Sections 1. General Questions 2. Setup 3. Common Problems 4. Troubleshooting 5. Security Aspects 6. Backup and Data Recovery 7. Interoperability with other Disk Encryption Tools 8. Issues with Specific Versions of cryptsetup A. Contributors 1. General Questions * What is this? This is the FAQ (Frequently Asked Questions) for cryptsetup. It covers Linux disk encryption with plain dm-crypt (one passphrase, no management, no metadata on disk) and LUKS (multiple user keys with one master key, anti-forensic features, metadata block at start of device, ...). The latest version of this FAQ should usually be available at http://code.google.com/p/cryptsetup/wiki/FrequentlyAskedQuestions * WARNINGS ATTENTION: If you are going to read just one thing, make it the section on Backup and Data Recovery. By far the most questions on the cryptsetup mailing list are from people that managed to damage the start of their LUKS partitions, i.e. the LUKS header. In most cases, there is nothing that can be done to help these poor souls recover their data. Make sure you understand the problem and limitations imposed by the LUKS security model BEFORE you face such a disaster! In particular, make sure you have a current header backup before doing any potentially dangerous operations. BACKUP: Yes, encrypted disks die, just as normal ones do. A full backup is mandatory, see Section "6. Backup and Data Recovery" on options for doing encrypted backup. DISTRIBUTION INSTALLERS: Some distribution installers offer to create LUKS containers in a way that can be mistaken as activation of an existing container. Creating a new LUKS container on top of an existing one leads to permanent, complete and irreversible data loss. It is strongly recommended to only use distribution installers after a complete backup of all LUKS containers has been made. NO WARNING ON NON-INERACTIVE FORMAT: If you feed cryptsetup from STDIN (e.g. via GPG) on LUKS format, it does not give you the warning that you are about to format (and e.g. will lose any pre-existing LUKS container on the target), as it assumes it is used form a script. In this scenario, the responsibility for warning the user and possibly checking for an existing LUKS header goes over to the script. This is a more general form of the previous item. LUKS PASSPHRASE IS NOT THE MASTER KEY: The LUKS passphrase is not used in deriving the master key. It is used in decrypting a master key that is randomly selected on header creation. This means that if you create a new LUKS header on top of an old one with exactly the same parameters and exactly the same passphrase as the old one, it will still have a different master key and your data will be permanently lost. PASSPHRASE CHARACTER SET: Some people have had difficulties with this when upgrading distributions. It is highly advisable to only use the 94 printable characters from the first 128 characters of the ASCII table, as they will always have the same binary representation. Other characters may have different encoding depending on system configuration and your passphrase will not work with a different encoding. A table of the standardized first 128 ASCII caracters can, e.g. be found on http://en.wikipedia.org/wiki/ASCII * System Specific warnings - Ubuntu as of 4/2011: It seems the installer offers to create LUKS partitions in a way that several people mistook for an offer to activate their existing LUKS partition. The installer gives no or an inadequate warning and will destroy your old LUKS header, causing permanent data loss. See also the section on Backup and Data Recovery. This issue has been acknowledged by the Ubuntu dev team, see here: http://launchpad.net/bugs/420080 * Who wrote this? Current FAQ maintainer is Arno Wagner <arno@xxxxxxxxxxx>. Other contributors are listed at the end. If you want to contribute, send your article, including a descriptive headline, to the maintainer, or the dm-crypt mailing list with something like "FAQ ..." in the subject. You can also send more raw information and have me write the section. Please note that by contributing to this FAQ, you accept the license described below. This work is under the "Attribution-Share Alike 3.0 Unported" license, which means distribution is unlimited, you may create derived works, but attributions to original authors and this license statement must be retained and the derived work must be under the same license. See http://creativecommons.org/licenses/by-sa/3.0/ for more details of the license. Side note: I did text license research some time ago and I think this license is best suited for the purpose at hand and creates the least problems. * Where is the project website? There is the project website at http://code.google.com/p/cryptsetup/ Please do not post questions there, nobody will read them. Use the mailing-list instead. * Is there a mailing-list? Instructions on how to subscribe to the mailing-list are at on the project website. People are generally helpful and friendly on the list. The question of how to unsubscribe from the list does crop up sometimes. For this you need your list management URL, which is sent to you initially and once at the start of each month. Go to the URL mentioned in the email and select "unsubscribe". This page also allows you to request a password reminder. Alternatively, you can send an Email to dm-crypt-request@xxxxxxxx with just the word "help" in the subject or message body. Make sure to send it from your list address. The mailing list archive is here: http://dir.gmane.org/gmane.linux.kernel.device-mapper.dm-crypt 2. Setup * What is the difference between "plain" and LUKS format? Plain format is just that: It has no metadata on disk, reads all paramters from the commandline (or the defaults), derives a master-key from the passphrase and then uses that to de-/encrypt the sectors of the device, with a direct 1:1 mapping between encrypted and decrypted sectors. Primary advantage is high resilience to damage, as one damaged encrypted sector results in exactly one damaged decrypted sector. Also, it is not readily apparent that there even is encrypted data on the device, as an overwrite with crypto-grade randomness (e.g. from /dev/urandom) looks exactly the same on disk. Side-note: That has limited value against the authorities. In civilized countries, they cannot force you to give up a crypto-key anyways. In the US, the UK and dictatorships around the world, they can force you to give up the keys (using imprisonment or worse to pressure you), and in the worst case, they only need a nebulous "suspicion" about the presence of encrypted data. My advice is to either be ready to give up the keys or to not have encrypted data when traveling to those countries, especially when crossing the borders. Disadvantages are that you do not have all the nice features that the LUKS metadata offers, like multiple passphrases that can be changed, the cipher being stored in the metadata, anti-forensic properties like key-slot diffusion and salts, etc.. LUKS format uses a metadata header and 8 key-slot areas that are being placed ath the begining of the disk, see below under "What does the LUKS on-disk format looks like?". The passphrases are used to decryt a single master key that is stored in the anti-forensic stripes. Advantages are a higher usability, automatic configuration of non-default crypto parameters, defenses against low-entropy passphrases like salting and iterated PBKDF2 passphrase hashing, the ability to change passhrases, and others. Disadvantages are that it is readily obvious there is encrypted data on disk (but see side note above) and that damage to the header or key-slots usually results in permanent data-loss. See below under "6. Backup and Data Recovery" on how to reduce that risk. Also the sector numbers get shifted by the length of the header and key-slots and there is a loss of that size in capacity (1MB+4096B for defaults and 2MB for the most commonly used non-default XTS mode). * Can I encrypt an already existing, non-empty partition to use LUKS? There is no converter, and it is not really needed. The way to do this is to make a backup of the device in question, securely wipe the device (as LUKS device initialization does not clear away old data), do a luksFormat, optionally overwrite the encrypted device, create a new filesystem and restore your backup on the now encrypted device. Also refer to sections "Security Aspects" and "Backup and Data Recovery". For backup, plain GNU tar works well and backs up anything likely to be in a filesystem. * How do I use LUKS with a loop-device? This can be very handy for experiments. Setup is just the same as with any block device. If you want, for example, to use a 100MiB file as LUKS container, do something like this: head -c 100M /dev/zero > luksfile # create empty file losetup /dev/loop0 luksfile # map luksfile to /dev/loop0 cryptsetup luksFormat /dev/loop0 # create LUKS on loop device Afterwards just use /dev/loop0 as a you would use a LUKS partition. To unmap the file when done, use "losetup -d /dev/loop0". * When I add a new key-slot to LUKS, it asks for a passphrase but then complains about there not being a key-slot with that passphrase? That is as intended. You are asked a passphrase of an existing key-slot first, before you can enter the passphrase for the new key-slot. Otherwise you could break the encryption by just adding a new key-slot. This way, you have to know the passphrase of one of the already configured key-slots in order to be able to configure a new key-slot. * Encrytion on top of RAID or the other way round? Unless you have special needs, place encryption between RAID and filesystem, i.e. encryption on top of RAID. You can do it the other way round, but you have to be aware that you then need to give the pasphrase for each individual disk and RAID autotetection will not work anymore. Therefore it is better to encrypt the RAID device, e.g. /dev/dm0 . * How do I read a dm-crypt key from file? Note that the file will still be hashed first, just like keyboard input. Use the --key-file option, like this: cryptsetup create --key-file keyfile e1 /dev/loop0 * How do I read a LUKS slot key from file? What you really do here is to read a passphrase from file, just as you would with manual entry of a passphrase for a key-slot. You can add a new passphrase to a free key-slot, set the passphrase of an specific key-slot or put an already configured passphrase into a file. In the last case make sure no trailing newline (0x0a) is contained in the key file, or the passphrase will not work because the whole file is used as input. To add a new passphrase to a free key slot from file, use something like this: cryptsetup luksAddKey /dev/loop0 keyfile To add a new passphrase to a specific key-slot, use something like this: cryptsetup luksAddKey --key-slot 7 /dev/loop0 keyfile To supply a key from file to any LUKS command, use the --key-file option, e.g. like this: cryptsetup luksOpen --key-file keyfile /dev/loop0 e1 * How do I read the LUKS master key from file? The question you should ask yourself first is why you would want to do this. The only legitimate reason I can think of is if you want to have two LUKS devices with the same master key. Even then, I think it would be preferable to just use key-slots with the same passphrase, or to use plain dm-crypt instead. If you really have a good reason, please tell me. If I am convinced, I will add how to do this here. * What are the security requirements for a key read from file? A file-stored key or passphrase has the same security requirements as one entered interactively, however you can use random bytes and thereby use bytes you cannot type on the keyboard. You can use any file you like as key file, for example a plain text file with a human readable passphrase. To generate a file with random bytes, use something like this: head -c 256 /dev/random > keyfile * If I map a journaled file system using dm-crypt/LUKS, does it still provide its usual transactional guarantees? As far as I know it does (but I may be wrong), but please note that these "guarantees" are far weaker than they appear to be. For example, you may not get a hard flush to disk surface even on a call to fsync. In addition, the HDD itself may do independent write reordering. Some other things can go wrong as well. The filesystem developers are aware of these problems and typically can make it work anyways. That said, dm-crypt/LUKS should not make things worse. Personally, I have several instances of ext3 on dm-crypt and have not noticed any specific problems. Update: I did run into frequent small freezes (1-2 sec) when putting a vmware image on ext3 over dm-crypt. This does indicate that the transactional guarantees are in place, but at a cost. When I went back to ext2, the problem went away. This also seems to have gotten better with kernel 2.6.36 and the reworking of filesystem flush locking. Kernel 2.6.38 is expected to have more improvements here. * Can I use LUKS or cryptsetup with a more secure (external) medium for key storage, e.g. TPM or a smartcard? Yes, see the answers on using a file-supplied key. You do have to write the glue-logic yourself though. Basically you can have cryptsetup read the key from STDIN and write it there with your own tool that in turn gets the key from the more secure key storage. * Can I resize a dm-crypt or LUKS partition? Yes, you can, as neither dm-crypt nor LUKS stores partition size. Whether you should is a different question. Personally I recommend backup, recreation of the encrypted partition with new size, recreation of the filesystem and restore. This gets around the tricky business of resizing the filesystem. Resizing a dm-crypt or LUKS container does not resize the filesystem in it. The backup is really non-optional here, as a lot can go wrong, resulting in partial or complete data loss. Using something like gparted to resize an encrypted partition is slow, but typicaly works. This will not change the size of the filesystem hidden under the encryption though. You also need to be aware of size-based limitations. The one currently relevant is that aes-xts-plain should not be used for encrypted container sizes larger than 2TiB. Use aes-xts-plain64 for that. 3. Common Problems * My dm-crypt/LUKS mapping does not work! What general steps are there to investigate the problem? If you get a specific error message, investigate what it claims first. If not, you may want to check the following things. - Check that "/dev", including "/dev/mapper/control" is there. If it is missing, you may have