8 6. Backup and Data Recovery
9 7. Interoperability with other Disk Encryption Tools
10 8. Issues with Specific Versions of cryptsetup
11 9. References and Further Reading
20 This is the FAQ (Frequently Asked Questions) for cryptsetup. It
21 covers Linux disk encryption with plain dm-crypt (one passphrase,
22 no management, no metadata on disk) and LUKS (multiple user keys
23 with one master key, anti-forensic features, metadata block at
24 start of device, ...). The latest version of this FAQ should
25 usually be available at
26 http://code.google.com/p/cryptsetup/wiki/FrequentlyAskedQuestions
31 ATTENTION: If you are going to read just one thing, make it the
32 section on Backup and Data Recovery. By far the most questions on
33 the cryptsetup mailing list are from people that managed to damage
34 the start of their LUKS partitions, i.e. the LUKS header. In
35 most cases, there is nothing that can be done to help these poor
36 souls recover their data. Make sure you understand the problem and
37 limitations imposed by the LUKS security model BEFORE you face
38 such a disaster! In particular, make sure you have a current header
39 backup before doing any potentially dangerous operations.
41 SSDs/FLASH DRIVES: SSDs and Flash are different. Currently it is
42 unclear how to get LUKS or plain dm-crypt to run on them with the
43 full set of security features intact. This may or may not be a
44 problem, depending on the attacher model. See Section 5.19.
46 BACKUP: Yes, encrypted disks die, just as normal ones do. A full
47 backup is mandatory, see Section "6. Backup and Data Recovery" on
48 options for doing encrypted backup.
50 CLONING/IMAGING: If you clone or image a LUKS container, you make a
51 copy of the LUKS header and the master key will stay the same!
52 That means that if you distribute an image to several machines, the
53 same master key will be used on all of them, regardless of whether
54 you change the passphrases. Do NOT do this! If you do, a root-user
55 on any of the machines with a mapped (decrypted) container or a
56 passphrase on that machine can decrypt all other copies, breaking
57 security. See also Item 6.15.
59 DISTRIBUTION INSTALLERS: Some distribution installers offer to
60 create LUKS containers in a way that can be mistaken as activation
61 of an existing container. Creating a new LUKS container on top of
62 an existing one leads to permanent, complete and irreversible data
63 loss. It is strongly recommended to only use distribution
64 installers after a complete backup of all LUKS containers has been
67 NO WARNING ON NON-INTERACTIVE FORMAT: If you feed cryptsetup from
68 STDIN (e.g. via GnuPG) on LUKS format, it does not give you the
69 warning that you are about to format (and e.g. will lose any
70 pre-existing LUKS container on the target), as it assumes it is
71 used from a script. In this scenario, the responsibility for
72 warning the user and possibly checking for an existing LUKS header
73 is shifted to the script. This is a more general form of the
76 LUKS PASSPHRASE IS NOT THE MASTER KEY: The LUKS passphrase is not
77 used in deriving the master key. It is used in decrypting a master
78 key that is randomly selected on header creation. This means that
79 if you create a new LUKS header on top of an old one with
80 exactly the same parameters and exactly the same passphrase as the
81 old one, it will still have a different master key and your data
82 will be permanently lost.
84 PASSPHRASE CHARACTER SET: Some people have had difficulties with
85 this when upgrading distributions. It is highly advisable to only
86 use the 95 printable characters from the first 128 characters of
87 the ASCII table, as they will always have the same binary
88 representation. Other characters may have different encoding
89 depending on system configuration and your passphrase will not
90 work with a different encoding. A table of the standardized first
91 128 ASCII characters can, e.g. be found on
92 http://en.wikipedia.org/wiki/ASCII
95 * 1.3 System specific warnings
97 - Ubuntu as of 4/2011: It seems the installer offers to create
98 LUKS partitions in a way that several people mistook for an offer
99 to activate their existing LUKS partition. The installer gives no
100 or an inadequate warning and will destroy your old LUKS header,
101 causing permanent data loss. See also the section on Backup and
104 This issue has been acknowledged by the Ubuntu dev team, see here:
105 http://launchpad.net/bugs/420080
107 Update 7/2012: I am unsure whether this has been fixed by now, best
111 * 1.4 My LUKS-device is broken! Help!
113 First: Do not panic! In many cases the data is still recoverable.
114 Do not do anything hasty! Steps:
116 - Take some deep breaths. Maybe add some relaxing music. This may
117 sound funny, but I am completely serious. Often, critical damage is
118 done only after the initial problem.
120 - Do not reboot. The keys mays still be in the kernel if the device
123 - Make sure others do not reboot the system.
125 - Do not write to your disk without a clear understanding why this
126 will not make matters worse. Do a sector-level backup before any
127 writes. Often you do not need to write at all to get enough access
128 to make a backup of the data.
132 - Read section 6 of this FAQ.
134 - Ask on the mailing-list if you need more help.
137 * 1.5 Who wrote this?
139 Current FAQ maintainer is Arno Wagner <arno@wagner.name>. Other
140 contributors are listed at the end. If you want to contribute, send
141 your article, including a descriptive headline, to the maintainer,
142 or the dm-crypt mailing list with something like "FAQ ..." in the
143 subject. You can also send more raw information and have me write
144 the section. Please note that by contributing to this FAQ, you
145 accept the license described below.
147 This work is under the "Attribution-Share Alike 3.0 Unported"
148 license, which means distribution is unlimited, you may create
149 derived works, but attributions to original authors and this
150 license statement must be retained and the derived work must be
151 under the same license. See
152 http://creativecommons.org/licenses/by-sa/3.0/ for more details of
155 Side note: I did text license research some time ago and I think
156 this license is best suited for the purpose at hand and creates the
160 * 1.5 Where is the project website?
162 There is the project website at http://code.google.com/p/cryptsetup/
163 Please do not post questions there, nobody will read them. Use
164 the mailing-list instead.
167 * 1.6 Is there a mailing-list?
169 Instructions on how to subscribe to the mailing-list are at on the
170 project website. People are generally helpful and friendly on the
173 The question of how to unsubscribe from the list does crop up
174 sometimes. For this you need your list management URL, which is
175 sent to you initially and once at the start of each month. Go to
176 the URL mentioned in the email and select "unsubscribe". This page
177 also allows you to request a password reminder.
179 Alternatively, you can send an Email to dm-crypt-request@saout.de
180 with just the word "help" in the subject or message body. Make sure
181 to send it from your list address.
183 The mailing list archive is here:
184 http://dir.gmane.org/gmane.linux.kernel.device-mapper.dm-crypt
187 * 1.7 Unsubscribe from the mailing-list
189 Send mail to dm-crypt-unsubscribe@saout.de from the subscribed
190 account. You will get an email with instructions.
192 Basically, you just have to respond to it unmodified to get
193 unsubscribed. The listserver admin functions are not very fast. It
194 can take 15 minutes or longer for a reply to arrive (I suspect
195 greylisting is in use), so be patient.
197 Also note that nobody on the list can unsubscribe you, sending
198 demands to be unsubscribed to the list just annoys people that are
199 entirely blameless for you being subscribed.
201 If you are subscribed, a subscription confirmation email was sent
202 to your email account and it had to be answered before the
203 subscription went active. The confirmation emails from the
204 listserver have subjects like these (with other numbers):
206 Subject: confirm 9964cf10.....
208 and are sent from dm-crypt-request@saout.de. You should check
209 whether you have anything like it in your sent email folder. If
210 you find nothing and are sure you did not confirm, then you should
211 look into a possible compromise of your email account.
217 * 2.1 LUKS Container Setup mini-HOWTO
219 This item tries to give you a very brief list of all the steps you
220 should go though when creating a new LUKS encrypted container, i.e.
221 encrypted disk, partition or loop-file.
223 01) All data will be lost, if there is data on the target, make a
226 02) Make very sure you have the right target disk, partition or
229 03) If the target was in use previously, it is a good idea to
230 wipe it before creating the LUKS container, in order to remove any
231 trace of old file systems, etc. For example, some users have
232 managed to run e2fsck on a partition containing a LUKS container,
233 possibly because of residual ext2 superblocks from an earlier use.
234 This can do arbitrary damage up to complete and permanent loss of
235 all data in the LUKS container.
237 To wipe, use something like
239 cat /dev/zero > target
241 This can take a while. To get a progress indicator, you can use
242 the tool dd_rescue (->google) instead or use my stream meter "wcs"
243 (source here: http://www.tansi.org/tools/index.html) in the
246 cat /dev/zero | wcs > target
248 Be very sure you have the right target, all data will be lost!
250 Note that automatic wiping is on the TODO list for cryptsetup, so
251 at some time in the future this will become unnecessary.
253 04) Create the LUKS container:
254 cryptsetup luksFormat target
256 Just follow the on-screen instructions.
258 05) Map the container. Here it will be mapped to /dev/mapper/c1:
259 cryptsetup luksOpen target c1
261 06) (Optionally) wipe the container
262 cat /dev/zero > /dev/mapper/c1
264 Note that this creates a small information leak, as an attacker can
265 determine whether a 512 byte block is zero if the attacker has
266 access to the encrypted container multiple times. Typically a
267 competent attacker that has access multiple times can install a
268 passphrase sniffer anyways, so this leakage is not very
269 significant. For getting a progress indicator, see step 03.
271 Note that at some time in the future, cryptsetup will do this for
272 you, but currently it is a TODO list item.
274 07) Create a file system in the mapped container, for example an
275 ext3 file system (any other file system is possible):
277 mke2fs -j /dev/mapper/c1
279 08) Mount your encrypted file system, here on /mnt:
280 mount /dev/mapper/c1 /mnt
282 Done. You can now use the encrypted file system to store data. Be
283 sure to read though the rest of the FAQ, these are just the very
284 basics. In particular, there are a number of mistakes that are
285 easy to make, but will compromise your security.
288 * 2.2 What is the difference between "plain" and LUKS format?
290 First, unless you happen to understand the cryptographic background
291 well, you should use LUKS. It does protect the user from a lot of
292 common mistakes. Plain dm-crypt is for experts.
294 Plain format is just that: It has no metadata on disk, reads all
295 parameters from the commandline (or the defaults), derives a
296 master-key from the passphrase and then uses that to de-/encrypt
297 the sectors of the device, with a direct 1:1 mapping between
298 encrypted and decrypted sectors.
300 Primary advantage is high resilience to damage, as one damaged
301 encrypted sector results in exactly one damaged decrypted sector.
302 Also, it is not readily apparent that there even is encrypted data
303 on the device, as an overwrite with crypto-grade randomness (e.g.
304 from /dev/urandom) looks exactly the same on disk.
306 Side-note: That has limited value against the authorities. In
307 civilized countries, they cannot force you to give up a crypto-key
308 anyways. In the US, the UK and dictatorships around the world,
309 they can force you to give up the keys (using imprisonment or worse
310 to pressure you), and in the worst case, they only need a
311 nebulous "suspicion" about the presence of encrypted data. My
312 advice is to either be ready to give up the keys or to not have
313 encrypted data when traveling to those countries, especially when
314 crossing the borders.
316 Disadvantages are that you do not have all the nice features that
317 the LUKS metadata offers, like multiple passphrases that can be
318 changed, the cipher being stored in the metadata, anti-forensic
319 properties like key-slot diffusion and salts, etc..
321 LUKS format uses a metadata header and 8 key-slot areas that are
322 being placed at the beginning of the disk, see below under "What
323 does the LUKS on-disk format looks like?". The passphrases are used
324 to decrypt a single master key that is stored in the anti-forensic
327 Advantages are a higher usability, automatic configuration of
328 non-default crypto parameters, defenses against low-entropy
329 passphrases like salting and iterated PBKDF2 passphrase hashing,
330 the ability to change passphrases, and others.
332 Disadvantages are that it is readily obvious there is encrypted
333 data on disk (but see side note above) and that damage to the
334 header or key-slots usually results in permanent data-loss. See
335 below under "6. Backup and Data Recovery" on how to reduce that
336 risk. Also the sector numbers get shifted by the length of the
337 header and key-slots and there is a loss of that size in capacity
338 (1MB+4096B for defaults and 2MB for the most commonly used
339 non-default XTS mode).
342 * 2.3 Can I encrypt an already existing, non-empty partition to use
345 There is no converter, and it is not really needed. The way to do
346 this is to make a backup of the device in question, securely wipe
347 the device (as LUKS device initialization does not clear away old
348 data), do a luksFormat, optionally overwrite the encrypted device,
349 create a new filesystem and restore your backup on the now
350 encrypted device. Also refer to sections "Security Aspects" and
351 "Backup and Data Recovery".
353 For backup, plain GNU tar works well and backs up anything likely
354 to be in a filesystem.
357 * 2.4 How do I use LUKS with a loop-device?
359 This can be very handy for experiments. Setup is just the same as
360 with any block device. If you want, for example, to use a 100MiB
361 file as LUKS container, do something like this:
363 head -c 100M /dev/zero > luksfile # create empty file
364 losetup /dev/loop0 luksfile # map luksfile to /dev/loop0
365 cryptsetup luksFormat /dev/loop0 # create LUKS on loop device
367 Afterwards just use /dev/loop0 as a you would use a LUKS partition.
368 To unmap the file when done, use "losetup -d /dev/loop0".
371 * 2.5 When I add a new key-slot to LUKS, it asks for a passphrase but
372 then complains about there not being a key-slot with that
375 That is as intended. You are asked a passphrase of an existing
376 key-slot first, before you can enter the passphrase for the new
377 key-slot. Otherwise you could break the encryption by just adding a
378 new key-slot. This way, you have to know the passphrase of one of
379 the already configured key-slots in order to be able to configure a
383 * 2.6 Encryption on top of RAID or the other way round?
385 Unless you have special needs, place encryption between RAID and
386 filesystem, i.e. encryption on top of RAID. You can do it the other
387 way round, but you have to be aware that you then need to give the
388 passphrase for each individual disk and RAID autodetection will
389 not work anymore. Therefore it is better to encrypt the RAID
390 device, e.g. /dev/dm0 .
393 * 2.7 How do I read a dm-crypt key from file?
395 Note that the file will still be hashed first, just like keyboard
396 input. Use the --key-file option, like this:
398 cryptsetup create --key-file keyfile e1 /dev/loop0
401 * 2.8 How do I read a LUKS slot key from file?
403 What you really do here is to read a passphrase from file, just as
404 you would with manual entry of a passphrase for a key-slot. You can
405 add a new passphrase to a free key-slot, set the passphrase of an
406 specific key-slot or put an already configured passphrase into a
407 file. In the last case make sure no trailing newline (0x0a) is
408 contained in the key file, or the passphrase will not work because
409 the whole file is used as input.
411 To add a new passphrase to a free key slot from file, use something
414 cryptsetup luksAddKey /dev/loop0 keyfile
416 To add a new passphrase to a specific key-slot, use something like
419 cryptsetup luksAddKey --key-slot 7 /dev/loop0 keyfile
421 To supply a key from file to any LUKS command, use the --key-file
422 option, e.g. like this:
424 cryptsetup luksOpen --key-file keyfile /dev/loop0 e1
427 * 2.9 How do I read the LUKS master key from file?
429 The question you should ask yourself first is why you would want to
430 do this. The only legitimate reason I can think of is if you want
431 to have two LUKS devices with the same master key. Even then, I
432 think it would be preferable to just use key-slots with the same
433 passphrase, or to use plain dm-crypt instead. If you really have a
434 good reason, please tell me. If I am convinced, I will add how to
438 * 2.10 What are the security requirements for a key read from file?
440 A file-stored key or passphrase has the same security requirements
441 as one entered interactively, however you can use random bytes and
442 thereby use bytes you cannot type on the keyboard. You can use any
443 file you like as key file, for example a plain text file with a
444 human readable passphrase. To generate a file with random bytes,
445 use something like this:
447 head -c 256 /dev/random > keyfile
450 * 2.11 If I map a journaled file system using dm-crypt/LUKS, does it
451 still provide its usual transactional guarantees?
453 Yes, it does, unless a very old kernel is used. The required flags
454 come from the filesystem layer and are processed and passed onwards
455 by dm-crypt. A bit more information on the process by which
456 transactional guarantees are implemented can be found here:
458 http://lwn.net/Articles/400541/
460 Please note that these "guarantees" are weaker than they appear to
461 be. One problem is that quite a few disks lie to the OS about
462 having flushed their buffers. Some other things can go wrong as
463 well. The filesystem developers are aware of these problems and
464 typically can make it work anyways. That said, dm-crypt/LUKS will
465 not make things worse.
467 One specific problem you can run into though is that you can get
468 short freezes and other slowdowns due to the encryption layer.
469 Encryption takes time and forced flushes will block for that time.
470 For example, I did run into frequent small freezes (1-2 sec) when
471 putting a vmware image on ext3 over dm-crypt. When I went back to
472 ext2, the problem went away. This seems to have gotten better with
473 kernel 2.6.36 and the reworking of filesystem flush locking
474 mechanism (less blocking of CPU activity during flushes). It
475 should improve further and eventually the problem should go away.
478 * 2.12 Can I use LUKS or cryptsetup with a more secure (external)
479 medium for key storage, e.g. TPM or a smartcard?
481 Yes, see the answers on using a file-supplied key. You do have to
482 write the glue-logic yourself though. Basically you can have
483 cryptsetup read the key from STDIN and write it there with your
484 own tool that in turn gets the key from the more secure key
488 * 2.13 Can I resize a dm-crypt or LUKS partition?
490 Yes, you can, as neither dm-crypt nor LUKS stores partition size.
491 Whether you should is a different question. Personally I recommend
492 backup, recreation of the encrypted partition with new size,
493 recreation of the filesystem and restore. This gets around the
494 tricky business of resizing the filesystem. Resizing a dm-crypt or
495 LUKS container does not resize the filesystem in it. The backup is
496 really non-optional here, as a lot can go wrong, resulting in
497 partial or complete data loss. Using something like gparted to
498 resize an encrypted partition is slow, but typically works. This
499 will not change the size of the filesystem hidden under the
502 You also need to be aware of size-based limitations. The one
503 currently relevant is that aes-xts-plain should not be used for
504 encrypted container sizes larger than 2TiB. Use aes-xts-plain64
511 * 3.1 My dm-crypt/LUKS mapping does not work! What general steps are
512 there to investigate the problem?
514 If you get a specific error message, investigate what it claims
515 first. If not, you may want to check the following things.
517 - Check that "/dev", including "/dev/mapper/control" is there. If it
518 is missing, you may have a problem with the "/dev" tree itself or
519 you may have broken udev rules.
521 - Check that you have the device mapper and the crypt target in your
522 kernel. The output of "dmsetup targets" should list a "crypt"
523 target. If it is not there or the command fails, add device mapper
524 and crypt-target to the kernel.
526 - Check that the hash-functions and ciphers you want to use are in
527 the kernel. The output of "cat /proc/crypto" needs to list them.
530 * 3.2 My dm-crypt mapping suddenly stopped when upgrading cryptsetup.
532 The default cipher, hash or mode may have changed (the mode changed
533 from 1.0.x to 1.1.x). See under "Issues With Specific Versions of
537 * 3.3 When I call cryptsetup from cron/CGI, I get errors about
540 If you get errors about unknown parameters or the like that are not
541 present when cryptsetup is called from the shell, make sure you
542 have no older version of cryptsetup on your system that then gets
543 called by cron/CGI. For example some distributions install
544 cryptsetup into /usr/sbin, while a manual install could go to
545 /usr/local/sbin. As a debugging aid, call "cryptsetup --version"
546 from cron/CGI or the non-shell mechanism to be sure the right
550 * 3.4 Unlocking a LUKS device takes very long. Why?
552 The iteration time for a key-slot (see Section 5 for an explanation
553 what iteration does) is calculated when setting a passphrase. By
554 default it is 1 second on the machine where the passphrase is set.
555 If you set a passphrase on a fast machine and then unlock it on a
556 slow machine, the unlocking time can be much longer. Also take into
557 account that up to 8 key-slots have to be tried in order to find the
560 If this is problem, you can add another key-slot using the slow
561 machine with the same passphrase and then remove the old key-slot.
562 The new key-slot will have an iteration count adjusted to 1 second
563 on the slow machine. Use luksKeyAdd and then luksKillSlot or
566 However, this operation will not change volume key iteration count
567 (MK iterations in output of "cryptsetup luksDump"). In order to
568 change that, you will have to backup the data in the LUKS
569 container (i.e. your encrypted data), luksFormat on the slow
570 machine and restore the data. Note that in the original LUKS
571 specification this value was fixed to 10, but it is now derived
572 from the PBKDF2 benchmark as well and set to iterations in 0.125
573 sec or 1000, whichever is larger. Also note that MK iterations
574 are not very security relevant. But as each key-slot already takes
575 1 second, spending the additional 0.125 seconds really does not
579 * 3.5 "blkid" sees a LUKS UUID and an ext2/swap UUID on the same
580 device. What is wrong?
582 Some old versions of cryptsetup have a bug where the header does
583 not get completely wiped during LUKS format and an older ext2/swap
584 signature remains on the device. This confuses blkid.
586 Fix: Wipe the unused header areas by doing a backup and restore of
587 the header with cryptsetup 1.1.x:
589 cryptsetup luksHeaderBackup --header-backup-file <file> <device>
590 cryptsetup luksHeaderRestore --header-backup-file <file> <device>
593 * 3.6 cryptsetup segfaults on Gentoo amd64 hardened ...
595 There seems to be some interference between the hardening and and
596 the way cryptsetup benchmarks PBKDF2. The solution to this is
597 currently not quite clear for an encrypted root filesystem. For
598 other uses, you can apparently specify USE="dynamic" as compile
599 flag, see http://bugs.gentoo.org/show_bug.cgi?id=283470
605 * 4.1 I get the error "LUKS keyslot x is invalid." What does that
608 This means that the given keyslot has an offset that points
609 outside the valid keyslot area. Typically, the reason is a
610 corrupted LUKS header because something was written to the start of
611 the device the LUKS container is on. Refer to Section "Backup and
612 Data Recovery" and ask on the mailing list if you have trouble
613 diagnosing and (if still possible) repairing this.
616 * 4.2 Can a bad RAM module cause problems?
618 LUKS and dm-crypt can give the RAM quite a workout, especially when
619 combined with software RAID. In particular the combination RAID5 +
620 LUKS + XFS seems to uncover RAM problems that never caused obvious
621 problems before. Symptoms vary, but often the problem manifest
622 itself when copying large amounts of data, typically several times
623 larger than your main memory.
625 Side note: One thing you should always do on large data
626 copy/movements is to run a verify, for example with the "-d"
627 option of "tar" or by doing a set of MD5 checksums on the source
630 find . -type f -exec md5sum \{\} \; > checksum-file
632 and then a "md5sum -c checksum-file" on the other side. If you get
633 mismatches here, RAM is the primary suspect. A lesser suspect is
634 an overclocked CPU. I have found countless hardware problems in
635 verify runs after copying or making backups. Bit errors are much
636 more common than most people think.
638 Some RAM issues are even worse and corrupt structures in one of the
639 layers. This typically results in lockups, CPU state dumps in the
640 system logs, kernel panic or other things. It is quite possible to
641 have the problem with an encrypted device, but not with an
642 otherwise the same unencrypted device. The reason for that is that
643 encryption has an error amplification property: You flip one bit
644 in an encrypted data block, and the decrypted version has half of
645 its bits flipped. This is an important security property for modern
646 ciphers. With the usual modes in cryptsetup (CBC, ESSIV, XTS), you
647 get up to a completely changed 512 byte block per bit error. A
648 corrupt block causes a lot more havoc than the occasionally
649 flipped single bit and can result in various obscure errors.
651 Note, that a verify run on copying between encrypted or
652 unencrypted devices will reliably detect corruption, even when the
653 copying itself did not report any problems. If you find defect
654 RAM, assume all backups and copied data to be suspect, unless you
658 * 4.3 How do I test RAM?
660 First you should know that overclocking often makes memory
661 problems worse. So if you overclock (which I strongly recommend
662 against in a system holding data that has some worth), run the
663 tests with the overclocking active.
665 There are two good options. One is Memtest86+ and the other is
666 "memtester" by Charles Cazabon. Memtest86+ requires a reboot and
667 then takes over the machine, while memtester runs from a
668 root-shell. Both use different testing methods and I have found
669 problems fast with each one that the other needed long to find. I
670 recommend running the following procedure until the first error is
673 - Run Memtest86+ for one cycle
675 - Run memtester for one cycle (shut down as many other applications
678 - Run Memtest86+ for 24h or more
680 - Run memtester for 24h or more
682 If all that does not produce error messages, your RAM may be sound,
683 but I have had one weak bit that Memtest86+ needed around 60 hours
684 to find. If you can reproduce the original problem reliably, a good
685 additional test may be to remove half of the RAM (if you have more
686 than one module) and try whether the problem is still there and if
687 so, try with the other half. If you just have one module, get a
688 different one and try with that. If you do overclocking, reduce
689 the settings to the most conservative ones available and try with
696 * 5.1 How long is a secure passphrase ?
698 This is just the short answer. For more info and explanation of
699 some of the terms used in this item, read the rest of Section 5.
700 The actual recommendation is at the end of this item.
702 First, passphrase length is not really the right measure,
703 passphrase entropy is. For example, a random lowercase letter (a-z)
704 gives you 4.7 bit of entropy, one element of a-z0-9 gives you 5.2
705 bits of entropy, an element of a-zA-Z0-9 gives you 5.9 bits and
706 a-zA-Z0-9!@#$%^&:-+ gives you 6.2 bits. On the other hand, a random
707 English word only gives you 0.6...1.3 bits of entropy per
708 character. Using sentences that make sense gives lower entropy,
709 series of random words gives higher entropy. Do not use sentences
710 that can be tied to you or found on your computer. This type of
711 attack is done routinely today.
713 That said, it does not matter too much what scheme you use, but it
714 does matter how much entropy your passphrase contains, because an
715 attacker has to try on average
717 1/2 * 2^(bits of entropy in passphrase)
719 different passphrases to guess correctly.
721 Historically, estimations tended to use computing time estimates,
722 but more modern approaches try to estimate cost of guessing a
725 As an example, I will try to get an estimate from the numbers in
726 http://it.slashdot.org/story/12/12/05/0623215/new-25-gpu-monster-devours-strong-passwords-in-minutes
727 More references can be found a the end of this document. Note that
728 these are estimates from the defender side, so assuming something
729 is easier than it actually is is fine. An attacker may still have
730 vastly higher cost than estimated here.
732 LUKS uses SHA1 for hasing per default. The claim in the reference is
733 63 billion tries/second for SHA1. We will leave aside the check
734 whether a try actually decrypts a key-slot. Now, the machine has 25
735 GPUs, which I will estimate at an overall lifetime cost of USD/EUR
736 1000 each, and an useful lifetime of 2 years. (This is on the low
737 side.) Disregarding downtime, the machine can then break
739 N = 63*10^9 * 3600 * 24 * 365 * 2 ~ 4*10^18
741 passphrases for EUR/USD 25k. That is one 62 bit passphrase hashed
742 once with SHA1 for EUR/USD 25k. Note that as this can be
743 parallelized, it can be done faster than 2 years with several of
746 For plain dm-crypt (no hash iteration) this is it. This gives (with
747 SHA1, plain dm-crypt default is ripemd160 which seems to be
748 slightly slower than SHA1):
750 Passphrase entropy Cost to break
759 For LUKS, you have to take into account hash iteration in PBKDF2.
760 For a current CPU, there are about 100k iterations (as can be
761 queried with ''cryptsetup luksDump''.
763 The table above then becomes:
765 Passphrase entropy Cost to break
776 To get reasonable security for the next 10 years, it is a good idea
777 to overestimate by a factor of at least 1000.
779 Then there is the question of how much the attacker is willing to
780 spend. That is up to your own security evaluation. For general use,
781 I will assume the attacker is willing to spend up to 1 million
782 EUR/USD. Then we get the following recommendations:
784 Plain dm-crypt: Use > 80 bit. That is e.g. 17 random chars from a-z
785 or a random English sentence of > 135 characters length.
787 LUKS: Use > 65 bit. That is e.g. 14 random chars from a-z or a
788 random English sentence of > 108 characters length.
790 If paranoid, add at least 20 bit. That is roughly four additional
791 characters for random passphrases and roughly 32 characters for a
792 random English sentence.
795 * 5.2 Is LUKS insecure? Everybody can see I have encrypted data!
797 In practice it does not really matter. In most civilized countries
798 you can just refuse to hand over the keys, no harm done. In some
799 countries they can force you to hand over the keys, if they suspect
800 encryption. However the suspicion is enough, they do not have to
801 prove anything. This is for practical reasons, as even the presence
802 of a header (like the LUKS header) is not enough to prove that you
803 have any keys. It might have been an experiment, for example. Or it
804 was used as encrypted swap with a key from /dev/random. So they
805 make you prove you do not have encrypted data. Of course that is
806 just as impossible as the other way round.
808 This means that if you have a large set of random-looking data,
809 they can already lock you up. Hidden containers (encryption hidden
810 within encryption), as possible with Truecrypt, do not help
811 either. They will just assume the hidden container is there and
812 unless you hand over the key, you will stay locked up. Don't have
813 a hidden container? Though luck. Anybody could claim that.
815 Still, if you are concerned about the LUKS header, use plain
816 dm-crypt with a good passphrase. See also Section 2, "What is the
817 difference between "plain" and LUKS format?"
820 * 5.3 Should I initialize (overwrite) a new LUKS/dm-crypt partition?
822 If you just create a filesystem on it, most of the old data will
823 still be there. If the old data is sensitive, you should overwrite
824 it before encrypting. In any case, not initializing will leave the
825 old data there until the specific sector gets written. That may
826 enable an attacker to determine how much and where on the
827 partition data was written. If you think this is a risk, you can
828 prevent this by overwriting the encrypted device (here assumed to
829 be named "e1") with zeros like this:
831 dd_rescue -w /dev/zero /dev/mapper/e1
833 or alternatively with one of the following more standard commands:
835 cat /dev/zero > /dev/mapper/e1
836 dd if=/dev/zero of=/dev/mapper/e1
839 * 5.4 How do I securely erase a LUKS (or other) partition?
841 For LUKS, if you are in a desperate hurry, overwrite the LUKS
842 header and key-slot area. This means overwriting the first
843 (keyslots x stripes x keysize) + offset bytes. For the default
844 parameters, this is the 1'052'672 bytes, i.e. 1MiB + 4096 of the
845 LUKS partition. For 512 bit key length (e.g. for aes-xts-plain with
846 512 bit key) this is 2MiB. (The different offset stems from
847 differences in the sector alignment of the key-slots.) If in doubt,
848 just be generous and overwrite the first 10MB or so, it will likely
849 still be fast enough. A single overwrite with zeros should be
850 enough. If you anticipate being in a desperate hurry, prepare the
851 command beforehand. Example with /dev/sde1 as the LUKS partition
852 and default parameters:
854 head -c 1052672 /dev/zero > /dev/sde1; sync
856 A LUKS header backup or full backup will still grant access to
857 most or all data, so make sure that an attacker does not have
858 access to backups or destroy them as well.
860 If you have time, overwrite the whole LUKS partition with a single
861 pass of zeros. This is enough for current HDDs. For SSDs or FLASH
862 (USB sticks) you may want to overwrite the whole drive several
863 times to be sure data is not retained by wear leveling. This is
864 possibly still insecure as SSD technology is not fully understood
865 in this regard. Still, due to the anti-forensic properties of the
866 LUKS key-slots, a single overwrite of an SSD or FLASH drive could
867 be enough. If in doubt, use physical destruction in addition. Here
868 is a link to some current research results on erasing SSDs and
870 http://www.usenix.org/events/fast11/tech/full_papers/Wei.pdf
872 Keep in mind to also erase all backups.
874 Example for a zero-overwrite erase of partition sde1 done with
877 dd_rescue -w /dev/zero /dev/sde1
880 * 5.5 How do I securely erase a backup of a LUKS partition or header?
882 That depends on the medium it is stored on. For HDD and SSD, use
883 overwrite with zeros. For an SSD or FLASH drive (USB stick), you
884 may want to overwrite the complete SSD several times and use
885 physical destruction in addition, see last item. For re-writable
886 CD/DVD, a single overwrite should also be enough, due to the
887 anti-forensic properties of the LUKS keyslots. For write-once
888 media, use physical destruction. For low security requirements,
889 just cut the CD/DVD into several parts. For high security needs,
890 shred or burn the medium. If your backup is on magnetic tape, I
891 advise physical destruction by shredding or burning, after
892 overwriting . The problem with magnetic tape is that it has a
893 higher dynamic range than HDDs and older data may well be
894 recoverable after overwrites. Also write-head alignment issues can
895 lead to data not actually being deleted at all during overwrites.
898 * 5.6 What about backup? Does it compromise security?
900 That depends. See item 6.7.
903 * 5.7 Why is all my data permanently gone if I overwrite the LUKS
906 Overwriting the LUKS header in part or in full is the most common
907 reason why access to LUKS containers is lost permanently.
908 Overwriting can be done in a number of fashions, like creating a
909 new filesystem on the raw LUKS partition, making the raw partition
910 part of a raid array and just writing to the raw partition.
912 The LUKS header contains a 256 bit "salt" value and without that no
913 decryption is possible. While the salt is not secret, it is
914 key-grade material and cannot be reconstructed. This is a
915 cryptographically strong "cannot". From observations on the
916 cryptsetup mailing-list, people typically go though the usual
917 stages of grief (Denial, Anger, Bargaining, Depression, Acceptance)
918 when this happens to them. Observed times vary between 1 day and 2
919 weeks to complete the cycle. Seeking help on the mailing-list is
920 fine. Even if we usually cannot help with getting back your data,
921 most people found the feedback comforting.
923 If your header does not contain an intact salt, best go directly
924 to the last stage ("Acceptance") and think about what to do now.
925 There is one exception that I know of: If your LUKS container is
926 still open, then it may be possible to extract the master key from
927 the running system. See Item "How do I recover the master key from
928 a mapped LUKS container?" in Section "Backup and Data Recovery".
931 * 5.8 What is a "salt"?
933 A salt is a random key-grade value added to the passphrase before
934 it is processed. It is not kept secret. The reason for using salts
935 is as follows: If an attacker wants to crack the password for a
936 single LUKS container, then every possible passphrase has to be
937 tried. Typically an attacker will not try every binary value, but
938 will try words and sentences from a dictionary.
940 If an attacker wants to attack several LUKS containers with the
941 same dictionary, then a different approach makes sense: Compute the
942 resulting slot-key for each dictionary element and store it on
943 disk. Then the test for each entry is just the slow unlocking with
944 the slot key (say 0.00001 sec) instead of calculating the slot-key
945 first (1 sec). For a single attack, this does not help. But if you
946 have more than one container to attack, this helps tremendously,
947 also because you can prepare your table before you even have the
948 container to attack! The calculation is also very simple to
949 parallelize. You could, for example, use the night-time unused CPU
950 power of your desktop PCs for this.
952 This is where the salt comes in. If the salt is combined with the
953 passphrase (in the simplest form, just appended to it), you
954 suddenly need a separate table for each salt value. With a
955 reasonably-sized salt value (256 bit, e.g.) this is quite
959 * 5.9 Is LUKS secure with a low-entropy (bad) passphrase?
961 Note: You should only use the 94 printable characters from 7 bit
962 ASCII code to prevent your passphrase from failing when the
963 character encoding changes, e.g. because of a system upgrade, see
964 also the note at the very start of this FAQ under "WARNINGS".
966 This needs a bit of theory. The quality of your passphrase is
967 directly related to its entropy (information theoretic, not
968 thermodynamic). The entropy says how many bits of "uncertainty" or
969 "randomness" are in you passphrase. In other words, that is how
970 difficult guessing the passphrase is.
972 Example: A random English sentence has about 1 bit of entropy per
973 character. A random lowercase (or uppercase) character has about
976 Now, if n is the number of bits of entropy in your passphrase and t
977 is the time it takes to process a passphrase in order to open the
978 LUKS container, then an attacker has to spend at maximum
980 attack_time_max = 2^n * t
982 time for a successful attack and on average half that. There is no
983 way getting around that relationship. However, there is one thing
984 that does help, namely increasing t, the time it takes to use a
985 passphrase, see next FAQ item.
987 Still, if you want good security, a high-entropy passphrase is the
988 only option. For example, a low-entropy passphrase can never be
989 considered secure against a TLA-level (Three Letter Agency level,
990 i.e. government-level) attacker, no matter what tricks are used in
991 the key-derivation function. Use at least 64 bits for secret stuff.
992 That is 64 characters of English text (but only if randomly chosen)
993 or a combination of 12 truly random letters and digits.
995 For passphrase generation, do not use lines from very well-known
996 texts (religious texts, Harry potter, etc.) as they are to easy to
997 guess. For example, the total Harry Potter has about 1'500'000
998 words (my estimation). Trying every 64 character sequence starting
999 and ending at a word boundary would take only something like 20
1000 days on a single CPU and is entirely feasible. To put that into
1001 perspective, using a number of Amazon EC2 High-CPU Extra Large
1002 instances (each gives about 8 real cores), this test costs
1003 currently about 50USD/EUR, but can be made to run arbitrarily fast.
1005 On the other hand, choosing 1.5 lines from, say, the Wheel of Time
1006 is in itself not more secure, but the book selection adds quite a
1007 bit of entropy. (Now that I have mentioned it here, don't use tWoT
1008 either!) If you add 2 or 3 typos or switch some words around, then
1009 this is good passphrase material.
1012 * 5.10 What is "iteration count" and why is decreasing it a bad idea?
1014 Iteration count is the number of PBKDF2 iterations a passphrase is
1015 put through before it is used to unlock a key-slot. Iterations are
1016 done with the explicit purpose to increase the time that it takes
1017 to unlock a key-slot. This provides some protection against use of
1018 low-entropy passphrases.
1020 The idea is that an attacker has to try all possible passphrases.
1021 Even if the attacker knows the passphrase is low-entropy (see last
1022 item), it is possible to make each individual try take longer. The
1023 way to do this is to repeatedly hash the passphrase for a certain
1024 time. The attacker then has to spend the same time (given the same
1025 computing power) as the user per try. With LUKS, the default is 1
1026 second of PBKDF2 hashing.
1028 Example 1: Lets assume we have a really bad passphrase (e.g. a
1029 girlfriends name) with 10 bits of entropy. With the same CPU, an
1030 attacker would need to spend around 500 seconds on average to
1031 break that passphrase. Without iteration, it would be more like
1032 0.0001 seconds on a modern CPU.
1034 Example 2: The user did a bit better and has 32 chars of English
1035 text. That would be about 32 bits of entropy. With 1 second
1036 iteration, that means an attacker on the same CPU needs around 136
1037 years. That is pretty impressive for such a weak passphrase.
1038 Without the iterations, it would be more like 50 days on a modern
1039 CPU, and possibly far less.
1041 In addition, the attacker can both parallelize and use special
1042 hardware like GPUs or FPGAs to speed up the attack. The attack can
1043 also happen quite some time after the luksFormat operation and CPUs
1044 can have become faster and cheaper. For that reason you want a
1045 bit of extra security. Anyways, in Example 1 your are screwed.
1046 In example 2, not necessarily. Even if the attack is faster, it
1047 still has a certain cost associated with it, say 10000 EUR/USD
1048 with iteration and 1 EUR/USD without iteration. The first can be
1049 prohibitively expensive, while the second is something you try
1050 even without solid proof that the decryption will yield something
1053 The numbers above are mostly made up, but show the idea. Of course
1054 the best thing is to have a high-entropy passphrase.
1056 Would a 100 sec iteration time be even better? Yes and no.
1057 Cryptographically it would be a lot better, namely 100 times better.
1058 However, usability is a very important factor for security
1059 technology and one that gets overlooked surprisingly often. For
1060 LUKS, if you have to wait 2 minutes to unlock the LUKS container,
1061 most people will not bother and use less secure storage instead. It
1062 is better to have less protection against low-entropy passphrases
1063 and people actually use LUKS, than having them do without
1064 encryption altogether.
1066 Now, what about decreasing the iteration time? This is generally a
1067 very bad idea, unless you know and can enforce that the users only
1068 use high-entropy passphrases. If you decrease the iteration time
1069 without ensuring that, then you put your users at increased risk,
1070 and considering how rarely LUKS containers are unlocked in a
1071 typical work-flow, you do so without a good reason. Don't do it.
1072 The iteration time is already low enough that users with entropy
1073 low passphrases are vulnerable. Lowering it even further increases
1074 this danger significantly.
1077 * 5.11 Some people say PBKDF2 is insecure?
1079 There is some discussion that a hash-function should have a "large
1080 memory" property, i.e. that it should require a lot of memory to be
1081 computed. This serves to prevent attacks using special programmable
1082 circuits, like FPGAs, and attacks using graphics cards. PBKDF2
1083 does not need a lot of memory and is vulnerable to these attacks.
1084 However, the publication usually referred in these discussions is
1085 not very convincing in proving that the presented hash really is
1086 "large memory" (that may change, email the FAQ maintainer when it
1087 does) and it is of limited usefulness anyways. Attackers that use
1088 clusters of normal PCs will not be affected at all by a "large
1089 memory" property. For example the US Secret Service is known to
1090 use the off-hour time of all the office PCs of the Treasury for
1091 password breaking. The Treasury has about 110'000 employees.
1092 Assuming every one has an office PC, that is significant computing
1093 power, all of it with plenty of memory for computing "large
1094 memory" hashes. Bot-net operators also have all the memory they
1095 want. The only protection against a resourceful attacker is a
1096 high-entropy passphrase, see items 5.9 and 5.10.
1099 * 5.12 What about iteration count with plain dm-crypt?
1101 Simple: There is none. There is also no salting. If you use plain
1102 dm-crypt, the only way to be secure is to use a high entropy
1103 passphrase. If in doubt, use LUKS instead.
1106 * 5.13 Is LUKS with default parameters less secure on a slow CPU?
1108 Unfortunately, yes. However the only aspect affected is the
1109 protection for low-entropy passphrase or master-key. All other
1110 security aspects are independent of CPU speed.
1112 The master key is less critical, as you really have to work at it
1113 to give it low entropy. One possibility is to supply the master key
1114 yourself. If that key is low-entropy, then you get what you
1115 deserve. The other known possibility is to use /dev/urandom for
1116 key generation in an entropy-starved situation (e.g. automatic
1117 installation on an embedded device without network and other entropy
1120 For the passphrase, don't use a low-entropy passphrase. If your
1121 passphrase is good, then a slow CPU will not matter. If you insist
1122 on a low-entropy passphrase on a slow CPU, use something like
1123 "--iter-time=10" or higher and wait a long time on each LUKS unlock
1124 and pray that the attacker does not find out in which way exactly
1125 your passphrase is low entropy. This also applies to low-entropy
1126 passphrases on fast CPUs. Technology can do only so much to
1127 compensate for problems in front of the keyboard.
1130 * 5.14 Why was the default aes-cbc-plain replaced with aes-cbc-essiv?
1132 Note: This item applies both to plain dm-crypt and to LUKS
1134 The problem is that cbc-plain has a fingerprint vulnerability, where
1135 a specially crafted file placed into the crypto-container can be
1136 recognized from the outside. The issue here is that for cbc-plain
1137 the initialization vector (IV) is the sector number. The IV gets
1138 XORed to the first data chunk of the sector to be encrypted. If you
1139 make sure that the first data block to be stored in a sector
1140 contains the sector number as well, the first data block to be
1141 encrypted is all zeros and always encrypted to the same ciphertext.
1142 This also works if the first data chunk just has a constant XOR
1143 with the sector number. By having several shifted patterns you can
1144 take care of the case of a non-power-of-two start sector number of
1147 This mechanism allows you to create a pattern of sectors that have
1148 the same first ciphertext block and signal one bit per sector to the
1149 outside, allowing you to e.g. mark media files that way for
1150 recognition without decryption. For large files this is a
1151 practical attack. For small ones, you do not have enough blocks to
1152 signal and take care of different file starting offsets.
1154 In order to prevent this attack, the default was changed to
1155 cbc-essiv. ESSIV uses a keyed hash of the sector number, with the
1156 encryption key as key. This makes the IV unpredictable without
1157 knowing the encryption key and the watermarking attack fails.
1160 * 5.15 Are there any problems with "plain" IV? What is "plain64"?
1162 First, "plain" and "plain64" are both not secure to use with CBC,
1163 see previous FAQ item.
1165 However there are modes, like XTS, that are secure with "plain" IV.
1166 The next limit is that "plain" is 64 bit, with the upper 32 bit set
1167 to zero. This means that on volumes larger than 2TiB, the IV
1168 repeats, creating a vulnerability that potentially leaks some
1169 data. To avoid this, use "plain64", which uses the full sector
1170 number up to 64 bit. Note that "plain64" requires a kernel >=
1171 2.6.33. Also note that "plain64" is backwards compatible for
1172 volume sizes <= 2TiB, but not for those > 2TiB. Finally, "plain64"
1173 does not cause any performance penalty compared to "plain".
1176 * 5.16 What about XTS mode?
1178 XTS mode is potentially even more secure than cbc-essiv (but only if
1179 cbc-essiv is insecure in your scenario). It is a NIST standard and
1180 used, e.g. in Truecrypt. At the moment, if you want to use it, you
1181 have to specify it manually as "aes-xts-plain", i.e.
1183 cryptsetup -c aes-xts-plain luksFormat <device>
1185 For volumes >2TiB and kernels >= 2.6.33 use "plain64" (see FAQ
1186 item on "plain" and "plain64"):
1188 cryptsetup -c aes-xts-plain64 luksFormat <device>
1190 There is a potential security issue with XTS mode and large blocks.
1191 LUKS and dm-crypt always use 512B blocks and the issue does not
1195 * 5.17 Is LUKS FIPS-140-2 certified?
1197 No. But that is more a problem of FIPS-140-2 than of LUKS. From a
1198 technical point-of-view, LUKS with the right parameters would be
1199 FIPS-140-2 compliant, but in order to make it certified, somebody
1200 has to pay real money for that. And then, whenever cryptsetup is
1201 changed or extended, the certification lapses and has to be
1204 From the aspect of actual security, LUKS with default parameters
1205 should be as good as most things that are FIPS-140-2 certified,
1206 although you may want to make sure to use /dev/random (by
1207 specifying --use-random on luksFormat) as randomness source for
1208 the master key to avoid being potentially insecure in an
1209 entropy-starved situation.
1212 * 5.18 What about Plausible Deniability?
1214 First let me attempt a definition for the case of encrypted
1215 filesystems: Plausible deniability is when you hide encrypted data
1216 inside an encrypted container and it is not possible to prove it is
1217 there. The idea is compelling and on first glance it seems
1218 possible to do it. And from a cryptographic point of view, it
1219 actually is possible.
1221 So, does it work in practice? No, unfortunately. The reasoning used
1222 by its proponents is fundamentally flawed in several ways and the
1223 cryptographic properties fail fatally when colliding with the real
1226 First, why should "I do not have a hidden partition" be any more
1227 plausible than "I forgot my crypto key" or "I wiped that partition
1228 with random data, nothing in there"? I do not see any reason.
1230 Second, there are two types of situations: Either they cannot force
1231 you to give them the key (then you simply do not) or the can. In
1232 the second case, they can always do bad things to you, because they
1233 cannot prove that you have the key in the first place! This means
1234 they do not have to prove you have the key, or that this random
1235 looking data on your disk is actually encrypted data. So the
1236 situation will allow them to waterboard/lock-up/deport you
1237 anyways, regardless of how "plausible" your deniability is. Do not
1238 have a hidden partition you could show to them, but there are
1239 indications you may? Too bad for you. Unfortunately "plausible
1240 deniability" also means you cannot prove there is no hidden data.
1242 Third, hidden partitions are not that hidden. There are basically
1243 just two possibilities: a) Make a large crypto container, but put a
1244 smaller filesystem in there and put the hidden partition into the
1245 free space. Unfortunately this is glaringly obvious and can be
1246 detected in an automated fashion. This means that the initial
1247 suspicion to put you under duress in order to make you reveal you
1248 hidden data is given. b) Make a filesystem that spans the whole
1249 encrypted partition, and put the hidden partition into space not
1250 currently used by that filesystem. Unfortunately that is also
1251 glaringly obvious, as you then cannot write to the filesystem
1252 without a high risk of destroying data in the hidden container.
1253 Have not written anything to the encrypted filesystem in a while?
1254 Too bad, they have the suspicion they need to do unpleasant things
1257 To be fair, if you prepare option b) carefully and directly before
1258 going into danger, it may work. But then, the mere presence of
1259 encrypted data may already be enough to get you into trouble in
1260 those places were they can demand encryption keys.
1262 Here is an additional reference for some problems with plausible
1263 deniability: http://www.schneier.com/paper-truecrypt-dfs.pdf I
1264 strongly suggest you read it.
1266 So, no, I will not provide any instructions on how to do it with
1267 plain dm-crypt or LUKS. If you insist on shooting yourself in the
1268 foot, you can figure out how to do it yourself.
1271 * 5.19 What about SSDs or Flash Drives?
1273 The problem is that you cannot reliably erase parts of these
1274 devices, mainly due to wear-leveling and possibly defect
1277 Basically, when overwriting a sector (of 512B), what the device
1278 does is to move an internal sector (may be 128kB or even larger) to
1279 some pool of discarded, not-yet erased unused sectors, take a
1280 fresh empty sector from the empty-sector pool and copy the old
1281 sector over with the changes to the small part you wrote. This is
1282 done in some fashion so that larger writes do not cause a lot of
1283 small internal updates.
1285 The thing is that the mappings between outside-adressable sectors
1286 and inside sectors is arbitrary (and the vendors are not talking).
1287 Also the discarded sectors are not necessarily erased immediately.
1288 They may linger a long time.
1290 For plain dm-crypt, the consequences are that older encrypted data
1291 may be lying around in some internal pools of the device. Thus may
1292 or may not be a problem and depends on the application. Remember
1293 the same can happen with a filesystem if consecutive writes to the
1294 same area of a file can go to different sectors.
1296 However, for LUKS, the worst case is that key-slots and LUKS
1297 header may end up in these internal pools. This means that password
1298 management functionality is compromised (the old passwords may
1299 still be around, potentially for a very long time) and that fast
1300 erase by overwriting the header and key-slot area is insecure.
1302 Also keep in mind that the discarded/used pool may be large. For
1303 example, a 240GB SSD has about 16GB of spare area in the chips that
1304 it is free to do with as it likes. You would need to make each
1305 individual key-slot larger than that to allow reliable overwriting.
1306 And that assumes the disk thinks all other space is in use.
1307 Reading the internal pools using forensic tools is not that hard,
1308 but may involve some soldering.
1312 If you trust the device vendor (you probably should not...) you can
1313 try an ATA "secure erase" command for SSDs. That does not work for
1314 USB keys though. And if it finishes after a few seconds, it was
1315 possibly faked by the SSD.
1317 If you can do without password management and are fine with doing
1318 physical destruction for permenently deleting data (allways after
1319 one or several full overwrites!), you can use plain dm-crypt or
1322 If you want or need the original LUKS security features to work,
1323 you can use a detached LUKS header and put that on a conventional,
1324 magnetic disk. That leaves potentially old encrypted data in the
1325 pools on the disk, but otherwise you get LUKS with the same
1326 security as on a magnetic disk.
1328 If you are concerned about your laptop being stolen, you are likely
1329 fine using LUKS on an SSD. An attacker would need to have access
1330 to an old passphrase (and the key-slot for this old passphrase
1331 would actually need to still be somewhere in the SSD) for your
1332 data to be at risk. So unless you pasted your old passphrase all
1333 over the Internet or the attacker has knowledge of it from some
1334 other source and does a targetted laptop theft to get at your
1335 data, you should be fine.
1338 6. Backup and Data Recovery
1341 * 6.1 Why do I need Backup?
1343 First, disks die. The rate for well-treated (!) disk is about 5%
1344 per year, which is high enough to worry about. There is some
1345 indication that this may be even worse for some SSDs. This applies
1346 both to LUKS and plain dm-crypt partitions.
1348 Second, for LUKS, if anything damages the LUKS header or the
1349 key-stripe area then decrypting the LUKS device can become
1350 impossible. This is a frequent occurrence. For example an
1351 accidental format as FAT or some software overwriting the first
1352 sector where it suspects a partition boot sector typically makes a
1353 LUKS partition permanently inaccessible. See more below on LUKS
1356 So, data-backup in some form is non-optional. For LUKS, you may
1357 also want to store a header backup in some secure location. This
1358 only needs an update if you change passphrases.
1361 * 6.2 How do I backup a LUKS header?
1363 While you could just copy the appropriate number of bytes from the
1364 start of the LUKS partition, the best way is to use command option
1365 "luksHeaderBackup" of cryptsetup. This protects also against
1366 errors when non-standard parameters have been used in LUKS
1367 partition creation. Example:
1370 cryptsetup luksHeaderBackup --header-backup-file <file> <device>
1372 To restore, use the inverse command, i.e.
1374 cryptsetup luksHeaderRestore --header-backup-file <file> <device>
1377 * 6.3 How do I test a LUKS header?
1381 cryptsetup -v isLuks <device>
1383 on the device. Without the "-v" it just signals its result via
1384 exit-status. You can also use the more general test
1388 which will also detect other types and give some more info. Omit
1389 "-p" for old versions of blkid that do not support it.
1392 * 6.4 How do I backup a LUKS or dm-crypt partition?
1394 There are two options, a sector-image and a plain file or
1395 filesystem backup of the contents of the partition. The sector
1396 image is already encrypted, but cannot be compressed and contains
1397 all empty space. The filesystem backup can be compressed, can
1398 contain only part of the encrypted device, but needs to be
1399 encrypted separately if so desired.
1401 A sector-image will contain the whole partition in encrypted form,
1402 for LUKS the LUKS header, the keys-slots and the data area. It can
1403 be done under Linux e.g. with dd_rescue (for a direct image copy)
1404 and with "cat" or "dd". Example:
1406 cat /dev/sda10 > sda10.img
1407 dd_rescue /dev/sda10 sda10.img
1409 You can also use any other backup software that is capable of making
1410 a sector image of a partition. Note that compression is
1411 ineffective for encrypted data, hence it does not make sense to
1414 For a filesystem backup, you decrypt and mount the encrypted
1415 partition and back it up as you would a normal filesystem. In this
1416 case the backup is not encrypted, unless your encryption method
1417 does that. For example you can encrypt a backup with "tar" as
1420 tar cjf - <path> | gpg --cipher-algo AES -c - > backup.tbz2.gpg
1422 And verify the backup like this if you are at "path":
1424 cat backup.tbz2.gpg | gpg - | tar djf -
1426 Note: Always verify backups, especially encrypted ones.
1428 In both cases GnuPG will ask you interactively for your symmetric
1429 key. The verify will only output errors. Use "tar dvjf -" to get
1430 all comparison results. To make sure no data is written to disk
1431 unencrypted, turn off swap if it is not encrypted before doing the
1434 You can of course use different or no compression and you can use
1435 an asymmetric key if you have one and have a backup of the secret
1436 key that belongs to it.
1438 A second option for a filesystem-level backup that can be used when
1439 the backup is also on local disk (e.g. an external USB drive) is
1440 to use a LUKS container there and copy the files to be backed up
1441 between both mounted containers. Also see next item.
1444 * 6.5 Do I need a backup of the full partition? Would the header and
1445 key-slots not be enough?
1447 Backup protects you against two things: Disk loss or corruption
1448 and user error. By far the most questions on the dm-crypt mailing
1449 list about how to recover a damaged LUKS partition are related
1450 to user error. For example, if you create a new filesystem on a
1451 LUKS partition, chances are good that all data is lost
1454 For this case, a header+key-slot backup would often be enough. But
1455 keep in mind that a well-treated (!) HDD has roughly a failure
1456 risk of 5% per year. It is highly advisable to have a complete
1457 backup to protect against this case.
1460 * *6.6 What do I need to backup if I use "decrypt_derived"?
1462 This is a script in Debian, intended for mounting /tmp or swap with
1463 a key derived from the master key of an already decrypted device.
1464 If you use this for an device with data that should be persistent,
1465 you need to make sure you either do not lose access to that master
1466 key or have a backup of the data. If you derive from a LUKS
1467 device, a header backup of that device would cover backing up the
1468 master key. Keep in mind that this does not protect against disk
1471 Note: If you recreate the LUKS header of the device you derive from
1472 (using luksFormat), the master key changes even if you use the same
1473 passphrase(s) and you will not be able to decrypt the derived
1474 device with the new LUKS header.
1477 * 6.7 Does a backup compromise security?
1479 Depends on how you do it. However if you do not have one, you are
1480 going to eventually lose your encrypted data.
1482 There are risks introduced by backups. For example if you
1483 change/disable a key-slot in LUKS, a binary backup of the partition
1484 will still have the old key-slot. To deal with this, you have to
1485 be able to change the key-slot on the backup as well, securely
1486 erase the backup or do a filesystem-level backup instead of a binary
1489 If you use dm-crypt, backup is simpler: As there is no key
1490 management, the main risk is that you cannot wipe the backup when
1491 wiping the original. However wiping the original for dm-crypt
1492 should consist of forgetting the passphrase and that you can do
1493 without actual access to the backup.
1495 In both cases, there is an additional (usually small) risk with
1496 binary backups: An attacker can see how many sectors and which
1497 ones have been changed since the backup. To prevent this, use a
1498 filesystem level backup method that encrypts the whole backup in
1499 one go, e.g. as described above with tar and GnuPG.
1501 My personal advice is to use one USB disk (low value data) or
1502 three disks (high value data) in rotating order for backups, and
1503 either use independent LUKS partitions on them, or use encrypted
1504 backup with tar and GnuPG.
1506 If you do network-backup or tape-backup, I strongly recommend to
1507 go the filesystem backup path with independent encryption, as you
1508 typically cannot reliably delete data in these scenarios,
1509 especially in a cloud setting. (Well, you can burn the tape if it
1510 is under your control...)
1513 * 6.8 What happens if I overwrite the start of a LUKS partition or
1514 damage the LUKS header or key-slots?
1516 There are two critical components for decryption: The salt values
1517 in the header itself and the key-slots. If the salt values are
1518 overwritten or changed, nothing (in the cryptographically strong
1519 sense) can be done to access the data, unless there is a backup
1520 of the LUKS header. If a key-slot is damaged, the data can still
1521 be read with a different key-slot, if there is a remaining
1522 undamaged and used key-slot. Note that in order to make a key-slot
1523 unrecoverable in a cryptographically strong sense, changing about
1524 4-6 bits in random locations of its 128kiB size is quite enough.
1527 * 6.9 What happens if I (quick) format a LUKS partition?
1529 I have not tried the different ways to do this, but very likely you
1530 will have written a new boot-sector, which in turn overwrites the
1531 LUKS header, including the salts, making your data permanently
1532 irretrievable, unless you have a LUKS header backup. You may also
1533 damage the key-slots in part or in full. See also last item.
1536 * 6.10 How do I recover the master key from a mapped LUKS container?
1538 This is typically only needed if you managed to damage your LUKS
1539 header, but the container is still mapped, i.e. "luksOpen"ed. It
1540 also helps if you have a mapped container that you forgot or do not
1541 know a passphrase for (e.g. on a long running server.)
1543 WARNING: Things go wrong, do a full backup before trying this!
1545 WARNING: This exposes the master key of the LUKS container. Note
1546 that both ways to recreate a LUKS header with the old master key
1547 described below will write the master key to disk. Unless you are
1548 sure you have securely erased it afterwards, e.g. by writing it to
1549 an encrypted partition, RAM disk or by erasing the filesystem you
1550 wrote it to by a complete overwrite, you should change the master
1551 key afterwards. Changing the master key requires a full data
1552 backup, luksFormat and then restore of the backup.
1554 First, there is a script by Milan that automates the whole
1555 process, except generating a new LUKS header with the old master
1556 key (it prints the command for that though):
1558 http://code.google.com/p/cryptsetup/source/browse/misc/luks-header-from-active
1560 You can also do this manually. Here is how:
1562 - Get the master key from the device mapper. This is done by the
1563 following command. Substitute c5 for whatever you mapped to:
1565 # dmsetup table --target crypt --showkey /dev/mapper/c5
1567 0 200704 crypt aes-cbc-essiv:sha256
1568 a1704d9715f73a1bb4db581dcacadaf405e700d591e93e2eaade13ba653d0d09
1571 The result is actually one line, wrapped here for clarity. The long
1572 hex string is the master key.
1574 - Convert the master key to a binary file representation. You can
1575 do this manually, e.g. with hexedit. You can also use the tool
1576 "xxd" from vim like this:
1578 echo "a1704d9....53d0d09" | xxd -r -p > <master-key-file>
1580 - Do a luksFormat to create a new LUKS header.
1582 NOTE: If your header is intact and you just forgot the
1583 passphrase, you can just set a new passphrase, see next
1586 Unmap the device before you do that (luksClose). Then do
1588 cryptsetup luksFormat --master-key-file=<master-key-file> <luks device>
1590 Note that if the container was created with other than the default
1591 settings of the cryptsetup version you are using, you need to give
1592 additional parameters specifying the deviations. If in doubt, try
1593 the script by Milan. It does recover the other parameters as well.
1595 Side note: This is the way the decrypt_derived script gets at the
1596 master key. It just omits the conversion and hashes the master key
1599 - If the header is intact and you just forgot the passphrase, just
1600 set a new passphrase like this:
1602 cryptsetup luksAddKey --master-key-file=<master-key-file> <luks device>
1604 You may want to disable the old one afterwards.
1607 * 6.11 What does the on-disk structure of dm-crypt look like?
1609 There is none. dm-crypt takes a block device and gives encrypted
1610 access to each of its blocks with a key derived from the passphrase
1611 given. If you use a cipher different than the default, you have to
1612 specify that as a parameter to cryptsetup too. If you want to
1613 change the password, you basically have to create a second
1614 encrypted device with the new passphrase and copy your data over.
1615 On the plus side, if you accidentally overwrite any part of a
1616 dm-crypt device, the damage will be limited to the are you
1620 * 6.12 What does the on-disk structure of LUKS look like?
1622 A LUKS partition consists of a header, followed by 8 key-slot
1623 descriptors, followed by 8 key slots, followed by the encrypted
1626 Header and key-slot descriptors fill the first 592 bytes. The
1627 key-slot size depends on the creation parameters, namely on the
1628 number of anti-forensic stripes, key material offset and master
1631 With the default parameters, each key-slot is a bit less than
1632 128kiB in size. Due to sector alignment of the key-slot start,
1633 that means the key block 0 is at offset 0x1000-0x20400, key
1634 block 1 at offset 0x21000-0x40400, and key block 7 at offset
1635 0xc1000-0xe0400. The space to the next full sector address is
1636 padded with zeros. Never used key-slots are filled with what the
1637 disk originally contained there, a key-slot removed with
1638 "luksRemoveKey" or "luksKillSlot" gets filled with 0xff. Due to
1639 2MiB default alignment, start of the data area for cryptsetup 1.3
1640 and later is at 2MiB, i.e. at 0x200000. For older versions, it is
1641 at 0x101000, i.e. at 1'052'672 bytes, i.e. at 1MiB + 4096 bytes
1642 from the start of the partition. Incidentally, "luksHeaderBackup"
1643 for a LUKS container created with default parameters dumps exactly
1644 the first 2MiB (or 1'052'672 bytes for headers created with
1645 cryptsetup versions < 1.3) to file and "luksHeaderRestore" restores
1648 For non-default parameters, you have to figure out placement
1649 yourself. "luksDump" helps. See also next item. For the most common
1650 non-default settings, namely aes-xts-plain with 512 bit key, the
1651 offsets are: 1st keyslot 0x1000-0x3f800, 2nd keyslot
1652 0x40000-0x7e000, 3rd keyslot 0x7e000-0xbd800, ..., and start of
1653 bulk data at 0x200000.
1655 The exact specification of the format is here:
1656 http://code.google.com/p/cryptsetup/wiki/Specification
1659 * 6.13 What is the smallest possible LUKS container?
1661 Note: From cryptsetup 1.3 onwards, alignment is set to 1MB. With
1662 modern Linux partitioning tools that also align to 1MB, this will
1663 result in alignment to 2k sectors and typical Flash/SSD sectors,
1664 which is highly desirable for a number of reasons. Changing the
1665 alignment is not recommended.
1667 That said, with default parameters, the data area starts at
1668 exactly 2MB offset (at 0x101000 for cryptsetup versions before
1669 1.3). The smallest data area you can have is one sector of 512
1670 bytes. Data areas of 0 bytes can be created, but fail on mapping.
1672 While you cannot put a filesystem into something this small, it may
1673 still be used to contain, for example, key. Note that with current
1674 formatting tools, a partition for a container this size will be
1675 3MiB anyways. If you put the LUKS container into a file (via
1676 losetup and a loopback device), the file needs to be 2097664 bytes
1677 in size, i.e. 2MiB + 512B.
1679 There two ways to influence the start of the data area are key-size
1682 For alignment, you can go down to 1 on the parameter. This will
1683 still leave you with a data-area starting at 0x101000, i.e.
1684 1MiB+4096B (default parameters) as alignment will be rounded up to
1685 the next multiple of 8 (i.e. 4096 bytes) If in doubt, do a dry-run
1686 on a larger file and dump the LUKS header to get actual
1689 For key-size, you can use 128 bit (e.g. AES-128 with CBC), 256 bit
1690 (e.g. AES-256 with CBC) or 512 bit (e.g. AES-256 with XTS mode).
1691 You can do 64 bit (e.g. blowfish-64 with CBC), but anything below
1692 128 bit has to be considered insecure today.
1694 Example 1 - AES 128 bit with CBC:
1696 cryptsetup luksFormat -s 128 --align-payload=8 <device>
1698 This results in a data offset of 0x81000, i.e. 516KiB or 528384
1699 bytes. Add one 512 byte sector and the smallest LUKS container size
1700 with these parameters is 516KiB + 512B or 528896 bytes.
1702 Example 2 - Blowfish 64 bit with CBC (WARNING: insecure):
1704 cryptsetup luksFormat -c blowfish -s 64 --align-payload=8 /dev/loop0
1706 This results in a data offset of 0x41000, i.e. 260kiB or 266240
1707 bytes, with a minimal LUKS container size of 260kiB + 512B or
1711 * 6.14 I think this is overly complicated. Is there an alternative?
1713 Not really. Encryption comes at a price. You can use plain
1714 dm-crypt to simplify things a bit. It does not allow multiple
1715 passphrases, but on the plus side, it has zero on disk description
1716 and if you overwrite some part of a plain dm-crypt partition,
1717 exactly the overwritten parts are lost (rounded up to sector
1721 * 6.15 Can I clone a LUKS container?
1723 You can, but it breaks security, because the cloned container has
1724 the same header and hence the same master key. You cannot change
1725 the master key on a LUKS container, even if you change the
1726 passphrase(s), the master key stays the same. That means whoever
1727 has access to one of the clones can decrypt them all, completely
1728 bypassing the passphrases.
1730 The right way to do this is to first luksFormat the target
1731 container, then to clone the contents of the source container, with
1732 both containers mapped, i.e. decrypted. You can clone the decrypted
1733 contents of a LUKS container in binary mode, although you may run
1734 into secondary issues with GUIDs in filesystems, partition tables,
1735 RAID-components and the like. These are just the normal problems
1736 binary cloning causes.
1738 Note that if you need to ship (e.g.) cloned LUKS containers with a
1739 default passphrase, that is fine as long as each container was
1740 individually created (and hence has its own master key). In this
1741 case, changing the default passphrase will make it secure again.
1744 7. Interoperability with other Disk Encryption Tools
1747 * 7.1 What is this section about?
1749 Cryptsetup for plain dm-crypt can be used to access a number of
1750 on-disk formats created by tools like loop-aes patched into
1751 losetup. This sometimes works and sometimes does not. This
1752 section collects insights into what works, what does not and where
1753 more information is required.
1755 Additional information may be found in the mailing-list archives,
1756 mentioned at the start of this FAQ document. If you have a
1757 solution working that is not yet documented here and think a wider
1758 audience may be interested, please email the FAQ maintainer.
1761 * 7.2 loop-aes: General observations.
1763 One problem is that there are different versions of losetup around.
1764 loop-aes is a patch for losetup. Possible problems and deviations
1765 from cryptsetup option syntax include:
1767 - Offsets specified in bytes (cryptsetup: 512 byte sectors)
1769 - The need to specify an IV offset
1771 - Encryption mode needs specifying (e.g. "-c twofish-cbc-plain")
1773 - Key size needs specifying (e.g. "-s 128" for 128 bit keys)
1775 - Passphrase hash algorithm needs specifying
1777 Also note that because plain dm-crypt and loop-aes format does not
1778 have metadata, and while the loopAES extension for cryptsetup tries
1779 autodetection (see command loopaesOpen), it may not always work.
1780 If you still have the old set-up, using a verbosity option (-v)
1781 on mapping with the old tool or having a look into the system logs
1782 after setup could give you the information you need. Below, there
1783 are also some things that worked for somebody.
1786 * 7.3 loop-aes patched into losetup on Debian 5.x, kernel 2.6.32
1788 In this case, the main problem seems to be that this variant of
1789 losetup takes the offset (-o option) in bytes, while cryptsetup
1790 takes it in sectors of 512 bytes each. Example: The losetup command
1792 losetup -e twofish -o 2560 /dev/loop0 /dev/sdb1
1793 mount /dev/loop0 mount-point
1797 cryptsetup create -c twofish -o 5 --skip 5 e1 /dev/sdb1
1798 mount /dev/mapper/e1 mount-point
1801 * 7.4 loop-aes with 160 bit key
1803 This seems to be sometimes used with twofish and blowfish and
1804 represents a 160 bit ripemed160 hash output padded to 196 bit key
1805 length. It seems the corresponding options for cryptsetup are
1807 --cipher twofish-cbc-null -s 192 -h ripemd160:20
1810 * 7.5 loop-aes v1 format OpenSUSE
1812 Apparently this is done by older OpenSUSE distros and stopped
1813 working from OpenSUSE 12.1 to 12.2. One user had success with the
1816 cryptsetup create <target> <device> -c aes -s 128 -h sha256
1819 * 7.6 Kernel encrypted loop device (cryptoloop)
1821 There are a number of different losetup implementations for using
1822 encrypted loop devices so getting this to work may need a bit of
1825 NOTE: Do NOT use this for new containers! Some of the existing
1826 implementations are insecure and future support is uncertain.
1828 Example for a compatible mapping:
1830 losetup -e twofish -N /dev/loop0 /image.img
1834 cryptsetup create image_plain /image.img -c twofish-cbc-plain -H plain
1836 with the mapping being done to /dev/mapper/image_plain instead of
1841 Cipher, mode and pasword hash (or no hash):
1843 -e cipher [-N] => -c cipher-cbc-plain -H plain [-s 256]
1844 -e cipher => -c cipher-cbc-plain -H ripemd160 [-s 256]
1846 Key size and offsets (losetup: bytes, cryptsetuop: sectors of 512
1850 -o 2560 => -o 5 -p 5 # 2560/512 = 5
1852 There is no replacement for --pass-fd, it has to be emulated using
1853 keyfiles, see the cryptsetup man-page.
1856 8. Issues with Specific Versions of cryptsetup
1859 * 8.1 When using the create command for plain dm-crypt with
1860 cryptsetup 1.1.x, the mapping is incompatible and my data is not
1863 With cryptsetup 1.1.x, the distro maintainer can define different
1864 default encryption modes for LUKS and plain devices. You can check
1865 these compiled-in defaults using "cryptsetup --help". Moreover, the
1866 plain device default changed because the old IV mode was
1867 vulnerable to a watermarking attack.
1869 If you are using a plain device and you need a compatible mode, just
1870 specify cipher, key size and hash algorithm explicitly. For
1871 compatibility with cryptsetup 1.0.x defaults, simple use the
1874 cryptsetup create -c aes-cbc-plain -s 256 -h ripemd160 <name> <dev>
1876 LUKS stores cipher and mode in the metadata on disk, avoiding this
1880 * 8.2 cryptsetup on SLED 10 has problems...
1882 SLED 10 is missing an essential kernel patch for dm-crypt, which
1883 is broken in its kernel as a result. There may be a very old
1884 version of cryptsetup (1.0.x) provided by SLED, which should also
1885 not be used anymore as well. My advice would be to drop SLED 10.
1888 9. References and Further Reading
1891 * Purpose of this Section
1893 The purpose of this section is to collect references to all
1894 materials that do not fit the FAQ but are relevant in some fashion.
1895 This can be core topics like the LUKS spec or disk encryption, but
1896 it can also be more tangential, like secure storage management or
1897 cryptography used in LUKS. It should still have relevance to
1898 cryptsetup and its applications.
1900 If you wan to see something added here, send email to the
1901 maintainer (or the cryptsetup mailing list) giving an URL, a
1902 description (1-3 lines preferred) and a section to put it in. You
1903 can also propose new sections.
1905 At this time I would like to limit the references to things that
1906 are available on the web.
1911 - LUKS on-disk format spec:
1912 http://code.google.com/p/cryptsetup/wiki/Specification
1917 - Some code examples are in the source package under docs/examples
1920 * Brute-forcing passphrases
1923 http://news.electricalchemy.net/2009/10/password-cracking-in-cloud-part-5.html
1926 http://it.slashdot.org/story/12/12/05/0623215/new-25-gpu-monster-devours-strong-passwords-in-minutes
1932 * SSD and Flash Disk Related
1938 * Attacks Against Disk Encryption
1941 * Risk Management as Relevant for Disk Encryption
1949 A. Contributors In no particular order: