"Fossies" - the Fresh Open Source Software Archive

Member "cryptsetup-2.4.3/FAQ" (13 Jan 2022, 137412 Bytes) of package /linux/misc/cryptsetup-2.4.3.tar.xz:


As a special service "Fossies" has tried to format the requested text file into HTML format (style: standard) with prefixed line numbers. Alternatively you can here view or download the uninterpreted source code file. See also the last Fossies "Diffs" side-by-side code changes report for "FAQ": 2.4.0_vs_2.4.1.

    1 Frequently Asked Questions Cryptsetup/LUKS
    2 
    3 Sections
    4 1. General Questions
    5 2. Setup
    6 3. Common Problems
    7 4. Troubleshooting
    8 5. Security Aspects
    9 6. Backup and Data Recovery
   10 7. Interoperability with other Disk Encryption Tools
   11 8. Issues with Specific Versions of cryptsetup
   12 9. The Initrd question
   13 10. LUKS2 Questions
   14 11. References and Further Reading
   15 A. Contributors
   16 
   17 1. General Questions
   18 
   19 
   20   * 1.1 What is this?
   21 
   22   This is the FAQ (Frequently Asked Questions) for cryptsetup.  It covers
   23   Linux disk encryption with plain dm-crypt (one passphrase, no
   24   management, no metadata on disk) and LUKS (multiple user keys with one
   25   master key, anti-forensic features, metadata block at start of device,
   26   ...).  The latest version of this FAQ should usually be available at
   27   https://gitlab.com/cryptsetup/cryptsetup/wikis/FrequentlyAskedQuestions
   28 
   29 
   30   * 1.2 WARNINGS
   31 
   32   LUKS2 COMPATIBILITY: This FAQ was originally written for LUKS1, not
   33   LUKS2.  Hence regarding LUKS2, some of the answers found here may not
   34   apply.  Updates for LUKS2 have been done and anything not applying to
   35   LUKS2 should clearly say LUKS1.  However, this is a Frequently Asked
   36   Questions, and questions for LUKS2 are limited at this time or at least
   37   those that have reached me are.  In the following, "LUKS" refers to both
   38   LUKS1 and LUKS2.
   39  
   40   The LUKS1 on-disk format specification is at  
   41   https://www.kernel.org/pub/linux/utils/cryptsetup/LUKS_docs/on-disk-format.pdf  
   42   The LUKS2 on-disk format specification is at  
   43   https://gitlab.com/cryptsetup/LUKS2-docs
   44 
   45   ATTENTION: If you are going to read just one thing, make it the section
   46   on Backup and Data Recovery.  By far the most questions on the
   47   cryptsetup mailing list are from people that managed to damage the start
   48   of their LUKS partitions, i.e.  the LUKS header.  In most cases, there
   49   is nothing that can be done to help these poor souls recover their data. 
   50   Make sure you understand the problem and limitations imposed by the LUKS
   51   security model BEFORE you face such a disaster!  In particular, make
   52   sure you have a current header backup before doing any potentially
   53   dangerous operations.  The LUKS2 header should be a bit more resilient
   54   as critical data starts later and is stored twice, but you can decidely
   55   still destroy it or a keyslot permanently by accident.
   56 
   57   DEBUG COMMANDS: While the --debug and --debug-json options should not
   58   leak secret data, "strace" and the like can leak your full passphrase. 
   59   Do not post an strace output with the correct passphrase to a
   60   mailing-list or online!  See Item 4.5 for more explanation.
   61 
   62   SSDs/FLASH DRIVES: SSDs and Flash are different.  Currently it is
   63   unclear how to get LUKS or plain dm-crypt to run on them with the full
   64   set of security assurances intact.  This may or may not be a problem,
   65   depending on the attacker model.  See Section 5.19.
   66 
   67   BACKUP: Yes, encrypted disks die, just as normal ones do.  A full backup
   68   is mandatory, see Section "6.  Backup and Data Recovery" on options for
   69   doing encrypted backup.
   70 
   71   CLONING/IMAGING: If you clone or image a LUKS container, you make a copy
   72   of the LUKS header and the master key will stay the same!  That means
   73   that if you distribute an image to several machines, the same master key
   74   will be used on all of them, regardless of whether you change the
   75   passphrases.  Do NOT do this!  If you do, a root-user on any of the
   76   machines with a mapped (decrypted) container or a passphrase on that
   77   machine can decrypt all other copies, breaking security.  See also Item
   78   6.15.
   79 
   80   DISTRIBUTION INSTALLERS: Some distribution installers offer to create
   81   LUKS containers in a way that can be mistaken as activation of an
   82   existing container.  Creating a new LUKS container on top of an existing
   83   one leads to permanent, complete and irreversible data loss.  It is
   84   strongly recommended to only use distribution installers after a
   85   complete backup of all LUKS containers has been made.
   86 
   87   UBUNTU INSTALLER: In particular the Ubuntu installer seems to be quite
   88   willing to kill LUKS containers in several different ways.  Those
   89   responsible at Ubuntu seem not to care very much (it is very easy to
   90   recognize a LUKS container), so treat the process of installing Ubuntu
   91   as a severe hazard to any LUKS container you may have.
   92 
   93   NO WARNING ON NON-INTERACTIVE FORMAT: If you feed cryptsetup from STDIN
   94   (e.g.  via GnuPG) on LUKS format, it does not give you the warning that
   95   you are about to format (and e.g.  will lose any pre-existing LUKS
   96   container on the target), as it assumes it is used from a script.  In
   97   this scenario, the responsibility for warning the user and possibly
   98   checking for an existing LUKS header is shifted to the script.  This is
   99   a more general form of the previous item.
  100 
  101   LUKS PASSPHRASE IS NOT THE MASTER KEY: The LUKS passphrase is not used
  102   in deriving the master key.  It is used in decrypting a master key that
  103   is randomly selected on header creation.  This means that if you create
  104   a new LUKS header on top of an old one with exactly the same parameters
  105   and exactly the same passphrase as the old one, it will still have a
  106   different master key and your data will be permanently lost.
  107 
  108   PASSPHRASE CHARACTER SET: Some people have had difficulties with this
  109   when upgrading distributions.  It is highly advisable to only use the 95
  110   printable characters from the first 128 characters of the ASCII table,
  111   as they will always have the same binary representation.  Other
  112   characters may have different encoding depending on system configuration
  113   and your passphrase will not work with a different encoding.  A table of
  114   the standardized first 128 ASCII characters can, e.g.  be found on
  115   https://en.wikipedia.org/wiki/ASCII
  116 
  117   KEYBOARD NUM-PAD: Apparently some pre-boot authentication environments
  118   (these are done by the distro, not by cryptsetup, so complain there)
  119   treat digits entered on the num-pad and ones entered regularly
  120   different.  This may be because the BIOS USB keyboard driver is used and
  121   that one may have bugs on some computers.  If you cannot open your
  122   device in pre-boot, try entering the digits over the regular digit keys.
  123 
  124 
  125   * 1.3 System specific warnings
  126 
  127   - The Ubuntu Natty uinstaller has a "won't fix" defect that may destroy
  128   LUKS containers.  This is quite old an not relevant for most people. 
  129   Reference:
  130   https://bugs.launchpad.net/ubuntu/+source/partman-crypto/+bug/420080
  131 
  132 
  133   * 1.4 My LUKS-device is broken! Help!
  134 
  135   First: Do not panic! In many cases the data is still recoverable.
  136   Do not do anything hasty! Steps:
  137 
  138   - Take some deep breaths. Maybe add some relaxing music.  This may
  139   sound funny, but I am completely serious.  Often, critical damage is
  140   done only after the initial problem.
  141 
  142   - Do not reboot. The keys may still be in the kernel if the device is
  143   mapped.
  144 
  145   - Make sure others do not reboot the system.
  146 
  147   - Do not write to your disk without a clear understanding why this will
  148   not make matters worse.  Do a sector-level backup before any writes. 
  149   Often you do not need to write at all to get enough access to make a
  150   backup of the data.
  151 
  152   - Relax some more.
  153 
  154   - Read section 6 of this FAQ.
  155 
  156   - Ask on the mailing-list if you need more help.
  157 
  158 
  159   * 1.5 Who wrote this?
  160 
  161   Current FAQ maintainer is Arno Wagner <arno@wagner.name>.  If you want
  162   to send me encrypted email, my current PGP key is DSA key CB5D9718,
  163   fingerprint 12D6 C03B 1B30 33BB 13CF B774 E35C 5FA1 CB5D 9718.
  164 
  165   Other contributors are listed at the end.  If you want to contribute,
  166   send your article, including a descriptive headline, to the maintainer,
  167   or the dm-crypt mailing list with something like "FAQ ..." 
  168   in the subject.  You can also send more raw information and have
  169   me write the section.  Please note that by contributing to this FAQ,
  170   you accept the license described below.
  171 
  172   This work is under the "Attribution-Share Alike 3.0 Unported" license,
  173   which means distribution is unlimited, you may create derived works, but
  174   attributions to original authors and this license statement must be
  175   retained and the derived work must be under the same license.  See
  176   https://creativecommons.org/licenses/by-sa/3.0/ for more details of the
  177   license.
  178 
  179   Side note: I did text license research some time ago and I think this
  180   license is best suited for the purpose at hand and creates the least
  181   problems.
  182 
  183 
  184   * 1.6 Where is the project website?
  185 
  186   There is the project website at
  187   https://gitlab.com/cryptsetup/cryptsetup/ Please do not post
  188   questions there, nobody will read them.  Use the mailing-list
  189   instead.
  190 
  191 
  192   * 1.7 Is there a mailing-list?
  193 
  194   Instructions on how to subscribe to the mailing-list are on the
  195   project website.  People are generally helpful and friendly on the
  196   list.
  197 
  198   The question of how to unsubscribe from the list does crop up sometimes. 
  199   For this you need your list management URL, which is sent to you
  200   initially and once at the start of each month.  Go to the URL mentioned
  201   in the email and select "unsubscribe".  This page also allows you to
  202   request a password reminder.
  203 
  204   Alternatively, you can send an Email to dm-crypt-request@saout.de with
  205   just the word "help" in the subject or message body.  Make sure to send
  206   it from your list address.
  207 
  208   The mailing list archive is here:
  209   https://marc.info/?l=dm-crypt
  210 
  211 
  212   * 1.8 Unsubscribe from the mailing-list
  213 
  214   Send mail to dm-crypt-unsubscribe@saout.de from the subscribed account. 
  215   You will get an email with instructions.
  216 
  217   Basically, you just have to respond to it unmodified to get
  218   unsubscribed.  The listserver admin functions are not very fast.  It can
  219   take 15 minutes or longer for a reply to arrive (I suspect greylisting
  220   is in use), so be patient.
  221 
  222   Also note that nobody on the list can unsubscribe you, sending demands
  223   to be unsubscribed to the list just annoys people that are entirely
  224   blameless for you being subscribed.
  225 
  226   If you are subscribed, a subscription confirmation email was sent to
  227   your email account and it had to be answered before the subscription
  228   went active.  The confirmation emails from the listserver have subjects
  229   like these (with other numbers):
  230 
  231     Subject: confirm 9964cf10.....
  232 
  233   and are sent from dm-crypt-request@saout.de.  You should check whether
  234   you have anything like it in your sent email folder.  If you find
  235   nothing and are sure you did not confirm, then you should look into a
  236   possible compromise of your email account.
  237 
  238   * 1.9 What can I do if cryptsetup is running out of memory?
  239 
  240   Memory issues are generally related to the key derivation function.  You may
  241   be able to tune usage with the options --pbkdf-memory or --pbkdf pbkdf2.
  242 
  243 
  244   * 1.10 Can cryptsetup be run without root access?
  245 
  246   Elevated privileges are required to use cryptsetup and LUKS.  Some operations
  247   require root access.  There are a few features which will work without root 
  248   access with the right switches but there are caveats.
  249 
  250 
  251   * 1.11 What are the problems with running as non root?
  252 
  253   The first issue is one of permissions to devices.  Generally, root or a group
  254   such as disk has ownership of the storage devices.  The non root user will
  255   need write access to the block device used for LUKS.
  256 
  257   Next, file locking is managed in /run/cryptsetup.  You may use 
  258   --disable-locks but cryptsetup will no longer protect you from race 
  259   conditions and problems with concurrent access to the same devices.
  260 
  261   Also, device mapper requires root access.  cryptsetup uses device mapper to 
  262   manage the decrypted container.
  263 
  264 2. Setup
  265 
  266   * 2.1 LUKS Container Setup mini-HOWTO
  267 
  268   This item tries to give you a very brief list of all the steps you
  269   should go through when creating a new LUKS encrypted container, i.e.
  270   encrypted disk, partition or loop-file.
  271 
  272   01) All data will be lost, if there is data on the target, make a
  273   backup.
  274 
  275   02) Make very sure you use the right target disk, partition or
  276   loop-file.
  277 
  278   03) If the target was in use previously, it is a good idea to wipe it
  279   before creating the LUKS container in order to remove any trace of old
  280   file systems and data.  For example, some users have managed to run
  281   e2fsck on a partition containing a LUKS container, possibly because of
  282   residual ext2 superblocks from an earlier use.  This can do arbitrary
  283   damage up to complete and permanent loss of all data in the LUKS
  284   container.
  285 
  286   To just quickly wipe file systems (old data may remain), use
  287 
  288     wipefs -a <target device>
  289 
  290   To wipe file system and data, use something like
  291 
  292     cat /dev/zero > <target device>
  293 
  294   This can take a while.  To get a progress indicator, you can use the
  295   tool dd_rescue (->google) instead or use my stream meter "wcs" (source
  296   here: https://www.tansi.org/tools/index.html) in the following fashion:
  297 
  298     cat /dev/zero | wcs > <target device>
  299 
  300   Plain "dd" also gives you the progress on a SIGUSR1, see its man-page.
  301 
  302   Be very sure you have the right target, all data will be lost!
  303 
  304   Note that automatic wiping is on the TODO list for cryptsetup, so at
  305   some time in the future this will become unnecessary.
  306 
  307   Alternatively, plain dm-crypt can be used for a very fast wipe with
  308   crypto-grade randomness, see Item 2.19
  309 
  310   04) Create the LUKS container.  
  311 
  312   LUKS1:
  313 
  314     cryptsetup luksFormat --type luks1 <target device>
  315 
  316   LUKS2:
  317 
  318     cryptsetup luksFormat --type luks2 <target device>
  319 
  320 
  321   Just follow the on-screen instructions.
  322 
  323   Note: Passphrase iteration count is based on time and hence security
  324   level depends on CPU power of the system the LUKS container is created
  325   on.  For example on a Raspberry Pi and LUKS1, I found some time ago that
  326   the iteration count is 15 times lower than for a regular PC (well, for
  327   my old one).  Depending on security requirements, this may need
  328   adjustment.  For LUKS1, you can just look at the iteration count on
  329   different systems and select one you like.  You can also change the
  330   benchmark time with the -i parameter to create a header for a slower
  331   system.
  332 
  333   For LUKS2, the parameters are more complex.  ARGON2 has iteration,
  334   parallelism and memory parameter.  cryptsetup actually may adjust the
  335   memory parameter for time scaling.  Hence to use -i is the easiest way
  336   to get slower or faster opening (default: 2000 = 2sec).  Just make sure
  337   to not drop this too low or you may get a memory parameter that is to
  338   small to be secure.  The luksDump command lists the memory parameter of
  339   a created LUKS2 keyslot in kB.  That parameter should probably be not
  340   much lower than 100000, i.e.  100MB, but don't take my word for it.
  341 
  342   05) Map the container. Here it will be mapped to /dev/mapper/c1:
  343 
  344     cryptsetup luksOpen <target device> c1
  345 
  346   06) (Optionally) wipe the container (make sure you have the right
  347       target!): 
  348 
  349     cat /dev/zero > /dev/mapper/c1
  350 
  351   This will take a while.  Note that this creates a small information
  352   leak, as an attacker can determine whether a 512 byte block is zero if
  353   the attacker has access to the encrypted container multiple times. 
  354   Typically a competent attacker that has access multiple times can
  355   install a passphrase sniffer anyways, so this leakage is not very
  356   significant.  For getting a progress indicator, see step 03.
  357 
  358   07) Create a file system in the mapped container, for example an
  359   ext3 file system (any other file system is possible):
  360 
  361     mke2fs -j /dev/mapper/c1
  362 
  363   08) Mount your encrypted file system, here on /mnt:
  364 
  365     mount /dev/mapper/c1 /mnt
  366 
  367   09) Make a LUKS header backup and plan for a container backup.
  368       See Section 6 for details.
  369 
  370   Done.  You can now use the encrypted file system to store data.  Be sure
  371   to read through the rest of the FAQ, these are just the very basics.  In
  372   particular, there are a number of mistakes that are easy to make, but
  373   will compromise your security.
  374 
  375 
  376   * 2.2 LUKS on partitions or raw disks? What about RAID?
  377 
  378   Also see Item 2.8.  
  379   This is a complicated question, and made more so by the availability of
  380   RAID and LVM.  I will try to give some scenarios and discuss advantages
  381   and disadvantages.  Note that I say LUKS for simplicity, but you can do
  382   all the things described with plain dm-crypt as well.  Also note that
  383   your specific scenario may be so special that most or even all things I
  384   say below do not apply.
  385 
  386   Be aware that if you add LVM into the mix, things can get very
  387   complicated.  Same with RAID but less so.  In particular, data recovery
  388   can get exceedingly difficult.  Only add LVM if you have a really good
  389   reason and always remember KISS is what separates an engineer from an
  390   amateur.  Of course, if you really need the added complexity, KISS is
  391   satisfied.  But be very sure as there is a price to pay for it.  In
  392   engineering, complexity is always the enemy and needs to be fought
  393   without mercy when encountered.
  394 
  395   Also consider using RAID instead of LVM, as at least with the old
  396   superblock format 0.90, the RAID superblock is in the place (end of
  397   disk) where the risk of it damaging the LUKS header is smallest and you
  398   can have your array assembled by the RAID controller (i.e.  the kernel),
  399   as it should be.  Use partition type 0xfd for that.  I recommend staying
  400   away from superblock formats 1.0, 1.1 and 1.2 unless you really need
  401   them.
  402 
  403   Scenarios:
  404 
  405   (1) Encrypted partition: Just make a partition to your liking, and put
  406   LUKS on top of it and a filesystem into the LUKS container.  This gives
  407   you isolation of differently-tasked data areas, just as ordinary
  408   partitioning does.  You can have confidential data, non-confidential
  409   data, data for some specific applications, user-homes, root, etc. 
  410   Advantages are simplicity as there is a 1:1 mapping between partitions
  411   and filesystems, clear security functionality and the ability to
  412   separate data into different, independent (!) containers.
  413 
  414   Note that you cannot do this for encrypted root, that requires an
  415   initrd.  On the other hand, an initrd is about as vulnerable to a
  416   competent attacker as a non-encrypted root, so there really is no
  417   security advantage to doing it that way.  An attacker that wants to
  418   compromise your system will just compromise the initrd or the kernel
  419   itself.  The better way to deal with this is to make sure the root
  420   partition does not store any critical data and to move that to
  421   additional encrypted partitions.  If you really are concerned your root
  422   partition may be sabotaged by somebody with physical access (who would
  423   however strangely not, say, sabotage your BIOS, keyboard, etc.), protect
  424   it in some other way.  The PC is just not set-up for a really secure
  425   boot-chain (whatever some people may claim).
  426 
  427   That said, if you want an encrypted root partition, you have to store 
  428   an initrd with cryptsetup somewhere else. The traditional approach is
  429   to have a separate partition under /boot for that. You can also put that 
  430   initrd on a bootable memory stick, bootable CD or bootable external
  431   drive as well. The kernel and Grub typically go to the same location 
  432   as that initrd. A minimal example what such an initrd can look like is 
  433   given in Section 9.
  434   
  435   (2) Fully encrypted raw block device: For this, put LUKS on the raw
  436   device (e.g.  /dev/sdb) and put a filesystem into the LUKS container, no
  437   partitioning whatsoever involved.  This is very suitable for things like
  438   external USB disks used for backups or offline data-storage.
  439 
  440   (3) Encrypted RAID: Create your RAID from partitions and/or full
  441   devices.  Put LUKS on top of the RAID device, just if it were an
  442   ordinary block device.  Applications are just the same as above, but you
  443   get redundancy.  (Side note as many people seem to be unaware of it: You
  444   can do RAID1 with an arbitrary number of components in Linux.) See also
  445   Item 2.8.
  446 
  447   (4) Now, some people advocate doing the encryption below the RAID layer. 
  448   That has several serious problems.  One is that suddenly debugging RAID
  449   issues becomes much harder.  You cannot do automatic RAID assembly
  450   anymore.  You need to keep the encryption keys for the different RAID
  451   components in sync or manage them somehow.  The only possible advantage
  452   is that things may run a little faster as more CPUs do the encryption,
  453   but if speed is a priority over security and simplicity, you are doing
  454   this wrong anyways.  A good way to mitigate a speed issue is to get a
  455   CPU that does hardware AES as most do today.
  456 
  457 
  458   * 2.3 How do I set up encrypted swap?
  459 
  460   As things that are confidential can end up in swap (keys, passphrases,
  461   etc.  are usually protected against being swapped to disk, but other
  462   things may not be), it may be advisable to do something about the issue. 
  463   One option is to run without swap, which generally works well in a
  464   desktop-context.  It may cause problems in a server-setting or under
  465   special circumstances.  The solution to that is to encrypt swap with a
  466   random key at boot-time.
  467 
  468   NOTE: This is for Debian, and should work for Debian-derived
  469   distributions.  For others you may have to write your own startup script
  470   or use other mechanisms.
  471 
  472   01) Add the swap partition to /etc/crypttab. A line like the
  473   following should do it:
  474 
  475     swap  /dev/<partition>  /dev/urandom   swap,noearly
  476 
  477   Warning: While Debian refuses to overwrite partitions with a filesystem
  478   or RAID signature on it, as your disk IDs may change (adding or removing
  479   disks, failure of disk during boot, etc.), you may want to take
  480   additional precautions.  Yes, this means that your kernel device names
  481   like sda, sdb, ...  can change between reboots!  This is not a concern
  482   if you have only one disk.  One possibility is to make sure the
  483   partition number is not present on additional disks or also swap there. 
  484   Another is to encapsulate the swap partition (by making it a 1-partition
  485   RAID1 or by using LVM), as that gets a persistent identifier. 
  486   Specifying it directly by UUID does not work, unfortunately, as the UUID
  487   is part of the swap signature and that is not visible from the outside
  488   due to the encryption and in addition changes on each reboot with this
  489   setup.
  490 
  491   Note: Use /dev/random if you are paranoid or in a potential low-entropy
  492   situation (embedded system, etc.).  This may cause the operation to take
  493   a long time during boot however.  If you are in a "no entropy"
  494   situation, you cannot encrypt swap securely.  In this situation you
  495   should find some entropy, also because nothing else using crypto will be
  496   secure, like ssh, ssl or GnuPG.
  497 
  498   Note: The "noearly" option makes sure things like LVM, RAID, etc.  are
  499   running.  As swap is non-critical for boot, it is fine to start it late.
  500 
  501   02) Add the swap partition to /etc/fstab. A line like the following
  502   should do it:
  503 
  504     /dev/mapper/swap none swap sw 0 0
  505 
  506   That is it. Reboot or start it manually to activate encrypted swap. 
  507   Manual start would look like this:
  508 
  509     /etc/init.d/cryptdisks start
  510     swapon /dev/mapper/swap
  511 
  512 
  513   * 2.4 What is the difference between "plain" and LUKS format?
  514 
  515   First, unless you happen to understand the cryptographic background
  516   well, you should use LUKS.  It does protect the user from a lot of
  517   common mistakes.  Plain dm-crypt is for experts.
  518 
  519   Plain format is just that: It has no metadata on disk, reads all
  520   parameters from the commandline (or the defaults), derives a master-key
  521   from the passphrase and then uses that to de-/encrypt the sectors of the
  522   device, with a direct 1:1 mapping between encrypted and decrypted
  523   sectors.
  524 
  525   Primary advantage is high resilience to damage, as one damaged encrypted
  526   sector results in exactly one damaged decrypted sector.  Also, it is not
  527   readily apparent that there even is encrypted data on the device, as an
  528   overwrite with crypto-grade randomness (e.g.  from
  529   /dev/urandom) looks exactly the same on disk.
  530 
  531   Side-note: That has limited value against the authorities.  In civilized
  532   countries, they cannot force you to give up a crypto-key anyways.  In
  533   quite a few countries around the world, they can force you to give up
  534   the keys (using imprisonment or worse to pressure you, sometimes without
  535   due process), and in the worst case, they only need a nebulous
  536   "suspicion" about the presence of encrypted data.  Sometimes this
  537   applies to everybody, sometimes only when you are suspected of having
  538   "illicit data" (definition subject to change) and sometimes specifically
  539   when crossing a border.  Note that this is going on in countries like
  540   the US and the UK to different degrees and sometimes with courts
  541   restricting what the authorities can actually demand.
  542 
  543   My advice is to either be ready to give up the keys or to not have
  544   encrypted data when traveling to those countries, especially when
  545   crossing the borders.  The latter also means not having any high-entropy
  546   (random) data areas on your disk, unless you can explain them and
  547   demonstrate that explanation.  Hence doing a zero-wipe of all free
  548   space, including unused space, may be a good idea.
  549 
  550   Disadvantages are that you do not have all the nice features that the
  551   LUKS metadata offers, like multiple passphrases that can be changed, the
  552   cipher being stored in the metadata, anti-forensic properties like
  553   key-slot diffusion and salts, etc..
  554 
  555   LUKS format uses a metadata header and 8 key-slot areas that are being
  556   placed at the beginning of the disk, see below under "What does the LUKS
  557   on-disk format looks like?".  The passphrases are used to decrypt a
  558   single master key that is stored in the anti-forensic stripes.  LUKS2
  559   adds some more flexibility.
  560 
  561   Advantages are a higher usability, automatic configuration of
  562   non-default crypto parameters, defenses against low-entropy passphrases
  563   like salting and iterated PBKDF2 or ARGON 2 passphrase hashing, the
  564   ability to change passphrases, and others.
  565 
  566   Disadvantages are that it is readily obvious there is encrypted data on
  567   disk (but see side note above) and that damage to the header or
  568   key-slots usually results in permanent data-loss.  See below under "6. 
  569   Backup and Data Recovery" on how to reduce that risk.  Also the sector
  570   numbers get shifted by the length of the header and key-slots and there
  571   is a loss of that size in capacity.  Unless you have a specific need,
  572   use LUKS2.
  573 
  574 
  575   * 2.5 Can I encrypt an existing, non-empty partition to use LUKS?
  576 
  577   There is no converter, and it is not really needed.  The way to do this
  578   is to make a backup of the device in question, securely wipe the device
  579   (as LUKS device initialization does not clear away old data), do a
  580   luksFormat, optionally overwrite the encrypted device, create a new
  581   filesystem and restore your backup on the now encrypted device.  Also
  582   refer to sections "Security Aspects" and "Backup and Data Recovery".
  583 
  584   For backup, plain GNU tar works well and backs up anything likely to be
  585   in a filesystem.
  586 
  587 
  588   * 2.6 How do I use LUKS with a loop-device?
  589 
  590   This can be very handy for experiments.  Setup is just the same as with
  591   any block device.  If you want, for example, to use a 100MiB file as
  592   LUKS container, do something like this:
  593 
  594     head -c 100M /dev/zero > luksfile               # create empty file
  595     losetup /dev/loop0 luksfile                     # map file to /dev/loop0
  596     cryptsetup luksFormat --type luks2 /dev/loop0   # create LUKS2 container
  597 
  598   Afterwards just use /dev/loop0 as a you would use a LUKS partition.
  599   To unmap the file when done, use "losetup -d /dev/loop0".
  600 
  601 
  602   * 2.7 When I add a new key-slot to LUKS, it asks for a passphrase
  603     but then complains about there not being a key-slot with that
  604     passphrase?
  605 
  606   That is as intended.  You are asked a passphrase of an existing key-slot
  607   first, before you can enter the passphrase for the new key-slot. 
  608   Otherwise you could break the encryption by just adding a new key-slot. 
  609   This way, you have to know the passphrase of one of the already
  610   configured key-slots in order to be able to configure a new key-slot.
  611 
  612 
  613   * 2.8 Encryption on top of RAID or the other way round?
  614 
  615   Also see Item 2.2.  
  616   Unless you have special needs, place encryption between RAID and
  617   filesystem, i.e.  encryption on top of RAID.  You can do it the other
  618   way round, but you have to be aware that you then need to give the
  619   passphrase for each individual disk and RAID auto-detection will not
  620   work anymore.  Therefore it is better to encrypt the RAID device, e.g. 
  621   /dev/dm0 .
  622 
  623   This means that the typical layering looks like this:
  624 
  625   Filesystem     <- top
  626   |
  627   Encryption (LUKS)
  628   |
  629   RAID
  630   |
  631   Raw partitions (optional)
  632   |
  633   Raw disks      <- bottom
  634 
  635   The big advantage of this is that you can manage the RAID container just
  636   like any other regular RAID container, it does not care that its content
  637   is encrypted.  This strongly cuts down on complexity, something very
  638   valuable with storage encryption.
  639 
  640 
  641   * 2.9 How do I read a dm-crypt key from file?
  642 
  643   Use the --key-file option, like this:
  644 
  645     cryptsetup create --key-file keyfile e1 /dev/loop0
  646 
  647   This will read the binary key from file, i.e.  no hashing or
  648   transformation will be applied to the keyfile before its bits are used
  649   as key.  Extra bits (beyond the length of the key) at the end are
  650   ignored.  Note that if you read from STDIN, the data will be hashed,
  651   just as a key read interactively from the terminal.  See the man-page
  652   sections "NOTES ON PASSPHRASE PROCESSING..." for more detail.
  653 
  654 
  655   * 2.10 How do I read a LUKS slot key from file?
  656 
  657   What you really do here is to read a passphrase from file, just as you
  658   would with manual entry of a passphrase for a key-slot.  You can add a
  659   new passphrase to a free key-slot, set the passphrase of an specific
  660   key-slot or put an already configured passphrase into a file.  Make sure
  661   no trailing newline (0x0a) is contained in the input key file, or the
  662   passphrase will not work because the whole file is used as input.
  663 
  664   To add a new passphrase to a free key slot from file, use something
  665   like this:
  666 
  667     cryptsetup luksAddKey /dev/loop0 keyfile
  668 
  669   To add a new passphrase to a specific key-slot, use something
  670   like this:
  671 
  672     cryptsetup luksAddKey --key-slot 7 /dev/loop0 keyfile
  673 
  674   To supply a key from file to any LUKS command, use the --key-file
  675   option, e.g. like this:
  676 
  677     cryptsetup luksOpen --key-file keyfile /dev/loop0 e1
  678 
  679 
  680 
  681   * 2.11 How do I read the LUKS master key from file?
  682 
  683   The question you should ask yourself first is why you would want to do
  684   this.  The only legitimate reason I can think of is if you want to have
  685   two LUKS devices with the same master key.  Even then, I think it would
  686   be preferable to just use key-slots with the same passphrase, or to use
  687   plain dm-crypt instead.  If you really have a good reason, please tell
  688   me.  If I am convinced, I will add how to do this here.
  689 
  690 
  691   * 2.12 What are the security requirements for a key read from file?
  692 
  693   A file-stored key or passphrase has the same security requirements as
  694   one entered interactively, however you can use random bytes and thereby
  695   use bytes you cannot type on the keyboard.  You can use any file you
  696   like as key file, for example a plain text file with a human readable
  697   passphrase.  To generate a file with random bytes, use something like
  698   this:
  699 
  700     head -c 256 /dev/random > keyfile
  701 
  702 
  703 
  704   * 2.13 If I map a journaled file system using dm-crypt/LUKS, does
  705     it still provide its usual transactional guarantees?
  706 
  707   Yes, it does, unless a very old kernel is used.  The required flags come
  708   from the filesystem layer and are processed and passed onward by
  709   dm-crypt (regardless of direct key management or LUKS key management). 
  710   A bit more information on the process by which transactional guarantees
  711   are implemented can be found here:
  712 
  713   https://lwn.net/Articles/400541/
  714 
  715   Please note that these "guarantees" are weaker than they appear to be. 
  716   One problem is that quite a few disks lie to the OS about having flushed
  717   their buffers.  This is likely still true with SSDs.  Some other things
  718   can go wrong as well.  The filesystem developers are aware of these
  719   problems and typically can make it work anyways.  That said,
  720   dm-crypt/LUKS will not make things worse.
  721 
  722   One specific problem you can run into is that you can get short freezes
  723   and other slowdowns due to the encryption layer.  Encryption takes time
  724   and forced flushes will block for that time.  For example, I did run
  725   into frequent small freezes (1-2 sec) when putting a vmware image on
  726   ext3 over dm-crypt.  When I went back to ext2, the problem went away. 
  727   This seems to have gotten better with kernel 2.6.36 and the reworking of
  728   filesystem flush locking mechanism (less blocking of CPU activity during
  729   flushes).  This should improve further and eventually the problem should
  730   go away.
  731 
  732 
  733   * 2.14 Can I use LUKS or cryptsetup with a more secure (external)
  734     medium for key storage, e.g. TPM or a smartcard?
  735 
  736   Yes, see the answers on using a file-supplied key.  You do have to write
  737   the glue-logic yourself though.  Basically you can have cryptsetup read
  738   the key from STDIN and write it there with your own tool that in turn
  739   gets the key from the more secure key storage.
  740 
  741 
  742   * 2.15 Can I resize a dm-crypt or LUKS container?
  743 
  744   Yes, you can, as neither dm-crypt nor LUKS1 stores partition size and
  745   LUKS2 uses a generic "whole device" size as default.  Note that LUKS2
  746   can use specified data-area sizes as a non-standard case and that these
  747   may cause issues when resizing a LUKS2 container if set to a specific
  748   value.
  749 
  750   Whether you should do this is a different question.  Personally I
  751   recommend backup, recreation of the dm-crypt or LUKS container with new
  752   size, recreation of the filesystem and restore.  This gets around the
  753   tricky business of resizing the filesystem.  Resizing a dm-crypt or LUKS
  754   container does not resize the filesystem in it.  A backup is really
  755   non-optional here, as a lot can go wrong, resulting in partial or
  756   complete data loss.  But if you have that backup, you can also just
  757   recreate everything.
  758 
  759   You also need to be aware of size-based limitations.  The one currently
  760   relevant is that aes-xts-plain should not be used for encrypted
  761   container sizes larger than 2TiB.  Use aes-xts-plain64 for that.
  762 
  763 
  764   * 2.16 How do I Benchmark the Ciphers, Hashes and Modes?
  765 
  766   Since version 1.60 cryptsetup supports the "benchmark" command. 
  767   Simply run as root:
  768 
  769     cryptsetup benchmark
  770 
  771   You can get more than the default benchmarks, see the man-page for the
  772   relevant parameters.  Note that XTS mode takes two keys, hence the
  773   listed key sizes are double that for other modes and half of it is the
  774   cipher key, the other half is the XTS key.
  775 
  776 
  777   * 2.17 How do I Verify I have an Authentic cryptsetup Source Package?
  778 
  779   Current maintainer is Milan Broz and he signs the release packages with
  780   his PGP key.  The key he currently uses is the "RSA key ID D93E98FC",
  781   fingerprint 2A29 1824 3FDE 4664 8D06 86F9 D9B0 577B D93E 98FC.  While I
  782   have every confidence this really is his key and that he is who he
  783   claims to be, don't depend on it if your life is at stake.  For that
  784   matter, if your life is at stake, don't depend on me being who I claim
  785   to be either.
  786 
  787   That said, as cryptsetup is under good version control and a malicious
  788   change should be noticed sooner or later, but it may take a while. 
  789   Also, the attacker model makes compromising the sources in a non-obvious
  790   way pretty hard.  Sure, you could put the master-key somewhere on disk,
  791   but that is rather obvious as soon as somebody looks as there would be
  792   data in an empty LUKS container in a place it should not be.  Doing this
  793   in a more nefarious way, for example hiding the master-key in the salts,
  794   would need a look at the sources to be discovered, but I think that
  795   somebody would find that sooner or later as well.
  796 
  797   That said, this discussion is really a lot more complicated and longer
  798   as an FAQ can sustain.  If in doubt, ask on the mailing list.
  799 
  800 
  801   * 2.18 Is there a concern with 4k Sectors?
  802 
  803   Not from dm-crypt itself.  Encryption will be done in 512B blocks, but
  804   if the partition and filesystem are aligned correctly and the filesystem
  805   uses multiples of 4kiB as block size, the dm-crypt layer will just
  806   process 8 x 512B = 4096B at a time with negligible overhead.  LUKS does
  807   place data at an offset, which is 2MiB per default and will not break
  808   alignment.  See also Item 6.12 of this FAQ for more details.  Note that
  809   if your partition or filesystem is misaligned, dm-crypt can make the
  810   effect worse though.  Also note that SSDs typically have much larger
  811   blocks internally (e.g.  128kB or even larger).
  812 
  813 
  814   * 2.19 How can I wipe a device with crypto-grade randomness?
  815 
  816   The conventional recommendation if you want to do more than just a
  817   zero-wipe is to use something like
  818 
  819     cat /dev/urandom >  <target-device>
  820 
  821   That used to very slow and painful at 10-20MB/s on a fast computer, but
  822   newer kernels can give you > 200MB/s (depending on hardware).  An
  823   alternative is using cryptsetup and a plain dm-crypt device with a
  824   random key, which is fast and on the same level of security.  The
  825   defaults are quite enough.
  826 
  827   For device set-up, do the following:
  828 
  829     cryptsetup open --type plain -d /dev/urandom /dev/<device> target
  830 
  831   This maps the container as plain under /dev/mapper/target with a random
  832   password.  For the actual wipe you have several options.  Basically, you
  833   pipe zeroes into the opened container that then get encrypted.  Simple
  834   wipe without progress-indicator:
  835 
  836     cat /dev/zero > /dev/mapper/to_be_wiped
  837 
  838   Progress-indicator by dd_rescue:
  839 
  840     dd_rescue -w /dev/zero /dev/mapper/to_be_wiped
  841 
  842   Progress-indicator by my "wcs" stream meter (available from
  843   https://www.tansi.org/tools/index.html ):
  844 
  845     cat /dev/zero | wcs > /dev/mapper/to_be_wiped
  846 
  847   Or use plain "dd", which gives you the progress when sent a SIGUSR1, see
  848   the dd man page.
  849 
  850   Remove the mapping at the end and you are done.
  851 
  852 
  853   * 2.20 How do I wipe only the LUKS header?
  854  
  855   This does _not_ describe an emergency wipe procedure, see Item 5.4 for
  856   that.  This procedure here is intended to be used when the data should
  857   stay intact, e.g.  when you change your LUKS container to use a detached
  858   header and want to remove the old one.  Please only do this if you have
  859   a current backup.
  860 
  861   LUKS1:  
  862   01) Determine header size in 512 Byte sectors with luksDump:
  863 
  864      cryptsetup luksDump <device with LUKS container>
  865 
  866 ->   ...
  867      Payload offset: <number> [of 512 byte sectors]
  868      ...
  869 
  870   02) Take the result number, multiply by 512 zeros and write to 
  871       the start of the device, e.g. using one of the following alternatives:
  872 
  873      dd bs=512 count=<number> if=/dev/zero of=<device>
  874 
  875 
  876      head -c <number * 512> /dev/zero > /dev/<device>
  877 
  878 
  879   LUKS2:  
  880   (warning, untested!  Remember that backup?) This assumes the
  881   LUKS2 container uses the defaults, in particular there is only one data
  882   segment.  
  883   01) Determine the data-segment offset using luksDump, same
  884       as above for LUKS1:
  885 
  886      cryptsetup luksDump <device with LUKS container>
  887 ->   ...  
  888      Data segments:
  889         0: crypt
  890            offset: <number> [bytes]
  891      ...
  892 
  893   02) Overwrite the stated number of bytes from the start of the device.
  894       Just to give yet another way to get a defined number of zeros:
  895 
  896      head -c <number> /dev/zero > /dev/<device>
  897 
  898 
  899 3. Common Problems
  900 
  901 
  902   * 3.1 My dm-crypt/LUKS mapping does not work! What general steps
  903     are there to investigate the problem?
  904 
  905   If you get a specific error message, investigate what it claims first. 
  906   If not, you may want to check the following things.
  907 
  908   - Check that "/dev", including "/dev/mapper/control" is there.  If it is
  909   missing, you may have a problem with the "/dev" tree itself or you may
  910   have broken udev rules.
  911 
  912   - Check that you have the device mapper and the crypt target in your
  913   kernel.  The output of "dmsetup targets" should list a "crypt" target. 
  914   If it is not there or the command fails, add device mapper and
  915   crypt-target to the kernel.
  916 
  917   - Check that the hash-functions and ciphers you want to use are in the
  918   kernel.  The output of "cat /proc/crypto" needs to list them.
  919 
  920 
  921   * 3.2 My dm-crypt mapping suddenly stopped when upgrading cryptsetup.
  922 
  923   The default cipher, hash or mode may have changed (the mode changed from
  924   1.0.x to 1.1.x).  See under "Issues With Specific Versions of
  925   cryptsetup".
  926 
  927 
  928   * 3.3 When I call cryptsetup from cron/CGI, I get errors about
  929     unknown features?
  930 
  931   If you get errors about unknown parameters or the like that are not
  932   present when cryptsetup is called from the shell, make sure you have no
  933   older version of cryptsetup on your system that then gets called by
  934   cron/CGI.  For example some distributions install cryptsetup into
  935   /usr/sbin, while a manual install could go to /usr/local/sbin.  As a
  936   debugging aid, call "cryptsetup --version" from cron/CGI or the
  937   non-shell mechanism to be sure the right version gets called.
  938 
  939 
  940   * 3.4 Unlocking a LUKS device takes very long. Why?
  941 
  942   The unlock time for a key-slot (see Section 5 for an explanation what
  943   iteration does) is calculated when setting a passphrase.  By default it
  944   is 1 second (2 seconds for LUKS2).  If you set a passphrase on a fast
  945   machine and then unlock it on a slow machine, the unlocking time can be
  946   much longer.  Also take into account that up to 8 key-slots (LUKS2: up
  947   to 32 key-slots) have to be tried in order to find the right one.
  948 
  949   If this is the problem, you can add another key-slot using the slow
  950   machine with the same passphrase and then remove the old key-slot.  The
  951   new key-slot will have the unlock time adjusted to the slow machine.
  952   Use luksKeyAdd and then luksKillSlot or luksRemoveKey.  You can also use
  953   the -i option to reduce iteration time (and security level) when setting 
  954   a passphrase.  Default is 1000 (1 sec) for LUKS1 and 2000 (2sec) for
  955   LUKS2.
  956 
  957   However, this operation will not change volume key iteration count ("MK
  958   iterations" for LUKS1, "Iterations" under "Digests" for LUKS2).  In
  959   order to change that, you will have to backup the data in the LUKS
  960   container (i.e.  your encrypted data), luksFormat on the slow machine
  961   and restore the data.  Note that MK iterations are not very security
  962   relevant.
  963 
  964 
  965   * 3.5 "blkid" sees a LUKS UUID and an ext2/swap UUID on the same
  966     device. What is wrong?
  967 
  968   Some old versions of cryptsetup have a bug where the header does not get
  969   completely wiped during LUKS format and an older ext2/swap signature
  970   remains on the device.  This confuses blkid.
  971 
  972   Fix: Wipe the unused header areas by doing a backup and restore of
  973   the header with cryptsetup 1.1.x or later:
  974 
  975     cryptsetup luksHeaderBackup --header-backup-file <file> <device>
  976     cryptsetup luksHeaderRestore --header-backup-file <file> <device>
  977 
  978 
  979 
  980 4. Troubleshooting
  981 
  982 
  983   * 4.1 I get the error "LUKS keyslot x is invalid." What does that mean?
  984 
  985   For LUKS1, this means that the given keyslot has an offset that points
  986   outside the valid keyslot area.  Typically, the reason is a corrupted
  987   LUKS1 header because something was written to the start of the device
  988   the LUKS1 container is on.  For LUKS2, I do not know when this error can
  989   happen, but I expect it will be something similar.  Refer to Section
  990   "Backup and Data Recovery" and ask on the mailing list if you have
  991   trouble diagnosing and (if still possible) repairing this.
  992 
  993 
  994   * 4.2 I cannot unlock my LUKS container! What could be the problem?
  995 
  996   First, make sure you have a correct passphrase.  Then make sure you have
  997   the correct key-map and correct keyboard.  And then make sure you have
  998   the correct character set and encoding, see also "PASSPHRASE CHARACTER
  999   SET" under Section 1.2.
 1000 
 1001   If you are sure you are entering the passphrase right, there is the
 1002   possibility that the respective key-slot has been damaged.  There is no
 1003   way to recover a damaged key-slot, except from a header backup (see
 1004   Section 6).  For security reasons, there is also no checksum in the
 1005   key-slots that could tell you whether a key-slot has been damaged.  The
 1006   only checksum present allows recognition of a correct passphrase, but
 1007   that only works with that correct passphrase and a respective key-slot
 1008   that is intact.
 1009 
 1010   In order to find out whether a key-slot is damaged one has to look for
 1011   "non-random looking" data in it.  There is a tool that automatizes this
 1012   for LUKS1 in the cryptsetup distribution from version 1.6.0 onwards.  It
 1013   is located in misc/keyslot_checker/.  Instructions how to use and how to
 1014   interpret results are in the README file.  Note that this tool requires
 1015   a libcryptsetup from cryptsetup 1.6.0 or later (which means
 1016   libcryptsetup.so.4.5.0 or later).  If the tool complains about missing
 1017   functions in libcryptsetup, you likely have an earlier version from your
 1018   distribution still installed.  You can either point the symbolic link(s)
 1019   from libcryptsetup.so.4 to the new version manually, or you can
 1020   uninstall the distribution version of cryptsetup and re-install that
 1021   from cryptsetup >= 1.6.0 again to fix this.
 1022 
 1023 
 1024   * 4.3 Can a bad RAM module cause problems?
 1025 
 1026   LUKS and dm-crypt can give the RAM quite a workout, especially when
 1027   combined with software RAID.  In particular the combination RAID5 +
 1028   LUKS1 + XFS seems to uncover RAM problems that do not cause obvious
 1029   problems otherwise.  Symptoms vary, but often the problem manifests
 1030   itself when copying large amounts of data, typically several times
 1031   larger than your main memory.
 1032 
 1033   Note: One thing you should always do on large data copying or movements
 1034   is to run a verify, for example with the "-d" option of "tar" or by
 1035   doing a set of MD5 checksums on the source or target with
 1036 
 1037     find . -type f -exec md5sum \{\} \; > checksum-file
 1038 
 1039   and then a "md5sum -c checksum-file" on the other side.  If you get
 1040   mismatches here, RAM is the primary suspect.  A lesser suspect is an
 1041   overclocked CPU.  I have found countless hardware problems in verify
 1042   runs after copying data or making backups.  Bit errors are much more
 1043   common than most people think.
 1044 
 1045   Some RAM issues are even worse and corrupt structures in one of the
 1046   layers.  This typically results in lockups, CPU state dumps in the
 1047   system logs, kernel panic or other things.  It is quite possible to have
 1048   a problem with an encrypted device, but not with an otherwise the same
 1049   unencrypted device.  The reason for that is that encryption has an error
 1050   amplification property: If you flip one bit in an encrypted data block,
 1051   the decrypted version has half of its bits flipped.  This is actually an
 1052   important security property for modern ciphers.  With the usual modes in
 1053   cryptsetup (CBC, ESSIV, XTS), you can get a completely changed 512 byte
 1054   block for a bit error.  A corrupt block causes a lot more havoc than the
 1055   occasionally flipped single bit and can result in various obscure
 1056   errors.
 1057 
 1058   Note that a verify run on copying between encrypted or unencrypted
 1059   devices will reliably detect corruption, even when the copying itself
 1060   did not report any problems.  If you find defect RAM, assume all backups
 1061   and copied data to be suspect, unless you did a verify.
 1062 
 1063 
 1064   * 4.4 How do I test RAM?
 1065 
 1066   First you should know that overclocking often makes memory problems
 1067   worse.  So if you overclock (which I strongly recommend against in a
 1068   system holding data that has any worth), run the tests with the
 1069   overclocking active.
 1070 
 1071   There are two good options.  One is Memtest86+ and the other is
 1072   "memtester" by Charles Cazabon.  Memtest86+ requires a reboot and then
 1073   takes over the machine, while memtester runs from a root-shell.  Both
 1074   use different testing methods and I have found problems fast with either
 1075   one that the other needed long to find.  I recommend running the
 1076   following procedure until the first error is found:
 1077 
 1078   - Run Memtest86+ for one cycle
 1079 
 1080   - Run memtester for one cycle (shut down as many other applications
 1081     as possible and use the largest memory area you can get)
 1082 
 1083   - Run Memtest86+ for 24h or more
 1084 
 1085   - Run memtester for 24h or more
 1086 
 1087   If all that does not produce error messages, your RAM may be sound,
 1088   but I have had one weak bit in the past that Memtest86+ needed around 
 1089   60 hours to find.  If you can reproduce the original problem reliably, 
 1090   a good additional test may be to remove half of the RAM (if you have 
 1091   more than one module) and try whether the problem is still there and if
 1092   so, try with the other half.  If you just have one module, get a
 1093   different one and try with that.  If you do overclocking, reduce the
 1094   settings to the most conservative ones available and try with that.
 1095 
 1096 
 1097   * 4.5 Is there a risk using debugging tools like strace?
 1098 
 1099   There most definitely is. A dump from strace and friends can contain
 1100   all data entered, including the full passphrase.  Example with strace
 1101   and passphrase "test":
 1102 
 1103     > strace cryptsetup luksOpen /dev/sda10 c1
 1104     ...
 1105     read(6, "test\n", 512)                  = 5
 1106     ...
 1107 
 1108   Depending on different factors and the tool used, the passphrase may
 1109   also be encoded and not plainly visible.  Hence it is never a good idea
 1110   to give such a trace from a live container to anybody.  Recreate the
 1111   problem with a test container or set a temporary passphrase like "test"
 1112   and use that for the trace generation.  Item 2.6 explains how to create
 1113   a loop-file backed LUKS container that may come in handy for this
 1114   purpose.
 1115 
 1116   See also Item 6.10 for another set of data you should not give to
 1117   others.
 1118 
 1119 
 1120 5. Security Aspects
 1121 
 1122 
 1123   * 5.1 How long is a secure passphrase?
 1124 
 1125   This is just the short answer.  For more info and explanation of some of
 1126   the terms used in this item, read the rest of Section 5.  The actual
 1127   recommendation is at the end of this item.
 1128 
 1129   First, passphrase length is not really the right measure, passphrase
 1130   entropy is.  If your passphrase is 200 times the letter "a", it is long
 1131   but has very low entropy and is pretty insecure.
 1132 
 1133   For example, a random lowercase letter (a-z) gives you 4.7 bit of
 1134   entropy, one element of a-z0-9 gives you 5.2 bits of entropy, an element
 1135   of a-zA-Z0-9 gives you 5.9 bits and a-zA-Z0-9!@#$%\^&:-+ gives you 6.2
 1136   bits.  On the other hand, a random English word only gives you 0.6...1.3
 1137   bits of entropy per character.  Using sentences that make sense gives
 1138   lower entropy, series of random words gives higher entropy.  Do not use
 1139   sentences that can be tied to you or found on your computer.  This type
 1140   of attack is done routinely today.
 1141 
 1142   That said, it does not matter too much what scheme you use, but it does
 1143   matter how much entropy your passphrase contains, because an attacker
 1144   has to try on average
 1145 
 1146     1/2 * 2^(bits of entropy in passphrase)
 1147 
 1148   different passphrases to guess correctly.
 1149 
 1150   Historically, estimations tended to use computing time estimates, but
 1151   more modern approaches try to estimate cost of guessing a passphrase.
 1152 
 1153   As an example, I will try to get an estimate from the numbers in
 1154   https://gist.github.com/epixoip/a83d38f412b4737e99bbef804a270c40 This
 1155   thing costs 23kUSD and does 68Ghashes/sec for SHA1.  This is in 2017.
 1156  
 1157   Incidentally, my older calculation for a machine around 1000 times
 1158   slower was off by a factor of about 1000, but in the right direction,
 1159   i.e.  I estimated the attack to be too easy.  Nobody noticed ;-) On the
 1160   plus side, the tables are now (2017) pretty much accurate.
 1161 
 1162   More references can be found at the end of this document.  Note that
 1163   these are estimates from the defender side, so assuming something is
 1164   easier than it actually is is fine.  An attacker may still have
 1165   significantly higher cost than estimated here.
 1166 
 1167   LUKS1 used SHA1 (since version 1.7.0 it uses SHA256) for hashing per
 1168   default.  We will leave aside the check whether a try actually decrypts 
 1169   a key-slot.  I will assume a useful lifetime of the hardware of 2 years. 
 1170   (This is on the low side.) Disregarding downtime, the machine can then
 1171   break
 1172 
 1173      N = 68*10^9 * 3600 * 24 * 365 * 2 ~ 4*10^18
 1174 
 1175   passphrases for EUR/USD 23k.  That is one 62 bit passphrase hashed once
 1176   with SHA1 for EUR/USD 23k.  This can be parallelized, it can be done
 1177   faster than 2 years with several of these machines.
 1178 
 1179   For LUKS2, things look a bit better, as the advantage of using graphics
 1180   cards is massively reduced.  Using the recommendations below should
 1181   hence be fine for LUKS2 as well and give a better security margin.
 1182 
 1183   For plain dm-crypt (no hash iteration) this is it.  This gives (with
 1184   SHA1, plain dm-crypt default is ripemd160 which seems to be slightly
 1185   slower than SHA1):
 1186 
 1187     Passphrase entropy  Cost to break
 1188     60 bit              EUR/USD     6k
 1189     65 bit              EUR/USD   200K
 1190     70 bit              EUR/USD     6M
 1191     75 bit              EUR/USD   200M
 1192     80 bit              EUR/USD     6B
 1193     85 bit              EUR/USD   200B
 1194     ...                      ...
 1195 
 1196 
 1197   For LUKS1, you have to take into account hash iteration in PBKDF2. 
 1198   For a current CPU, there are about 100k iterations (as can be queried
 1199   with ''cryptsetup luksDump''. 
 1200 
 1201   The table above then becomes:
 1202 
 1203     Passphrase entropy  Cost to break
 1204     50 bit              EUR/USD   600k
 1205     55 bit              EUR/USD    20M
 1206     60 bit              EUR/USD   600M
 1207     65 bit              EUR/USD    20B
 1208     70 bit              EUR/USD   600B
 1209     75 bit              EUR/USD    20T
 1210     ...                      ...
 1211 
 1212 
 1213   Recommendation:
 1214 
 1215   To get reasonable security for the  next 10 years, it is a good idea
 1216   to overestimate by a factor of at least 1000.
 1217 
 1218   Then there is the question of how much the attacker is willing to spend. 
 1219   That is up to your own security evaluation.  For general use, I will
 1220   assume the attacker is willing to spend up to 1 million EUR/USD.  Then
 1221   we get the following recommendations:
 1222 
 1223   Plain dm-crypt: Use > 80 bit. That is e.g. 17 random chars from a-z
 1224   or a random English sentence of > 135 characters length.
 1225 
 1226   LUKS1 and LUKS2: Use > 65 bit. That is e.g. 14 random chars from a-z 
 1227   or a random English sentence of > 108 characters length.
 1228 
 1229   If paranoid, add at least 20 bit. That is roughly four additional
 1230   characters for random passphrases and roughly 32 characters for a
 1231   random English sentence. 
 1232 
 1233 
 1234   * 5.2 Is LUKS insecure? Everybody can see I have encrypted data!
 1235 
 1236   In practice it does not really matter.  In most civilized countries you
 1237   can just refuse to hand over the keys, no harm done.  In some countries
 1238   they can force you to hand over the keys if they suspect encryption. 
 1239   The suspicion is enough, they do not have to prove anything.  This is
 1240   for practical reasons, as even the presence of a header (like the LUKS
 1241   header) is not enough to prove that you have any keys.  It might have
 1242   been an experiment, for example.  Or it was used as encrypted swap with
 1243   a key from /dev/random.  So they make you prove you do not have
 1244   encrypted data.  Of course, if true, that is impossible and hence the
 1245   whole idea is not compatible with fair laws.  Note that in this context,
 1246   countries like the US or the UK are not civilized and do not have fair
 1247   laws.
 1248 
 1249   As a side-note, standards for biometrics (fingerprint, retina, 
 1250   vein-pattern, etc.) are often different and much lower. If you put
 1251   your LUKS passphrase into a device that can be unlocked using biometrics,
 1252   they may force a biometric sample in many countries where they could not
 1253   force you to give them a passphrase you solely have in your memory and
 1254   can claim to have forgotten if needed (it happens). If you need protection
 1255   on this level, make sure you know what the respective legal situation is,
 1256   also while traveling, and make sure you decide beforehand what you
 1257   will do if push comes to shove as they will definitely put you under
 1258   as much pressure as they can legally apply. 
 1259 
 1260   This means that if you have a large set of random-looking data, they can
 1261   already lock you up.  Hidden containers (encryption hidden within
 1262   encryption), as possible with Truecrypt, do not help either.  They will
 1263   just assume the hidden container is there and unless you hand over the
 1264   key, you will stay locked up.  Don't have a hidden container?  Tough
 1265   luck.  Anybody could claim that.
 1266 
 1267   Still, if you are concerned about the LUKS header, use plain dm-crypt
 1268   with a good passphrase.  See also Section 2, "What is the difference
 1269   between "plain" and LUKS format?"
 1270 
 1271 
 1272   * 5.3 Should I initialize (overwrite) a new LUKS/dm-crypt partition?
 1273 
 1274   If you just create a filesystem on it, most of the old data will still
 1275   be there.  If the old data is sensitive, you should overwrite it before
 1276   encrypting.  In any case, not initializing will leave the old data there
 1277   until the specific sector gets written.  That may enable an attacker to
 1278   determine how much and where on the partition data was written.  If you
 1279   think this is a risk, you can prevent this by overwriting the encrypted
 1280   device (here assumed to be named "e1") with zeros like this:
 1281 
 1282     dd_rescue -w /dev/zero /dev/mapper/e1
 1283 
 1284   or alternatively with one of the following more standard commands:
 1285 
 1286     cat /dev/zero > /dev/mapper/e1
 1287     dd if=/dev/zero of=/dev/mapper/e1
 1288 
 1289 
 1290 
 1291   * 5.4 How do I securely erase a LUKS container?
 1292 
 1293   For LUKS, if you are in a desperate hurry, overwrite the LUKS header and
 1294   key-slot area.  For LUKS1 and LUKS2, just be generous and overwrite the
 1295   first 100MB.  A single overwrite with zeros should be enough.  If you
 1296   anticipate being in a desperate hurry, prepare the command beforehand. 
 1297   Example with /dev/sde1 as the LUKS partition and default parameters:
 1298 
 1299     head -c 100000000 /dev/zero > /dev/sde1; sync
 1300 
 1301   A LUKS header backup or full backup will still grant access to most or
 1302   all data, so make sure that an attacker does not have access to backups
 1303   or destroy them as well.
 1304 
 1305   Also note that SSDs and also some HDDs (SMR and hybrid HDDs, for
 1306   example) may not actually overwrite the header and only do that an
 1307   unspecified and possibly very long time later.  The only way to be sure
 1308   there is physical destruction.  If the situation permits, do both
 1309   overwrite and physical destruction.
 1310 
 1311   If you have time, overwrite the whole drive with a single pass of random
 1312   data.  This is enough for most HDDs.  For SSDs or FLASH (USB sticks) or
 1313   SMR or hybrid drives, you may want to overwrite the whole drive several
 1314   times to be sure data is not retained.  This is possibly still insecure
 1315   as the respective technologies are not fully understood in this regard. 
 1316   Still, due to the anti-forensic properties of the LUKS key-slots, a
 1317   single overwrite could be enough.  If in doubt, use physical destruction
 1318   in addition.  Here is a link to some current research results on erasing
 1319   SSDs and FLASH drives:
 1320   https://www.usenix.org/events/fast11/tech/full_papers/Wei.pdf
 1321 
 1322   Keep in mind to also erase all backups.
 1323 
 1324   Example for a random-overwrite erase of partition sde1 done with
 1325   dd_rescue:
 1326 
 1327     dd_rescue -w /dev/urandom /dev/sde1
 1328 
 1329 
 1330 
 1331   * 5.5 How do I securely erase a backup of a LUKS partition or header?
 1332 
 1333   That depends on the medium it is stored on.  For HDD and SSD, use
 1334   overwrite with random data.  For an SSD, FLASH drive (USB stick) hybrid
 1335   HDD or SMR HDD, you may want to overwrite the complete drive several
 1336   times and use physical destruction in addition, see last item.  For
 1337   re-writable CD/DVD, a single overwrite should be enough, due to the
 1338   anti-forensic properties of the LUKS keyslots.  For write-once media,
 1339   use physical destruction.  For low security requirements, just cut the
 1340   CD/DVD into several parts.  For high security needs, shred or burn the
 1341   medium.
 1342 
 1343   If your backup is on magnetic tape, I advise physical destruction by
 1344   shredding or burning, after (!) overwriting.  The problem with magnetic
 1345   tape is that it has a higher dynamic range than HDDs and older data may
 1346   well be recoverable after overwrites.  Also write-head alignment issues
 1347   can lead to data not actually being deleted during overwrites.
 1348 
 1349   The best option is to actually encrypt the backup, for example with
 1350   PGP/GnuPG and then just destroy all copies of the encryption key if
 1351   needed.  Best keep them on paper, as that has excellent durability and
 1352   secure destruction is easy, for example by burning and then crushing the
 1353   ashes to a fine powder.  A blender and water also works nicely.
 1354 
 1355 
 1356   * 5.6 What about backup? Does it compromise security?
 1357 
 1358   That depends. See item 6.7.
 1359 
 1360 
 1361   * 5.7 Why is all my data permanently gone if I overwrite the LUKS header?
 1362 
 1363   Overwriting the LUKS header in part or in full is the most common reason
 1364   why access to LUKS containers is lost permanently.  Overwriting can be
 1365   done in a number of fashions, like creating a new filesystem on the raw
 1366   LUKS partition, making the raw partition part of a RAID array and just
 1367   writing to the raw partition.
 1368 
 1369   The LUKS1 header contains a 256 bit "salt" per key-slot and without that
 1370   no decryption is possible.  While the salts are not secret, they are
 1371   key-grade material and cannot be reconstructed.  This is a
 1372   cryptographically strong "cannot".  From observations on the cryptsetup
 1373   mailing-list, people typically go though the usual stages of grief
 1374   (Denial, Anger, Bargaining, Depression, Acceptance) when this happens to
 1375   them.  Observed times vary between 1 day and 2 weeks to complete the
 1376   cycle.  Seeking help on the mailing-list is fine.  Even if we usually
 1377   cannot help with getting back your data, most people found the feedback
 1378   comforting.
 1379 
 1380   If your header does not contain an intact key-slot salt, best go
 1381   directly to the last stage ("Acceptance") and think about what to do
 1382   now.  There is one exception that I know of: If your LUKS1 container is
 1383   still open, then it may be possible to extract the master key from the
 1384   running system.  See Item "How do I recover the master key from a mapped
 1385   LUKS1 container?" in Section "Backup and Data Recovery".
 1386 
 1387   For LUKS2, things are both better and worse.  First, the salts are in a
 1388   less vulnerable position now.  But, on the other hand, the keys of a
 1389   mapped (open) container are now stored in the kernel key-store, and
 1390   while there probably is some way to get them out of there, I am not sure
 1391   how much effort that needs.
 1392 
 1393 
 1394   * 5.8 What is a "salt"?
 1395 
 1396   A salt is a random key-grade value added to the passphrase before it is
 1397   processed.  It is not kept secret.  The reason for using salts is as
 1398   follows: If an attacker wants to crack the password for a single LUKS
 1399   container, then every possible passphrase has to be tried.  Typically an
 1400   attacker will not try every binary value, but will try words and
 1401   sentences from a dictionary.
 1402 
 1403   If an attacker wants to attack several LUKS containers with the same
 1404   dictionary, then a different approach makes sense: Compute the resulting
 1405   slot-key for each dictionary element and store it on disk.  Then the
 1406   test for each entry is just the slow unlocking with the slot key (say
 1407   0.00001 sec) instead of calculating the slot-key first (1 sec).  For a
 1408   single attack, this does not help.  But if you have more than one
 1409   container to attack, this helps tremendously, also because you can
 1410   prepare your table before you even have the container to attack!  The
 1411   calculation is also very simple to parallelize.  You could, for example,
 1412   use the night-time unused CPU power of your desktop PCs for this.
 1413 
 1414   This is where the salt comes in.  If the salt is combined with the
 1415   passphrase (in the simplest form, just appended to it), you suddenly
 1416   need a separate table for each salt value.  With a reasonably-sized salt
 1417   value (256 bit, e.g.) this is quite infeasible.
 1418 
 1419 
 1420   * 5.9 Is LUKS secure with a low-entropy (bad) passphrase?
 1421 
 1422   Short answer: yes. Do not use a low-entropy passphrase.
 1423 
 1424   Note: For LUKS2, protection for bad passphrases is a bit better
 1425   due to the use of Argon2, but that is only a gradual improvement.
 1426 
 1427   Longer answer:  
 1428   This needs a bit of theory.  The quality of your passphrase is directly
 1429   related to its entropy (information theoretic, not thermodynamic).  The
 1430   entropy says how many bits of "uncertainty" or "randomness" are in you
 1431   passphrase.  In other words, that is how difficult guessing the
 1432   passphrase is.
 1433 
 1434   Example: A random English sentence has about 1 bit of entropy per
 1435   character.  A random lowercase (or uppercase) character has about 4.7
 1436   bit of entropy.
 1437 
 1438   Now, if n is the number of bits of entropy in your passphrase and t
 1439   is the time it takes to process a passphrase in order to open the
 1440   LUKS container, then an attacker has to spend at maximum
 1441 
 1442     attack_time_max = 2^n * t
 1443 
 1444   time for a successful attack and on average half that.  There is no way
 1445   getting around that relationship.  However, there is one thing that does
 1446   help, namely increasing t, the time it takes to use a passphrase, see
 1447   next FAQ item.
 1448 
 1449   Still, if you want good security, a high-entropy passphrase is the only
 1450   option.  For example, a low-entropy passphrase can never be considered
 1451   secure against a TLA-level (Three Letter Agency level, i.e. 
 1452   government-level) attacker, no matter what tricks are used in the
 1453   key-derivation function.  Use at least 64 bits for secret stuff.  That
 1454   is 64 characters of English text (but only if randomly chosen) or a
 1455   combination of 12 truly random letters and digits.
 1456 
 1457   For passphrase generation, do not use lines from very well-known texts
 1458   (religious texts, Harry Potter, etc.) as they are too easy to guess.
 1459   For example, the total Harry Potter has about 1'500'000 words (my
 1460   estimation).  Trying every 64 character sequence starting and ending at
 1461   a word boundary would take only something like 20 days on a single CPU
 1462   and is entirely feasible.  To put that into perspective, using a number
 1463   of Amazon EC2 High-CPU Extra Large instances (each gives about 8 real
 1464   cores), this test costs currently about 50USD/EUR, but can be made to
 1465   run arbitrarily fast.
 1466 
 1467   On the other hand, choosing 1.5 lines from, say, the Wheel of Time, is
 1468   in itself not more secure, but the book selection adds quite a bit of
 1469   entropy.  (Now that I have mentioned it here, don't use tWoT either!) If
 1470   you add 2 or 3 typos and switch some words around, then this is good
 1471   passphrase material.
 1472 
 1473 
 1474   * 5.10 What is "iteration count" and why is decreasing it a bad idea?
 1475 
 1476   LUKS1:  
 1477   Iteration count is the number of PBKDF2 iterations a passphrase is put
 1478   through before it is used to unlock a key-slot.  Iterations are done
 1479   with the explicit purpose to increase the time that it takes to unlock a
 1480   key-slot.  This provides some protection against use of low-entropy
 1481   passphrases.
 1482 
 1483   The idea is that an attacker has to try all possible passphrases.  Even
 1484   if the attacker knows the passphrase is low-entropy (see last item), it
 1485   is possible to make each individual try take longer.  The way to do this
 1486   is to repeatedly hash the passphrase for a certain time.  The attacker
 1487   then has to spend the same time (given the same computing power) as the
 1488   user per try.  With LUKS1, the default is 1 second of PBKDF2 hashing.
 1489 
 1490   Example 1: Lets assume we have a really bad passphrase (e.g.  a
 1491   girlfriends name) with 10 bits of entropy.  With the same CPU, an
 1492   attacker would need to spend around 500 seconds on average to break that
 1493   passphrase.  Without iteration, it would be more like 0.0001 seconds on
 1494   a modern CPU.
 1495 
 1496   Example 2: The user did a bit better and has 32 chars of English text. 
 1497   That would be about 32 bits of entropy.  With 1 second iteration, that
 1498   means an attacker on the same CPU needs around 136 years.  That is
 1499   pretty impressive for such a weak passphrase.  Without the iterations,
 1500   it would be more like 50 days on a modern CPU, and possibly far less.
 1501 
 1502   In addition, the attacker can both parallelize and use special hardware
 1503   like GPUs or FPGAs to speed up the attack.  The attack can also happen
 1504   quite some time after the luksFormat operation and CPUs can have become
 1505   faster and cheaper.  For that reason you want a bit of extra security. 
 1506   Anyways, in Example 1 your are screwed.  In example 2, not necessarily. 
 1507   Even if the attack is faster, it still has a certain cost associated
 1508   with it, say 10000 EUR/USD with iteration and 1 EUR/USD without
 1509   iteration.  The first can be prohibitively expensive, while the second
 1510   is something you try even without solid proof that the decryption will
 1511   yield something useful.
 1512 
 1513   The numbers above are mostly made up, but show the idea.  Of course the
 1514   best thing is to have a high-entropy passphrase.
 1515 
 1516   Would a 100 sec iteration time be even better?  Yes and no. 
 1517   Cryptographically it would be a lot better, namely 100 times better. 
 1518   However, usability is a very important factor for security technology
 1519   and one that gets overlooked surprisingly often.  For LUKS, if you have
 1520   to wait 2 minutes to unlock the LUKS container, most people will not
 1521   bother and use less secure storage instead.  It is better to have less
 1522   protection against low-entropy passphrases and people actually use LUKS,
 1523   than having them do without encryption altogether.
 1524 
 1525   Now, what about decreasing the iteration time?  This is generally a very
 1526   bad idea, unless you know and can enforce that the users only use
 1527   high-entropy passphrases.  If you decrease the iteration time without
 1528   ensuring that, then you put your users at increased risk, and
 1529   considering how rarely LUKS containers are unlocked in a typical
 1530   work-flow, you do so without a good reason.  Don't do it.  The iteration
 1531   time is already low enough that users with low entropy passphrases are
 1532   vulnerable.  Lowering it even further increases this danger
 1533   significantly.
 1534 
 1535   LUKS2: Pretty much the same reasoning applies. The advantages of using
 1536   GPUs or FPGAs in an attack have been significantly reduced, but that 
 1537   is the only main difference.
 1538 
 1539 
 1540   * 5.11 Some people say PBKDF2 is insecure?
 1541 
 1542   There is some discussion that a hash-function should have a "large
 1543   memory" property, i.e.  that it should require a lot of memory to be
 1544   computed.  This serves to prevent attacks using special programmable
 1545   circuits, like FPGAs, and attacks using graphics cards.  PBKDF2 does not
 1546   need a lot of memory and is vulnerable to these attacks.  However, the
 1547   publication usually referred in these discussions is not very convincing
 1548   in proving that the presented hash really is "large memory" (that may
 1549   change, email the FAQ maintainer when it does) and it is of limited
 1550   usefulness anyways.  Attackers that use clusters of normal PCs will not
 1551   be affected at all by a "large memory" property.  For example the US
 1552   Secret Service is known to use the off-hour time of all the office PCs
 1553   of the Treasury for password breaking.  The Treasury has about 110'000
 1554   employees.  Assuming every one has an office PC, that is significant
 1555   computing power, all of it with plenty of memory for computing "large
 1556   memory" hashes.  Bot-net operators also have all the memory they want. 
 1557   The only protection against a resourceful attacker is a high-entropy
 1558   passphrase, see items 5.9 and 5.10.
 1559 
 1560   That said, LUKS2 defaults to Argon2, which has a large-memory property
 1561   and massively reduces the advantages of GPUs and FPGAs.
 1562 
 1563 
 1564   * 5.12 What about iteration count with plain dm-crypt?
 1565 
 1566   Simple: There is none.  There is also no salting.  If you use plain
 1567   dm-crypt, the only way to be secure is to use a high entropy passphrase. 
 1568   If in doubt, use LUKS instead.
 1569 
 1570 
 1571   * 5.13 Is LUKS with default parameters less secure on a slow CPU?
 1572 
 1573   Unfortunately, yes.  However the only aspect affected is the protection
 1574   for low-entropy passphrase or master-key.  All other security aspects
 1575   are independent of CPU speed.
 1576 
 1577   The master key is less critical, as you really have to work at it to
 1578   give it low entropy.  One possibility to mess this up is to supply the
 1579   master key yourself.  If that key is low-entropy, then you get what you
 1580   deserve.  The other known possibility to create a LUKS container with a
 1581   bad master key is to use /dev/urandom for key generation in an
 1582   entropy-starved situation (e.g.  automatic installation on an embedded
 1583   device without network and other entropy sources or installation in a VM
 1584   under certain circumstances).
 1585 
 1586   For the passphrase, don't use a low-entropy passphrase.  If your
 1587   passphrase is good, then a slow CPU will not matter.  If you insist on a
 1588   low-entropy passphrase on a slow CPU, use something like
 1589   "--iter-time=10000" or higher and wait a long time on each LUKS unlock
 1590   and pray that the attacker does not find out in which way exactly your
 1591   passphrase is low entropy.  This also applies to low-entropy passphrases
 1592   on fast CPUs.  Technology can do only so much to compensate for problems
 1593   in front of the keyboard.
 1594 
 1595   Also note that power-saving modes will make your CPU slower.  This will
 1596   reduce iteration count on LUKS container creation.  It will keep unlock
 1597   times at the expected values though at this CPU speed.
 1598 
 1599 
 1600   * 5.14 Why was the default aes-cbc-plain replaced with aes-cbc-essiv?
 1601 
 1602   Note: This item applies both to plain dm-crypt and to LUKS
 1603 
 1604   The problem is that cbc-plain has a fingerprint vulnerability, where a
 1605   specially crafted file placed into the crypto-container can be
 1606   recognized from the outside.  The issue here is that for cbc-plain the
 1607   initialization vector (IV) is the sector number.  The IV gets XORed to
 1608   the first data chunk of the sector to be encrypted.  If you make sure
 1609   that the first data block to be stored in a sector contains the sector
 1610   number as well, the first data block to be encrypted is all zeros and
 1611   always encrypted to the same ciphertext.  This also works if the first
 1612   data chunk just has a constant XOR with the sector number.  By having
 1613   several shifted patterns you can take care of the case of a
 1614   non-power-of-two start sector number of the file.
 1615 
 1616   This mechanism allows you to create a pattern of sectors that have the
 1617   same first ciphertext block and signal one bit per sector to the
 1618   outside, allowing you to e.g.  mark media files that way for recognition
 1619   without decryption.  For large files this is a practical attack.  For
 1620   small ones, you do not have enough blocks to signal and take care of
 1621   different file starting offsets.
 1622 
 1623   In order to prevent this attack, the default was changed to cbc-essiv. 
 1624   ESSIV uses a keyed hash of the sector number, with the encryption key as
 1625   key.  This makes the IV unpredictable without knowing the encryption key
 1626   and the watermarking attack fails.
 1627 
 1628 
 1629   * 5.15 Are there any problems with "plain" IV? What is "plain64"?
 1630 
 1631   First, "plain" and "plain64" are both not secure to use with CBC, see
 1632   previous FAQ item.
 1633 
 1634   However there are modes, like XTS, that are secure with "plain" IV.  The
 1635   next limit is that "plain" is 64 bit, with the upper 32 bit set to zero. 
 1636   This means that on volumes larger than 2TiB, the IV repeats, creating a
 1637   vulnerability that potentially leaks some data.  To avoid this, use
 1638   "plain64", which uses the full sector number up to 64 bit.  Note that
 1639   "plain64" requires a kernel 2.6.33 or more recent.  Also note that
 1640   "plain64" is backwards compatible for volume sizes of maximum size 2TiB,
 1641   but not for those > 2TiB.  Finally, "plain64" does not cause any
 1642   performance penalty compared to "plain".
 1643 
 1644 
 1645   * 5.16 What about XTS mode?
 1646 
 1647   XTS mode is potentially even more secure than cbc-essiv (but only if
 1648   cbc-essiv is insecure in your scenario).  It is a NIST standard and
 1649   used, e.g.  in Truecrypt.  From version 1.6.0 of cryptsetup onwards,
 1650   aes-xts-plain64 is the default for LUKS.  If you want to use it with a
 1651   cryptsetup before version 1.6.0 or with plain dm-crypt, you have to
 1652   specify it manually as "aes-xts-plain", i.e.
 1653 
 1654     cryptsetup -c aes-xts-plain luksFormat <device>
 1655 
 1656   For volumes >2TiB and kernels >= 2.6.33 use "plain64" (see FAQ item
 1657   on "plain" and "plain64"):
 1658 
 1659     cryptsetup -c aes-xts-plain64 luksFormat <device>
 1660 
 1661   There is a potential security issue with XTS mode and large blocks. 
 1662   LUKS and dm-crypt always use 512B blocks and the issue does not apply.
 1663 
 1664 
 1665   * 5.17 Is LUKS FIPS-140-2 certified?
 1666 
 1667   No.  But that is more a problem of FIPS-140-2 than of LUKS.  From a
 1668   technical point-of-view, LUKS with the right parameters would be
 1669   FIPS-140-2 compliant, but in order to make it certified, somebody has to
 1670   pay real money for that.  And then, whenever cryptsetup is changed or
 1671   extended, the certification lapses and has to be obtained again.
 1672 
 1673   From the aspect of actual security, LUKS with default parameters should
 1674   be as good as most things that are FIPS-140-2 certified, although you
 1675   may want to make sure to use /dev/random (by specifying --use-random on
 1676   luksFormat) as randomness source for the master key to avoid being
 1677   potentially insecure in an entropy-starved situation.
 1678 
 1679 
 1680   * 5.18 What about Plausible Deniability?
 1681 
 1682   First let me attempt a definition for the case of encrypted filesystems:
 1683   Plausible deniability is when you store data inside an encrypted
 1684   container and it is not possible to prove it is there without having a
 1685   special passphrase.  And at the same time it must be "plausible" that
 1686   there actually is no hidden data there.
 1687 
 1688   As a simple entropy-analysis will show that here may be data there, the
 1689   second part is what makes it tricky.
 1690 
 1691   There seem to  be a lot of misunderstandings about this idea, so let me
 1692   make it clear that this refers to the situation where the attackers can
 1693   prove that there is data that either may be random or may be part of a
 1694   plausible-deniability scheme, they just cannot prove which one it is. 
 1695   Hence a plausible-deniability scheme must hold up when the attackers
 1696   know there is something potentially fishy.  If you just hide data and
 1697   rely on it not being found, that is just simple deniability, not
 1698   "plausible" deniability and I am not talking about that in the
 1699   following.  Simple deniability against a low-competence attacker may be
 1700   as simple as renaming a file or putting data into an unused part of a
 1701   disk.  Simple deniability against a high-skill attacker with time to
 1702   invest is usually pointless unless you go for advanced steganographic
 1703   techniques, which have their own drawbacks, such as low data capacity.
 1704 
 1705   Now, the idea of plausible deniability is compelling and on a first
 1706   glance it seems possible to do it.  And from a cryptographic point of
 1707   view, it actually is possible.
 1708 
 1709   So, does the idea work in practice?  No, unfortunately.  The reasoning
 1710   used by its proponents is fundamentally flawed in several ways and the
 1711   cryptographic properties fail fatally when colliding with the real
 1712   world.
 1713 
 1714   First, why should "I do not have a hidden partition" be any more
 1715   plausible than "I forgot my crypto key" or "I wiped that partition with
 1716   random data, nothing in there"?  I do not see any reason.
 1717 
 1718   Second, there are two types of situations: Either they cannot force you
 1719   to give them the key (then you simply do not) or they can.  In the
 1720   second case, they can always do bad things to you, because they cannot
 1721   prove that you have the key in the first place!  This means they do not
 1722   have to prove you have the key, or that this random looking data on your
 1723   disk is actually encrypted data.  So the situation will allow them to
 1724   waterboard/lock-up/deport you anyways, regardless of how "plausible"
 1725   your deniability is.  Do not have a hidden partition you could show to
 1726   them, but there are indications you may?  Too bad for you. 
 1727   Unfortunately "plausible deniability" also means you cannot prove there
 1728   is no hidden data.
 1729 
 1730   Third, hidden partitions are not that hidden.  There are basically just
 1731   two possibilities: a) Make a large crypto container, but put a smaller
 1732   filesystem in there and put the hidden partition into the free space. 
 1733   Unfortunately this is glaringly obvious and can be detected in an
 1734   automated fashion.  This means that the initial suspicion to put you
 1735   under duress in order to make you reveal your hidden data is given.  b)
 1736   Make a filesystem that spans the whole encrypted partition, and put the
 1737   hidden partition into space not currently used by that filesystem. 
 1738   Unfortunately that is also glaringly obvious, as you then cannot write
 1739   to the filesystem without a high risk of destroying data in the hidden
 1740   container.  Have not written anything to the encrypted filesystem in a
 1741   while?  Too bad, they have the suspicion they need to do unpleasant
 1742   things to you.
 1743 
 1744   To be fair, if you prepare option b) carefully and directly before going
 1745   into danger, it may work.  But then, the mere presence of encrypted data
 1746   may already be enough to get you into trouble in those places were they
 1747   can demand encryption keys.
 1748 
 1749   Here is an additional reference for some problems with plausible
 1750   deniability:
 1751   https://www.schneier.com/academic/paperfiles/paper-truecrypt-dfs.pdf
 1752   I strongly suggest you read it.
 1753 
 1754   So, no, I will not provide any instructions on how to do it with plain
 1755   dm-crypt or LUKS.  If you insist on shooting yourself in the foot, you
 1756   can figure out how to do it yourself.
 1757 
 1758 
 1759  * 5.19 What about SSDs, Flash, Hybrid and SMR Drives?
 1760 
 1761   The problem is that you cannot reliably erase parts of these devices,
 1762   mainly due to wear-leveling and possibly defect management and delayed
 1763   writes to the main data area.
 1764 
 1765   For example for SSDs, when overwriting a sector, what the device does is
 1766   to move an internal sector (may be 128kB or even larger) to some pool of
 1767   discarded, not-yet erased unused sectors, take a fresh empty sector from
 1768   the empty-sector pool and copy the old sector over with the changes to
 1769   the small part you wrote.  This is done in some fashion so that larger
 1770   writes do not cause a lot of small internal updates.
 1771 
 1772   The thing is that the mappings between outside-addressable sectors and
 1773   inside sectors is arbitrary (and the vendors are not talking).  Also the
 1774   discarded sectors are not necessarily erased immediately.  They may
 1775   linger a long time.
 1776 
 1777   For plain dm-crypt, the consequences are that older encrypted data may
 1778   be lying around in some internal pools of the device.  Thus may or may
 1779   not be a problem and depends on the application.  Remember the same can
 1780   happen with a filesystem if consecutive writes to the same area of a
 1781   file can go to different sectors.
 1782 
 1783   However, for LUKS, the worst case is that key-slots and LUKS header may
 1784   end up in these internal pools.  This means that password management
 1785   functionality is compromised (the old passwords may still be around,
 1786   potentially for a very long time) and that fast erase by overwriting the
 1787   header and key-slot area is insecure.
 1788 
 1789   Also keep in mind that the discarded/used pool may be large.  For
 1790   example, a 240GB SSD has about 16GB of spare area in the chips that it
 1791   is free to do with as it likes.  You would need to make each individual
 1792   key-slot larger than that to allow reliable overwriting.  And that
 1793   assumes the disk thinks all other space is in use.  Reading the internal
 1794   pools using forensic tools is not that hard, but may involve some
 1795   soldering.
 1796 
 1797   What to do?
 1798 
 1799   If you trust the device vendor (you probably should not...) you can try
 1800   an ATA "secure erase" command.  That is not present in USB keys though
 1801   and may or may not be secure for a hybrid drive.
 1802 
 1803   If you can do without password management and are fine with doing
 1804   physical destruction for permanently deleting data (always after one or
 1805   several full overwrites!), you can use plain dm-crypt.
 1806 
 1807   If you want or need all the original LUKS security features to work, you
 1808   can use a detached LUKS header and put that on a conventional, magnetic
 1809   disk.  That leaves potentially old encrypted data in the pools on the
 1810   main disk, but otherwise you get LUKS with the same security as on a
 1811   traditional magnetic disk.  Note however that storage vendors are prone
 1812   to lying to their customers.  For example, it recently came out that
 1813   HDDs sold without any warning or mentioning in the data-sheets were
 1814   actually using SMR and that will write data first to a faster area and
 1815   only overwrite the original data area some time later when things are
 1816   quiet.
 1817 
 1818   If you are concerned about your laptop being stolen, you are likely fine
 1819   using LUKS on an SSD or hybrid drive.  An attacker would need to have
 1820   access to an old passphrase (and the key-slot for this old passphrase
 1821   would actually need to still be somewhere in the SSD) for your data to
 1822   be at risk.  So unless you pasted your old passphrase all over the
 1823   Internet or the attacker has knowledge of it from some other source and
 1824   does a targeted laptop theft to get at your data, you should be fine.
 1825 
 1826 
 1827  * 5.20 LUKS1 is broken! It uses SHA-1!
 1828 
 1829   No, it is not.  SHA-1 is (academically) broken for finding collisions,
 1830   but not for using it in a key-derivation function.  And that collision
 1831   vulnerability is for non-iterated use only.  And you need the hash-value
 1832   in verbatim.
 1833 
 1834   This basically means that if you already have a slot-key, and you have
 1835   set the PBKDF2 iteration count to 1 (it is > 10'000 normally), you could
 1836   (maybe) derive a different passphrase that gives you the the same
 1837   slot-key.  But if you have the slot-key, you can already unlock the
 1838   key-slot and get the master key, breaking everything.  So basically,
 1839   this SHA-1 vulnerability allows you to open a LUKS1 container with high
 1840   effort when you already have it open.
 1841 
 1842   The real problem here is people that do not understand crypto and claim
 1843   things are broken just because some mechanism is used that has been
 1844   broken for a specific different use.  The way the mechanism is used
 1845   matters very much.  A hash that is broken for one use can be completely
 1846   secure for other uses and here it is.
 1847 
 1848   Since version 1.7.0, cryptsetup uses SHA-256 as default to ensure that
 1849   it will be compatible in the future. There are already some systems 
 1850   where SHA-1 is completely phased out or disabled by a security policy.
 1851 
 1852 
 1853  * 5.21 Why is there no "Nuke-Option"?
 1854 
 1855   A "Nuke-Option" or "Kill-switch" is a password that when entered upon
 1856   unlocking instead wipes the header and all passwords.  So when somebody
 1857   forces you to enter your password, you can destroy the data instead.
 1858 
 1859   While this sounds attractive at first glance, it does not make sense
 1860   once a real security analysis is done.  One problem is that you have to
 1861   have some kind of HSM (Hardware Security Module) in order to implement
 1862   it securely.  In the movies, a HSM starts to smoke and melt once the
 1863   Nuke-Option has been activated.  In actual reality, it just wipes some
 1864   battery-backed RAM cells.  A proper HSM costs something like
 1865   20'000...100'000 EUR/USD and there a Nuke-Option may make some sense. 
 1866   BTW, a chipcard or a TPM is not a HSM, although some vendors are
 1867   promoting that myth.
 1868 
 1869   Now, a proper HSMs will have a wipe option but not a Nuke-Option, i.e. 
 1870   you can explicitly wipe the HSM, but by a different process than
 1871   unlocking it takes.  Why is that?  Simple: If somebody can force you to
 1872   reveal passwords, then they can also do bad things to you if you do not
 1873   or if you enter a nuke password instead.  Think locking you up for a few
 1874   years for "destroying evidence" or for far longer and without trial for
 1875   being a "terrorist suspect".  No HSM maker will want to expose its
 1876   customers to that risk.
 1877 
 1878   Now think of the typical LUKS application scenario, i.e.  disk
 1879   encryption.  Usually the ones forcing you to hand over your password
 1880   will have access to the disk as well, and, if they have any real
 1881   suspicion, they will mirror your disk before entering anything supplied
 1882   by you.  This neatly negates any Nuke-Option.  If they have no suspicion
 1883   (just harassing people that cross some border for example), the
 1884   Nuke-Option would work, but see above about likely negative consequences
 1885   and remember that a Nuke-Option may not work reliably on SSD and hybrid
 1886   drives anyways.
 1887 
 1888   Hence my advice is to never take data that you do not want to reveal
 1889   into any such situation in the first place.  There is no need to
 1890   transfer data on physical carriers today.  The Internet makes it quite
 1891   possible to transfer data between arbitrary places and modern encryption
 1892   makes it secure.  If you do it right, nobody will even be able to
 1893   identify source or destination.  (How to do that is out of scope of this
 1894   document.  It does require advanced skills in this age of pervasive
 1895   surveillance.)
 1896 
 1897   Hence, LUKS has no kill option because it would do much more harm than
 1898   good.
 1899 
 1900   Still, if you have a good use-case (i.e.  non-abstract real-world
 1901   situation) where a Nuke-Option would actually be beneficial, please let
 1902   me know.
 1903 
 1904 
 1905  * 5.22 Does cryptsetup open network connections to websites, etc. ?
 1906 
 1907   This question seems not to make much sense at first glance, but here is
 1908   an example form the real world: The TrueCrypt GUI has a "Donation"
 1909   button.  Press it, and a web-connection to the TrueCrypt website is
 1910   opened via the default browser, telling everybody that listens that you
 1911   use TrueCrypt.  In the worst case, things like this can get people
 1912   tortured or killed.
 1913 
 1914   So: Cryptsetup will never open any network connections except the
 1915   local netlink socket it needs to talk to the kernel crypto API.
 1916 
 1917   In addition, the installation package should contain all documentation,
 1918   including this FAQ, so that you do not have to go to a web-site to read
 1919   it.  (If your distro cuts the docu, please complain to them.) In
 1920   security software, any connection initiated to anywhere outside your
 1921   machine should always be the result of an explicit request for such a
 1922   connection by the user and cryptsetup will stay true to that principle.
 1923 
 1924 
 1925 6. Backup and Data Recovery
 1926 
 1927 
 1928  * 6.1 Why do I need Backup?
 1929 
 1930   First, disks die.  The rate for well-treated (!) disk is about 5% per
 1931   year, which is high enough to worry about.  There is some indication
 1932   that this may be even worse for some SSDs.  This applies both to LUKS
 1933   and plain dm-crypt partitions.
 1934 
 1935   Second, for LUKS, if anything damages the LUKS header or the key-stripe
 1936   area then decrypting the LUKS device can become impossible.  This is a
 1937   frequent occurrence.  For example an accidental format as FAT or some
 1938   software overwriting the first sector where it suspects a partition boot
 1939   sector typically makes a LUKS1 partition permanently inaccessible.  See
 1940   more below on LUKS header damage.
 1941 
 1942   So, data-backup in some form is non-optional.  For LUKS, you may also
 1943   want to store a header backup in some secure location.  This only needs
 1944   an update if you change passphrases.
 1945 
 1946 
 1947  * 6.2 How do I backup a LUKS header?
 1948 
 1949   While you could just copy the appropriate number of bytes from the start
 1950   of the LUKS partition, the best way is to use command option
 1951   "luksHeaderBackup" of cryptsetup.  This protects also against errors
 1952   when non-standard parameters have been used in LUKS partition creation.  
 1953   Example:
 1954 
 1955     cryptsetup luksHeaderBackup --header-backup-file <file> <device>
 1956 
 1957   To restore, use the inverse command, i.e.
 1958 
 1959     cryptsetup luksHeaderRestore --header-backup-file <file> <device>
 1960 
 1961   If you are unsure about a header to be restored, make a backup of the
 1962   current one first!  You can also test the header-file without restoring
 1963   it by using the --header option for a detached header like this:
 1964 
 1965     cryptsetup --header <file> luksOpen <device> </dev/mapper/name>
 1966 
 1967   If that unlocks your key-slot, you are good. Do not forget to close
 1968   the device again.
 1969 
 1970   Under some circumstances (damaged header), this fails.  Then use the
 1971   following steps in case it is LUKS1:
 1972 
 1973   First determine the master-key size:
 1974 
 1975     cryptsetup luksDump <device>
 1976 
 1977   gives a line of the form
 1978 
 1979     MK bits:        <bits>
 1980 
 1981   with bits equal to 256 for the old defaults and 512 for the new
 1982   defaults.  256 bits equals a total header size of 1'052'672 Bytes and
 1983   512 bits one of 2MiB.  (See also Item 6.12) If luksDump fails, assume
 1984   2MiB, but be aware that if you restore that, you may also restore the
 1985   first 1M or so of the filesystem.  Do not change the filesystem if you
 1986   were unable to determine the header size!  With that, restoring a
 1987   too-large header backup is still safe.
 1988 
 1989   Second, dump the header to file. There are many ways to do it, I
 1990   prefer the following:
 1991 
 1992     head -c 1052672 <device>  >  header_backup.dmp
 1993 
 1994   or
 1995 
 1996     head -c 2M <device>  >  header_backup.dmp
 1997 
 1998   for a 2MiB header. Verify the size of the dump-file to be sure.
 1999 
 2000   To restore such a backup, you can try luksHeaderRestore or do a more
 2001   basic
 2002 
 2003     cat header_backup.dmp  >  <device>
 2004 
 2005 
 2006 
 2007   * 6.3 How do I test for a LUKS header?
 2008 
 2009   Use
 2010 
 2011     cryptsetup -v isLuks <device>
 2012 
 2013   on the device.  Without the "-v" it just signals its result via
 2014   exit-status.  You can also use the more general test
 2015 
 2016     blkid -p <device>
 2017 
 2018   which will also detect other types and give some more info.  Omit
 2019   "-p" for old versions of blkid that do not support it.
 2020 
 2021 
 2022   * 6.4 How do I backup a LUKS or dm-crypt partition?
 2023 
 2024   There are two options, a sector-image and a plain file or filesystem
 2025   backup of the contents of the partition.  The sector image is already
 2026   encrypted, but cannot be compressed and contains all empty space.  The
 2027   filesystem backup can be compressed, can contain only part of the
 2028   encrypted device, but needs to be encrypted separately if so desired.
 2029 
 2030   A sector-image will contain the whole partition in encrypted form, for
 2031   LUKS the LUKS header, the keys-slots and the data area.  It can be done
 2032   under Linux e.g.  with dd_rescue (for a direct image copy) and with
 2033   "cat" or "dd".  Examples:
 2034 
 2035     cat /dev/sda10 > sda10.img
 2036     dd_rescue /dev/sda10 sda10.img
 2037 
 2038   You can also use any other backup software that is capable of making a
 2039   sector image of a partition.  Note that compression is ineffective for
 2040   encrypted data, hence it does not make sense to use it.
 2041 
 2042   For a filesystem backup, you decrypt and mount the encrypted partition
 2043   and back it up as you would a normal filesystem.  In this case the
 2044   backup is not encrypted, unless your encryption method does that.  For
 2045   example you can encrypt a backup with "tar" as follows with GnuPG:
 2046 
 2047     tar cjf - <path> | gpg --cipher-algo AES -c - > backup.tbz2.gpg
 2048 
 2049   And verify the backup like this if you are at "path":
 2050 
 2051     cat backup.tbz2.gpg | gpg - | tar djf -
 2052 
 2053   Note: Always verify backups, especially encrypted ones!
 2054 
 2055   There is one problem with verifying like this: The kernel may still have
 2056   some files cached and in fact verify them against RAM or may even verify
 2057   RAM against RAM, which defeats the purpose of the exercise.  The
 2058   following command empties the kernel caches:
 2059 
 2060     echo 3 > /proc/sys/vm/drop_caches
 2061 
 2062   Run it after backup and before verify.
 2063 
 2064   In both cases GnuPG will ask you interactively for your symmetric key. 
 2065   The verify will only output errors.  Use "tar dvjf -" to get all
 2066   comparison results.  To make sure no data is written to disk
 2067   unencrypted, turn off swap if it is not encrypted before doing the
 2068   backup.
 2069 
 2070   Restore works like certification with the 'd' ('difference') replaced 
 2071   by 'x' ('eXtract').  Refer to the man-page of tar for more explanations 
 2072   and instructions.  Note that with default options tar will overwrite 
 2073   already existing files without warning.  If you are unsure about how 
 2074   to use tar, experiment with it in a location where you cannot do damage.
 2075 
 2076   You can of course use different or no compression and you can use an
 2077   asymmetric key if you have one and have a backup of the secret key that
 2078   belongs to it.
 2079 
 2080   A second option for a filesystem-level backup that can be used when the
 2081   backup is also on local disk (e.g.  an external USB drive) is to use a
 2082   LUKS container there and copy the files to be backed up between both
 2083   mounted containers.  Also see next item.
 2084 
 2085 
 2086   * 6.5 Do I need a backup of the full partition? Would the header
 2087     and key-slots not be enough?
 2088 
 2089   Backup protects you against two things: Disk loss or corruption and user
 2090   error.  By far the most questions on the dm-crypt mailing list about how
 2091   to recover a damaged LUKS partition are related to user error.  For
 2092   example, if you create a new filesystem on a non-mapped LUKS container,
 2093   chances are good that all data is lost permanently.
 2094 
 2095   For this case, a header+key-slot backup would often be enough.  But keep
 2096   in mind that a well-treated (!) HDD has roughly a failure risk of 5% per
 2097   year.  It is highly advisable to have a complete backup to protect
 2098   against this case.
 2099 
 2100 
 2101   * 6.6 What do I need to backup if I use "decrypt_derived"?
 2102 
 2103   This is a script in Debian, intended for mounting /tmp or swap with a
 2104   key derived from the master key of an already decrypted device.  If you
 2105   use this for an device with data that should be persistent, you need to
 2106   make sure you either do not lose access to that master key or have a
 2107   backup of the data.  If you derive from a LUKS device, a header backup
 2108   of that device would cover backing up the master key.  Keep in mind that
 2109   this does not protect against disk loss.
 2110 
 2111   Note: If you recreate the LUKS header of the device you derive from
 2112   (using luksFormat), the master key changes even if you use the same
 2113   passphrase(s) and you will not be able to decrypt the derived device
 2114   with the new LUKS header.
 2115 
 2116 
 2117   * 6.7 Does a backup compromise security?
 2118 
 2119   Depends on how you do it.  However if you do not have one, you are going
 2120   to eventually lose your encrypted data.
 2121 
 2122   There are risks introduced by backups.  For example if you
 2123   change/disable a key-slot in LUKS, a binary backup of the partition will
 2124   still have the old key-slot.  To deal with this, you have to be able to
 2125   change the key-slot on the backup as well, securely erase the backup or
 2126   do a filesystem-level backup instead of a binary one.
 2127 
 2128   If you use dm-crypt, backup is simpler: As there is no key management,
 2129   the main risk is that you cannot wipe the backup when wiping the
 2130   original.  However wiping the original for dm-crypt should consist of
 2131   forgetting the passphrase and that you can do without actual access to
 2132   the backup.
 2133 
 2134   In both cases, there is an additional (usually small) risk with binary
 2135   backups: An attacker can see how many sectors and which ones have been
 2136   changed since the backup.  To prevent this, use a filesystem level
 2137   backup method that encrypts the whole backup in one go, e.g.  as
 2138   described above with tar and GnuPG.
 2139 
 2140   My personal advice is to use one USB disk (low value data) or three
 2141   disks (high value data) in rotating order for backups, and either use
 2142   independent LUKS partitions on them, or use encrypted backup with tar
 2143   and GnuPG.
 2144 
 2145   If you do network-backup or tape-backup, I strongly recommend to go
 2146   the filesystem backup path with independent encryption, as you
 2147   typically cannot reliably delete data in these scenarios, especially
 2148   in a cloud setting.  (Well, you can burn the tape if it is under your
 2149   control...)
 2150 
 2151 
 2152   * 6.8 What happens if I overwrite the start of a LUKS partition or
 2153     damage the LUKS header or key-slots?
 2154 
 2155   There are two critical components for decryption: The salt values in the
 2156   key-slot descriptors of the header and the key-slots.  For LUKS2 they
 2157   are a bit better protected.  but for LUKS1, these are right in the first
 2158   sector.  If the salt values are overwritten or changed, nothing (in the
 2159   cryptographically strong sense) can be done to access the data, unless
 2160   there is a backup of the LUKS header.  If a key-slot is damaged, the
 2161   data can still be read with a different key-slot, if there is a
 2162   remaining undamaged and used key-slot.  Note that in order to make a
 2163   key-slot completely unrecoverable, changing about 4-6 bits in random
 2164   locations of its 128kiB size is quite enough.
 2165 
 2166 
 2167   * 6.9 What happens if I (quick) format a LUKS partition?
 2168 
 2169   I have not tried the different ways to do this, but very likely you will
 2170   have written a new boot-sector, which in turn overwrites the LUKS
 2171   header, including the salts, making your data permanently irretrievable,
 2172   unless you have a LUKS header backup.  For LUKS2 this may still be
 2173   recoverable without that header backup, for LUKS1 it is not.  You may
 2174   also damage the key-slots in part or in full.  See also last item.
 2175 
 2176 
 2177   * 6.10 How do I recover the master key from a mapped LUKS1 container?
 2178 
 2179   Note: LUKS2 uses the kernel keyring to store keys and hence this
 2180   procedure does not work unless you have explicitly disabled the use of
 2181   the keyring with "--disable-keyring" on opening.
 2182  
 2183   This is typically only needed if you managed to damage your LUKS1
 2184   header, but the container is still mapped, i.e.  "luksOpen"ed.  It also
 2185   helps if you have a mapped container that you forgot or do not know a
 2186   passphrase for (e.g.  on a long running server.)
 2187 
 2188   WARNING: Things go wrong, do a full backup before trying this!
 2189 
 2190   WARNING: This exposes the master key of the LUKS1 container.  Note that
 2191   both ways to recreate a LUKS header with the old master key described
 2192   below will write the master key to disk.  Unless you are sure you have
 2193   securely erased it afterwards, e.g.  by writing it to an encrypted
 2194   partition, RAM disk or by erasing the filesystem you wrote it to by a
 2195   complete overwrite, you should change the master key afterwards. 
 2196   Changing the master key requires a full data backup, luksFormat and then
 2197   restore of the backup.  Alternatively the tool cryptsetup-reencrypt from
 2198   the cryptsetup package can be used to change the master key (see its
 2199   man-page), but a full backup is still highly recommended.
 2200 
 2201   First, there is a script by Milan that automates the whole process,
 2202   except generating a new LUKS1 header with the old master key (it prints
 2203   the command for that though):
 2204 
 2205   https://gitlab.com/cryptsetup/cryptsetup/blob/master/misc/luks-header-from-active
 2206 
 2207   You can also do this manually. Here is how:
 2208 
 2209   - Get the master key from the device mapper.  This is done by the
 2210   following command.  Substitute c5 for whatever you mapped to:
 2211 
 2212     # dmsetup table --target crypt --showkey /dev/mapper/c5
 2213 
 2214     Result:
 2215     0 200704 crypt aes-cbc-essiv:sha256
 2216     a1704d9715f73a1bb4db581dcacadaf405e700d591e93e2eaade13ba653d0d09
 2217     0 7:0 4096
 2218 
 2219   The result is actually one line, wrapped here for clarity.  The long
 2220   hex string is the master key.
 2221 
 2222   - Convert the master key to a binary file representation.  You can do
 2223   this manually, e.g.  with hexedit.  You can also use the tool "xxd"
 2224   from vim like this:
 2225 
 2226     echo "a1704d9....53d0d09" | xxd -r -p > <master-key-file>
 2227 
 2228 
 2229   - Do a luksFormat to create a new LUKS1 header.
 2230 
 2231     NOTE: If your header is intact and you just forgot the passphrase,
 2232     you can just set a new passphrase, see next sub-item.
 2233 
 2234   Unmap the device before you do that (luksClose). Then do
 2235 
 2236     cryptsetup luksFormat --master-key-file=<master-key-file> <luks device>
 2237 
 2238   Note that if the container was created with other than the default
 2239   settings of the cryptsetup version you are using, you need to give
 2240   additional parameters specifying the deviations.  If in doubt, try the
 2241   script by Milan.  It does recover the other parameters as well.
 2242 
 2243   Side note: This is the way the decrypt_derived script gets at the master
 2244   key.  It just omits the conversion and hashes the master key string.
 2245 
 2246   - If the header is intact and you just forgot the passphrase, just
 2247   set a new passphrase like this:
 2248 
 2249       cryptsetup luksAddKey --master-key-file=<master-key-file> <luks device>
 2250 
 2251   You may want to disable the old one afterwards.
 2252 
 2253 
 2254   * 6.11 What does the on-disk structure of dm-crypt look like?
 2255 
 2256   There is none.  dm-crypt takes a block device and gives encrypted access
 2257   to each of its blocks with a key derived from the passphrase given.  If
 2258   you use a cipher different than the default, you have to specify that as
 2259   a parameter to cryptsetup too.  If you want to change the password, you
 2260   basically have to create a second encrypted device with the new
 2261   passphrase and copy your data over.  On the plus side, if you
 2262   accidentally overwrite any part of a dm-crypt device, the damage will be
 2263   limited to the area you overwrote.
 2264 
 2265 
 2266   * 6.12 What does the on-disk structure of LUKS1 look like?
 2267 
 2268   Note: For LUKS2, refer to the LUKS2 document referenced in Item 1.2
 2269 
 2270   A LUKS1 partition consists of a header, followed by 8 key-slot
 2271   descriptors, followed by 8 key slots, followed by the encrypted data
 2272   area.
 2273 
 2274   Header and key-slot descriptors fill the first 592 bytes.  The key-slot
 2275   size depends on the creation parameters, namely on the number of
 2276   anti-forensic stripes, key material offset and master key size.
 2277 
 2278   With the default parameters, each key-slot is a bit less than 128kiB in
 2279   size.  Due to sector alignment of the key-slot start, that means the key
 2280   block 0 is at offset 0x1000-0x20400, key block 1 at offset
 2281   0x21000-0x40400, and key block 7 at offset 0xc1000-0xe0400.  The space
 2282   to the next full sector address is padded with zeros.  Never used
 2283   key-slots are filled with what the disk originally contained there, a
 2284   key-slot removed with "luksRemoveKey" or "luksKillSlot" gets filled with
 2285   0xff.  Due to 2MiB default alignment, start of the data area for
 2286   cryptsetup 1.3 and later is at 2MiB, i.e.  at 0x200000.  For older
 2287   versions, it is at 0x101000, i.e.  at 1'052'672 bytes, i.e.  at 1MiB +
 2288   4096 bytes from the start of the partition.  Incidentally,
 2289   "luksHeaderBackup" for a LUKS container created with default parameters
 2290   dumps exactly the first 2MiB (or 1'052'672 bytes for headers created
 2291   with cryptsetup versions < 1.3) to file and "luksHeaderRestore" restores
 2292   them.
 2293 
 2294   For non-default parameters, you have to figure out placement yourself. 
 2295   "luksDump" helps.  See also next item.  For the most common non-default
 2296   settings, namely aes-xts-plain with 512 bit key, the offsets are: 1st
 2297   keyslot 0x1000-0x3f800, 2nd keyslot 0x40000-0x7e000, 3rd keyslot
 2298   0x7e000-0xbd800, ..., and start of bulk data at 0x200000.
 2299 
 2300   The exact specification of the format is here:
 2301      https://gitlab.com/cryptsetup/cryptsetup/wikis/Specification
 2302 
 2303   For your convenience, here is the LUKS1 header with hex offsets.  
 2304   NOTE:
 2305   The spec counts key-slots from 1 to 8, but the cryptsetup tool counts
 2306   from 0 to 7.  The numbers here refer to the cryptsetup numbers.
 2307 
 2308 
 2309 Refers to LUKS1 On-Disk Format Specification Version 1.2.3
 2310 
 2311 LUKS1 header:
 2312 
 2313 offset  length  name             data type  description
 2314 -----------------------------------------------------------------------
 2315 0x0000   0x06   magic            byte[]     'L','U','K','S', 0xba, 0xbe
 2316      0      6
 2317 0x0006   0x02   version          uint16_t   LUKS version
 2318      6      3
 2319 0x0008   0x20   cipher-name      char[]     cipher name spec.
 2320      8     32
 2321 0x0028   0x20   cipher-mode      char[]     cipher mode spec.
 2322     40     32
 2323 0x0048   0x20   hash-spec        char[]     hash spec.
 2324     72     32
 2325 0x0068   0x04   payload-offset   uint32_t   bulk data offset in sectors
 2326    104      4                               (512 bytes per sector)
 2327 0x006c   0x04   key-bytes        uint32_t   number of bytes in key
 2328    108      4
 2329 0x0070   0x14   mk-digest        byte[]     master key checksum
 2330    112     20                               calculated with PBKDF2
 2331 0x0084   0x20   mk-digest-salt   byte[]     salt for PBKDF2 when
 2332    132     32                               calculating mk-digest
 2333 0x00a4   0x04   mk-digest-iter   uint32_t   iteration count for PBKDF2
 2334    164      4                               when calculating mk-digest
 2335 0x00a8   0x28   uuid             char[]     partition UUID
 2336    168     40
 2337 0x00d0   0x30   key-slot-0       key slot   key slot 0
 2338    208     48
 2339 0x0100   0x30   key-slot-1       key slot   key slot 1
 2340    256     48
 2341 0x0130   0x30   key-slot-2       key slot   key slot 2
 2342    304     48
 2343 0x0160   0x30   key-slot-3       key slot   key slot 3
 2344    352     48
 2345 0x0190   0x30   key-slot-4       key slot   key slot 4
 2346    400     48
 2347 0x01c0   0x30   key-slot-5       key slot   key slot 5
 2348    448     48
 2349 0x01f0   0x30   key-slot-6       key slot   key slot 6
 2350    496     48
 2351 0x0220   0x30   key-slot-7       key slot   key slot 7
 2352    544     48
 2353 
 2354 
 2355 Key slot:
 2356 
 2357 offset  length  name                  data type  description
 2358 -------------------------------------------------------------------------
 2359 0x0000   0x04   active                uint32_t   key slot enabled/disabled
 2360      0      4
 2361 0x0004   0x04   iterations            uint32_t   PBKDF2 iteration count
 2362      4      4
 2363 0x0008   0x20   salt                  byte[]     PBKDF2 salt
 2364      8     32
 2365 0x0028   0x04   key-material-offset   uint32_t   key start sector
 2366     40      4                                    (512 bytes/sector)
 2367 0x002c   0x04   stripes               uint32_t   number of anti-forensic
 2368     44      4                                    stripes
 2369 
 2370 
 2371 
 2372   * 6.13 What is the smallest possible LUKS1 container?
 2373 
 2374   Note: From cryptsetup 1.3 onwards, alignment is set to 1MB.  With modern
 2375   Linux partitioning tools that also align to 1MB, this will result in
 2376   alignment to 2k sectors and typical Flash/SSD sectors, which is highly
 2377   desirable for a number of reasons.  Changing the alignment is not
 2378   recommended.
 2379 
 2380   That said, with default parameters, the data area starts at exactly 2MB
 2381   offset (at 0x101000 for cryptsetup versions before 1.3).  The smallest
 2382   data area you can have is one sector of 512 bytes.  Data areas of 0
 2383   bytes can be created, but fail on mapping.
 2384 
 2385   While you cannot put a filesystem into something this small, it may
 2386   still be used to contain, for example, key.  Note that with current
 2387   formatting tools, a partition for a container this size will be 3MiB
 2388   anyways.  If you put the LUKS container into a file (via losetup and a
 2389   loopback device), the file needs to be 2097664 bytes in size, i.e.  2MiB
 2390   + 512B.
 2391 
 2392   The two ways to influence the start of the data area are key-size and
 2393   alignment.
 2394 
 2395   For alignment, you can go down to 1 on the parameter.  This will still
 2396   leave you with a data-area starting at 0x101000, i.e.  1MiB+4096B
 2397   (default parameters) as alignment will be rounded up to the next
 2398   multiple of 8 (i.e.  4096 bytes) If in doubt, do a dry-run on a larger
 2399   file and dump the LUKS header to get actual information.
 2400 
 2401   For key-size, you can use 128 bit (e.g.  AES-128 with CBC), 256 bit
 2402   (e.g.  AES-256 with CBC) or 512 bit (e.g.  AES-256 with XTS mode).  You
 2403   can do 64 bit (e.g.  blowfish-64 with CBC), but anything below 128 bit
 2404   has to be considered insecure today.
 2405 
 2406   Example 1 - AES 128 bit with CBC:
 2407 
 2408       cryptsetup luksFormat -s 128 --align-payload=8 <device>
 2409 
 2410   This results in a data offset of 0x81000, i.e. 516KiB or 528384
 2411   bytes.  Add one 512 byte sector and the smallest LUKS container size
 2412   with these parameters is 516KiB + 512B or 528896 bytes.
 2413 
 2414   Example 2 - Blowfish 64 bit with CBC (WARNING: insecure):
 2415 
 2416       cryptsetup luksFormat -c blowfish -s 64 --align-payload=8 /dev/loop0
 2417 
 2418   This results in a data offset of 0x41000, i.e. 260kiB or 266240
 2419   bytes, with a minimal LUKS1 container size of 260kiB + 512B or 266752
 2420   bytes.
 2421 
 2422 
 2423   * 6.14 I think this is overly complicated. Is there an alternative?
 2424 
 2425   Not really.  Encryption comes at a price.  You can use plain dm-crypt to
 2426   simplify things a bit.  It does not allow multiple passphrases, but on
 2427   the plus side, it has zero on disk description and if you overwrite some
 2428   part of a plain dm-crypt partition, exactly the overwritten parts are
 2429   lost (rounded up to full sectors).
 2430 
 2431   * 6.15 Can I clone a LUKS container?
 2432 
 2433   You can, but it breaks security, because the cloned container has the
 2434   same header and hence the same master key.  Even if you change the 
 2435   passphrase(s), the master key stays the same.  That means whoever has 
 2436   access to one of the clones can decrypt them all, completely bypassing 
 2437   the passphrases. 
 2438 
 2439   While you can use cryptsetup-reencrypt to change the master key, 
 2440   this is probably more effort than to create separate LUKS containers
 2441   in the first place.
 2442 
 2443   The right way to do this is to first luksFormat the target container,
 2444   then to clone the contents of the source container, with both containers
 2445   mapped, i.e.  decrypted.  You can clone the decrypted contents of a LUKS
 2446   container in binary mode, although you may run into secondary issues
 2447   with GUIDs in filesystems, partition tables, RAID-components and the
 2448   like.  These are just the normal problems binary cloning causes.
 2449 
 2450   Note that if you need to ship (e.g.) cloned LUKS containers with a
 2451   default passphrase, that is fine as long as each container was
 2452   individually created (and hence has its own master key).  In this case,
 2453   changing the default passphrase will make it secure again.
 2454 
 2455 
 2456 7. Interoperability with other Disk Encryption Tools
 2457 
 2458 
 2459   * 7.1 What is this section about?
 2460 
 2461   Cryptsetup for plain dm-crypt can be used to access a number of on-disk
 2462   formats created by tools like loop-aes patched into losetup.  This
 2463   sometimes works and sometimes does not.  This section collects insights
 2464   into what works, what does not and where more information is required.
 2465 
 2466   Additional information may be found in the mailing-list archives,
 2467   mentioned at the start of this FAQ document.  If you have a solution
 2468   working that is not yet documented here and think a wider audience may
 2469   be interested, please email the FAQ maintainer.
 2470 
 2471 
 2472   * 7.2 loop-aes: General observations.
 2473 
 2474   One problem is that there are different versions of losetup around. 
 2475   loop-aes is a patch for losetup.  Possible problems and deviations
 2476   from cryptsetup option syntax include:
 2477 
 2478   - Offsets specified in bytes (cryptsetup: 512 byte sectors)
 2479 
 2480   - The need to specify an IV offset
 2481 
 2482   - Encryption mode needs specifying (e.g. "-c twofish-cbc-plain")
 2483 
 2484   - Key size needs specifying (e.g. "-s 128" for 128 bit keys)
 2485 
 2486   - Passphrase hash algorithm needs specifying
 2487 
 2488   Also note that because plain dm-crypt and loop-aes format does not have
 2489   metadata, and while the loopAES extension for cryptsetup tries
 2490   autodetection (see command loopaesOpen), it may not always work.  If you
 2491   still have the old set-up, using a verbosity option (-v) on mapping with
 2492   the old tool or having a look into the system logs after setup could
 2493   give you the information you need.  Below, there are also some things
 2494   that worked for somebody.
 2495 
 2496 
 2497   * 7.3 loop-aes patched into losetup on Debian 5.x, kernel 2.6.32
 2498 
 2499   In this case, the main problem seems to be that this variant of
 2500   losetup takes the offset (-o option) in bytes, while cryptsetup takes
 2501   it in sectors of 512 bytes each.  
 2502 
 2503   Example: The losetup command
 2504 
 2505     losetup -e twofish -o 2560 /dev/loop0 /dev/sdb1
 2506     mount /dev/loop0 mount-point
 2507 
 2508   translates to
 2509 
 2510     cryptsetup create -c twofish -o 5 --skip 5 e1 /dev/sdb1
 2511     mount /dev/mapper/e1 mount-point
 2512 
 2513 
 2514 
 2515   * 7.4 loop-aes with 160 bit key
 2516 
 2517   This seems to be sometimes used with twofish and blowfish and represents
 2518   a 160 bit ripemed160 hash output padded to 196 bit key length.  It seems
 2519   the corresponding options for cryptsetup are
 2520 
 2521     --cipher twofish-cbc-null -s 192 -h ripemd160:20
 2522 
 2523 
 2524 
 2525   * 7.5 loop-aes v1 format OpenSUSE
 2526 
 2527   Apparently this is done by older OpenSUSE distros and stopped working
 2528   from OpenSUSE 12.1 to 12.2.  One user had success with the following:
 2529 
 2530     cryptsetup create <target> <device> -c aes -s 128 -h sha256
 2531 
 2532 
 2533 
 2534   * 7.6 Kernel encrypted loop device (cryptoloop)
 2535 
 2536   There are a number of different losetup implementations for using
 2537   encrypted loop devices so getting this to work may need a bit of
 2538   experimentation.
 2539 
 2540   NOTE: Do NOT use this for new containers! Some of the existing
 2541   implementations are insecure and future support is uncertain.
 2542 
 2543   Example for a compatible mapping:
 2544 
 2545     losetup -e twofish -N /dev/loop0 /image.img
 2546 
 2547   translates to
 2548 
 2549     cryptsetup create image_plain /image.img -c twofish-cbc-plain -H plain
 2550 
 2551   with the mapping being done to /dev/mapper/image_plain instead of
 2552   to /dev/loop0.
 2553 
 2554   More details:
 2555 
 2556   Cipher, mode and password hash (or no hash):
 2557 
 2558   -e cipher [-N]        => -c cipher-cbc-plain -H plain [-s 256]
 2559   -e cipher             => -c cipher-cbc-plain -H ripemd160 [-s 256]
 2560 
 2561 
 2562   Key size and offsets (losetup: bytes, cryptsetuop: sectors of 512 bytes):
 2563 
 2564   -k 128                 => -s 128
 2565   -o 2560                => -o 5 -p 5       # 2560/512 = 5
 2566 
 2567 
 2568   There is no replacement for --pass-fd, it has to be emulated using
 2569   keyfiles, see the cryptsetup man-page.
 2570 
 2571 
 2572 8. Issues with Specific Versions of cryptsetup
 2573 
 2574 
 2575   * 8.1 When using the create command for plain dm-crypt with
 2576     cryptsetup 1.1.x, the mapping is incompatible and my data is not
 2577     accessible anymore!
 2578 
 2579   With cryptsetup 1.1.x, the distro maintainer can define different
 2580   default encryption modes.  You can check the compiled-in defaults using
 2581   "cryptsetup --help".  Moreover, the plain device default changed because
 2582   the old IV mode was vulnerable to a watermarking attack.
 2583 
 2584   If you are using a plain device and you need a compatible mode, just
 2585   specify cipher, key size and hash algorithm explicitly.  For
 2586   compatibility with cryptsetup 1.0.x defaults, simple use the following:
 2587 
 2588     cryptsetup create -c aes-cbc-plain -s 256 -h ripemd160 <name> <dev>
 2589 
 2590   LUKS stores cipher and mode in the metadata on disk, avoiding this
 2591   problem.
 2592 
 2593 
 2594   * 8.2 cryptsetup on SLED 10 has problems...
 2595 
 2596   SLED 10 is missing an essential kernel patch for dm-crypt, which is
 2597   broken in its kernel as a result.  There may be a very old version of
 2598   cryptsetup (1.0.x) provided by SLED, which should also not be used
 2599   anymore as well.  My advice would be to drop SLED 10.
 2600 
 2601 
 2602   * 8.3 Gcrypt 1.6.x and later break Whirlpool
 2603 
 2604   It is the other way round: In gcrypt 1.5.x, Whirlpool is broken and it
 2605   was fixed in 1.6.0 and later.  If you selected whirlpool as hash on
 2606   creation of a LUKS container, it does not work anymore with the fixed
 2607   library.  This shows one serious risk of using rarely used settings.
 2608 
 2609   Note that at the time this FAQ item was written, 1.5.4 was the latest
 2610   1.5.x version and it has the flaw, i.e.  works with the old Whirlpool
 2611   version.  Possibly later 1.5.x versions will work as well.  If not,
 2612   please let me know.
 2613 
 2614   The only two ways to access older LUKS containers created with Whirlpool
 2615   are to either decrypt with an old gcrypt version that has the flaw or to
 2616   use a compatibility feature introduced in cryptsetup 1.6.4 and gcrypt
 2617   1.6.1 or later.  Version 1.6.0 cannot be used.
 2618 
 2619   Steps:
 2620 
 2621   - Make at least a header backup or better, refresh your full backup. 
 2622   (You have a full backup, right?  See Item 6.1 and following.)
 2623 
 2624   - Make sure you have cryptsetup 1.6.4 or later and check the gcrypt
 2625     version:
 2626 
 2627      cryptsetup luksDump <your luks device> --debug | grep backend
 2628 
 2629   If gcrypt is at version 1.5.x or before:
 2630 
 2631   - Reencrypt the LUKS header with a different hash. (Requires entering
 2632   all keyslot passphrases.  If you do not have all, remove the ones you
 2633   do not have before.):
 2634 
 2635      cryptsetup-reencrypt --keep-key --hash sha256 <your luks device>
 2636 
 2637   If gcrypt is at version 1.6.1 or later:
 2638 
 2639   - Patch the hash name in the LUKS header from "whirlpool" to
 2640   "whirlpool_gcryptbug".  This activates the broken implementation. 
 2641   The detailed header layout is in Item 6.12 of this FAQ and in the
 2642   LUKS on-disk format specification.  One way to change the hash is
 2643   with the following command:
 2644 
 2645      echo -n -e 'whirlpool_gcryptbug\0' | dd of=<luks device> bs=1 seek=72 conv=notrunc
 2646 
 2647   - You can now open the device again. It is highly advisable to change
 2648   the hash now with cryptsetup-reencrypt as described above.  While you
 2649   can reencrypt to use the fixed whirlpool, that may not be a good idea
 2650   as almost nobody seems to use it and hence the long time until the
 2651   bug was discovered.
 2652 
 2653 
 2654 9. The Initrd question
 2655 
 2656 
 2657   * 9.1 My initrd is broken with cryptsetup
 2658 
 2659   That is not nice!  However the initrd is supplied by your distribution,
 2660   not by the cryptsetup project and hence you should complain to them.  We
 2661   cannot really do anything about it.
 2662 
 2663 
 2664   * 9.2 CVE-2016-4484 says cryptsetup is broken!
 2665 
 2666   Not really. It says the initrd in some Debian versions have a behavior 
 2667   that under some very special and unusual conditions may be considered
 2668   a vulnerability. 
 2669 
 2670   What happens is that you can trick the initrd to go to a rescue-shell if
 2671   you enter the LUKS password wrongly in a specific way.  But falling back
 2672   to a rescue shell on initrd errors is a sensible default behavior in the
 2673   first place.  It gives you about as much access as booting a rescue
 2674   system from CD or USB-Stick or as removing the disk would give you.  So
 2675   this only applies when an attacker has physical access, but cannot boot
 2676   anything else or remove the disk.  These will be rare circumstances
 2677   indeed, and if you rely on the default distribution initrd to keep you
 2678   safe under these circumstances, then you have bigger problems than this
 2679   somewhat expected behavior.
 2680 
 2681   The CVE was exagerrated and should not be assigned to upstream
 2682   cryptsetup in the first place (it is a distro specific initrd issue). 
 2683   It was driven more by a try to make a splash for self-aggrandizement,
 2684   than by any actual security concerns.  Ignore it.
 2685 
 2686 
 2687   * 9.3 How do I do my own initrd with cryptsetup?
 2688 
 2689   Note: The instructions here apply to an initrd in initramfs format, not
 2690   to an initrd in initrd format.  The latter is a filesystem image, not a
 2691   cpio-archive, and seems to not be widely used anymore.
 2692  
 2693   It depends on the distribution.  Below, I give a very simple example and
 2694   step-by-step instructions for Debian.  With a bit of work, it should be
 2695   possible to adapt this to other distributions.  Note that the
 2696   description is pretty general, so if you want to do other things with an
 2697   initrd it provides a useful starting point for that too.
 2698 
 2699   01) Unpacking an existing initrd to use as template
 2700 
 2701   A Linux initrd is in gzip'ed cpio format. To unpack it, use something
 2702   like this: 
 2703 
 2704      md tmp; cd tmp; cat ../initrd | gunzip | cpio -id
 2705 
 2706   After this, you have the full initrd content in tmp/
 2707 
 2708   02) Inspecting the init-script
 2709 
 2710   The init-script is the only thing the kernel cares about.  All activity
 2711   starts there.  Its traditional location is /sbin/init on disk, but /init
 2712   in an initrd.  In an initrd unpacked as above it is tmp/init.
 2713 
 2714   While init can be a binary despite usually being called "init script",
 2715   in Debian the main init on the root partition is a binary, but the init
 2716   in the initrd (and only that one is called by the kernel) is a script
 2717   and starts like this:
 2718 
 2719     #!/bin/sh
 2720     ....
 2721 
 2722   The "sh" used here is in tmp/bin/sh as just unpacked, and in Debian it
 2723   currently is a busybox.
 2724 
 2725   03) Creating your own initrd
 2726 
 2727   The two examples below should give you most of what is needed.  This is
 2728   tested with LUKS1 and should work with LUKS2 as well.  If not, please
 2729   let me know.
 2730 
 2731   Here is a really minimal example.  It does nothing but set up some
 2732   things and then drop to an interactive shell.  It is perfect to try out
 2733   things that you want to go into the init-script.
 2734 
 2735    #!/bin/sh
 2736    export PATH=/sbin:/bin  
 2737    [ -d /sys ] || mkdir /sys
 2738    [ -d /proc ] || mkdir /proc
 2739    [ -d /tmp ] || mkdir /tmp
 2740    mount -t sysfs -o nodev,noexec,nosuid sysfs /sys
 2741    mount -t proc -o nodev,noexec,nosuid proc /proc
 2742    echo "initrd is running, starting BusyBox..."
 2743    exec /bin/sh --login
 2744 
 2745 
 2746   Here is an example that opens the first LUKS-partition it finds with the
 2747   hard-coded password "test2" and then mounts it as root-filesystem.  This
 2748   is intended to be used on an USB-stick that after boot goes into a safe,
 2749   as it contains the LUKS-passphrase in plain text and is not secure to be
 2750   left in the system.  The script contains debug-output that should make it
 2751   easier to see what is going on.  Note that the final hand-over to the init
 2752   on the encrypted root-partition is done by "exec switch_root /mnt/root
 2753   /sbin/init", after mounting the decrypted LUKS container with "mount
 2754   /dev/mapper/c1 /mnt/root".  The second argument of switch_root is relative
 2755   to the first argument, i.e.  the init started with this command is really
 2756   /mnt/sbin/init before switch_root runs.
 2757 
 2758    #!/bin/sh
 2759    export PATH=/sbin:/bin
 2760    [ -d /sys ] || mkdir /sys
 2761    [ -d /proc ] || mkdir /proc
 2762    [ -d /tmp ] || mkdir /tmp
 2763    mount -t sysfs -o nodev,noexec,nosuid sysfs /sys
 2764    mount -t proc -o nodev,noexec,nosuid proc /proc
 2765    echo "detecting LUKS containers in sda1-10, sdb1-10"; sleep 1
 2766    for i in a b
 2767    do
 2768      for j in 1 2 3 4 5 6 7 8 9 10
 2769      do
 2770        sleep 0.5
 2771        d="/dev/sd"$i""$j
 2772        echo -n $d
 2773        cryptsetup isLuks $d >/dev/null 2>&1
 2774        r=$?
 2775        echo -n "  result: "$r""
 2776        # 0 = is LUKS, 1 = is not LUKS, 4 = other error
 2777        if expr $r = 0 > /dev/null
 2778        then
 2779          echo "  is LUKS, attempting unlock"
 2780          echo -n "test2" | cryptsetup luksOpen --key-file=- $d c1
 2781          r=$?
 2782          echo "  result of unlock attempt: "$r""
 2783          sleep 2
 2784          if expr $r = 0 > /dev/null
 2785          then
 2786            echo "*** LUKS partition unlocked, switching root *** 
 2787            echo "    (waiting 30 seconds before doing that)"
 2788            mount /dev/mapper/c1 /mnt/root
 2789            sleep 30
 2790            exec switch_root /mnt/root /sbin/init
 2791          fi
 2792        else
 2793          echo "  is not LUKS"
 2794        fi
 2795      done
 2796    done
 2797    echo "FAIL finding root on LUKS, loading BusyBox..."; sleep 5
 2798    exec /bin/sh --login
 2799 
 2800 
 2801   04) What if I want a binary in the initrd, but libraries are missing?
 2802 
 2803   That is a bit tricky.  One option is to compile statically, but that
 2804   does not work for everything.  Debian puts some libraries into lib/ and
 2805   lib64/ which are usually enough.  If you need more, you can add the
 2806   libraries you need there.  That may or may not need a configuration
 2807   change for the dynamic linker "ld" as well.  Refer to standard Linux
 2808   documentation on how to add a library to a Linux system.  A running
 2809   initrd is just a running Linux system after all, it is not special in
 2810   any way.
 2811 
 2812   05) How do I repack the initrd?
 2813 
 2814   Simply repack the changed directory. While in tmp/, do
 2815   the following:
 2816   ```
 2817   find . | cpio --create --format='newc' | gzip > ../new_initrd
 2818   ```
 2819   Rename "new_initrd" to however you want it called (the name of
 2820   the initrd is a kernel-parameter) and move to /boot. That is it.
 2821 
 2822 
 2823 10. LUKS2 Questions
 2824 
 2825 
 2826   * 10.1 Is the cryptography of LUKS2 different?
 2827 
 2828   Mostly not.  The header has changed in its structure, but the
 2829   crytpgraphy is the same.  The one exception is that PBKDF2 has been
 2830   replaced by Argon2 to give better resilience against attacks by
 2831   graphics cards and other hardware with lots of computing power but
 2832   limited local memory per computing element.
 2833 
 2834 
 2835   * 10.2 What new features does LUKS2 have?
 2836   
 2837   There are quite a few.  I recommend reading the man-page and the on-disk
 2838   format specification, see Item 1.2.
 2839 
 2840   To list just some:
 2841   - A lot of the metadata is JSON, allowing for easier extension
 2842   - Max 32 key-slots per default
 2843   - Better protection for bad passphrases now available with Argon2
 2844   - Authenticated encryption 
 2845   - The LUKS2 header is less vulnerable to corruption and has a 2nd copy
 2846   
 2847   
 2848   * 10.3 Why does LUKS2 need so much memory?
 2849 
 2850   LUKS2 uses Argon2 instead of PBKDF2.  That causes the increase in memory. 
 2851   See next item.
 2852 
 2853 
 2854   * 10.4  Why use Argon2 in LUKS 2 instead of PBKDF2?
 2855 
 2856   LUKS tries to be secure with not-so-good passwords.  Bad passwords need to
 2857   be protected in some way against an attacker that just tries all possible
 2858   combinations.  (For good passwords, you can just wait for the attacker to
 2859   die of old age...) The situation with LUKS is not quite the same as with a
 2860   password stored in a database, but there are similarities.
 2861 
 2862   LUKS does not store passwords on disk.  Instead, the passwords are used to
 2863   decrypt the master-key with it and that one is stored on disk in encrypted
 2864   form.  If you have a good password, with, say, more than 80 bits of
 2865   entropy, you could just put the password through a single crypto-hash (to
 2866   turn it into something that can be used as a key) and that would be secure. 
 2867   This is what plain dm-crypt does.
 2868 
 2869   If the password has lower entropy, you want to make this process cost some
 2870   effort, so that each try takes time and resources and slows the attacker
 2871   down.  LUKS1 uses PBKDF2 for that, adding an iteration count and a salt. 
 2872   The iteration count is per default set to that it takes 1 second per try on
 2873   the CPU of the device where the respective passphrase was set.  The salt is
 2874   there to prevent precomputation.
 2875 
 2876   The problem with that is that if you use a graphics card, you can massively
 2877   speed up these computations as PBKDF2 needs very little memory to compute
 2878   it.  A graphics card is (grossly simplified) a mass of small CPUs with some
 2879   small very fast local memory per CPU and a large slow memory (the 4/6/8 GB
 2880   a current card may have).  If you can keep a computation in the small,
 2881   CPU-local memory, you can gain a speed factor of 1000 or more when trying
 2882   passwords with PBKDF2.
 2883 
 2884   Argon2 was created to address this problem.  It adds a "large memory
 2885   property" where computing the result with less memory than the memory
 2886   parameter requires is massively (exponentially) slowed down.  That means,
 2887   if you set, for example, 4GB of memory, computing Argon2 on a graphics card
 2888   with around 100kB of memory per "CPU" makes no sense at all because it is
 2889   far too slow.  An attacker has hence to use real CPUs and furthermore is
 2890   limited by main memory bandwidth.
 2891 
 2892   Hence the large amount of memory used is a security feature and should not
 2893   be turned off or reduced.  If you really (!) understand what you are doing
 2894   and can assure good passwords, you can either go back to PBKDF2 or set a
 2895   low amount of memory used for Argon2 when creating the header.
 2896 
 2897 
 2898   * 10.5 LUKS2 is insecure! It uses less memory than the Argon2 RFC say!
 2899 
 2900   Well, not really.  The RFC recommends 6GiB of memory for use with disk
 2901   encryption.  That is a bit insane and something clearly went wrong in the
 2902   standardization process here.  First, that makes Argon2 unusable on any 32
 2903   bit Linux and that is clearly a bad thing.  Second, there are many small
 2904   Linux devices around that do not have 6GiB of RAM in the first place.  For
 2905   example, the current Raspberry Pi has 1GB, 2GB or 4GB of RAM, and with the
 2906   RFC recommendations, none of these could compute Argon2 hashes.
 2907 
 2908   Hence LUKS2 uses a more real-world approach.  Iteration is set to a
 2909   minimum of 4 because there are some theoretical attacks that work up to an
 2910   iteration count of 3.  The thread parameter is set to 4.  To achieve 2
 2911   second/slot unlock time, LUKS2 adjusts the memory parameter down if
 2912   needed.  In the other direction, it will respect available memory and not
 2913   exceed it.  On a current PC, the memory parameter will be somewhere around
 2914   1GB, which should be quite generous.  The minimum I was able to set in an
 2915   experiment with "-i 1" was 400kB of memory and that is too low to be
 2916   secure.  A Raspberry Pi would probably end up somewhere around 50MB (have
 2917   not tried it) and that should still be plenty.
 2918 
 2919   That said, if you have a good, high-entropy passphrase, LUKS2 is secure
 2920   with any memory parameter.
 2921 
 2922 
 2923   * 10.6 How does re-encryption store data while it is running?
 2924 
 2925   All metadata necessary to perform a recovery of said segment (in case of 
 2926   crash) is stored in the LUKS2 metadata area. No matter if the LUKS2 
 2927   reencryption was run in online or offline mode.
 2928 
 2929   
 2930   * 10.7 What do I do if re-encryption crashes?
 2931   
 2932   In case of a reencryption application crash, try to close the original
 2933   device via following command first: 
 2934 
 2935     cryptsetup close <my_crypt_device>. 
 2936 
 2937   Cryptsetup assesses if it's safe to teardown the reencryption device stack
 2938   or not.  It will also cut off I/O (via dm-error mapping) to current
 2939   hotzone segment (to make later recovery possible).  If it can't be torn
 2940   down, i.e.  due to a mounted fs, you must unmount the filesystem first. 
 2941   Never try to tear down reencryption dm devices manually using e.g. 
 2942   dmsetup tool, at least not unless cryptsetup says it's safe to do so.  It
 2943   could damage the data beyond repair.
 2944 
 2945 
 2946   * 10.8 Do I need to enter two passphrases to recover a crashed
 2947     re-encryption? 
 2948 
 2949   Cryptsetup (command line utility) expects the passphrases to be identical
 2950   for the keyslot containing old volume key and for the keyslot containing
 2951   new one.  So the recovery happens during normal the "cryptsetup open" 
 2952   operation or the equivalent during boot.
 2953 
 2954   Re-encryption recovery can be also performed in offline mode by 
 2955   the "cryptsetup repair" command.
 2956 
 2957 
 2958   * 10.9 What is an unbound keyslot and what is it used for?
 2959 
 2960   Quite simply, an 'unbound key' is an independent 'key' stored in a luks2 
 2961   keyslot that cannot be used to unlock a LUKS2 data device. More specifically, 
 2962   an 'unbound key' or 'unbound luks2 keyslot' contains a secret that is not
 2963   currently associated with any data/crypt segment (encrypted area) in the 
 2964   LUKS2 'Segments' section (displayed by luksDump).
 2965 
 2966   This is a bit of a more general idea. It basically allows to use a keyslot
 2967   as a container for a key to be used in other things than decrypting a 
 2968   data segment.
 2969 
 2970   As of April 2020, the following uses are defined:
 2971 
 2972   1) LUKS2 re-encryption. The new volume key is stored in an unbound keyslot 
 2973      which becomes a regular LUKS2 keyslot later when re-encryption is 
 2974      finished.
 2975   
 2976   2) Somewhat similar is the use with a wrapped key scheme (e.g. with the 
 2977      paes cipher). In this case, the VK (Volume Key) stored in a keyslot 
 2978      is an encrypted binary binary blob. The KEK (Key Encryption Key) for 
 2979      that blob may be refreshed (Note that this KEK is not managed by 
 2980      cryptsetup!) and the binary blob gets changed. The KEK refresh process 
 2981      uses an 'unbound keyslot'. First the future effective VK is placed 
 2982      in the unbound keyslot and later it gets turned into the new real VK 
 2983      (and bound to the respective crypt segment).
 2984 
 2985 
 2986   * 10.10 What about the size of the LUKS2 header?
 2987 
 2988   While the LUKS1 header has a fixed size that is determined by the cipher
 2989   spec (see Item 6.12), LUKS2 is more variable. The default size is 16MB,
 2990   but it can be adjusted on creation by using the --luks2-metadata-size 
 2991   and --luks2-keyslots-size options. Refer to the man-page for details.
 2992   While adjusting the size in an existing LUKS2 container is possible,
 2993   it is somewhat complicated and risky. My advice is to do a backup, 
 2994   recreate the container with changed parameters and restore that backup.
 2995 
 2996 
 2997   * 10.11 Does LUKS2 store metadata anywhere except in the header?
 2998  
 2999   It does not. But note that if you use the experimental integrity support,
 3000   there will be an integrity header as well at the start of the data area 
 3001   and things  get a bit more complicated. All metadata will still be at the 
 3002   start of the device, nothing gets stored somewhere in the middle or at 
 3003   the end. 
 3004   
 3005   * 10.12 What is a LUKS2 Token?
 3006 
 3007   A LUKS2 token is an object that describes "how to get a passphrase or 
 3008   key" to unlock particular keyslot. A LUKS2 token is stored as json data 
 3009   in the LUKS2 header. The token can be related to all keyslots or a 
 3010   specific one. As the token is stored in JSON formay it is text by 
 3011   default but binary data can be encoded into it according to the JSON 
 3012   conventions.
 3013  
 3014   Documentation on the last changes to LUKS2 tokens can be found in the 
 3015   release notes. As of version 2.4 of cryptsetup, there are significant 
 3016   features. The standard documentation for working with tokens is 
 3017   in the luks2 reference available as PDF on the project page.
 3018 
 3019 
 3020 11. References and Further Reading
 3021 
 3022 
 3023   * Purpose of this Section
 3024 
 3025   The purpose of this section is to collect references to all materials
 3026   that do not fit the FAQ but are relevant in some fashion.  This can be
 3027   core topics like the LUKS spec or disk encryption, but it can also be
 3028   more tangential, like secure storage management or cryptography used in
 3029   LUKS.  It should still have relevance to cryptsetup and its
 3030   applications.
 3031 
 3032   If you want to see something added here, send email to the maintainer
 3033   (or the cryptsetup mailing list) giving an URL, a description (1-3 lines
 3034   preferred) and a section to put it in.  You can also propose new
 3035   sections.
 3036 
 3037   At this time I would like to limit the references to things that are
 3038   available on the web.
 3039 
 3040   * Specifications
 3041 
 3042   - LUKS on-disk format spec: See Item 1.2
 3043 
 3044   * Other Documentation
 3045   
 3046   - Arch Linux on LUKS, LVM and full-disk encryption: 
 3047     https://wiki.archlinux.org/index.php/Dm-crypt/Encrypting_an_entire_system
 3048 
 3049   * Code Examples
 3050 
 3051   - Some code examples are in the source package under docs/examples
 3052 
 3053   - LUKS AF Splitter in Ruby by John Lane: https://rubygems.org/gems/afsplitter
 3054 
 3055   * Brute-forcing passphrases
 3056 
 3057   - http://news.electricalchemy.net/2009/10/password-cracking-in-cloud-part-5.html
 3058 
 3059   - https://it.slashdot.org/story/12/12/05/0623215/new-25-gpu-monster-devours-strong-passwords-in-minutes
 3060 
 3061   * Tools
 3062 
 3063   * SSD and Flash Disk Related
 3064 
 3065   * Disk Encryption
 3066 
 3067   * Attacks Against Disk Encryption
 3068 
 3069   * Risk Management as Relevant for Disk Encryption
 3070 
 3071   * Cryptography
 3072 
 3073   * Secure Storage
 3074 
 3075 
 3076 A. Contributors
 3077 In no particular order:
 3078 
 3079   - Arno Wagner
 3080 
 3081   - Milan Broz
 3082 
 3083 ___