a problem with the "/dev" tree itself or you may have broken udev rules. - Check that you have the device mapper and the crypt target in your kernel. The output of "dmsetup targets" should list a "crypt" target. If it is not there or the command fails, add device mapper and crypt-target to the kernel. - Check that the hash-functions and ciphers you want to use are in the kernel. The output of "cat /proc/crypto" needs to list them. * My dm-crypt mapping suddenly stopped when upgrading cryptsetup. The default cipher, hash or mode may have changed (the mode changed from 1.0.x to 1.1.x). See under "Issues With Specific Versions of cryptsetup". * When I call cryptsetup from cron/CGI, I get errors about unknown features? If you get errors about unknown parameters or the like that are not present when cryptsetup is called from the shell, make sure you have no older version of cryptsetup on your system that then gets called by cron/CGI. For example some distributions install cryptsetup into /usr/sbin, while a manual install could go to /usr/local/sbin. As a debugging aid, call "cryptsetup --version" from cron/CGI or the non-shell mechanism to be sure the right version gets called. * Unlocking a LUKS device takes very long. Why? The iteration time for a key-slot (see Section 5 for an explanation what iteration does) is calculated when setting a passphrase. By default it is 1 second on the machine where the passphrase is set. If you set a passphrase on a fast machine and then unlock it on a slow machine, the unlocking time can be much longer. Also take into account that up to 8 key-slots have to be tried in order to find the right one. If this is problem, you can add another key-slot using the slow machine with the same passphrase and then remove the old key-slot. The new key-slot will have an iteration count adjusted to 1 second on the slow machine. Use luksKeyAdd and then luksKillSlot or luksRemoveKey. However, this operation will not change volume key iteration count (MK iterations in output of "cryptsetup luksDump"). In order to change that, you will have to backup the data in the LUKS container (i.e. your encrypted data), luksFormat on the slow machine and restore the data. Note that in the original LUKS specification this value was fixed to 10, but it is now derived from the PBKDF2 benchmark as well and set to iterations in 0.125 sec or 1000, whichever is larger. Also note that MK iterations are not very security relevant. But as each key-slot already takes 1 second, spending the additional 0.125 seconds really does not matter. * "blkid" sees a LUKS UUID and an ext2/swap UUID on the same device. What is wrong? Some old versions of cryptsetup have a bug where the header does not get completely wiped during LUKS format and an older ext2/swap signature remains on the device. This confuses blkid. Fix: Wipe the unused header areas by doing a backup and restore of the header with cryptsetup 1.1.x: cryptsetup luksHeaderBackup --header-backup-file <file> <device> cryptsetup luksHeaderRestore --header-backup-file <file> <device> * cryptsetup segfaults on Gentoo amd64 hardened ... There seems to be some inteference between the hardening and and the way cryptsetup benchmarks PBKDF2. The solution to this is currently not quite clear for an encrypted root filesystem. For other uses, you can apparently specify USE="dynamic" as compile flag, see http://bugs.gentoo.org/show_bug.cgi?id=283470 4. Troubleshooting * I get the error "LUKS keyslot x is invalid." What does that mean? This means that the given keyslot has an offset that points outside the valid keyslot area. Typically, the reason is a corrupted LUKS header because something was written to the start of the device the LUKS contaner is on. Refer to Section "Backup and Data Recovery" and ask on the mailing list if you have trouble diagnosing and (if still possible) repairing this. * Can a bad RAM module cause problems? LUKS and dm-crypt can give the RAM quite a workout, especially when combined with software RAID. In particular the combination RAID5 + LUKS + XFS seems to uncover RAM problems that never caused obvious problems before. Symptoms vary, but often the problem manifest itself when copying large amounts of data, typically several times larger than your main memory. Side note: One thing you should always do on large data copy/movements is to run a verify, for example with the "-d" option of "tar" or by doing a set of MD5 checksums on the source or target with find . -type f -exec md5sum \{\} \; > checksum-file and then a "md5sum -c checksum-file" on the other side. If you get mismatches here, RAM is the primary suspect. A lesser suspect is an overclocked CPU. I have found countless hardware problems in verify runs after copying or making backups. Bit errors are much more common than most people think. Some RAM issues are even worse and corrupt structures in one of the layers. This typically results in lockups, CPU state dumps in the system logs, kernel panic or other things. It is quite possible to have the problem with an encrypted device, but not with an otherwise the same unencrypted device. The reason for that is that encryption has an error amplification property: You flip one bit in an encrypted data block, and the decrypted version has half of its bits flipped. This is an important security property for modern ciphers. With the usual modes in cryptsetup (CBC, ESSIV, XTS), you get up to a completely changed 512 byte block per bit error. A corrupt block causes a lot more havoc than the occasionally flipped single bit and can result in various obscure errors. Note, that a verify run on copying between encrypted or unencrypted devices will reliably detect corruption, even when the copying itself did not report any problems. If you find defect RAM, assume all backups and copied data to be suspect, unless you did a verify. * How do I test RAM? First you should know that overclocking often makes memory problems worse. So if you overclock (which I strongly recommend against in a system holding data that has some worth), run the tests with the overclocking active. There are two good options. One is Memtest86+ and the other is "memtester" by Charles Cazabon. Memtest86+ requires a reboot and then takes over the machine, while memtester runs from a root-shell. Both use different testing methods and I have found problems fast with each one that the other needed long to find. I recommend running the following procedure until the first error is found: - Run Memtest86+ for one cycle - Run memterster for one cycle (shut down as many other applications as possible) - Run Memtest86+ for 24h or more - Run memtester for 24h or more If all that does not produce error messages, your RAM may be sound, but I have had one weak bit that Memtest86+ needed around 60 hours to find. If you can reproduce the original problem reliably, a good additional test may be to remove half of the RAM (if you have more than one module) and try whether the problem is still there and if so, try with the other half. If you just have one module, get a different one and try with that. If you do overclocking, reduce the settings to the most conservative ones available and try with that. 5. Security Aspects * Is LUKS insecure? Everybody can see I have encrypted data! In practice it does not really matter. In most civilized countries you can just refuse to hand over the keys, no harm done. In some countries they can force you to hand over the keys, if they suspect encryption. However the suspicion is enough, they do not have to prove anything. This is for practical reasons, as even the presence of a header (like the LUKS header) is not enough to prove that you have any keys. It might have been an experiment, for example. Or it was used as encrypted swap with a key from /dev/random. So they make you prove you do not have encrypted data. Of course that is just as impossible as the other way round. This means that if you have a large set of random-looking data, they can already lock you up. Hidden containers (encryption hidden within encryption), as possible with Truecrypt, do not help either. They will just assume the hidden container is there and unless you hand over the key, you will stay locked up. Don't have a hidden container? Though luck. Anybody could claim that. Still, if you are concerned about the LUKS header, use plain dm-crypt with a good passphrase. See also Section 2, "What is the difference between "plain" and LUKS format?" * Should I initialize (overwrite) a new LUKS/dm-crypt partition? If you just create a filesystem on it, most of the old data will still be there. If the old data is sensitive, you should overwrite it before encrypting. In any case, not initializing will leave the old data there until the specific sector gets written. That may enable an attacker to determine how much and where on the partition data was written. If you think this is a risk, you can prevent this by overwriting the encrypted device (here assumed to be named "e1") with zeros like this: dd_rescue -w /dev/zero /dev/mapper/e1 or alternatively with one of the following more standard commands: cat /dev/zero > /dev/mapper/e1 dd if=/dev/zero of=/dev/mapper/e1 * How do I securely erase a LUKS (or other) partition? For LUKS, if you are in a desperate hurry, overwrite the LUKS header and key-slot area. This means overwriting the first (keyslots x stripes x keysize) + offset bytes. For the default parameters, this is the 1'052'672 bytes, i.e. 1MiB + 4096 of the LUKS partition. For 512 bit key length (e.g. for aes-xts-plain with 512 bit key) this is 2MiB. (The diferent offset stems from differences in the sector alignment of the key-slots.) If in doubt, just be generous and overwrite the first 10MB or so, it will likely still be fast enough. A single overwrite with zeros should be enough. If you anticipate being in a desperate hurry, prepare the command beforehand. Example with /dev/sde1 as the LUKS partition and default parameters: head -c 1052672 /dev/zero > /dev/sde1; sync A LUKS header backup or full backup will still grant access to most or all data, so make sure that an attacker does not have access to backups or destroy them as well. If you have time, overwrite the whole LUKS partition with a single pass of zeros. This is enough for current HDDs. For SSDs or FLASH (USB sticks) you may want to overwrite the whole drive several times to be sure data is not retained by wear leveling. This is possibly still insecure as SSD technology is not fully understood in this regard. Still, due to the anti-forensic properties of the LUKS key-slots, a single overwrite of an SSD or FLASH drive could be enough. If in doubt, use physical destruction in addition. Here is a link to some current reseach results on erasing SSDs and FLASH drives: http://www.usenix.org/events/fast11/tech/full_papers/Wei.pdf Keep in mind to also erase all backups. Example for a zero-overwrite erase of partition sde1 done with dd_rescue: dd_rescue -w /dev/zero /dev/sde1 * How do I securely erase a backup of a LUKS partition or header? That depends on the medium it is stored on. For HDD and SSD, use overwrite with zeros. For an SSD or FLASH drive (USB stick), you may want to overwrite the complete SSD several times and use physical destruction in addition, see last item. For re-writable CD/DVD, a single overwrite should also be enough, due to the anti-forensic properties of the LUKS keyslots. For write-once media, use physical destruction. For low security requirements, just cut the CD/DVD into several parts. For high security needs, shred or burn the medium. If your backup is on magnetic tape, I advise physical destruction by shredding or burning, after overwriting . The problem with magnetic tape is that it has a higher dynamic range than HDDs and older data may well be recoverable after overwrites. Also write-head alignment issues can lead to data not actually being deleted at all during overwrites. * What about backup? Does it compromise security? That depends. See next section. * Why is all my data permanently gone if I overwrite the LUKS header? Overwriting the LUKS header in part or in full is the most common reason why access to LUKS containers is lost permanently. Overwriting can be done in a number of fashions, like creating a new filesystem on the raw LUKS partition, making the raw partition part of a raid array and just writing to the raw partition. The LUKS header contains a 256 bit "salt" value and without that no decryption is possible. While the salt is not secret, it is key-grade material and cannot be reconstructed. This is a cryptographically strong "cannot". From observations on the cryptsetup mailing-list, people typically go though the usual stages of grief (Denial, Anger, Bargaining, Depression, Acceptance) when this happens to them. Observed times vary between 1 day and 2 weeks to complete the cycle. Seeking help on the mailing-list is fine. Even if we usually cannot help with getting back your data, most people found the feedback comforting. If your header does not contain an intact salt, best go directly to the last stage ("Acceptance") and think about what to do now. There is one exception that I know of: If your LUKS container is still open, then it may be possible to extract the master key from the running system. See Item "How do I recover the master key from a mapped LUKS container?" in Section "Backup and Data Recovery". * What is a "salt"? A salt is a random key-grade value added to the passphrase before it is processed. It is not kept secret. The reason for using salts is as follows: If an attacker wants to crack the password for a single LUKS container, then every possible passphrase has to be tried. Typically an attacker will not try every binary value, but will try words and sentences from a dictionary. If an attacker wants to attack several LUKS containers with the same dictionary, then a different approach makes sense: Compute the resulting slot-key for each dictionary element and store it on disk. Then the test for each entry is just the slow unlocking with the slot key (say 0.00001 sec) instead of calculating the slot-key first (1 sec). For a single attack, this does not help. But if you have more than one container to attack, this helps tremendously, also because you can prepare your table before you even have the container to attack! The calculation is also very simple to parallelize. You could, for example, use the night-time unused CPU power of your desktop PCs for this. This is where the salt comes in. If the salt is combined with the passphrase (in the simplest form, just appended to it), you suddenly need a separate table for each salt value. With a reasonably-sized salt value (256 bit, e.g.) this is quite infeasible. * Is LUKS secure with a low-entropy (bad) passphrase? Note: You should only use the 94 printable characters from 7 bit ASCII code to prevent your passphrase from failing when the character encoding changes, e.g. because of a system upgrade, see also the note at the very start of this FAQ under "WARNINGS". This needs a bit of theory. The quality of your passphrase is directly related to its entropy (information theoretic, not thermodynamic). The entropy says how many bits of "uncertainty" or "randomness" are in you passphrase. In other words, that is how difficult guessing the passphrase is. Example: A random English sentence has about 1 bit of entropy per character. A random lowercase (or uppercase) character has about 4.7 bit of entropy. Now, if n is the number of bits of entropy in your passphrase and t is the time it takes to process a passphrase in order to open the LUKS container, then an attacker has to spend at maximum attack_time_max = 2^n * t time for a successful attack and on average half that. There is no way getting around that relationship. However, there is one thing that does help, namely increasing t, the time it takes to use a passphrase, see next FAQ item. Still, if you want good security, a high-entropy passphrase is the only option. For example, a low-entropy passphrase can never be considered secure against a TLA-level (Three Letter Agency level, i.e. government-level) attacker, no matter what tricks are used in the key-derivation function. Use at least 64 bits for secret stuff. That is 64 characters of English text (but only if randomly chosen) or a combination of 12 truly random letters and digits. For passphrase generation, do not use lines from very well-known texts (religious texts, Harry potter, etc.) as they are to easy to guess. For example, the total Harry Potter has about 1'500'000 words (my estimation). Trying every 64 character sequence starting and ending at a word boundary would take only something like 20 days on a single CPU and is entirely feasible. To put that into perspective, using a number of Amazon EC2 High-CPU Extra Large instances (each gives about 8 real cores), this test costs currently about 50USD/EUR, but can be made to run arbitrarily fast. On the other hand, choosing 1.5 lines from, say, the Wheel of Time is in itself not more secure, but the book selection adds quite a bit of entropy. (Now that I have mentioned it here, don't use tWoT either!) If you add 2 or 3 typos or switch some words around, then this is good passphrase material. * What is "iteration count" and why is decreasing it a bad idea? Iteration count is the number of PBKDF2 iterations a passphrase is put through before it is used to unlock a key-slot. Iterations are done with the explicit purpose to increase the time that it takes to unlock a key-slot. This provides some protection against use of low-entropy passphrases. The idea is that an attacker has to try all possible passphrases. Even if the attacker knows the passphrase is low-entropy (see last item), it is possible to make each individual try take longer. The way to do this is to repeatedly hash the passphrase for a certain time. The attacker then has to spend the same time (given the same computing power) as the user per try. With LUKS, the default is 1 second of PBKDF2 hashing. Example 1: Lets assume we have a really bad passphrase (e.g. a girlfriends name) with 10 bits of entropy. With the same CPU, an attacker would need to spend around 500 seconds on average to break that passphrase. Without iteration, it would be more like 0.0001 seconds on a modern CPU. Example 2: The user did a bit better and has 32 chars of English text. That would be about 32 bits of entropy. With 1 second iteration, that means an attacker on the same CPU needs around 136 years. That is pretty impressive for such a weak passphrase. Without the iterations, it would be more like 50 days on a modern CPU, and possibly far less. In addition, the attacker can both parallelize and use special hardware like GPUs or FPGAs to speed up the attack. The attack can also happen quite some time after the luksFormat operation and CPUs can have become faster and cheaper. For that reason you want a bit of extra security. Anyways, in Example 1 your are screwed. In example 2, not necessarily. Even if the attack is faster, it still has a certain cost associated with it, say 10000 EUR/USD with iteration and 1 EUR/USD without iteration. The first can be prohibitively expensive, while the second is something you try even without solid proof that the decryption will yield something useful. The numbers above are mostly made up, but show the idea. Of course the best thing is to have a high-entropy passphrase. Would a 100 sec iteration time be even better? Yes and no. Cryptographically it would be a lot better, namely 100 times better. However, usability is a very important factor for security technology and one that gets overlooked surprisingly often. For LUKS, if you have to wait 2 minutes to unlock the LUKS container, most people will not bother and use less secure storage instead. It is better to have less protection against low-entropy passphrases and people actually use LUKS, than having them do without encryption altogether. Now, what about decreasing the iteration time? This is generally a very bad idea, unless you know and can enforce that the users only use high-entropy passphrases. If you decrease the iteration time without ensuring that, then you put your users at increased risk, and considering how rarely LUKS containers are unlocked in a typical work-flow, you do so without a good reason. Don't do it. The iteration time is already low enough that users with entropy low passphrases are vulnerable. Lowering it even further increases this danger significantly. * What about iteration count with plain dm-crypt? Simple: There is none. There is also no salting. If you use plain dm-crypt, the only way to be secure is to use a high entropy passphrase. If in doubt, use LUKS instead. * Is LUKS with default parameters less secure on a slow CPU? Unfortunately, yes. However the only aspect affected is the protection for low-entropy passphrase or master-key. All other security aspects are independent of CPU speed. The master key is less critical, as you really have to work at it to give it low entropy. One possibility is to supply the master key yourself. If that key is low-entropy, then you get what you deserve. The other known possibility is to use /dev/urandom for key generation in an entropy-startved situation (e.g. automatic installation on an embedded device without network and other entropy sources). For the passphrase, don't use a low-entropy passphrase. If your passphrase is good, then a slow CPU will not matter. If you insist on a low-entropy passphrase on a slow CPU, use something like "--iter-time=10" or higher and wait a long time on each LUKS unlock and pray that the attacker does not find out in which way exactly your passphrase is low entropy. This also applies to low-entropy passphrases on fast CPUs. Technology can do only so much to compensate for problems in front of the keyboard. * Why was the default aes-cbc-plain replaced with aes-cbc-essiv? The problem is that cbc-plain has a fingerprint vulnerability, where a specially crafted file placed into the crypto-container can be recognized from the outside. The issue here is that for cbc-plain the initialization vector (IV) is the sector number. The IV gets XORed to the first data chunk of the sector to be encrypted. If you make sure that the first data block to be stored in a sector contains the sector number as well, the first data block to be encrypted is all zeros and always encrypted to the same ciphertext. This also works if the first data chunk just has a constant XOR with the sector number. By having several shifted patterns you can take care of the case of a non-power-of-two start sector number of the file. This mechanism allows you to create a pattern of sectors that have the same first ciphertext block and signal one bit per sector to the outside, allowing you to e.g. mark media files that way for recognition without decryption. For large files this is a practical attack. For small ones, you do not have enough blocks to signal and take care of different file starting offsets. In order to prevent this attack, the default was changed to cbc-essiv. ESSIV uses a keyed hash of the sector number, with the encryption key as key. This makes the IV unpredictable without knowing the encryption key and the watermarking attack fails. * Are there any problems with "plain" IV? What is "plain64"? First, "plain" and "plain64" are both not secure to use with CBC, see previous FAQ item. However there are modes, like XTS, that are secure with "plain" IV. The next limit is that "plain" is 64 bit, with the upper 32 bit set to zero. This means that on volumes larger than 2TiB, the IV repeats, creating a vulnerability that potentially leaks some data. To avoid this, use "plain64", which uses the full sector number up to 64 bit. Note that "plain64" requires a kernel >= 2.6.33. Also note that "plain64" is backwards compatible for volume sizes <= 2TiB, but not for those > 2TiB. Finally, "plain64" does not cause any performance penalty compared to "plain". * What about XTS mode? XTS mode is potentially even more secure than cbc-essiv (but only if cbc-essiv is insecure in your scenario). It is a NIST standard and used, e.g. in Truecrypt. At the moment, if you want to use it, you have to specify it manually as "aes-xts-plain", i.e. cryptsetup -c aes-xts-plain luksFormat <device> For volumes >2TiB and kernels >= 2.6.33 use "plain64" (see FAQ item on "plain" and "plain64"): cryptsetup -c aes-xts-plain64 luksFormat <device> There is a potential security issue with XTS mode and large blocks. LUKS and dm-crypt always use 512B blocks and the issue does not apply. 6. Backup and Data Recovery * Why do I need Backup? First, disks die. The rate for well-treated (!) disk is about 5% per year, which is high enough to worry about. There is some indication that this may be even worse for some SSDs. This applies both to LUKS and plain dm-crypt partitions. Second, for LUKS, if anything damages the LUKS header or the key-stripe area then decrypting the LUKS device can become impossible. This is a frequent occuurence. For example an accidental format as FAT or some software overwriting the first sector where it suspects a partition boot sector typically makes a LUKS partition permanently inacessible. See more below on LUKS header damage. So, data-backup in some form is non-optional. For LUKS, you may also want to store a header backup in some secure location. This only needs an update if you change passphrases. * How do I backup a LUKS header? While you could just copy the appropriate number of bytes from the start of the LUKS partition, the best way is to use command option "luksHeaderBackup" of cryptsetup. This protects also against errors when non-standard parameters have been used in LUKS partition creation. Example: cryptsetup luksHeaderBackup --header-backup-file h <device> To restore, use the inverse command, i.e. cryptsetup luksHeaderRestore --header-backup-file h <device> * How do I test a LUKS header? Use cryptsetup -v isLuks <device> on the device. Without the "-v" it just signals its result via exit-status. You can alos use the more general test blkid -p <device> which will also detect other types and give some more info. Omit "-p" for old versions of blkid that do not support it. * How do I backup a LUKS or dm-crypt partition? There are two options, a sector-image and a plain file or filesystem backup of the contents of the partition. The sector image is already encrypted, but cannot be compressed and contains all empty space. The filesystem backup can be compressed, can contain only part of the encrypted device, but needs to be encrypted separately if so desired. A sector-image will contain the whole partition in encrypted form, for LUKS the LUKS header, the keys-slots and the data area. It can be done under Linux e.g. with dd_rescue (for a direct image copy) and with "cat" or "dd". Example: cat /dev/sda10 > sda10.img dd_rescue /dev/sda10 sda10.img You can also use any other backup software that is capable of making a sector image of a partition. Note that compression is ineffective for encrypted data, hence it does not make sense to use it. For a filesystem backup, you decrypt and mount the encrypted partition and back it up as you would a normal filesystem. In this case the backup is not encrypted, unless your encryption method does that. For example you can encrypt a backup with "tar" as follows with GnuPG: tar cjf - <path> | gpg --cipher-algo AES -c - > backup.tbz2.gpg And verify the backup like this if you are at "path": cat backup.tbz2.gpg | gpg - | tar djf - Note: Allways verify backups, especially encrypted ones. In both cases GnuPG will ask you interactively for your symmetric key. The verify will only output errors. Use "tar dvjf -" to get all comparison results. To make sure no data is written to disk unencrypted, turn off swap if it is not encrypted before doing the backup. You can of course use different or no compression and you can use an asymmetric key if you have one and have a backup of the secret key that belongs to it. A second option for a filestem-level backup that can be used when the backup is also on local disk (e.g. an external USB drive) is to use a LUKS container there and copy the files to be backed up between both mounted containers. Also see next item. * Do I need a backup of the full partition? Would the header and key-slots not be enough? Backup protects you against two things: Disk loss or corruption and user error. By far the most questions on the dm-crypt mailing list about how to recover a damaged LUKS partition are related to user error. For example, if you create a new filesystem on a LUKS partition, chances are good that all data is lost permanently. For this case, a header+key-slot backup would often be enough. But keep in mind that a well-treated (!) HDD has roughly a failure risk of 5% per year. It is highly advisable to have a complete backup to protect against this case. * *What do I need to backup if I use "decrypt_derived"? This is a script in Debian, intended for mounting /tmp or swap with a key derived from the master key of an already decrypted device. If you use this for an device with data that should be persistent, you need to make sure you either do not lose access to that master key or have a backup of the data. If you derive from a LUKS device, a header backup of that device would cover backing up the master key. Keep in mind that this does not protect against disk loss. Note: If you recreate the LUKS header of the device you derive from (using luksFormat), the master key changes even if you use the same passphrase(s) and you will not be able to decrypt the derived device with the new LUKS header. * Does a backup compromise security? Depends on how you do it. However if you do not have one, you are going to eventually lose your encrypted data. There are risks introduced by backups. For example if you change/disable a key-slot in LUKS, a binary backup of the partition will still have the old key-slot. To deal with this, you have to be able to change the key-slot on the backup as well, securely erase the backup or do a filesystem-level backup instead of a binary one. If you use dm-crypt, backup is simpler: As there is no key management, the main risk is that you cannot wipe the backup when wiping the original. However wiping the original for dm-crypt should consist of forgetting the passphrase and that you can do without actual access to the backup. In both cases, there is an additional (usually small) risk with binary backups: An attacker can see how many sectors and which ones have been changed since the backup. To prevent this, use a filesystem level backup methid that encrypts the whole backup in one go, e.g. as described above with tar and GnuPG. My personal advice is to use one USB disk (low value data) or three disks (high value data) in rotating order for backups, and either use independent LUKS partitions on them, or use encrypted backup with tar and GnuPG. If you do network-backup or tape-backup, I strongly recommend to go the filesystem backup path with independent encryption, as you typically cannot reliably delete data in these scenarios, especially in a cloud setting. (Well, you can burn the tape if it is under your control...) * What happens if I overwrite the start of a LUKS partition or damage the LUKS header or key-slots? There are two critical components for decryption: The salt values in the header itself and the key-slots. If the salt values are overwritten or changed, nothing (in the cryptographically strong sense) can be done to access the data, unless there is a backup of the LUKS header. If a key-slot is damaged, the data can still be read with a different key-slot, if there is a remaining undamaged and used key-slot. Note that in order to make a key-slot unrecoverable in a cryptographically strong sense, changing about 4-6 bits in random locations of its 128kiB size is quite enough. * What happens if I (quick) format a LUKS partition? I have not tried the different ways to do this, but very likely you will have written a new boot-sector, which in turn overwrites the LUKS header, including the salts, making your data permanently irretrivable, unless you have a LUKS header backup. You may also damage the key-slots in part or in full. See also last item. * How do I recover the master key from a mapped LUKS container? This is typically only needed if you managed to damage your LUKS header, but the container is still mapped, i.e. "luksOpen"ed. It also helps if you have a mapped container that you forgot or do not know a passphrase for (e.g. on a long running server.) WARNING: Things go wrong, do a full backup before trying this! WARNING: This exposes the master key of the LUKS container. Note that both ways to recreate a LUKS header with the old master key described below will write the master key to disk. Unless you are sure you have securely erased it afterwards, e.g. by writing it to an encrypted partition, RAM disk or by erasing the filesystem you wrote it to by a complete overwrite, you should change the master key afterwards. Changing the master key requires a full data backup, luksFormat and then restore of the backup. First, there is a script by Milan that automatizes the whole process, except generating a new LUKS header with the old master key (it prints the command for that though): http://code.google.com/p/cryptsetup/source/browse/trunk/misc/luks-header-from-active You can also do this manually. Here is how: - Get the master key from the device mapper. This is done by the following command. Substitute c5 for whatever you mapped to: # dmsetup table --target crypt --showkey /dev/mapper/c5 Result: 0 200704 crypt aes-cbc-essiv:sha256 a1704d9715f73a1bb4db581dcacadaf405e700d591e93e2eaade13ba653d0d09 0 7:0 4096 The result is actually one line, wrapped here for clarity. The long hex string is the master key. - Convert the master key to a binary file representation. You can do this manually, e.g. with hexedit. You can also use the tool "xxd" from vim like this: echo "a1704d9....53d0d09" | xxd -r -p > <master-key-file> - Do a luksFormat to create a new LUKS header. NOTE: If your header is intact and you just forgot the passphrase, you can just set a new passphrase, see next subitem. Unmap the device before you do that (luksClose). Then do cryptsetup luksFormat --master-key-file=<master-key-file> <luks device> Note that if the container was created with other than the default settings of the cryptsetup version you are using, you need to give additional parameters specifying the deviations. If in doubt, try the script by Milan. It does recover the other parameters as well. Side note: This is the way the decrypt_derived script gets at the master key. It just omits the conversion and hashes the master key string. - If the header is intact and you just forgot the passphrase, just set a new passphrase like this: cryptsetup luksAddKey --master-key-file=<master-key-file> <luks device> You may want to disable the old one afterwards. * What does the on-disk structure of dm-crypt look like? There is none. dm-crypt takes a block device and gives encrypted access to each of its blocks with a key derived from the passphrase given. If you use a cipher different than the default, you have to specify that as a parameter to cryptsetup too. If you want to change the password, you basically have to create a second encrypted device with the new passphrase and copy your data over. On the plus side, if you accidentally overwrite any part of a dm-crypt device, the damage will be limited to the are you overwrote. * What does the on-disk structure of LUKS look like? A LUKS partition consists of a header, followed by 8 key-slot descriptors, followed by 8 key slots, followed by the encrypted data area. Header and key-slot descriptors fill the first 592 bytes. The key-slot size depends on the creation parameters, namely on the number of anti-forensic stripes, key material offset and master key size. With the default parameters, each key-slot is a bit less than 128kiB in size. Due to sector alignment of the key-slot start, that means the key block 0 is at offset 0x1000-0x20400, key block 1 at offset 0x21000-0x40400, and key block 7 at offset 0xc1000-0xe0400. The space to the next full sector address is padded with zeros. Never used key-slots are filled with what the disk originally contained there, a key-slot removed with "luksRemoveKey" or "luksKillSlot" gets filled with 0xff. Due to 2MiB default alignment, start of the data area for cryptsetup 1.3 and later is at 2MiB, i.e. at 0x200000. For older versions, it is at 0x101000, i.e. at 1'052'672 bytes, i.e. at 1MiB + 4096 bytes from the start of the partition. Incidentally, "luksHeaderBackup" for a LUKS container created with default parameters dumps exactly the first 2MiB (or 1'052'672 bytes for headers created with cryptsetup versions < 1.3) to file and "luksHeaderRestore" restores them. For non-default parameters, you have to figure out placement yourself. "luksDump" helps. See also next item. For the most common non-default settings, namely aes-xts-plain with 512 bit key, the offsets are: 1st keyslot 0x1000-0x3f800, 2nd keyslot 0x40000-0x7e000, 3rd keyslot 0x7e000-0xbd800, ..., and start of bulk data at 0x200000. The exact specification of the format is here: http://code.google.com/p/cryptsetup/wiki/Specification * What is the smallest possible LUKS container? Note: From cryptsetup 1.3 onwards, alignment is set to 1MB. With modern Linux partitioning tools that also align to 1MB, this will result in aligmnet to 2k secors and typical Flash/SSD sectors, which is highly desirable for a number of reasons. Changing the alignment is not recomended. That said, with default parameters, the data area starts at exactly 2MB offset (at 0x101000 for cryptsetup versions before 1.3). The smallest data area you can have is one sector of 512 bytes. Data areas of 0 bytes can be created, but fail on mapping. While you cannot put a filesystem into something this small, it may still be used to contain, for example, key. Note that with current formatting tools, a partition for a container this size will be 3MiB anyways. If you put the LUKS container into a file (via losetup and a loopback device), the file needs to be 2097664 bytes in size, i.e. 2MiB + 512B. There two ways to influence the start of the data area are key-size and alignment. For alignment, you can go down to 1 on the parameter. This will still leave you with a data-area starting at 0x101000, i.e. 1MiB+4096B (default parameters) as alignment will be rounded up to the next multiple of 8 (i.e. 4096 bytes) If in doubt, do a dry-run on a larger file and dump the LUKS header to get actual information. For key-size, you can use 128 bit (e.g. AES-128 with CBC), 256 bit (e.g. AES-256 with CBC) or 512 bit (e.g. AES-256 with XTS mode). You can do 64 bit (e.g. blowfish-64 with CBC), but anything below 128 bit has to be considered insecure today. Example 1 - AES 128 bit with CBC: cryptsetup luksFormat -s 128 --align-payload=8 <device> This results in a data offset of 0x81000, i.e. 516KiB or 528384 bytes. Add one 512 byte sector and the smallest LUKS container size with these parameters is 516KiB + 512B or 528896 bytes. Example 2 - Blowfish 64 bit with CBC (WARNING: insecure): cryptsetup luksFormat -c blowfish -s 64 --align-payload=8 /dev/loop0 This results in a data offset of 0x41000, i.e. 260kiB or 266240 bytes, with a minimal LUKS conatiner size of 260kiB + 512B or 266752 bytes. * I think this is overly complicated. Is there an alternative? Not really. Encryption comes at a price. You can use plain dm-crypt to simplify things a bit. It does not allow multiple passphrases, but on the plus side, it has zero on disk description and if you overwrite some part of a plain dm-crypt partition, exactly the overwritten parts are lost (rounded up to sector borders). 7. Interoperability with other Disk Encryption Tools * What is this section about? Cryptsetup for plain dm-crypt can be used to access a number of on-disk formats created by tools like loop-aes patched into losetup. This somtimes works and sometimes does not. This section collects insights into what works, what does not and where more information is required. Additional information may be found in the mailing-list archives, mentioned at the start of this FAQ document. If you have a solution working that is not yet documented here and think a wider audience may be intertested, please email the FAQ maintainer. * loop-aes: General observations. One problem is that there are different versions of losetup around. loop-aes is a patch for losetup. Possible problems and deviations from cryptsetup option syntax include: - Offsets specifed in bytes (cryptsetup: 512 byte sectors) - The need to specify an IV offset - Encryption mode needs specifying (e.g. "-c twofish-cbc-plain") - Key size needs specifying (e.g. "-s 128" for 128 bit keys) - Passphrase hash algorithm needs specifying Also note that because plain dm-crypt and loop-aes format does not have metadata, autodetection, while feasible in most cases, would be a lot of work that nobody really wants to do. If you still have the old set-up, using a verbosity option (-v) on mapping with the old tool or having a look into the system logs after setup could give you the information you need. * loop-aes patched into losetup on debian 5.x, kernel 2.6.32 In this case, the main problem seems to be that this variant of losetup takes the offset (-o option) in bytes, while cryptsetup takes it in sectors of 512 bytes each. Example: The losetupp command losetup -e twofish -o 2560 /dev/loop0 /dev/sdb1 mount /dev/loop0 mountpoint translates to cryptsetup create -c twofish -o 5 --skip 5 e1 /dev/sdb1 mount /dev/mapper/e1 mountpoint * loop-aes with 160 bit key This seems to be sometimes used with twofish and blowfish and represents a 160 bit ripemed160 hash output padded to 196 bit key length. It seems the corresponding options for cryptsetup are --cipher twofish-cbc-null -s 192 -h ripemd160:20 8. Issues with Specific Versions of cryptsetup * When using the create command for plain dm-crypt with cryptsetup 1.1.x, the mapping is incompatible and my data is not accessible anymore! With cryptsetup 1.1.x, the distro maintainer can define different default encryption modes for LUKS and plain devices. You can check these compiled-in defaults using "cryptsetup --help". Moreover, the plain device default changed because the old IV mode was vulnerable to a watermarking attack. If you are using a plain device and you need a compatible mode, just specify cipher, key size and hash algorithm explicitly. For compatibility with cryptsetup 1.0.x defaults, simple use the following: cryptsetup create -c aes-cbc-plain -s 256 -h ripemd160 <name> <dev> LUKS stores cipher and mode in the metadata on disk, avoiding this problem. * cryptsetup on SLED 10 has problems... SLED 10 is missing an essential kernel patch for dm-crypt, which is broken in its kernel as a result. There may be a very old version of cryptsetup (1.0.x) provided by SLED, which should also not be used anymore as well. My advice would be to drop SLED 10. A. Contributors In no particular order: - Arno Wagner - Milan Broz -- Arno Wagner, Dr. sc. techn., Dipl. Inform., CISSP -- Email: arno@xxxxxxxxxxx GnuPG: ID: 1E25338F FP: 0C30 5782 9D93 F785 E79C 0296 797F 6B50 1E25 338F ---- One of the painful things about our time is that those who feel certainty are stupid, and those with any imagination and understanding are filled with doubt and indecision. -- Bertrand Russell _______________________________________________ dm-crypt mailing list dm-crypt@xxxxxxxx http://www.saout.de/mailman/listinfo/dm-crypt
