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 and data. 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 just quickly wipe file systems (old data may remain), use
239 wipefs -a <target device>
241 To wipe file system and data, use something like
243 cat /dev/zero > <target device>
245 This can take a while. To get a progress indicator, you can use
246 the tool dd_rescue (->google) instead or use my stream meter "wcs"
247 (source here: http://www.tansi.org/tools/index.html) in the
250 cat /dev/zero | wcs > <target device>
252 Be very sure you have the right target, all data will be lost!
254 Note that automatic wiping is on the TODO list for cryptsetup, so
255 at some time in the future this will become unnecessary.
257 04) Create the LUKS container:
258 cryptsetup luksFormat <target device>
260 Just follow the on-screen instructions.
262 05) Map the container. Here it will be mapped to /dev/mapper/c1:
263 cryptsetup luksOpen <target device> c1
265 06) (Optionally) wipe the container (make sure you have the right target!):
266 cat /dev/zero > /dev/mapper/c1
268 Note that this creates a small information leak, as an attacker can
269 determine whether a 512 byte block is zero if the attacker has
270 access to the encrypted container multiple times. Typically a
271 competent attacker that has access multiple times can install a
272 passphrase sniffer anyways, so this leakage is not very
273 significant. For getting a progress indicator, see step 03.
275 Note that at some time in the future, cryptsetup will do this for
276 you, but currently it is a TODO list item.
278 07) Create a file system in the mapped container, for example an
279 ext3 file system (any other file system is possible):
281 mke2fs -j /dev/mapper/c1
283 08) Mount your encrypted file system, here on /mnt:
284 mount /dev/mapper/c1 /mnt
286 Done. You can now use the encrypted file system to store data. Be
287 sure to read though the rest of the FAQ, these are just the very
288 basics. In particular, there are a number of mistakes that are
289 easy to make, but will compromise your security.
292 * 2.2 What is the difference between "plain" and LUKS format?
294 First, unless you happen to understand the cryptographic background
295 well, you should use LUKS. It does protect the user from a lot of
296 common mistakes. Plain dm-crypt is for experts.
298 Plain format is just that: It has no metadata on disk, reads all
299 parameters from the commandline (or the defaults), derives a
300 master-key from the passphrase and then uses that to de-/encrypt
301 the sectors of the device, with a direct 1:1 mapping between
302 encrypted and decrypted sectors.
304 Primary advantage is high resilience to damage, as one damaged
305 encrypted sector results in exactly one damaged decrypted sector.
306 Also, it is not readily apparent that there even is encrypted data
307 on the device, as an overwrite with crypto-grade randomness (e.g.
308 from /dev/urandom) looks exactly the same on disk.
310 Side-note: That has limited value against the authorities. In
311 civilized countries, they cannot force you to give up a crypto-key
312 anyways. In the US, the UK and dictatorships around the world,
313 they can force you to give up the keys (using imprisonment or worse
314 to pressure you), and in the worst case, they only need a
315 nebulous "suspicion" about the presence of encrypted data. My
316 advice is to either be ready to give up the keys or to not have
317 encrypted data when traveling to those countries, especially when
318 crossing the borders.
320 Disadvantages are that you do not have all the nice features that
321 the LUKS metadata offers, like multiple passphrases that can be
322 changed, the cipher being stored in the metadata, anti-forensic
323 properties like key-slot diffusion and salts, etc..
325 LUKS format uses a metadata header and 8 key-slot areas that are
326 being placed at the beginning of the disk, see below under "What
327 does the LUKS on-disk format looks like?". The passphrases are used
328 to decrypt a single master key that is stored in the anti-forensic
331 Advantages are a higher usability, automatic configuration of
332 non-default crypto parameters, defenses against low-entropy
333 passphrases like salting and iterated PBKDF2 passphrase hashing,
334 the ability to change passphrases, and others.
336 Disadvantages are that it is readily obvious there is encrypted
337 data on disk (but see side note above) and that damage to the
338 header or key-slots usually results in permanent data-loss. See
339 below under "6. Backup and Data Recovery" on how to reduce that
340 risk. Also the sector numbers get shifted by the length of the
341 header and key-slots and there is a loss of that size in capacity
342 (1MB+4096B for defaults and 2MB for the most commonly used
343 non-default XTS mode).
346 * 2.3 Can I encrypt an already existing, non-empty partition to use
349 There is no converter, and it is not really needed. The way to do
350 this is to make a backup of the device in question, securely wipe
351 the device (as LUKS device initialization does not clear away old
352 data), do a luksFormat, optionally overwrite the encrypted device,
353 create a new filesystem and restore your backup on the now
354 encrypted device. Also refer to sections "Security Aspects" and
355 "Backup and Data Recovery".
357 For backup, plain GNU tar works well and backs up anything likely
358 to be in a filesystem.
361 * 2.4 How do I use LUKS with a loop-device?
363 This can be very handy for experiments. Setup is just the same as
364 with any block device. If you want, for example, to use a 100MiB
365 file as LUKS container, do something like this:
367 head -c 100M /dev/zero > luksfile # create empty file
368 losetup /dev/loop0 luksfile # map luksfile to /dev/loop0
369 cryptsetup luksFormat /dev/loop0 # create LUKS on loop device
371 Afterwards just use /dev/loop0 as a you would use a LUKS partition.
372 To unmap the file when done, use "losetup -d /dev/loop0".
375 * 2.5 When I add a new key-slot to LUKS, it asks for a passphrase but
376 then complains about there not being a key-slot with that
379 That is as intended. You are asked a passphrase of an existing
380 key-slot first, before you can enter the passphrase for the new
381 key-slot. Otherwise you could break the encryption by just adding a
382 new key-slot. This way, you have to know the passphrase of one of
383 the already configured key-slots in order to be able to configure a
387 * 2.6 Encryption on top of RAID or the other way round?
389 Unless you have special needs, place encryption between RAID and
390 filesystem, i.e. encryption on top of RAID. You can do it the other
391 way round, but you have to be aware that you then need to give the
392 passphrase for each individual disk and RAID autodetection will
393 not work anymore. Therefore it is better to encrypt the RAID
394 device, e.g. /dev/dm0 .
397 * 2.7 How do I read a dm-crypt key from file?
399 Note that the file will still be hashed first, just like keyboard
400 input. Use the --key-file option, like this:
402 cryptsetup create --key-file keyfile e1 /dev/loop0
405 * 2.8 How do I read a LUKS slot key from file?
407 What you really do here is to read a passphrase from file, just as
408 you would with manual entry of a passphrase for a key-slot. You can
409 add a new passphrase to a free key-slot, set the passphrase of an
410 specific key-slot or put an already configured passphrase into a
411 file. In the last case make sure no trailing newline (0x0a) is
412 contained in the key file, or the passphrase will not work because
413 the whole file is used as input.
415 To add a new passphrase to a free key slot from file, use something
418 cryptsetup luksAddKey /dev/loop0 keyfile
420 To add a new passphrase to a specific key-slot, use something like
423 cryptsetup luksAddKey --key-slot 7 /dev/loop0 keyfile
425 To supply a key from file to any LUKS command, use the --key-file
426 option, e.g. like this:
428 cryptsetup luksOpen --key-file keyfile /dev/loop0 e1
431 * 2.9 How do I read the LUKS master key from file?
433 The question you should ask yourself first is why you would want to
434 do this. The only legitimate reason I can think of is if you want
435 to have two LUKS devices with the same master key. Even then, I
436 think it would be preferable to just use key-slots with the same
437 passphrase, or to use plain dm-crypt instead. If you really have a
438 good reason, please tell me. If I am convinced, I will add how to
442 * 2.10 What are the security requirements for a key read from file?
444 A file-stored key or passphrase has the same security requirements
445 as one entered interactively, however you can use random bytes and
446 thereby use bytes you cannot type on the keyboard. You can use any
447 file you like as key file, for example a plain text file with a
448 human readable passphrase. To generate a file with random bytes,
449 use something like this:
451 head -c 256 /dev/random > keyfile
454 * 2.11 If I map a journaled file system using dm-crypt/LUKS, does it
455 still provide its usual transactional guarantees?
457 Yes, it does, unless a very old kernel is used. The required flags
458 come from the filesystem layer and are processed and passed onwards
459 by dm-crypt. A bit more information on the process by which
460 transactional guarantees are implemented can be found here:
462 http://lwn.net/Articles/400541/
464 Please note that these "guarantees" are weaker than they appear to
465 be. One problem is that quite a few disks lie to the OS about
466 having flushed their buffers. Some other things can go wrong as
467 well. The filesystem developers are aware of these problems and
468 typically can make it work anyways. That said, dm-crypt/LUKS will
469 not make things worse.
471 One specific problem you can run into though is that you can get
472 short freezes and other slowdowns due to the encryption layer.
473 Encryption takes time and forced flushes will block for that time.
474 For example, I did run into frequent small freezes (1-2 sec) when
475 putting a vmware image on ext3 over dm-crypt. When I went back to
476 ext2, the problem went away. This seems to have gotten better with
477 kernel 2.6.36 and the reworking of filesystem flush locking
478 mechanism (less blocking of CPU activity during flushes). It
479 should improve further and eventually the problem should go away.
482 * 2.12 Can I use LUKS or cryptsetup with a more secure (external)
483 medium for key storage, e.g. TPM or a smartcard?
485 Yes, see the answers on using a file-supplied key. You do have to
486 write the glue-logic yourself though. Basically you can have
487 cryptsetup read the key from STDIN and write it there with your
488 own tool that in turn gets the key from the more secure key
492 * 2.13 Can I resize a dm-crypt or LUKS partition?
494 Yes, you can, as neither dm-crypt nor LUKS stores partition size.
495 Whether you should is a different question. Personally I recommend
496 backup, recreation of the encrypted partition with new size,
497 recreation of the filesystem and restore. This gets around the
498 tricky business of resizing the filesystem. Resizing a dm-crypt or
499 LUKS container does not resize the filesystem in it. The backup is
500 really non-optional here, as a lot can go wrong, resulting in
501 partial or complete data loss. Using something like gparted to
502 resize an encrypted partition is slow, but typically works. This
503 will not change the size of the filesystem hidden under the
506 You also need to be aware of size-based limitations. The one
507 currently relevant is that aes-xts-plain should not be used for
508 encrypted container sizes larger than 2TiB. Use aes-xts-plain64
515 * 3.1 My dm-crypt/LUKS mapping does not work! What general steps are
516 there to investigate the problem?
518 If you get a specific error message, investigate what it claims
519 first. If not, you may want to check the following things.
521 - Check that "/dev", including "/dev/mapper/control" is there. If it
522 is missing, you may have a problem with the "/dev" tree itself or
523 you may have broken udev rules.
525 - Check that you have the device mapper and the crypt target in your
526 kernel. The output of "dmsetup targets" should list a "crypt"
527 target. If it is not there or the command fails, add device mapper
528 and crypt-target to the kernel.
530 - Check that the hash-functions and ciphers you want to use are in
531 the kernel. The output of "cat /proc/crypto" needs to list them.
534 * 3.2 My dm-crypt mapping suddenly stopped when upgrading cryptsetup.
536 The default cipher, hash or mode may have changed (the mode changed
537 from 1.0.x to 1.1.x). See under "Issues With Specific Versions of
541 * 3.3 When I call cryptsetup from cron/CGI, I get errors about
544 If you get errors about unknown parameters or the like that are not
545 present when cryptsetup is called from the shell, make sure you
546 have no older version of cryptsetup on your system that then gets
547 called by cron/CGI. For example some distributions install
548 cryptsetup into /usr/sbin, while a manual install could go to
549 /usr/local/sbin. As a debugging aid, call "cryptsetup --version"
550 from cron/CGI or the non-shell mechanism to be sure the right
554 * 3.4 Unlocking a LUKS device takes very long. Why?
556 The iteration time for a key-slot (see Section 5 for an explanation
557 what iteration does) is calculated when setting a passphrase. By
558 default it is 1 second on the machine where the passphrase is set.
559 If you set a passphrase on a fast machine and then unlock it on a
560 slow machine, the unlocking time can be much longer. Also take into
561 account that up to 8 key-slots have to be tried in order to find the
564 If this is problem, you can add another key-slot using the slow
565 machine with the same passphrase and then remove the old key-slot.
566 The new key-slot will have an iteration count adjusted to 1 second
567 on the slow machine. Use luksKeyAdd and then luksKillSlot or
570 However, this operation will not change volume key iteration count
571 (MK iterations in output of "cryptsetup luksDump"). In order to
572 change that, you will have to backup the data in the LUKS
573 container (i.e. your encrypted data), luksFormat on the slow
574 machine and restore the data. Note that in the original LUKS
575 specification this value was fixed to 10, but it is now derived
576 from the PBKDF2 benchmark as well and set to iterations in 0.125
577 sec or 1000, whichever is larger. Also note that MK iterations
578 are not very security relevant. But as each key-slot already takes
579 1 second, spending the additional 0.125 seconds really does not
583 * 3.5 "blkid" sees a LUKS UUID and an ext2/swap UUID on the same
584 device. What is wrong?
586 Some old versions of cryptsetup have a bug where the header does
587 not get completely wiped during LUKS format and an older ext2/swap
588 signature remains on the device. This confuses blkid.
590 Fix: Wipe the unused header areas by doing a backup and restore of
591 the header with cryptsetup 1.1.x:
593 cryptsetup luksHeaderBackup --header-backup-file <file> <device>
594 cryptsetup luksHeaderRestore --header-backup-file <file> <device>
597 * 3.6 cryptsetup segfaults on Gentoo amd64 hardened ...
599 There seems to be some interference between the hardening and and
600 the way cryptsetup benchmarks PBKDF2. The solution to this is
601 currently not quite clear for an encrypted root filesystem. For
602 other uses, you can apparently specify USE="dynamic" as compile
603 flag, see http://bugs.gentoo.org/show_bug.cgi?id=283470
609 * 4.1 I get the error "LUKS keyslot x is invalid." What does that
612 This means that the given keyslot has an offset that points
613 outside the valid keyslot area. Typically, the reason is a
614 corrupted LUKS header because something was written to the start of
615 the device the LUKS container is on. Refer to Section "Backup and
616 Data Recovery" and ask on the mailing list if you have trouble
617 diagnosing and (if still possible) repairing this.
620 * 4.2 Can a bad RAM module cause problems?
622 LUKS and dm-crypt can give the RAM quite a workout, especially when
623 combined with software RAID. In particular the combination RAID5 +
624 LUKS + XFS seems to uncover RAM problems that never caused obvious
625 problems before. Symptoms vary, but often the problem manifest
626 itself when copying large amounts of data, typically several times
627 larger than your main memory.
629 Side note: One thing you should always do on large data
630 copy/movements is to run a verify, for example with the "-d"
631 option of "tar" or by doing a set of MD5 checksums on the source
634 find . -type f -exec md5sum \{\} \; > checksum-file
636 and then a "md5sum -c checksum-file" on the other side. If you get
637 mismatches here, RAM is the primary suspect. A lesser suspect is
638 an overclocked CPU. I have found countless hardware problems in
639 verify runs after copying or making backups. Bit errors are much
640 more common than most people think.
642 Some RAM issues are even worse and corrupt structures in one of the
643 layers. This typically results in lockups, CPU state dumps in the
644 system logs, kernel panic or other things. It is quite possible to
645 have the problem with an encrypted device, but not with an
646 otherwise the same unencrypted device. The reason for that is that
647 encryption has an error amplification property: You flip one bit
648 in an encrypted data block, and the decrypted version has half of
649 its bits flipped. This is an important security property for modern
650 ciphers. With the usual modes in cryptsetup (CBC, ESSIV, XTS), you
651 get up to a completely changed 512 byte block per bit error. A
652 corrupt block causes a lot more havoc than the occasionally
653 flipped single bit and can result in various obscure errors.
655 Note, that a verify run on copying between encrypted or
656 unencrypted devices will reliably detect corruption, even when the
657 copying itself did not report any problems. If you find defect
658 RAM, assume all backups and copied data to be suspect, unless you
662 * 4.3 How do I test RAM?
664 First you should know that overclocking often makes memory
665 problems worse. So if you overclock (which I strongly recommend
666 against in a system holding data that has some worth), run the
667 tests with the overclocking active.
669 There are two good options. One is Memtest86+ and the other is
670 "memtester" by Charles Cazabon. Memtest86+ requires a reboot and
671 then takes over the machine, while memtester runs from a
672 root-shell. Both use different testing methods and I have found
673 problems fast with each one that the other needed long to find. I
674 recommend running the following procedure until the first error is
677 - Run Memtest86+ for one cycle
679 - Run memtester for one cycle (shut down as many other applications
682 - Run Memtest86+ for 24h or more
684 - Run memtester for 24h or more
686 If all that does not produce error messages, your RAM may be sound,
687 but I have had one weak bit that Memtest86+ needed around 60 hours
688 to find. If you can reproduce the original problem reliably, a good
689 additional test may be to remove half of the RAM (if you have more
690 than one module) and try whether the problem is still there and if
691 so, try with the other half. If you just have one module, get a
692 different one and try with that. If you do overclocking, reduce
693 the settings to the most conservative ones available and try with
700 * 5.1 How long is a secure passphrase ?
702 This is just the short answer. For more info and explanation of
703 some of the terms used in this item, read the rest of Section 5.
704 The actual recommendation is at the end of this item.
706 First, passphrase length is not really the right measure,
707 passphrase entropy is. For example, a random lowercase letter (a-z)
708 gives you 4.7 bit of entropy, one element of a-z0-9 gives you 5.2
709 bits of entropy, an element of a-zA-Z0-9 gives you 5.9 bits and
710 a-zA-Z0-9!@#$%^&:-+ gives you 6.2 bits. On the other hand, a random
711 English word only gives you 0.6...1.3 bits of entropy per
712 character. Using sentences that make sense gives lower entropy,
713 series of random words gives higher entropy. Do not use sentences
714 that can be tied to you or found on your computer. This type of
715 attack is done routinely today.
717 That said, it does not matter too much what scheme you use, but it
718 does matter how much entropy your passphrase contains, because an
719 attacker has to try on average
721 1/2 * 2^(bits of entropy in passphrase)
723 different passphrases to guess correctly.
725 Historically, estimations tended to use computing time estimates,
726 but more modern approaches try to estimate cost of guessing a
729 As an example, I will try to get an estimate from the numbers in
730 http://it.slashdot.org/story/12/12/05/0623215/new-25-gpu-monster-devours-strong-passwords-in-minutes
731 More references can be found a the end of this document. Note that
732 these are estimates from the defender side, so assuming something
733 is easier than it actually is is fine. An attacker may still have
734 vastly higher cost than estimated here.
736 LUKS uses SHA1 for hasing per default. The claim in the reference is
737 63 billion tries/second for SHA1. We will leave aside the check
738 whether a try actually decrypts a key-slot. Now, the machine has 25
739 GPUs, which I will estimate at an overall lifetime cost of USD/EUR
740 1000 each, and an useful lifetime of 2 years. (This is on the low
741 side.) Disregarding downtime, the machine can then break
743 N = 63*10^9 * 3600 * 24 * 365 * 2 ~ 4*10^18
745 passphrases for EUR/USD 25k. That is one 62 bit passphrase hashed
746 once with SHA1 for EUR/USD 25k. Note that as this can be
747 parallelized, it can be done faster than 2 years with several of
750 For plain dm-crypt (no hash iteration) this is it. This gives (with
751 SHA1, plain dm-crypt default is ripemd160 which seems to be
752 slightly slower than SHA1):
754 Passphrase entropy Cost to break
763 For LUKS, you have to take into account hash iteration in PBKDF2.
764 For a current CPU, there are about 100k iterations (as can be
765 queried with ''cryptsetup luksDump''.
767 The table above then becomes:
769 Passphrase entropy Cost to break
780 To get reasonable security for the next 10 years, it is a good idea
781 to overestimate by a factor of at least 1000.
783 Then there is the question of how much the attacker is willing to
784 spend. That is up to your own security evaluation. For general use,
785 I will assume the attacker is willing to spend up to 1 million
786 EUR/USD. Then we get the following recommendations:
788 Plain dm-crypt: Use > 80 bit. That is e.g. 17 random chars from a-z
789 or a random English sentence of > 135 characters length.
791 LUKS: Use > 65 bit. That is e.g. 14 random chars from a-z or a
792 random English sentence of > 108 characters length.
794 If paranoid, add at least 20 bit. That is roughly four additional
795 characters for random passphrases and roughly 32 characters for a
796 random English sentence.
799 * 5.2 Is LUKS insecure? Everybody can see I have encrypted data!
801 In practice it does not really matter. In most civilized countries
802 you can just refuse to hand over the keys, no harm done. In some
803 countries they can force you to hand over the keys, if they suspect
804 encryption. However the suspicion is enough, they do not have to
805 prove anything. This is for practical reasons, as even the presence
806 of a header (like the LUKS header) is not enough to prove that you
807 have any keys. It might have been an experiment, for example. Or it
808 was used as encrypted swap with a key from /dev/random. So they
809 make you prove you do not have encrypted data. Of course that is
810 just as impossible as the other way round.
812 This means that if you have a large set of random-looking data,
813 they can already lock you up. Hidden containers (encryption hidden
814 within encryption), as possible with Truecrypt, do not help
815 either. They will just assume the hidden container is there and
816 unless you hand over the key, you will stay locked up. Don't have
817 a hidden container? Though luck. Anybody could claim that.
819 Still, if you are concerned about the LUKS header, use plain
820 dm-crypt with a good passphrase. See also Section 2, "What is the
821 difference between "plain" and LUKS format?"
824 * 5.3 Should I initialize (overwrite) a new LUKS/dm-crypt partition?
826 If you just create a filesystem on it, most of the old data will
827 still be there. If the old data is sensitive, you should overwrite
828 it before encrypting. In any case, not initializing will leave the
829 old data there until the specific sector gets written. That may
830 enable an attacker to determine how much and where on the
831 partition data was written. If you think this is a risk, you can
832 prevent this by overwriting the encrypted device (here assumed to
833 be named "e1") with zeros like this:
835 dd_rescue -w /dev/zero /dev/mapper/e1
837 or alternatively with one of the following more standard commands:
839 cat /dev/zero > /dev/mapper/e1
840 dd if=/dev/zero of=/dev/mapper/e1
843 * 5.4 How do I securely erase a LUKS (or other) partition?
845 For LUKS, if you are in a desperate hurry, overwrite the LUKS
846 header and key-slot area. This means overwriting the first
847 (keyslots x stripes x keysize) + offset bytes. For the default
848 parameters, this is the 1'052'672 bytes, i.e. 1MiB + 4096 of the
849 LUKS partition. For 512 bit key length (e.g. for aes-xts-plain with
850 512 bit key) this is 2MiB. (The different offset stems from
851 differences in the sector alignment of the key-slots.) If in doubt,
852 just be generous and overwrite the first 10MB or so, it will likely
853 still be fast enough. A single overwrite with zeros should be
854 enough. If you anticipate being in a desperate hurry, prepare the
855 command beforehand. Example with /dev/sde1 as the LUKS partition
856 and default parameters:
858 head -c 1052672 /dev/zero > /dev/sde1; sync
860 A LUKS header backup or full backup will still grant access to
861 most or all data, so make sure that an attacker does not have
862 access to backups or destroy them as well.
864 If you have time, overwrite the whole LUKS partition with a single
865 pass of zeros. This is enough for current HDDs. For SSDs or FLASH
866 (USB sticks) you may want to overwrite the whole drive several
867 times to be sure data is not retained by wear leveling. This is
868 possibly still insecure as SSD technology is not fully understood
869 in this regard. Still, due to the anti-forensic properties of the
870 LUKS key-slots, a single overwrite of an SSD or FLASH drive could
871 be enough. If in doubt, use physical destruction in addition. Here
872 is a link to some current research results on erasing SSDs and
874 http://www.usenix.org/events/fast11/tech/full_papers/Wei.pdf
876 Keep in mind to also erase all backups.
878 Example for a zero-overwrite erase of partition sde1 done with
881 dd_rescue -w /dev/zero /dev/sde1
884 * 5.5 How do I securely erase a backup of a LUKS partition or header?
886 That depends on the medium it is stored on. For HDD and SSD, use
887 overwrite with zeros. For an SSD or FLASH drive (USB stick), you
888 may want to overwrite the complete SSD several times and use
889 physical destruction in addition, see last item. For re-writable
890 CD/DVD, a single overwrite should also be enough, due to the
891 anti-forensic properties of the LUKS keyslots. For write-once
892 media, use physical destruction. For low security requirements,
893 just cut the CD/DVD into several parts. For high security needs,
894 shred or burn the medium. If your backup is on magnetic tape, I
895 advise physical destruction by shredding or burning, after
896 overwriting . The problem with magnetic tape is that it has a
897 higher dynamic range than HDDs and older data may well be
898 recoverable after overwrites. Also write-head alignment issues can
899 lead to data not actually being deleted at all during overwrites.
902 * 5.6 What about backup? Does it compromise security?
904 That depends. See item 6.7.
907 * 5.7 Why is all my data permanently gone if I overwrite the LUKS
910 Overwriting the LUKS header in part or in full is the most common
911 reason why access to LUKS containers is lost permanently.
912 Overwriting can be done in a number of fashions, like creating a
913 new filesystem on the raw LUKS partition, making the raw partition
914 part of a raid array and just writing to the raw partition.
916 The LUKS header contains a 256 bit "salt" value and without that no
917 decryption is possible. While the salt is not secret, it is
918 key-grade material and cannot be reconstructed. This is a
919 cryptographically strong "cannot". From observations on the
920 cryptsetup mailing-list, people typically go though the usual
921 stages of grief (Denial, Anger, Bargaining, Depression, Acceptance)
922 when this happens to them. Observed times vary between 1 day and 2
923 weeks to complete the cycle. Seeking help on the mailing-list is
924 fine. Even if we usually cannot help with getting back your data,
925 most people found the feedback comforting.
927 If your header does not contain an intact salt, best go directly
928 to the last stage ("Acceptance") and think about what to do now.
929 There is one exception that I know of: If your LUKS container is
930 still open, then it may be possible to extract the master key from
931 the running system. See Item "How do I recover the master key from
932 a mapped LUKS container?" in Section "Backup and Data Recovery".
935 * 5.8 What is a "salt"?
937 A salt is a random key-grade value added to the passphrase before
938 it is processed. It is not kept secret. The reason for using salts
939 is as follows: If an attacker wants to crack the password for a
940 single LUKS container, then every possible passphrase has to be
941 tried. Typically an attacker will not try every binary value, but
942 will try words and sentences from a dictionary.
944 If an attacker wants to attack several LUKS containers with the
945 same dictionary, then a different approach makes sense: Compute the
946 resulting slot-key for each dictionary element and store it on
947 disk. Then the test for each entry is just the slow unlocking with
948 the slot key (say 0.00001 sec) instead of calculating the slot-key
949 first (1 sec). For a single attack, this does not help. But if you
950 have more than one container to attack, this helps tremendously,
951 also because you can prepare your table before you even have the
952 container to attack! The calculation is also very simple to
953 parallelize. You could, for example, use the night-time unused CPU
954 power of your desktop PCs for this.
956 This is where the salt comes in. If the salt is combined with the
957 passphrase (in the simplest form, just appended to it), you
958 suddenly need a separate table for each salt value. With a
959 reasonably-sized salt value (256 bit, e.g.) this is quite
963 * 5.9 Is LUKS secure with a low-entropy (bad) passphrase?
965 Note: You should only use the 94 printable characters from 7 bit
966 ASCII code to prevent your passphrase from failing when the
967 character encoding changes, e.g. because of a system upgrade, see
968 also the note at the very start of this FAQ under "WARNINGS".
970 This needs a bit of theory. The quality of your passphrase is
971 directly related to its entropy (information theoretic, not
972 thermodynamic). The entropy says how many bits of "uncertainty" or
973 "randomness" are in you passphrase. In other words, that is how
974 difficult guessing the passphrase is.
976 Example: A random English sentence has about 1 bit of entropy per
977 character. A random lowercase (or uppercase) character has about
980 Now, if n is the number of bits of entropy in your passphrase and t
981 is the time it takes to process a passphrase in order to open the
982 LUKS container, then an attacker has to spend at maximum
984 attack_time_max = 2^n * t
986 time for a successful attack and on average half that. There is no
987 way getting around that relationship. However, there is one thing
988 that does help, namely increasing t, the time it takes to use a
989 passphrase, see next FAQ item.
991 Still, if you want good security, a high-entropy passphrase is the
992 only option. For example, a low-entropy passphrase can never be
993 considered secure against a TLA-level (Three Letter Agency level,
994 i.e. government-level) attacker, no matter what tricks are used in
995 the key-derivation function. Use at least 64 bits for secret stuff.
996 That is 64 characters of English text (but only if randomly chosen)
997 or a combination of 12 truly random letters and digits.
999 For passphrase generation, do not use lines from very well-known
1000 texts (religious texts, Harry potter, etc.) as they are to easy to
1001 guess. For example, the total Harry Potter has about 1'500'000
1002 words (my estimation). Trying every 64 character sequence starting
1003 and ending at a word boundary would take only something like 20
1004 days on a single CPU and is entirely feasible. To put that into
1005 perspective, using a number of Amazon EC2 High-CPU Extra Large
1006 instances (each gives about 8 real cores), this test costs
1007 currently about 50USD/EUR, but can be made to run arbitrarily fast.
1009 On the other hand, choosing 1.5 lines from, say, the Wheel of Time
1010 is in itself not more secure, but the book selection adds quite a
1011 bit of entropy. (Now that I have mentioned it here, don't use tWoT
1012 either!) If you add 2 or 3 typos or switch some words around, then
1013 this is good passphrase material.
1016 * 5.10 What is "iteration count" and why is decreasing it a bad idea?
1018 Iteration count is the number of PBKDF2 iterations a passphrase is
1019 put through before it is used to unlock a key-slot. Iterations are
1020 done with the explicit purpose to increase the time that it takes
1021 to unlock a key-slot. This provides some protection against use of
1022 low-entropy passphrases.
1024 The idea is that an attacker has to try all possible passphrases.
1025 Even if the attacker knows the passphrase is low-entropy (see last
1026 item), it is possible to make each individual try take longer. The
1027 way to do this is to repeatedly hash the passphrase for a certain
1028 time. The attacker then has to spend the same time (given the same
1029 computing power) as the user per try. With LUKS, the default is 1
1030 second of PBKDF2 hashing.
1032 Example 1: Lets assume we have a really bad passphrase (e.g. a
1033 girlfriends name) with 10 bits of entropy. With the same CPU, an
1034 attacker would need to spend around 500 seconds on average to
1035 break that passphrase. Without iteration, it would be more like
1036 0.0001 seconds on a modern CPU.
1038 Example 2: The user did a bit better and has 32 chars of English
1039 text. That would be about 32 bits of entropy. With 1 second
1040 iteration, that means an attacker on the same CPU needs around 136
1041 years. That is pretty impressive for such a weak passphrase.
1042 Without the iterations, it would be more like 50 days on a modern
1043 CPU, and possibly far less.
1045 In addition, the attacker can both parallelize and use special
1046 hardware like GPUs or FPGAs to speed up the attack. The attack can
1047 also happen quite some time after the luksFormat operation and CPUs
1048 can have become faster and cheaper. For that reason you want a
1049 bit of extra security. Anyways, in Example 1 your are screwed.
1050 In example 2, not necessarily. Even if the attack is faster, it
1051 still has a certain cost associated with it, say 10000 EUR/USD
1052 with iteration and 1 EUR/USD without iteration. The first can be
1053 prohibitively expensive, while the second is something you try
1054 even without solid proof that the decryption will yield something
1057 The numbers above are mostly made up, but show the idea. Of course
1058 the best thing is to have a high-entropy passphrase.
1060 Would a 100 sec iteration time be even better? Yes and no.
1061 Cryptographically it would be a lot better, namely 100 times better.
1062 However, usability is a very important factor for security
1063 technology and one that gets overlooked surprisingly often. For
1064 LUKS, if you have to wait 2 minutes to unlock the LUKS container,
1065 most people will not bother and use less secure storage instead. It
1066 is better to have less protection against low-entropy passphrases
1067 and people actually use LUKS, than having them do without
1068 encryption altogether.
1070 Now, what about decreasing the iteration time? This is generally a
1071 very bad idea, unless you know and can enforce that the users only
1072 use high-entropy passphrases. If you decrease the iteration time
1073 without ensuring that, then you put your users at increased risk,
1074 and considering how rarely LUKS containers are unlocked in a
1075 typical work-flow, you do so without a good reason. Don't do it.
1076 The iteration time is already low enough that users with entropy
1077 low passphrases are vulnerable. Lowering it even further increases
1078 this danger significantly.
1081 * 5.11 Some people say PBKDF2 is insecure?
1083 There is some discussion that a hash-function should have a "large
1084 memory" property, i.e. that it should require a lot of memory to be
1085 computed. This serves to prevent attacks using special programmable
1086 circuits, like FPGAs, and attacks using graphics cards. PBKDF2
1087 does not need a lot of memory and is vulnerable to these attacks.
1088 However, the publication usually referred in these discussions is
1089 not very convincing in proving that the presented hash really is
1090 "large memory" (that may change, email the FAQ maintainer when it
1091 does) and it is of limited usefulness anyways. Attackers that use
1092 clusters of normal PCs will not be affected at all by a "large
1093 memory" property. For example the US Secret Service is known to
1094 use the off-hour time of all the office PCs of the Treasury for
1095 password breaking. The Treasury has about 110'000 employees.
1096 Assuming every one has an office PC, that is significant computing
1097 power, all of it with plenty of memory for computing "large
1098 memory" hashes. Bot-net operators also have all the memory they
1099 want. The only protection against a resourceful attacker is a
1100 high-entropy passphrase, see items 5.9 and 5.10.
1103 * 5.12 What about iteration count with plain dm-crypt?
1105 Simple: There is none. There is also no salting. If you use plain
1106 dm-crypt, the only way to be secure is to use a high entropy
1107 passphrase. If in doubt, use LUKS instead.
1110 * 5.13 Is LUKS with default parameters less secure on a slow CPU?
1112 Unfortunately, yes. However the only aspect affected is the
1113 protection for low-entropy passphrase or master-key. All other
1114 security aspects are independent of CPU speed.
1116 The master key is less critical, as you really have to work at it
1117 to give it low entropy. One possibility is to supply the master key
1118 yourself. If that key is low-entropy, then you get what you
1119 deserve. The other known possibility is to use /dev/urandom for
1120 key generation in an entropy-starved situation (e.g. automatic
1121 installation on an embedded device without network and other entropy
1124 For the passphrase, don't use a low-entropy passphrase. If your
1125 passphrase is good, then a slow CPU will not matter. If you insist
1126 on a low-entropy passphrase on a slow CPU, use something like
1127 "--iter-time=10" or higher and wait a long time on each LUKS unlock
1128 and pray that the attacker does not find out in which way exactly
1129 your passphrase is low entropy. This also applies to low-entropy
1130 passphrases on fast CPUs. Technology can do only so much to
1131 compensate for problems in front of the keyboard.
1134 * 5.14 Why was the default aes-cbc-plain replaced with aes-cbc-essiv?
1136 Note: This item applies both to plain dm-crypt and to LUKS
1138 The problem is that cbc-plain has a fingerprint vulnerability, where
1139 a specially crafted file placed into the crypto-container can be
1140 recognized from the outside. The issue here is that for cbc-plain
1141 the initialization vector (IV) is the sector number. The IV gets
1142 XORed to the first data chunk of the sector to be encrypted. If you
1143 make sure that the first data block to be stored in a sector
1144 contains the sector number as well, the first data block to be
1145 encrypted is all zeros and always encrypted to the same ciphertext.
1146 This also works if the first data chunk just has a constant XOR
1147 with the sector number. By having several shifted patterns you can
1148 take care of the case of a non-power-of-two start sector number of
1151 This mechanism allows you to create a pattern of sectors that have
1152 the same first ciphertext block and signal one bit per sector to the
1153 outside, allowing you to e.g. mark media files that way for
1154 recognition without decryption. For large files this is a
1155 practical attack. For small ones, you do not have enough blocks to
1156 signal and take care of different file starting offsets.
1158 In order to prevent this attack, the default was changed to
1159 cbc-essiv. ESSIV uses a keyed hash of the sector number, with the
1160 encryption key as key. This makes the IV unpredictable without
1161 knowing the encryption key and the watermarking attack fails.
1164 * 5.15 Are there any problems with "plain" IV? What is "plain64"?
1166 First, "plain" and "plain64" are both not secure to use with CBC,
1167 see previous FAQ item.
1169 However there are modes, like XTS, that are secure with "plain" IV.
1170 The next limit is that "plain" is 64 bit, with the upper 32 bit set
1171 to zero. This means that on volumes larger than 2TiB, the IV
1172 repeats, creating a vulnerability that potentially leaks some
1173 data. To avoid this, use "plain64", which uses the full sector
1174 number up to 64 bit. Note that "plain64" requires a kernel >=
1175 2.6.33. Also note that "plain64" is backwards compatible for
1176 volume sizes <= 2TiB, but not for those > 2TiB. Finally, "plain64"
1177 does not cause any performance penalty compared to "plain".
1180 * 5.16 What about XTS mode?
1182 XTS mode is potentially even more secure than cbc-essiv (but only if
1183 cbc-essiv is insecure in your scenario). It is a NIST standard and
1184 used, e.g. in Truecrypt. At the moment, if you want to use it, you
1185 have to specify it manually as "aes-xts-plain", i.e.
1187 cryptsetup -c aes-xts-plain luksFormat <device>
1189 For volumes >2TiB and kernels >= 2.6.33 use "plain64" (see FAQ
1190 item on "plain" and "plain64"):
1192 cryptsetup -c aes-xts-plain64 luksFormat <device>
1194 There is a potential security issue with XTS mode and large blocks.
1195 LUKS and dm-crypt always use 512B blocks and the issue does not
1199 * 5.17 Is LUKS FIPS-140-2 certified?
1201 No. But that is more a problem of FIPS-140-2 than of LUKS. From a
1202 technical point-of-view, LUKS with the right parameters would be
1203 FIPS-140-2 compliant, but in order to make it certified, somebody
1204 has to pay real money for that. And then, whenever cryptsetup is
1205 changed or extended, the certification lapses and has to be
1208 From the aspect of actual security, LUKS with default parameters
1209 should be as good as most things that are FIPS-140-2 certified,
1210 although you may want to make sure to use /dev/random (by
1211 specifying --use-random on luksFormat) as randomness source for
1212 the master key to avoid being potentially insecure in an
1213 entropy-starved situation.
1216 * 5.18 What about Plausible Deniability?
1218 First let me attempt a definition for the case of encrypted
1219 filesystems: Plausible deniability is when you hide encrypted data
1220 inside an encrypted container and it is not possible to prove it is
1221 there. The idea is compelling and on first glance it seems
1222 possible to do it. And from a cryptographic point of view, it
1223 actually is possible.
1225 So, does it work in practice? No, unfortunately. The reasoning used
1226 by its proponents is fundamentally flawed in several ways and the
1227 cryptographic properties fail fatally when colliding with the real
1230 First, why should "I do not have a hidden partition" be any more
1231 plausible than "I forgot my crypto key" or "I wiped that partition
1232 with random data, nothing in there"? I do not see any reason.
1234 Second, there are two types of situations: Either they cannot force
1235 you to give them the key (then you simply do not) or the can. In
1236 the second case, they can always do bad things to you, because they
1237 cannot prove that you have the key in the first place! This means
1238 they do not have to prove you have the key, or that this random
1239 looking data on your disk is actually encrypted data. So the
1240 situation will allow them to waterboard/lock-up/deport you
1241 anyways, regardless of how "plausible" your deniability is. Do not
1242 have a hidden partition you could show to them, but there are
1243 indications you may? Too bad for you. Unfortunately "plausible
1244 deniability" also means you cannot prove there is no hidden data.
1246 Third, hidden partitions are not that hidden. There are basically
1247 just two possibilities: a) Make a large crypto container, but put a
1248 smaller filesystem in there and put the hidden partition into the
1249 free space. Unfortunately this is glaringly obvious and can be
1250 detected in an automated fashion. This means that the initial
1251 suspicion to put you under duress in order to make you reveal you
1252 hidden data is given. b) Make a filesystem that spans the whole
1253 encrypted partition, and put the hidden partition into space not
1254 currently used by that filesystem. Unfortunately that is also
1255 glaringly obvious, as you then cannot write to the filesystem
1256 without a high risk of destroying data in the hidden container.
1257 Have not written anything to the encrypted filesystem in a while?
1258 Too bad, they have the suspicion they need to do unpleasant things
1261 To be fair, if you prepare option b) carefully and directly before
1262 going into danger, it may work. But then, the mere presence of
1263 encrypted data may already be enough to get you into trouble in
1264 those places were they can demand encryption keys.
1266 Here is an additional reference for some problems with plausible
1267 deniability: http://www.schneier.com/paper-truecrypt-dfs.pdf I
1268 strongly suggest you read it.
1270 So, no, I will not provide any instructions on how to do it with
1271 plain dm-crypt or LUKS. If you insist on shooting yourself in the
1272 foot, you can figure out how to do it yourself.
1275 * 5.19 What about SSDs or Flash Drives?
1277 The problem is that you cannot reliably erase parts of these
1278 devices, mainly due to wear-leveling and possibly defect
1281 Basically, when overwriting a sector (of 512B), what the device
1282 does is to move an internal sector (may be 128kB or even larger) to
1283 some pool of discarded, not-yet erased unused sectors, take a
1284 fresh empty sector from the empty-sector pool and copy the old
1285 sector over with the changes to the small part you wrote. This is
1286 done in some fashion so that larger writes do not cause a lot of
1287 small internal updates.
1289 The thing is that the mappings between outside-adressable sectors
1290 and inside sectors is arbitrary (and the vendors are not talking).
1291 Also the discarded sectors are not necessarily erased immediately.
1292 They may linger a long time.
1294 For plain dm-crypt, the consequences are that older encrypted data
1295 may be lying around in some internal pools of the device. Thus may
1296 or may not be a problem and depends on the application. Remember
1297 the same can happen with a filesystem if consecutive writes to the
1298 same area of a file can go to different sectors.
1300 However, for LUKS, the worst case is that key-slots and LUKS
1301 header may end up in these internal pools. This means that password
1302 management functionality is compromised (the old passwords may
1303 still be around, potentially for a very long time) and that fast
1304 erase by overwriting the header and key-slot area is insecure.
1306 Also keep in mind that the discarded/used pool may be large. For
1307 example, a 240GB SSD has about 16GB of spare area in the chips that
1308 it is free to do with as it likes. You would need to make each
1309 individual key-slot larger than that to allow reliable overwriting.
1310 And that assumes the disk thinks all other space is in use.
1311 Reading the internal pools using forensic tools is not that hard,
1312 but may involve some soldering.
1316 If you trust the device vendor (you probably should not...) you can
1317 try an ATA "secure erase" command for SSDs. That does not work for
1318 USB keys though. And if it finishes after a few seconds, it was
1319 possibly faked by the SSD.
1321 If you can do without password management and are fine with doing
1322 physical destruction for permenently deleting data (allways after
1323 one or several full overwrites!), you can use plain dm-crypt or
1326 If you want or need the original LUKS security features to work,
1327 you can use a detached LUKS header and put that on a conventional,
1328 magnetic disk. That leaves potentially old encrypted data in the
1329 pools on the disk, but otherwise you get LUKS with the same
1330 security as on a magnetic disk.
1332 If you are concerned about your laptop being stolen, you are likely
1333 fine using LUKS on an SSD. An attacker would need to have access
1334 to an old passphrase (and the key-slot for this old passphrase
1335 would actually need to still be somewhere in the SSD) for your
1336 data to be at risk. So unless you pasted your old passphrase all
1337 over the Internet or the attacker has knowledge of it from some
1338 other source and does a targetted laptop theft to get at your
1339 data, you should be fine.
1342 6. Backup and Data Recovery
1345 * 6.1 Why do I need Backup?
1347 First, disks die. The rate for well-treated (!) disk is about 5%
1348 per year, which is high enough to worry about. There is some
1349 indication that this may be even worse for some SSDs. This applies
1350 both to LUKS and plain dm-crypt partitions.
1352 Second, for LUKS, if anything damages the LUKS header or the
1353 key-stripe area then decrypting the LUKS device can become
1354 impossible. This is a frequent occurrence. For example an
1355 accidental format as FAT or some software overwriting the first
1356 sector where it suspects a partition boot sector typically makes a
1357 LUKS partition permanently inaccessible. See more below on LUKS
1360 So, data-backup in some form is non-optional. For LUKS, you may
1361 also want to store a header backup in some secure location. This
1362 only needs an update if you change passphrases.
1365 * 6.2 How do I backup a LUKS header?
1367 While you could just copy the appropriate number of bytes from the
1368 start of the LUKS partition, the best way is to use command option
1369 "luksHeaderBackup" of cryptsetup. This protects also against
1370 errors when non-standard parameters have been used in LUKS
1371 partition creation. Example:
1374 cryptsetup luksHeaderBackup --header-backup-file <file> <device>
1376 To restore, use the inverse command, i.e.
1378 cryptsetup luksHeaderRestore --header-backup-file <file> <device>
1381 * 6.3 How do I test a LUKS header?
1385 cryptsetup -v isLuks <device>
1387 on the device. Without the "-v" it just signals its result via
1388 exit-status. You can also use the more general test
1392 which will also detect other types and give some more info. Omit
1393 "-p" for old versions of blkid that do not support it.
1396 * 6.4 How do I backup a LUKS or dm-crypt partition?
1398 There are two options, a sector-image and a plain file or
1399 filesystem backup of the contents of the partition. The sector
1400 image is already encrypted, but cannot be compressed and contains
1401 all empty space. The filesystem backup can be compressed, can
1402 contain only part of the encrypted device, but needs to be
1403 encrypted separately if so desired.
1405 A sector-image will contain the whole partition in encrypted form,
1406 for LUKS the LUKS header, the keys-slots and the data area. It can
1407 be done under Linux e.g. with dd_rescue (for a direct image copy)
1408 and with "cat" or "dd". Example:
1410 cat /dev/sda10 > sda10.img
1411 dd_rescue /dev/sda10 sda10.img
1413 You can also use any other backup software that is capable of making
1414 a sector image of a partition. Note that compression is
1415 ineffective for encrypted data, hence it does not make sense to
1418 For a filesystem backup, you decrypt and mount the encrypted
1419 partition and back it up as you would a normal filesystem. In this
1420 case the backup is not encrypted, unless your encryption method
1421 does that. For example you can encrypt a backup with "tar" as
1424 tar cjf - <path> | gpg --cipher-algo AES -c - > backup.tbz2.gpg
1426 And verify the backup like this if you are at "path":
1428 cat backup.tbz2.gpg | gpg - | tar djf -
1430 Note: Always verify backups, especially encrypted ones.
1432 In both cases GnuPG will ask you interactively for your symmetric
1433 key. The verify will only output errors. Use "tar dvjf -" to get
1434 all comparison results. To make sure no data is written to disk
1435 unencrypted, turn off swap if it is not encrypted before doing the
1438 You can of course use different or no compression and you can use
1439 an asymmetric key if you have one and have a backup of the secret
1440 key that belongs to it.
1442 A second option for a filesystem-level backup that can be used when
1443 the backup is also on local disk (e.g. an external USB drive) is
1444 to use a LUKS container there and copy the files to be backed up
1445 between both mounted containers. Also see next item.
1448 * 6.5 Do I need a backup of the full partition? Would the header and
1449 key-slots not be enough?
1451 Backup protects you against two things: Disk loss or corruption
1452 and user error. By far the most questions on the dm-crypt mailing
1453 list about how to recover a damaged LUKS partition are related
1454 to user error. For example, if you create a new filesystem on a
1455 LUKS partition, chances are good that all data is lost
1458 For this case, a header+key-slot backup would often be enough. But
1459 keep in mind that a well-treated (!) HDD has roughly a failure
1460 risk of 5% per year. It is highly advisable to have a complete
1461 backup to protect against this case.
1464 * *6.6 What do I need to backup if I use "decrypt_derived"?
1466 This is a script in Debian, intended for mounting /tmp or swap with
1467 a key derived from the master key of an already decrypted device.
1468 If you use this for an device with data that should be persistent,
1469 you need to make sure you either do not lose access to that master
1470 key or have a backup of the data. If you derive from a LUKS
1471 device, a header backup of that device would cover backing up the
1472 master key. Keep in mind that this does not protect against disk
1475 Note: If you recreate the LUKS header of the device you derive from
1476 (using luksFormat), the master key changes even if you use the same
1477 passphrase(s) and you will not be able to decrypt the derived
1478 device with the new LUKS header.
1481 * 6.7 Does a backup compromise security?
1483 Depends on how you do it. However if you do not have one, you are
1484 going to eventually lose your encrypted data.
1486 There are risks introduced by backups. For example if you
1487 change/disable a key-slot in LUKS, a binary backup of the partition
1488 will still have the old key-slot. To deal with this, you have to
1489 be able to change the key-slot on the backup as well, securely
1490 erase the backup or do a filesystem-level backup instead of a binary
1493 If you use dm-crypt, backup is simpler: As there is no key
1494 management, the main risk is that you cannot wipe the backup when
1495 wiping the original. However wiping the original for dm-crypt
1496 should consist of forgetting the passphrase and that you can do
1497 without actual access to the backup.
1499 In both cases, there is an additional (usually small) risk with
1500 binary backups: An attacker can see how many sectors and which
1501 ones have been changed since the backup. To prevent this, use a
1502 filesystem level backup method that encrypts the whole backup in
1503 one go, e.g. as described above with tar and GnuPG.
1505 My personal advice is to use one USB disk (low value data) or
1506 three disks (high value data) in rotating order for backups, and
1507 either use independent LUKS partitions on them, or use encrypted
1508 backup with tar and GnuPG.
1510 If you do network-backup or tape-backup, I strongly recommend to
1511 go the filesystem backup path with independent encryption, as you
1512 typically cannot reliably delete data in these scenarios,
1513 especially in a cloud setting. (Well, you can burn the tape if it
1514 is under your control...)
1517 * 6.8 What happens if I overwrite the start of a LUKS partition or
1518 damage the LUKS header or key-slots?
1520 There are two critical components for decryption: The salt values
1521 in the header itself and the key-slots. If the salt values are
1522 overwritten or changed, nothing (in the cryptographically strong
1523 sense) can be done to access the data, unless there is a backup
1524 of the LUKS header. If a key-slot is damaged, the data can still
1525 be read with a different key-slot, if there is a remaining
1526 undamaged and used key-slot. Note that in order to make a key-slot
1527 unrecoverable in a cryptographically strong sense, changing about
1528 4-6 bits in random locations of its 128kiB size is quite enough.
1531 * 6.9 What happens if I (quick) format a LUKS partition?
1533 I have not tried the different ways to do this, but very likely you
1534 will have written a new boot-sector, which in turn overwrites the
1535 LUKS header, including the salts, making your data permanently
1536 irretrievable, unless you have a LUKS header backup. You may also
1537 damage the key-slots in part or in full. See also last item.
1540 * 6.10 How do I recover the master key from a mapped LUKS container?
1542 This is typically only needed if you managed to damage your LUKS
1543 header, but the container is still mapped, i.e. "luksOpen"ed. It
1544 also helps if you have a mapped container that you forgot or do not
1545 know a passphrase for (e.g. on a long running server.)
1547 WARNING: Things go wrong, do a full backup before trying this!
1549 WARNING: This exposes the master key of the LUKS container. Note
1550 that both ways to recreate a LUKS header with the old master key
1551 described below will write the master key to disk. Unless you are
1552 sure you have securely erased it afterwards, e.g. by writing it to
1553 an encrypted partition, RAM disk or by erasing the filesystem you
1554 wrote it to by a complete overwrite, you should change the master
1555 key afterwards. Changing the master key requires a full data
1556 backup, luksFormat and then restore of the backup.
1558 First, there is a script by Milan that automates the whole
1559 process, except generating a new LUKS header with the old master
1560 key (it prints the command for that though):
1562 http://code.google.com/p/cryptsetup/source/browse/misc/luks-header-from-active
1564 You can also do this manually. Here is how:
1566 - Get the master key from the device mapper. This is done by the
1567 following command. Substitute c5 for whatever you mapped to:
1569 # dmsetup table --target crypt --showkey /dev/mapper/c5
1571 0 200704 crypt aes-cbc-essiv:sha256
1572 a1704d9715f73a1bb4db581dcacadaf405e700d591e93e2eaade13ba653d0d09
1575 The result is actually one line, wrapped here for clarity. The long
1576 hex string is the master key.
1578 - Convert the master key to a binary file representation. You can
1579 do this manually, e.g. with hexedit. You can also use the tool
1580 "xxd" from vim like this:
1582 echo "a1704d9....53d0d09" | xxd -r -p > <master-key-file>
1584 - Do a luksFormat to create a new LUKS header.
1586 NOTE: If your header is intact and you just forgot the
1587 passphrase, you can just set a new passphrase, see next
1590 Unmap the device before you do that (luksClose). Then do
1592 cryptsetup luksFormat --master-key-file=<master-key-file> <luks device>
1594 Note that if the container was created with other than the default
1595 settings of the cryptsetup version you are using, you need to give
1596 additional parameters specifying the deviations. If in doubt, try
1597 the script by Milan. It does recover the other parameters as well.
1599 Side note: This is the way the decrypt_derived script gets at the
1600 master key. It just omits the conversion and hashes the master key
1603 - If the header is intact and you just forgot the passphrase, just
1604 set a new passphrase like this:
1606 cryptsetup luksAddKey --master-key-file=<master-key-file> <luks device>
1608 You may want to disable the old one afterwards.
1611 * 6.11 What does the on-disk structure of dm-crypt look like?
1613 There is none. dm-crypt takes a block device and gives encrypted
1614 access to each of its blocks with a key derived from the passphrase
1615 given. If you use a cipher different than the default, you have to
1616 specify that as a parameter to cryptsetup too. If you want to
1617 change the password, you basically have to create a second
1618 encrypted device with the new passphrase and copy your data over.
1619 On the plus side, if you accidentally overwrite any part of a
1620 dm-crypt device, the damage will be limited to the are you
1624 * 6.12 What does the on-disk structure of LUKS look like?
1626 A LUKS partition consists of a header, followed by 8 key-slot
1627 descriptors, followed by 8 key slots, followed by the encrypted
1630 Header and key-slot descriptors fill the first 592 bytes. The
1631 key-slot size depends on the creation parameters, namely on the
1632 number of anti-forensic stripes, key material offset and master
1635 With the default parameters, each key-slot is a bit less than
1636 128kiB in size. Due to sector alignment of the key-slot start,
1637 that means the key block 0 is at offset 0x1000-0x20400, key
1638 block 1 at offset 0x21000-0x40400, and key block 7 at offset
1639 0xc1000-0xe0400. The space to the next full sector address is
1640 padded with zeros. Never used key-slots are filled with what the
1641 disk originally contained there, a key-slot removed with
1642 "luksRemoveKey" or "luksKillSlot" gets filled with 0xff. Due to
1643 2MiB default alignment, start of the data area for cryptsetup 1.3
1644 and later is at 2MiB, i.e. at 0x200000. For older versions, it is
1645 at 0x101000, i.e. at 1'052'672 bytes, i.e. at 1MiB + 4096 bytes
1646 from the start of the partition. Incidentally, "luksHeaderBackup"
1647 for a LUKS container created with default parameters dumps exactly
1648 the first 2MiB (or 1'052'672 bytes for headers created with
1649 cryptsetup versions < 1.3) to file and "luksHeaderRestore" restores
1652 For non-default parameters, you have to figure out placement
1653 yourself. "luksDump" helps. See also next item. For the most common
1654 non-default settings, namely aes-xts-plain with 512 bit key, the
1655 offsets are: 1st keyslot 0x1000-0x3f800, 2nd keyslot
1656 0x40000-0x7e000, 3rd keyslot 0x7e000-0xbd800, ..., and start of
1657 bulk data at 0x200000.
1659 The exact specification of the format is here:
1660 http://code.google.com/p/cryptsetup/wiki/Specification
1663 * 6.13 What is the smallest possible LUKS container?
1665 Note: From cryptsetup 1.3 onwards, alignment is set to 1MB. With
1666 modern Linux partitioning tools that also align to 1MB, this will
1667 result in alignment to 2k sectors and typical Flash/SSD sectors,
1668 which is highly desirable for a number of reasons. Changing the
1669 alignment is not recommended.
1671 That said, with default parameters, the data area starts at
1672 exactly 2MB offset (at 0x101000 for cryptsetup versions before
1673 1.3). The smallest data area you can have is one sector of 512
1674 bytes. Data areas of 0 bytes can be created, but fail on mapping.
1676 While you cannot put a filesystem into something this small, it may
1677 still be used to contain, for example, key. Note that with current
1678 formatting tools, a partition for a container this size will be
1679 3MiB anyways. If you put the LUKS container into a file (via
1680 losetup and a loopback device), the file needs to be 2097664 bytes
1681 in size, i.e. 2MiB + 512B.
1683 There two ways to influence the start of the data area are key-size
1686 For alignment, you can go down to 1 on the parameter. This will
1687 still leave you with a data-area starting at 0x101000, i.e.
1688 1MiB+4096B (default parameters) as alignment will be rounded up to
1689 the next multiple of 8 (i.e. 4096 bytes) If in doubt, do a dry-run
1690 on a larger file and dump the LUKS header to get actual
1693 For key-size, you can use 128 bit (e.g. AES-128 with CBC), 256 bit
1694 (e.g. AES-256 with CBC) or 512 bit (e.g. AES-256 with XTS mode).
1695 You can do 64 bit (e.g. blowfish-64 with CBC), but anything below
1696 128 bit has to be considered insecure today.
1698 Example 1 - AES 128 bit with CBC:
1700 cryptsetup luksFormat -s 128 --align-payload=8 <device>
1702 This results in a data offset of 0x81000, i.e. 516KiB or 528384
1703 bytes. Add one 512 byte sector and the smallest LUKS container size
1704 with these parameters is 516KiB + 512B or 528896 bytes.
1706 Example 2 - Blowfish 64 bit with CBC (WARNING: insecure):
1708 cryptsetup luksFormat -c blowfish -s 64 --align-payload=8 /dev/loop0
1710 This results in a data offset of 0x41000, i.e. 260kiB or 266240
1711 bytes, with a minimal LUKS container size of 260kiB + 512B or
1715 * 6.14 I think this is overly complicated. Is there an alternative?
1717 Not really. Encryption comes at a price. You can use plain
1718 dm-crypt to simplify things a bit. It does not allow multiple
1719 passphrases, but on the plus side, it has zero on disk description
1720 and if you overwrite some part of a plain dm-crypt partition,
1721 exactly the overwritten parts are lost (rounded up to sector
1725 * 6.15 Can I clone a LUKS container?
1727 You can, but it breaks security, because the cloned container has
1728 the same header and hence the same master key. You cannot change
1729 the master key on a LUKS container, even if you change the
1730 passphrase(s), the master key stays the same. That means whoever
1731 has access to one of the clones can decrypt them all, completely
1732 bypassing the passphrases.
1734 The right way to do this is to first luksFormat the target
1735 container, then to clone the contents of the source container, with
1736 both containers mapped, i.e. decrypted. You can clone the decrypted
1737 contents of a LUKS container in binary mode, although you may run
1738 into secondary issues with GUIDs in filesystems, partition tables,
1739 RAID-components and the like. These are just the normal problems
1740 binary cloning causes.
1742 Note that if you need to ship (e.g.) cloned LUKS containers with a
1743 default passphrase, that is fine as long as each container was
1744 individually created (and hence has its own master key). In this
1745 case, changing the default passphrase will make it secure again.
1748 7. Interoperability with other Disk Encryption Tools
1751 * 7.1 What is this section about?
1753 Cryptsetup for plain dm-crypt can be used to access a number of
1754 on-disk formats created by tools like loop-aes patched into
1755 losetup. This sometimes works and sometimes does not. This
1756 section collects insights into what works, what does not and where
1757 more information is required.
1759 Additional information may be found in the mailing-list archives,
1760 mentioned at the start of this FAQ document. If you have a
1761 solution working that is not yet documented here and think a wider
1762 audience may be interested, please email the FAQ maintainer.
1765 * 7.2 loop-aes: General observations.
1767 One problem is that there are different versions of losetup around.
1768 loop-aes is a patch for losetup. Possible problems and deviations
1769 from cryptsetup option syntax include:
1771 - Offsets specified in bytes (cryptsetup: 512 byte sectors)
1773 - The need to specify an IV offset
1775 - Encryption mode needs specifying (e.g. "-c twofish-cbc-plain")
1777 - Key size needs specifying (e.g. "-s 128" for 128 bit keys)
1779 - Passphrase hash algorithm needs specifying
1781 Also note that because plain dm-crypt and loop-aes format does not
1782 have metadata, and while the loopAES extension for cryptsetup tries
1783 autodetection (see command loopaesOpen), it may not always work.
1784 If you still have the old set-up, using a verbosity option (-v)
1785 on mapping with the old tool or having a look into the system logs
1786 after setup could give you the information you need. Below, there
1787 are also some things that worked for somebody.
1790 * 7.3 loop-aes patched into losetup on Debian 5.x, kernel 2.6.32
1792 In this case, the main problem seems to be that this variant of
1793 losetup takes the offset (-o option) in bytes, while cryptsetup
1794 takes it in sectors of 512 bytes each. Example: The losetup command
1796 losetup -e twofish -o 2560 /dev/loop0 /dev/sdb1
1797 mount /dev/loop0 mount-point
1801 cryptsetup create -c twofish -o 5 --skip 5 e1 /dev/sdb1
1802 mount /dev/mapper/e1 mount-point
1805 * 7.4 loop-aes with 160 bit key
1807 This seems to be sometimes used with twofish and blowfish and
1808 represents a 160 bit ripemed160 hash output padded to 196 bit key
1809 length. It seems the corresponding options for cryptsetup are
1811 --cipher twofish-cbc-null -s 192 -h ripemd160:20
1814 * 7.5 loop-aes v1 format OpenSUSE
1816 Apparently this is done by older OpenSUSE distros and stopped
1817 working from OpenSUSE 12.1 to 12.2. One user had success with the
1820 cryptsetup create <target> <device> -c aes -s 128 -h sha256
1823 * 7.6 Kernel encrypted loop device (cryptoloop)
1825 There are a number of different losetup implementations for using
1826 encrypted loop devices so getting this to work may need a bit of
1829 NOTE: Do NOT use this for new containers! Some of the existing
1830 implementations are insecure and future support is uncertain.
1832 Example for a compatible mapping:
1834 losetup -e twofish -N /dev/loop0 /image.img
1838 cryptsetup create image_plain /image.img -c twofish-cbc-plain -H plain
1840 with the mapping being done to /dev/mapper/image_plain instead of
1845 Cipher, mode and pasword hash (or no hash):
1847 -e cipher [-N] => -c cipher-cbc-plain -H plain [-s 256]
1848 -e cipher => -c cipher-cbc-plain -H ripemd160 [-s 256]
1850 Key size and offsets (losetup: bytes, cryptsetuop: sectors of 512
1854 -o 2560 => -o 5 -p 5 # 2560/512 = 5
1856 There is no replacement for --pass-fd, it has to be emulated using
1857 keyfiles, see the cryptsetup man-page.
1860 8. Issues with Specific Versions of cryptsetup
1863 * 8.1 When using the create command for plain dm-crypt with
1864 cryptsetup 1.1.x, the mapping is incompatible and my data is not
1867 With cryptsetup 1.1.x, the distro maintainer can define different
1868 default encryption modes for LUKS and plain devices. You can check
1869 these compiled-in defaults using "cryptsetup --help". Moreover, the
1870 plain device default changed because the old IV mode was
1871 vulnerable to a watermarking attack.
1873 If you are using a plain device and you need a compatible mode, just
1874 specify cipher, key size and hash algorithm explicitly. For
1875 compatibility with cryptsetup 1.0.x defaults, simple use the
1878 cryptsetup create -c aes-cbc-plain -s 256 -h ripemd160 <name> <dev>
1880 LUKS stores cipher and mode in the metadata on disk, avoiding this
1884 * 8.2 cryptsetup on SLED 10 has problems...
1886 SLED 10 is missing an essential kernel patch for dm-crypt, which
1887 is broken in its kernel as a result. There may be a very old
1888 version of cryptsetup (1.0.x) provided by SLED, which should also
1889 not be used anymore as well. My advice would be to drop SLED 10.
1892 9. References and Further Reading
1895 * Purpose of this Section
1897 The purpose of this section is to collect references to all
1898 materials that do not fit the FAQ but are relevant in some fashion.
1899 This can be core topics like the LUKS spec or disk encryption, but
1900 it can also be more tangential, like secure storage management or
1901 cryptography used in LUKS. It should still have relevance to
1902 cryptsetup and its applications.
1904 If you wan to see something added here, send email to the
1905 maintainer (or the cryptsetup mailing list) giving an URL, a
1906 description (1-3 lines preferred) and a section to put it in. You
1907 can also propose new sections.
1909 At this time I would like to limit the references to things that
1910 are available on the web.
1915 - LUKS on-disk format spec:
1916 http://code.google.com/p/cryptsetup/wiki/Specification
1921 - Some code examples are in the source package under docs/examples
1924 * Brute-forcing passphrases
1927 http://news.electricalchemy.net/2009/10/password-cracking-in-cloud-part-5.html
1930 http://it.slashdot.org/story/12/12/05/0623215/new-25-gpu-monster-devours-strong-passwords-in-minutes
1936 * SSD and Flash Disk Related
1942 * Attacks Against Disk Encryption
1945 * Risk Management as Relevant for Disk Encryption
1953 A. Contributors In no particular order: