
228 risk. Also the sector numbers get shifted by the length of the
229 header and key-slots and there is a loss of that size in capacity
230 (1MB+4096B for defaults and 2MB for the most commonly used
231 non-default XTS mode).
234 * 2.2 Can I encrypt an already existing, non-empty partition to use
237 There is no converter, and it is not really needed. The way to do
238 this is to make a backup of the device in question, securely wipe
239 the device (as LUKS device initialization does not clear away old
240 data), do a luksFormat, optionally overwrite the encrypted device,
241 create a new filesystem and restore your backup on the now
242 encrypted device. Also refer to sections "Security Aspects" and
243 "Backup and Data Recovery".
245 For backup, plain GNU tar works well and backs up anything likely
246 to be in a filesystem.
249 * 2.3 How do I use LUKS with a loop-device?
251 This can be very handy for experiments. Setup is just the same as
252 with any block device. If you want, for example, to use a 100MiB
253 file as LUKS container, do something like this:
255 head -c 100M /dev/zero > luksfile # create empty file
256 losetup /dev/loop0 luksfile # map luksfile to /dev/loop0
257 cryptsetup luksFormat /dev/loop0 # create LUKS on loop device
259 Afterwards just use /dev/loop0 as a you would use a LUKS partition.
260 To unmap the file when done, use "losetup -d /dev/loop0".
263 * 2.4 When I add a new key-slot to LUKS, it asks for a passphrase but
264 then complains about there not being a key-slot with that
267 That is as intended. You are asked a passphrase of an existing
268 key-slot first, before you can enter the passphrase for the new
269 key-slot. Otherwise you could break the encryption by just adding a
270 new key-slot. This way, you have to know the passphrase of one of
271 the already configured key-slots in order to be able to configure a
275 * 2.5 Encryption on top of RAID or the other way round?
277 Unless you have special needs, place encryption between RAID and
278 filesystem, i.e. encryption on top of RAID. You can do it the other
279 way round, but you have to be aware that you then need to give the
280 passphrase for each individual disk and RAID autodetection will
281 not work anymore. Therefore it is better to encrypt the RAID
282 device, e.g. /dev/dm0 .
285 * 2.6 How do I read a dm-crypt key from file?
287 Note that the file will still be hashed first, just like keyboard
288 input. Use the --key-file option, like this:
290 cryptsetup create --key-file keyfile e1 /dev/loop0
293 * 2.7 How do I read a LUKS slot key from file?
295 What you really do here is to read a passphrase from file, just as
296 you would with manual entry of a passphrase for a key-slot. You can
297 add a new passphrase to a free key-slot, set the passphrase of an
298 specific key-slot or put an already configured passphrase into a
299 file. In the last case make sure no trailing newline (0x0a) is
300 contained in the key file, or the passphrase will not work because
301 the whole file is used as input.
303 To add a new passphrase to a free key slot from file, use something
306 cryptsetup luksAddKey /dev/loop0 keyfile
308 To add a new passphrase to a specific key-slot, use something like
311 cryptsetup luksAddKey --key-slot 7 /dev/loop0 keyfile
313 To supply a key from file to any LUKS command, use the --key-file
314 option, e.g. like this:
316 cryptsetup luksOpen --key-file keyfile /dev/loop0 e1
319 * 2.8 How do I read the LUKS master key from file?
321 The question you should ask yourself first is why you would want to
322 do this. The only legitimate reason I can think of is if you want
323 to have two LUKS devices with the same master key. Even then, I
324 think it would be preferable to just use key-slots with the same
325 passphrase, or to use plain dm-crypt instead. If you really have a
326 good reason, please tell me. If I am convinced, I will add how to
330 * 2.9 What are the security requirements for a key read from file?
332 A file-stored key or passphrase has the same security requirements
333 as one entered interactively, however you can use random bytes and
334 thereby use bytes you cannot type on the keyboard. You can use any
335 file you like as key file, for example a plain text file with a
336 human readable passphrase. To generate a file with random bytes,
337 use something like this:
339 head -c 256 /dev/random > keyfile
342 * 2.10 If I map a journaled file system using dm-crypt/LUKS, does it
343 still provide its usual transactional guarantees?
345 Yes, it does, unless a very old kernel is used. The required flags
346 come from the filesystem layer and are processed and passed onwards
347 by dm-crypt. A bit more information on the process by which
348 transactional guarantees are implemented can be found here:
350 http://lwn.net/Articles/400541/
352 Please note that these "guarantees" are weaker than they appear to
353 be. One problem is that quite a few disks lie to the OS about
354 having flushed their buffers. Some other things can go wrong as
355 well. The filesystem developers are aware of these problems and
356 typically can make it work anyways. That said, dm-crypt/LUKS will
357 not make things worse.
359 One specific problem you can run into though is that you can get
360 short freezes and other slowdowns due to the encryption layer.
361 Encryption takes time and forced flushes will block for that time.
362 For example, I did run into frequent small freezes (1-2 sec) when
363 putting a vmware image on ext3 over dm-crypt. When I went back to
364 ext2, the problem went away. This seems to have gotten better with
365 kernel 2.6.36 and the reworking of filesystem flush locking
366 mechanism (less blocking of CPU activity during flushes). It
367 should improve further and eventually the problem should go away.
370 * 2.11 Can I use LUKS or cryptsetup with a more secure (external)
371 medium for key storage, e.g. TPM or a smartcard?
373 Yes, see the answers on using a file-supplied key. You do have to
374 write the glue-logic yourself though. Basically you can have
375 cryptsetup read the key from STDIN and write it there with your
376 own tool that in turn gets the key from the more secure key
380 * 2.12 Can I resize a dm-crypt or LUKS partition?
382 Yes, you can, as neither dm-crypt nor LUKS stores partition size.
383 Whether you should is a different question. Personally I recommend
384 backup, recreation of the encrypted partition with new size,
385 recreation of the filesystem and restore. This gets around the
386 tricky business of resizing the filesystem. Resizing a dm-crypt or
387 LUKS container does not resize the filesystem in it. The backup is
388 really non-optional here, as a lot can go wrong, resulting in
389 partial or complete data loss. Using something like gparted to
390 resize an encrypted partition is slow, but typically works. This
391 will not change the size of the filesystem hidden under the
394 You also need to be aware of size-based limitations. The one
395 currently relevant is that aes-xts-plain should not be used for
396 encrypted container sizes larger than 2TiB. Use aes-xts-plain64
403 * 3.1 My dm-crypt/LUKS mapping does not work! What general steps are
404 there to investigate the problem?
406 If you get a specific error message, investigate what it claims
407 first. If not, you may want to check the following things.
409 - Check that "/dev", including "/dev/mapper/control" is there. If it
410 is missing, you may have a problem with the "/dev" tree itself or
411 you may have broken udev rules.
413 - Check that you have the device mapper and the crypt target in your
414 kernel. The output of "dmsetup targets" should list a "crypt"
415 target. If it is not there or the command fails, add device mapper
416 and crypt-target to the kernel.
418 - Check that the hash-functions and ciphers you want to use are in
419 the kernel. The output of "cat /proc/crypto" needs to list them.
422 * 3.2 My dm-crypt mapping suddenly stopped when upgrading cryptsetup.
424 The default cipher, hash or mode may have changed (the mode changed
425 from 1.0.x to 1.1.x). See under "Issues With Specific Versions of
429 * 3.3 When I call cryptsetup from cron/CGI, I get errors about
432 If you get errors about unknown parameters or the like that are not
433 present when cryptsetup is called from the shell, make sure you
434 have no older version of cryptsetup on your system that then gets
435 called by cron/CGI. For example some distributions install
436 cryptsetup into /usr/sbin, while a manual install could go to
437 /usr/local/sbin. As a debugging aid, call "cryptsetup --version"
438 from cron/CGI or the non-shell mechanism to be sure the right
442 * 3.4 Unlocking a LUKS device takes very long. Why?
444 The iteration time for a key-slot (see Section 5 for an explanation
445 what iteration does) is calculated when setting a passphrase. By
446 default it is 1 second on the machine where the passphrase is set.
447 If you set a passphrase on a fast machine and then unlock it on a
448 slow machine, the unlocking time can be much longer. Also take into
449 account that up to 8 key-slots have to be tried in order to find the
452 If this is problem, you can add another key-slot using the slow
453 machine with the same passphrase and then remove the old key-slot.
454 The new key-slot will have an iteration count adjusted to 1 second
455 on the slow machine. Use luksKeyAdd and then luksKillSlot or
458 However, this operation will not change volume key iteration count
459 (MK iterations in output of "cryptsetup luksDump"). In order to
460 change that, you will have to backup the data in the LUKS
461 container (i.e. your encrypted data), luksFormat on the slow
462 machine and restore the data. Note that in the original LUKS
463 specification this value was fixed to 10, but it is now derived
464 from the PBKDF2 benchmark as well and set to iterations in 0.125
465 sec or 1000, whichever is larger. Also note that MK iterations
466 are not very security relevant. But as each key-slot already takes
467 1 second, spending the additional 0.125 seconds really does not
471 * 3.5 "blkid" sees a LUKS UUID and an ext2/swap UUID on the same
472 device. What is wrong?
474 Some old versions of cryptsetup have a bug where the header does
475 not get completely wiped during LUKS format and an older ext2/swap
476 signature remains on the device. This confuses blkid.
478 Fix: Wipe the unused header areas by doing a backup and restore of
479 the header with cryptsetup 1.1.x:
481 cryptsetup luksHeaderBackup --header-backup-file <file> <device>
482 cryptsetup luksHeaderRestore --header-backup-file <file> <device>
485 * 3.6 cryptsetup segfaults on Gentoo amd64 hardened ...
487 There seems to be some interference between the hardening and and
488 the way cryptsetup benchmarks PBKDF2. The solution to this is
489 currently not quite clear for an encrypted root filesystem. For
490 other uses, you can apparently specify USE="dynamic" as compile
491 flag, see http://bugs.gentoo.org/show_bug.cgi?id=283470
497 * 4.1 I get the error "LUKS keyslot x is invalid." What does that
500 This means that the given keyslot has an offset that points
501 outside the valid keyslot area. Typically, the reason is a
502 corrupted LUKS header because something was written to the start of
503 the device the LUKS container is on. Refer to Section "Backup and
504 Data Recovery" and ask on the mailing list if you have trouble
505 diagnosing and (if still possible) repairing this.
508 * 4.2 Can a bad RAM module cause problems?
510 LUKS and dm-crypt can give the RAM quite a workout, especially when
511 combined with software RAID. In particular the combination RAID5 +
512 LUKS + XFS seems to uncover RAM problems that never caused obvious
513 problems before. Symptoms vary, but often the problem manifest
514 itself when copying large amounts of data, typically several times
515 larger than your main memory.
517 Side note: One thing you should always do on large data
518 copy/movements is to run a verify, for example with the "-d"
519 option of "tar" or by doing a set of MD5 checksums on the source
522 find . -type f -exec md5sum \{\} \; > checksum-file
524 and then a "md5sum -c checksum-file" on the other side. If you get
525 mismatches here, RAM is the primary suspect. A lesser suspect is
526 an overclocked CPU. I have found countless hardware problems in
527 verify runs after copying or making backups. Bit errors are much
528 more common than most people think.
530 Some RAM issues are even worse and corrupt structures in one of the
531 layers. This typically results in lockups, CPU state dumps in the
532 system logs, kernel panic or other things. It is quite possible to
533 have the problem with an encrypted device, but not with an
534 otherwise the same unencrypted device. The reason for that is that
535 encryption has an error amplification property: You flip one bit
536 in an encrypted data block, and the decrypted version has half of
537 its bits flipped. This is an important security property for modern
538 ciphers. With the usual modes in cryptsetup (CBC, ESSIV, XTS), you
539 get up to a completely changed 512 byte block per bit error. A
540 corrupt block causes a lot more havoc than the occasionally
541 flipped single bit and can result in various obscure errors.
543 Note, that a verify run on copying between encrypted or
544 unencrypted devices will reliably detect corruption, even when the
545 copying itself did not report any problems. If you find defect
546 RAM, assume all backups and copied data to be suspect, unless you
550 * 4.3 How do I test RAM?
552 First you should know that overclocking often makes memory
553 problems worse. So if you overclock (which I strongly recommend
554 against in a system holding data that has some worth), run the
555 tests with the overclocking active.
557 There are two good options. One is Memtest86+ and the other is
558 "memtester" by Charles Cazabon. Memtest86+ requires a reboot and
559 then takes over the machine, while memtester runs from a
560 root-shell. Both use different testing methods and I have found
561 problems fast with each one that the other needed long to find. I
562 recommend running the following procedure until the first error is
565 - Run Memtest86+ for one cycle
567 - Run memtester for one cycle (shut down as many other applications
570 - Run Memtest86+ for 24h or more
572 - Run memtester for 24h or more
574 If all that does not produce error messages, your RAM may be sound,
575 but I have had one weak bit that Memtest86+ needed around 60 hours
576 to find. If you can reproduce the original problem reliably, a good
577 additional test may be to remove half of the RAM (if you have more
578 than one module) and try whether the problem is still there and if
579 so, try with the other half. If you just have one module, get a
580 different one and try with that. If you do overclocking, reduce
581 the settings to the most conservative ones available and try with
588 * 5.1 Is LUKS insecure? Everybody can see I have encrypted data!
590 In practice it does not really matter. In most civilized countries
591 you can just refuse to hand over the keys, no harm done. In some
592 countries they can force you to hand over the keys, if they suspect
593 encryption. However the suspicion is enough, they do not have to
594 prove anything. This is for practical reasons, as even the presence
595 of a header (like the LUKS header) is not enough to prove that you
596 have any keys. It might have been an experiment, for example. Or it
597 was used as encrypted swap with a key from /dev/random. So they
598 make you prove you do not have encrypted data. Of course that is
599 just as impossible as the other way round.
601 This means that if you have a large set of random-looking data,
602 they can already lock you up. Hidden containers (encryption hidden
603 within encryption), as possible with Truecrypt, do not help
604 either. They will just assume the hidden container is there and
605 unless you hand over the key, you will stay locked up. Don't have
606 a hidden container? Though luck. Anybody could claim that.
608 Still, if you are concerned about the LUKS header, use plain
609 dm-crypt with a good passphrase. See also Section 2, "What is the
610 difference between "plain" and LUKS format?"
613 * 5.2 Should I initialize (overwrite) a new LUKS/dm-crypt partition?
615 If you just create a filesystem on it, most of the old data will
616 still be there. If the old data is sensitive, you should overwrite
617 it before encrypting. In any case, not initializing will leave the
618 old data there until the specific sector gets written. That may
619 enable an attacker to determine how much and where on the
620 partition data was written. If you think this is a risk, you can
621 prevent this by overwriting the encrypted device (here assumed to
622 be named "e1") with zeros like this:
624 dd_rescue -w /dev/zero /dev/mapper/e1
626 or alternatively with one of the following more standard commands:
628 cat /dev/zero > /dev/mapper/e1
629 dd if=/dev/zero of=/dev/mapper/e1
632 * 5.3 How do I securely erase a LUKS (or other) partition?
634 For LUKS, if you are in a desperate hurry, overwrite the LUKS
635 header and key-slot area. This means overwriting the first
636 (keyslots x stripes x keysize) + offset bytes. For the default
637 parameters, this is the 1'052'672 bytes, i.e. 1MiB + 4096 of the
638 LUKS partition. For 512 bit key length (e.g. for aes-xts-plain with
639 512 bit key) this is 2MiB. (The different offset stems from
640 differences in the sector alignment of the key-slots.) If in doubt,
641 just be generous and overwrite the first 10MB or so, it will likely
642 still be fast enough. A single overwrite with zeros should be
643 enough. If you anticipate being in a desperate hurry, prepare the
644 command beforehand. Example with /dev/sde1 as the LUKS partition
645 and default parameters:
647 head -c 1052672 /dev/zero > /dev/sde1; sync
649 A LUKS header backup or full backup will still grant access to
650 most or all data, so make sure that an attacker does not have
651 access to backups or destroy them as well.
653 If you have time, overwrite the whole LUKS partition with a single
654 pass of zeros. This is enough for current HDDs. For SSDs or FLASH
655 (USB sticks) you may want to overwrite the whole drive several
656 times to be sure data is not retained by wear leveling. This is
657 possibly still insecure as SSD technology is not fully understood
658 in this regard. Still, due to the anti-forensic properties of the
659 LUKS key-slots, a single overwrite of an SSD or FLASH drive could
660 be enough. If in doubt, use physical destruction in addition. Here
661 is a link to some current research results on erasing SSDs and
663 http://www.usenix.org/events/fast11/tech/full_papers/Wei.pdf
665 Keep in mind to also erase all backups.
667 Example for a zero-overwrite erase of partition sde1 done with
670 dd_rescue -w /dev/zero /dev/sde1
673 * 5.4 How do I securely erase a backup of a LUKS partition or header?
675 That depends on the medium it is stored on. For HDD and SSD, use
676 overwrite with zeros. For an SSD or FLASH drive (USB stick), you
677 may want to overwrite the complete SSD several times and use
678 physical destruction in addition, see last item. For re-writable
679 CD/DVD, a single overwrite should also be enough, due to the
680 anti-forensic properties of the LUKS keyslots. For write-once
681 media, use physical destruction. For low security requirements,
682 just cut the CD/DVD into several parts. For high security needs,
683 shred or burn the medium. If your backup is on magnetic tape, I
684 advise physical destruction by shredding or burning, after
685 overwriting . The problem with magnetic tape is that it has a
686 higher dynamic range than HDDs and older data may well be
687 recoverable after overwrites. Also write-head alignment issues can
688 lead to data not actually being deleted at all during overwrites.
691 * 5.5 What about backup? Does it compromise security?
693 That depends. See item 6.7.
696 * 5.6 Why is all my data permanently gone if I overwrite the LUKS
699 Overwriting the LUKS header in part or in full is the most common
700 reason why access to LUKS containers is lost permanently.
701 Overwriting can be done in a number of fashions, like creating a
702 new filesystem on the raw LUKS partition, making the raw partition
703 part of a raid array and just writing to the raw partition.
705 The LUKS header contains a 256 bit "salt" value and without that no
706 decryption is possible. While the salt is not secret, it is
707 key-grade material and cannot be reconstructed. This is a
708 cryptographically strong "cannot". From observations on the
709 cryptsetup mailing-list, people typically go though the usual
710 stages of grief (Denial, Anger, Bargaining, Depression, Acceptance)
711 when this happens to them. Observed times vary between 1 day and 2
712 weeks to complete the cycle. Seeking help on the mailing-list is
713 fine. Even if we usually cannot help with getting back your data,
714 most people found the feedback comforting.
716 If your header does not contain an intact salt, best go directly
717 to the last stage ("Acceptance") and think about what to do now.
718 There is one exception that I know of: If your LUKS container is
719 still open, then it may be possible to extract the master key from
720 the running system. See Item "How do I recover the master key from
721 a mapped LUKS container?" in Section "Backup and Data Recovery".
724 * 5.7 What is a "salt"?
726 A salt is a random key-grade value added to the passphrase before
727 it is processed. It is not kept secret. The reason for using salts
728 is as follows: If an attacker wants to crack the password for a
729 single LUKS container, then every possible passphrase has to be
730 tried. Typically an attacker will not try every binary value, but
731 will try words and sentences from a dictionary.
733 If an attacker wants to attack several LUKS containers with the
734 same dictionary, then a different approach makes sense: Compute the
735 resulting slot-key for each dictionary element and store it on
736 disk. Then the test for each entry is just the slow unlocking with
737 the slot key (say 0.00001 sec) instead of calculating the slot-key
738 first (1 sec). For a single attack, this does not help. But if you
739 have more than one container to attack, this helps tremendously,
740 also because you can prepare your table before you even have the
741 container to attack! The calculation is also very simple to
742 parallelize. You could, for example, use the night-time unused CPU
743 power of your desktop PCs for this.
745 This is where the salt comes in. If the salt is combined with the
746 passphrase (in the simplest form, just appended to it), you
747 suddenly need a separate table for each salt value. With a
748 reasonably-sized salt value (256 bit, e.g.) this is quite
752 * 5.8 Is LUKS secure with a low-entropy (bad) passphrase?
754 Note: You should only use the 94 printable characters from 7 bit
755 ASCII code to prevent your passphrase from failing when the
756 character encoding changes, e.g. because of a system upgrade, see
757 also the note at the very start of this FAQ under "WARNINGS".
759 This needs a bit of theory. The quality of your passphrase is
760 directly related to its entropy (information theoretic, not
761 thermodynamic). The entropy says how many bits of "uncertainty" or
762 "randomness" are in you passphrase. In other words, that is how
763 difficult guessing the passphrase is.
765 Example: A random English sentence has about 1 bit of entropy per
766 character. A random lowercase (or uppercase) character has about
769 Now, if n is the number of bits of entropy in your passphrase and t
770 is the time it takes to process a passphrase in order to open the
771 LUKS container, then an attacker has to spend at maximum
773 attack_time_max = 2^n * t
775 time for a successful attack and on average half that. There is no
776 way getting around that relationship. However, there is one thing
777 that does help, namely increasing t, the time it takes to use a
778 passphrase, see next FAQ item.
780 Still, if you want good security, a high-entropy passphrase is the
781 only option. For example, a low-entropy passphrase can never be
782 considered secure against a TLA-level (Three Letter Agency level,
783 i.e. government-level) attacker, no matter what tricks are used in
784 the key-derivation function. Use at least 64 bits for secret stuff.
785 That is 64 characters of English text (but only if randomly chosen)
786 or a combination of 12 truly random letters and digits.
788 For passphrase generation, do not use lines from very well-known
789 texts (religious texts, Harry potter, etc.) as they are to easy to
790 guess. For example, the total Harry Potter has about 1'500'000
791 words (my estimation). Trying every 64 character sequence starting
792 and ending at a word boundary would take only something like 20
793 days on a single CPU and is entirely feasible. To put that into
794 perspective, using a number of Amazon EC2 High-CPU Extra Large
795 instances (each gives about 8 real cores), this test costs
796 currently about 50USD/EUR, but can be made to run arbitrarily fast.
798 On the other hand, choosing 1.5 lines from, say, the Wheel of Time
799 is in itself not more secure, but the book selection adds quite a
800 bit of entropy. (Now that I have mentioned it here, don't use tWoT
801 either!) If you add 2 or 3 typos or switch some words around, then
802 this is good passphrase material.
805 * 5.9 What is "iteration count" and why is decreasing it a bad idea?
807 Iteration count is the number of PBKDF2 iterations a passphrase is
808 put through before it is used to unlock a key-slot. Iterations are
809 done with the explicit purpose to increase the time that it takes
810 to unlock a key-slot. This provides some protection against use of
811 low-entropy passphrases.
813 The idea is that an attacker has to try all possible passphrases.
814 Even if the attacker knows the passphrase is low-entropy (see last
815 item), it is possible to make each individual try take longer. The
816 way to do this is to repeatedly hash the passphrase for a certain
817 time. The attacker then has to spend the same time (given the same
818 computing power) as the user per try. With LUKS, the default is 1
819 second of PBKDF2 hashing.
821 Example 1: Lets assume we have a really bad passphrase (e.g. a
822 girlfriends name) with 10 bits of entropy. With the same CPU, an
823 attacker would need to spend around 500 seconds on average to
824 break that passphrase. Without iteration, it would be more like
825 0.0001 seconds on a modern CPU.
827 Example 2: The user did a bit better and has 32 chars of English
828 text. That would be about 32 bits of entropy. With 1 second
829 iteration, that means an attacker on the same CPU needs around 136
830 years. That is pretty impressive for such a weak passphrase.
831 Without the iterations, it would be more like 50 days on a modern
832 CPU, and possibly far less.
834 In addition, the attacker can both parallelize and use special
835 hardware like GPUs or FPGAs to speed up the attack. The attack can
836 also happen quite some time after the luksFormat operation and CPUs
837 can have become faster and cheaper. For that reason you want a
838 bit of extra security. Anyways, in Example 1 your are screwed.
839 In example 2, not necessarily. Even if the attack is faster, it
840 still has a certain cost associated with it, say 10000 EUR/USD
841 with iteration and 1 EUR/USD without iteration. The first can be
842 prohibitively expensive, while the second is something you try
843 even without solid proof that the decryption will yield something
846 The numbers above are mostly made up, but show the idea. Of course
847 the best thing is to have a high-entropy passphrase.
849 Would a 100 sec iteration time be even better? Yes and no.
850 Cryptographically it would be a lot better, namely 100 times better.
851 However, usability is a very important factor for security
852 technology and one that gets overlooked surprisingly often. For
853 LUKS, if you have to wait 2 minutes to unlock the LUKS container,
854 most people will not bother and use less secure storage instead. It
855 is better to have less protection against low-entropy passphrases
856 and people actually use LUKS, than having them do without
857 encryption altogether.
859 Now, what about decreasing the iteration time? This is generally a
860 very bad idea, unless you know and can enforce that the users only
861 use high-entropy passphrases. If you decrease the iteration time
862 without ensuring that, then you put your users at increased risk,
863 and considering how rarely LUKS containers are unlocked in a
864 typical work-flow, you do so without a good reason. Don't do it.
865 The iteration time is already low enough that users with entropy
866 low passphrases are vulnerable. Lowering it even further increases
867 this danger significantly.
870 * 5.10 Some people say PBKDF2 is insecure?
872 There is some discussion that a hash-function should have a "large
873 memory" property, i.e. that it should require a lot of memory to be
874 computed. This serves to prevent attacks using special programmable
875 circuits, like FPGAs, and attacks using graphics cards. PBKDF2
876 does not need a lot of memory and is vulnerable to these attacks.
877 However, the publication usually referred in these discussions is
878 not very convincing in proving that the presented hash really is
879 "large memory" (that may change, email the FAQ maintainer when it
880 does) and it is of limited usefulness anyways. Attackers that use
881 clusters of normal PCs will not be affected at all by a "large
882 memory" property. For example the US Secret Service is known to
883 use the off-hour time of all the office PCs of the Treasury for
884 password breaking. The Treasury has about 110'000 employees.
885 Assuming every one has an office PC, that is significant computing
886 power, all of it with plenty of memory for computing "large
887 memory" hashes. Bot-net operators also have all the memory they
888 want. The only protection against a resourceful attacker is a
889 high-entropy passphrase, see items 5.8 and 5.9.
892 * 5.11 What about iteration count with plain dm-crypt?
894 Simple: There is none. There is also no salting. If you use plain
895 dm-crypt, the only way to be secure is to use a high entropy
896 passphrase. If in doubt, use LUKS instead.
899 * 5.12 Is LUKS with default parameters less secure on a slow CPU?
901 Unfortunately, yes. However the only aspect affected is the
902 protection for low-entropy passphrase or master-key. All other
903 security aspects are independent of CPU speed.
905 The master key is less critical, as you really have to work at it
906 to give it low entropy. One possibility is to supply the master key
907 yourself. If that key is low-entropy, then you get what you
908 deserve. The other known possibility is to use /dev/urandom for
909 key generation in an entropy-starved situation (e.g. automatic
910 installation on an embedded device without network and other entropy
913 For the passphrase, don't use a low-entropy passphrase. If your
914 passphrase is good, then a slow CPU will not matter. If you insist
915 on a low-entropy passphrase on a slow CPU, use something like
916 "--iter-time=10" or higher and wait a long time on each LUKS unlock
917 and pray that the attacker does not find out in which way exactly
918 your passphrase is low entropy. This also applies to low-entropy
919 passphrases on fast CPUs. Technology can do only so much to
920 compensate for problems in front of the keyboard.
923 * 5.13 Why was the default aes-cbc-plain replaced with aes-cbc-essiv?
925 Note: This item applies both to plain dm-crypt and to LUKS
927 The problem is that cbc-plain has a fingerprint vulnerability, where
928 a specially crafted file placed into the crypto-container can be
929 recognized from the outside. The issue here is that for cbc-plain
930 the initialization vector (IV) is the sector number. The IV gets
931 XORed to the first data chunk of the sector to be encrypted. If you
932 make sure that the first data block to be stored in a sector
933 contains the sector number as well, the first data block to be
934 encrypted is all zeros and always encrypted to the same ciphertext.
935 This also works if the first data chunk just has a constant XOR
936 with the sector number. By having several shifted patterns you can
937 take care of the case of a non-power-of-two start sector number of
940 This mechanism allows you to create a pattern of sectors that have
941 the same first ciphertext block and signal one bit per sector to the
942 outside, allowing you to e.g. mark media files that way for
943 recognition without decryption. For large files this is a
944 practical attack. For small ones, you do not have enough blocks to
945 signal and take care of different file starting offsets.
947 In order to prevent this attack, the default was changed to
948 cbc-essiv. ESSIV uses a keyed hash of the sector number, with the
949 encryption key as key. This makes the IV unpredictable without
950 knowing the encryption key and the watermarking attack fails.
953 * 5.14 Are there any problems with "plain" IV? What is "plain64"?
955 First, "plain" and "plain64" are both not secure to use with CBC,
956 see previous FAQ item.
958 However there are modes, like XTS, that are secure with "plain" IV.
959 The next limit is that "plain" is 64 bit, with the upper 32 bit set
960 to zero. This means that on volumes larger than 2TiB, the IV
961 repeats, creating a vulnerability that potentially leaks some
962 data. To avoid this, use "plain64", which uses the full sector
963 number up to 64 bit. Note that "plain64" requires a kernel >=
964 2.6.33. Also note that "plain64" is backwards compatible for
965 volume sizes <= 2TiB, but not for those > 2TiB. Finally, "plain64"
966 does not cause any performance penalty compared to "plain".
969 * 5.15 What about XTS mode?
971 XTS mode is potentially even more secure than cbc-essiv (but only if
972 cbc-essiv is insecure in your scenario). It is a NIST standard and
973 used, e.g. in Truecrypt. At the moment, if you want to use it, you
974 have to specify it manually as "aes-xts-plain", i.e.
976 cryptsetup -c aes-xts-plain luksFormat <device>
978 For volumes >2TiB and kernels >= 2.6.33 use "plain64" (see FAQ
979 item on "plain" and "plain64"):
981 cryptsetup -c aes-xts-plain64 luksFormat <device>
983 There is a potential security issue with XTS mode and large blocks.
984 LUKS and dm-crypt always use 512B blocks and the issue does not
988 * 5.16 Is LUKS FIPS-140-2 certified?
990 No. But that is more a problem of FIPS-140-2 than of LUKS. From a
991 technical point-of-view, LUKS with the right parameters would be
992 FIPS-140-2 compliant, but in order to make it certified, somebody
993 has to pay real money for that. And then, whenever cryptsetup is
994 changed or extended, the certification lapses and has to be
997 From the aspect of actual security, LUKS with default parameters
998 should be as good as most things that are FIPS-140-2 certified,
999 although you may want to make sure to use /dev/random (by
1000 specifying --use-random on luksFormat) as randomness source for
1001 the master key to avoid being potentially insecure in an
1002 entropy-starved situation.
1005 * 5.16 What about Plausible Deniability?
1007 First let me attempt a definition for the case of encrypted
1008 filesystems: Plausible deniability is when you hide encrypted data
1009 inside an encrypted container and it is not possible to prove it is
1010 there. The idea is compelling and on first glance it seems
1011 possible to do it. And from a cryptographic point of view, it
1012 actually is possible.
1014 So, does it work in practice? No, unfortunately. The reasoning used
1015 by its proponents is fundamentally flawed in several ways and the
1016 cryptographic properties fail fatally when colliding with the real
1019 First, why should "I do not have a hidden partition" be any more
1020 plausible than "I forgot my crypto key" or "I wiped that partition
1021 with random data, nothing in there"? I do not see any reason.
1023 Second, there are two types of situations: Either they cannot force
1024 you to give them the key (then you simply do not) or the can. In
1025 the second case, they can always do bad things to you, because they
1026 cannot prove that you have the key in the first place! This means
1027 they do not have to prove you have the key, or that this random
1028 looking data on your disk is actually encrypted data. So the
1029 situation will allow them to waterboard/lock-up/deport you
1030 anyways, regardless of how "plausible" your deniability is. Do not
1031 have a hidden partition you could show to them, but there are
1032 indications you may? Too bad for you. Unfortunately "plausible
1033 deniability" also means you cannot prove there is no hidden data.
1035 Third, hidden partitions are not that hidden. There are basically
1036 just two possibilities: a) Make a large crypto container, but put a
1037 smaller filesystem in there and put the hidden partition into the
1038 free space. Unfortunately this is glaringly obvious and can be
1039 detected in an automated fashion. This means that the initial
1040 suspicion to put you under duress in order to make you reveal you
1041 hidden data is given. b) Make a filesystem that spans the whole
1042 encrypted partition, and put the hidden partition into space not
1043 currently used by that filesystem. Unfortunately that is also
1044 glaringly obvious, as you then cannot write to the filesystem
1045 without a high risk of destroying data in the hidden container.
1046 Have not written anything to the encrypted filesystem in a while?
1047 Too bad, they have the suspicion they need to do unpleasant things
1050 To be fair, if you prepare option b) carefully and directly before
1051 going into danger, it may work. But then, the mere presence of
1052 encrypted data may already be enough to get you into trouble in
1053 those places were they can demand encryption keys.
1055 Here is an additional reference for some problems with plausible
1056 deniability: http://www.schneier.com/paper-truecrypt-dfs.pdf I
1057 strongly suggest you read it.
1059 So, no, I will not provide any instructions on how to do it with
1060 plain dm-crypt or LUKS. If you insist on shooting yourself in the
1061 foot, you can figure out how to do it yourself.
1064 6. Backup and Data Recovery
1067 * 6.1 Why do I need Backup?
1069 First, disks die. The rate for well-treated (!) disk is about 5%
1070 per year, which is high enough to worry about. There is some
1071 indication that this may be even worse for some SSDs. This applies
1072 both to LUKS and plain dm-crypt partitions.
1074 Second, for LUKS, if anything damages the LUKS header or the
1075 key-stripe area then decrypting the LUKS device can become
1076 impossible. This is a frequent occurrence. For example an
1077 accidental format as FAT or some software overwriting the first
1078 sector where it suspects a partition boot sector typically makes a
1079 LUKS partition permanently inaccessible. See more below on LUKS
1082 So, data-backup in some form is non-optional. For LUKS, you may
1083 also want to store a header backup in some secure location. This
1084 only needs an update if you change passphrases.
1087 * 6.2 How do I backup a LUKS header?
1089 While you could just copy the appropriate number of bytes from the
1090 start of the LUKS partition, the best way is to use command option
1091 "luksHeaderBackup" of cryptsetup. This protects also against
1092 errors when non-standard parameters have been used in LUKS
1093 partition creation. Example:
1096 cryptsetup luksHeaderBackup --header-backup-file <file> <device>
1098 To restore, use the inverse command, i.e.
1100 cryptsetup luksHeaderRestore --header-backup-file <file> <device>
1103 * 6.3 How do I test a LUKS header?
1107 cryptsetup -v isLuks <device>
1109 on the device. Without the "-v" it just signals its result via
1110 exit-status. You can also use the more general test
1114 which will also detect other types and give some more info. Omit
1115 "-p" for old versions of blkid that do not support it.
1118 * 6.4 How do I backup a LUKS or dm-crypt partition?
1120 There are two options, a sector-image and a plain file or
1121 filesystem backup of the contents of the partition. The sector
1122 image is already encrypted, but cannot be compressed and contains
1123 all empty space. The filesystem backup can be compressed, can
1124 contain only part of the encrypted device, but needs to be
1125 encrypted separately if so desired.
1127 A sector-image will contain the whole partition in encrypted form,
1128 for LUKS the LUKS header, the keys-slots and the data area. It can
1129 be done under Linux e.g. with dd_rescue (for a direct image copy)
1130 and with "cat" or "dd". Example:
1132 cat /dev/sda10 > sda10.img
1133 dd_rescue /dev/sda10 sda10.img
1135 You can also use any other backup software that is capable of making
1136 a sector image of a partition. Note that compression is
1137 ineffective for encrypted data, hence it does not make sense to
1140 For a filesystem backup, you decrypt and mount the encrypted
1141 partition and back it up as you would a normal filesystem. In this
1142 case the backup is not encrypted, unless your encryption method
1143 does that. For example you can encrypt a backup with "tar" as
1146 tar cjf - <path> | gpg --cipher-algo AES -c - > backup.tbz2.gpg
1148 And verify the backup like this if you are at "path":
1150 cat backup.tbz2.gpg | gpg - | tar djf -
1152 Note: Always verify backups, especially encrypted ones.
1154 In both cases GnuPG will ask you interactively for your symmetric
1155 key. The verify will only output errors. Use "tar dvjf -" to get
1156 all comparison results. To make sure no data is written to disk
1157 unencrypted, turn off swap if it is not encrypted before doing the
1160 You can of course use different or no compression and you can use
1161 an asymmetric key if you have one and have a backup of the secret
1162 key that belongs to it.
1164 A second option for a filesystem-level backup that can be used when
1165 the backup is also on local disk (e.g. an external USB drive) is
1166 to use a LUKS container there and copy the files to be backed up
1167 between both mounted containers. Also see next item.
1170 * 6.5 Do I need a backup of the full partition? Would the header and
1171 key-slots not be enough?
1173 Backup protects you against two things: Disk loss or corruption
1174 and user error. By far the most questions on the dm-crypt mailing
1175 list about how to recover a damaged LUKS partition are related
1176 to user error. For example, if you create a new filesystem on a
1177 LUKS partition, chances are good that all data is lost
1180 For this case, a header+key-slot backup would often be enough. But
1181 keep in mind that a well-treated (!) HDD has roughly a failure
1182 risk of 5% per year. It is highly advisable to have a complete
1183 backup to protect against this case.
1186 * *6.6 What do I need to backup if I use "decrypt_derived"?
1188 This is a script in Debian, intended for mounting /tmp or swap with
1189 a key derived from the master key of an already decrypted device.
1190 If you use this for an device with data that should be persistent,
1191 you need to make sure you either do not lose access to that master
1192 key or have a backup of the data. If you derive from a LUKS
1193 device, a header backup of that device would cover backing up the
1194 master key. Keep in mind that this does not protect against disk
1197 Note: If you recreate the LUKS header of the device you derive from
1198 (using luksFormat), the master key changes even if you use the same
1199 passphrase(s) and you will not be able to decrypt the derived
1200 device with the new LUKS header.
1203 * 6.7 Does a backup compromise security?
1205 Depends on how you do it. However if you do not have one, you are
1206 going to eventually lose your encrypted data.
1208 There are risks introduced by backups. For example if you
1209 change/disable a key-slot in LUKS, a binary backup of the partition
1210 will still have the old key-slot. To deal with this, you have to
1211 be able to change the key-slot on the backup as well, securely
1212 erase the backup or do a filesystem-level backup instead of a binary
1215 If you use dm-crypt, backup is simpler: As there is no key
1216 management, the main risk is that you cannot wipe the backup when
1217 wiping the original. However wiping the original for dm-crypt
1218 should consist of forgetting the passphrase and that you can do
1219 without actual access to the backup.
1221 In both cases, there is an additional (usually small) risk with
1222 binary backups: An attacker can see how many sectors and which
1223 ones have been changed since the backup. To prevent this, use a
1224 filesystem level backup method that encrypts the whole backup in
1225 one go, e.g. as described above with tar and GnuPG.
1227 My personal advice is to use one USB disk (low value data) or
1228 three disks (high value data) in rotating order for backups, and
1229 either use independent LUKS partitions on them, or use encrypted
1230 backup with tar and GnuPG.
1232 If you do network-backup or tape-backup, I strongly recommend to
1233 go the filesystem backup path with independent encryption, as you
1234 typically cannot reliably delete data in these scenarios,
1235 especially in a cloud setting. (Well, you can burn the tape if it
1236 is under your control...)
1239 * 6.8 What happens if I overwrite the start of a LUKS partition or
1240 damage the LUKS header or key-slots?
1242 There are two critical components for decryption: The salt values
1243 in the header itself and the key-slots. If the salt values are
1244 overwritten or changed, nothing (in the cryptographically strong
1245 sense) can be done to access the data, unless there is a backup
1246 of the LUKS header. If a key-slot is damaged, the data can still
1247 be read with a different key-slot, if there is a remaining
1248 undamaged and used key-slot. Note that in order to make a key-slot
1249 unrecoverable in a cryptographically strong sense, changing about
1250 4-6 bits in random locations of its 128kiB size is quite enough.
1253 * 6.9 What happens if I (quick) format a LUKS partition?
1255 I have not tried the different ways to do this, but very likely you
1256 will have written a new boot-sector, which in turn overwrites the
1257 LUKS header, including the salts, making your data permanently
1258 irretrievable, unless you have a LUKS header backup. You may also
1259 damage the key-slots in part or in full. See also last item.
1262 * 6.10 How do I recover the master key from a mapped LUKS container?
1264 This is typically only needed if you managed to damage your LUKS
1265 header, but the container is still mapped, i.e. "luksOpen"ed. It
1266 also helps if you have a mapped container that you forgot or do not
1267 know a passphrase for (e.g. on a long running server.)
1269 WARNING: Things go wrong, do a full backup before trying this!
1271 WARNING: This exposes the master key of the LUKS container. Note
1272 that both ways to recreate a LUKS header with the old master key
1273 described below will write the master key to disk. Unless you are
1274 sure you have securely erased it afterwards, e.g. by writing it to
1275 an encrypted partition, RAM disk or by erasing the filesystem you
1276 wrote it to by a complete overwrite, you should change the master
1277 key afterwards. Changing the master key requires a full data
1278 backup, luksFormat and then restore of the backup.
1280 First, there is a script by Milan that automates the whole
1281 process, except generating a new LUKS header with the old master
1282 key (it prints the command for that though):
1284 http://code.google.com/p/cryptsetup/source/browse/misc/luks-header-from-active
1286 You can also do this manually. Here is how:
1288 - Get the master key from the device mapper. This is done by the
1289 following command. Substitute c5 for whatever you mapped to:
1291 # dmsetup table --target crypt --showkey /dev/mapper/c5
1293 0 200704 crypt aes-cbc-essiv:sha256
1294 a1704d9715f73a1bb4db581dcacadaf405e700d591e93e2eaade13ba653d0d09
1297 The result is actually one line, wrapped here for clarity. The long
1298 hex string is the master key.
1300 - Convert the master key to a binary file representation. You can
1301 do this manually, e.g. with hexedit. You can also use the tool
1302 "xxd" from vim like this:
1304 echo "a1704d9....53d0d09" | xxd -r -p > <master-key-file>
1306 - Do a luksFormat to create a new LUKS header.
1308 NOTE: If your header is intact and you just forgot the
1309 passphrase, you can just set a new passphrase, see next
1312 Unmap the device before you do that (luksClose). Then do
1314 cryptsetup luksFormat --master-key-file=<master-key-file> <luks device>
1316 Note that if the container was created with other than the default
1317 settings of the cryptsetup version you are using, you need to give
1318 additional parameters specifying the deviations. If in doubt, try
1319 the script by Milan. It does recover the other parameters as well.
1321 Side note: This is the way the decrypt_derived script gets at the
1322 master key. It just omits the conversion and hashes the master key
1325 - If the header is intact and you just forgot the passphrase, just
1326 set a new passphrase like this:
1328 cryptsetup luksAddKey --master-key-file=<master-key-file> <luks device>
1330 You may want to disable the old one afterwards.
1333 * 6.11 What does the on-disk structure of dm-crypt look like?
1335 There is none. dm-crypt takes a block device and gives encrypted
1336 access to each of its blocks with a key derived from the passphrase
1337 given. If you use a cipher different than the default, you have to
1338 specify that as a parameter to cryptsetup too. If you want to
1339 change the password, you basically have to create a second
1340 encrypted device with the new passphrase and copy your data over.
1341 On the plus side, if you accidentally overwrite any part of a
1342 dm-crypt device, the damage will be limited to the are you
1346 * 6.12 What does the on-disk structure of LUKS look like?
1348 A LUKS partition consists of a header, followed by 8 key-slot
1349 descriptors, followed by 8 key slots, followed by the encrypted
1352 Header and key-slot descriptors fill the first 592 bytes. The
1353 key-slot size depends on the creation parameters, namely on the
1354 number of anti-forensic stripes, key material offset and master
1357 With the default parameters, each key-slot is a bit less than
1358 128kiB in size. Due to sector alignment of the key-slot start,
1359 that means the key block 0 is at offset 0x1000-0x20400, key
1360 block 1 at offset 0x21000-0x40400, and key block 7 at offset
1361 0xc1000-0xe0400. The space to the next full sector address is
1362 padded with zeros. Never used key-slots are filled with what the
1363 disk originally contained there, a key-slot removed with
1364 "luksRemoveKey" or "luksKillSlot" gets filled with 0xff. Due to
1365 2MiB default alignment, start of the data area for cryptsetup 1.3
1366 and later is at 2MiB, i.e. at 0x200000. For older versions, it is
1367 at 0x101000, i.e. at 1'052'672 bytes, i.e. at 1MiB + 4096 bytes
1368 from the start of the partition. Incidentally, "luksHeaderBackup"
1369 for a LUKS container created with default parameters dumps exactly
1370 the first 2MiB (or 1'052'672 bytes for headers created with
1371 cryptsetup versions < 1.3) to file and "luksHeaderRestore" restores
1374 For non-default parameters, you have to figure out placement
1375 yourself. "luksDump" helps. See also next item. For the most common
1376 non-default settings, namely aes-xts-plain with 512 bit key, the
1377 offsets are: 1st keyslot 0x1000-0x3f800, 2nd keyslot
1378 0x40000-0x7e000, 3rd keyslot 0x7e000-0xbd800, ..., and start of
1379 bulk data at 0x200000.
1381 The exact specification of the format is here:
1382 http://code.google.com/p/cryptsetup/wiki/Specification
1385 * 6.13 What is the smallest possible LUKS container?
1387 Note: From cryptsetup 1.3 onwards, alignment is set to 1MB. With
1388 modern Linux partitioning tools that also align to 1MB, this will
1389 result in alignment to 2k sectors and typical Flash/SSD sectors,
1390 which is highly desirable for a number of reasons. Changing the
1391 alignment is not recommended.
1393 That said, with default parameters, the data area starts at
1394 exactly 2MB offset (at 0x101000 for cryptsetup versions before
1395 1.3). The smallest data area you can have is one sector of 512
1396 bytes. Data areas of 0 bytes can be created, but fail on mapping.
1398 While you cannot put a filesystem into something this small, it may
1399 still be used to contain, for example, key. Note that with current
1400 formatting tools, a partition for a container this size will be
1401 3MiB anyways. If you put the LUKS container into a file (via
1402 losetup and a loopback device), the file needs to be 2097664 bytes
1403 in size, i.e. 2MiB + 512B.
1405 There two ways to influence the start of the data area are key-size
1408 For alignment, you can go down to 1 on the parameter. This will
1409 still leave you with a data-area starting at 0x101000, i.e.
1410 1MiB+4096B (default parameters) as alignment will be rounded up to
1411 the next multiple of 8 (i.e. 4096 bytes) If in doubt, do a dry-run
1412 on a larger file and dump the LUKS header to get actual
1415 For key-size, you can use 128 bit (e.g. AES-128 with CBC), 256 bit
1416 (e.g. AES-256 with CBC) or 512 bit (e.g. AES-256 with XTS mode).
1417 You can do 64 bit (e.g. blowfish-64 with CBC), but anything below
1418 128 bit has to be considered insecure today.
1420 Example 1 - AES 128 bit with CBC:
1422 cryptsetup luksFormat -s 128 --align-payload=8 <device>
1424 This results in a data offset of 0x81000, i.e. 516KiB or 528384
1425 bytes. Add one 512 byte sector and the smallest LUKS container size
1426 with these parameters is 516KiB + 512B or 528896 bytes.
1428 Example 2 - Blowfish 64 bit with CBC (WARNING: insecure):
1430 cryptsetup luksFormat -c blowfish -s 64 --align-payload=8 /dev/loop0
1432 This results in a data offset of 0x41000, i.e. 260kiB or 266240
1433 bytes, with a minimal LUKS container size of 260kiB + 512B or
1437 * 6.14 I think this is overly complicated. Is there an alternative?
1439 Not really. Encryption comes at a price. You can use plain
1440 dm-crypt to simplify things a bit. It does not allow multiple
1441 passphrases, but on the plus side, it has zero on disk description
1442 and if you overwrite some part of a plain dm-crypt partition,
1443 exactly the overwritten parts are lost (rounded up to sector
1447 * 6.15 Can I clone a LUKS container?
1449 You can, but it breaks security, because the cloned container has
1450 the same header and hence the same master key. You cannot change
1451 the master key on a LUKS container, even if you change the
1452 passphrase(s), the master key stays the same. That means whoever
1453 has access to one of the clones can decrypt them all, completely
1454 bypassing the passphrases.
1456 The right way to do this is to first luksFormat the target
1457 container, then to clone the contents of the source container, with
1458 both containers mapped, i.e. decrypted. You can clone the decrypted
1459 contents of a LUKS container in binary mode, although you may run
1460 into secondary issues with GUIDs in filesystems, partition tables,
1461 RAID-components and the like. These are just the normal problems
1462 binary cloning causes.
1464 Note that if you need to ship (e.g.) cloned LUKS containers with a
1465 default passphrase, that is fine as long as each container was
1466 individually created (and hence has its own master key). In this
1467 case, changing the default passphrase will make it secure again.
1470 7. Interoperability with other Disk Encryption Tools
1473 * 7.1 What is this section about?
1475 Cryptsetup for plain dm-crypt can be used to access a number of
1476 on-disk formats created by tools like loop-aes patched into
1477 losetup. This sometimes works and sometimes does not. This
1478 section collects insights into what works, what does not and where
1479 more information is required.
1481 Additional information may be found in the mailing-list archives,
1482 mentioned at the start of this FAQ document. If you have a
1483 solution working that is not yet documented here and think a wider
1484 audience may be interested, please email the FAQ maintainer.
1487 * 7.2 loop-aes: General observations.
1489 One problem is that there are different versions of losetup around.
1490 loop-aes is a patch for losetup. Possible problems and deviations
1491 from cryptsetup option syntax include:
1493 - Offsets specified in bytes (cryptsetup: 512 byte sectors)
1495 - The need to specify an IV offset
1497 - Encryption mode needs specifying (e.g. "-c twofish-cbc-plain")
1499 - Key size needs specifying (e.g. "-s 128" for 128 bit keys)
1501 - Passphrase hash algorithm needs specifying
1503 Also note that because plain dm-crypt and loop-aes format does not
1504 have metadata, autodetection, while feasible in most cases, would
1505 be a lot of work that nobody really wants to do. If you still have
1506 the old set-up, using a verbosity option (-v) on mapping with the
1507 old tool or having a look into the system logs after setup could
1508 give you the information you need.
1511 * 7.3 loop-aes patched into losetup on Debian 5.x, kernel 2.6.32
1513 In this case, the main problem seems to be that this variant of
1514 losetup takes the offset (-o option) in bytes, while cryptsetup
1515 takes it in sectors of 512 bytes each. Example: The losetup command
1517 losetup -e twofish -o 2560 /dev/loop0 /dev/sdb1
1518 mount /dev/loop0 mount-point
1522 cryptsetup create -c twofish -o 5 --skip 5 e1 /dev/sdb1
1523 mount /dev/mapper/e1 mount-point
1526 * 7.4 loop-aes with 160 bit key
1528 This seems to be sometimes used with twofish and blowfish and
1529 represents a 160 bit ripemed160 hash output padded to 196 bit key
1530 length. It seems the corresponding options for cryptsetup are
1532 --cipher twofish-cbc-null -s 192 -h ripemd160:20
1535 8. Issues with Specific Versions of cryptsetup
1538 * 8.1 When using the create command for plain dm-crypt with
1539 cryptsetup 1.1.x, the mapping is incompatible and my data is not
1542 With cryptsetup 1.1.x, the distro maintainer can define different
1543 default encryption modes for LUKS and plain devices. You can check
1544 these compiled-in defaults using "cryptsetup --help". Moreover, the
1545 plain device default changed because the old IV mode was
1546 vulnerable to a watermarking attack.
1548 If you are using a plain device and you need a compatible mode, just
1549 specify cipher, key size and hash algorithm explicitly. For
1550 compatibility with cryptsetup 1.0.x defaults, simple use the
1553 cryptsetup create -c aes-cbc-plain -s 256 -h ripemd160 <name> <dev>
1555 LUKS stores cipher and mode in the metadata on disk, avoiding this
1559 * 8.2 cryptsetup on SLED 10 has problems...
1561 SLED 10 is missing an essential kernel patch for dm-crypt, which
1562 is broken in its kernel as a result. There may be a very old
1563 version of cryptsetup (1.0.x) provided by SLED, which should also
1564 not be used anymore as well. My advice would be to drop SLED 10.
1566 A. Contributors In no particular order: