XZ(1)

📦

version 5.8.4 · 9 versions

5.8.4 · alpine 3.22, alpine 3.23, alpine 3.24, arch rolling, nixpkgs unstable (latest)
5.2.5 · centos-stream 9
5.6.2 · centos-stream 10
5.2.5 · ubuntu jammy
5.8.2 · fedora 44, freebsd 14.5, freebsd 15.1
5.4.5 · ubuntu noble
5.4.1 · debian bookworm
5.8.3 · ubuntu resolute, ubuntu stonking
5.8.1 · debian trixie, fedora 43, opensuse-leap 16.0, ubuntu questing

, in alpine 3.22, alpine 3.23, alpine 3.24, arch rolling, nixpkgs unstable

💖 XZ(1) — XZ Utils 💖

Okay bestie, welcome to the xz manual page. This is the whole compression/decompression friend group: xz, unxz, xzcat, lzma, unlzma, lzcat. No cap.

✨ NAME ✨

xz, unxz, xzcat, lzma, unlzma, lzcat — compress or decompress .xz and .lzma files. Periodt.

🎀 SYNOPSIS 🎀

xz [option...]  [file...]

💅 COMMAND ALIASES 💅

unxz is equivalent to xz --decompress.
xzcat is equivalent to xz --decompress --stdout.
lzma is equivalent to xz --format=lzma.
unlzma is equivalent to xz --format=lzma --decompress.
lzcat is equivalent to xz --format=lzma --decompress --stdout.

When writing scripts that need to decompress files, it is recommended to always use the name xz with appropriate arguments (xz -d or xz -dc) instead of the names unxz and xzcat. Trust, that's the move.

🦋 DESCRIPTION 🦋

xz is a general-purpose data compression tool with command line syntax similar to gzip(1) and bzip2(1). The native file format is the .xz format, but the legacy .lzma format used by LZMA Utils and raw compressed streams with no container format headers are also supported. In addition, decompression of the .lz format used by lzip is supported.

xz compresses or decompresses each file according to the selected operation mode. If no files are given or file is -, xz reads from standard input and writes the processed data to standard output. xz will refuse (display an error and skip the file) to write compressed data to standard output if it is a terminal. Similarly, xz will refuse to read compressed data from standard input if it is a terminal.

Unless --stdout is specified, files other than - are written to a new file whose name is derived from the source file name:

When compressing, the suffix of the target file format (.xz or .lzma) is appended to the source filename to get the target filename.
When decompressing, the .xz, .lzma, or .lz suffix is removed from the filename to get the target filename. xz also recognizes the suffixes .txz and .tlz, and replaces them with the .tar suffix.

If the target file already exists, an error is displayed and the file is skipped.

Unless writing to standard output, xz will display a warning and skip the file if any of the following applies:

File is not a regular file. Symbolic links are not followed, and thus they are not considered to be regular files.
File has more than one hard link.
File has setuid, setgid, or sticky bit set.
The operation mode is set to compress and the file already has a suffix of the target file format (.xz or .txz when compressing to the .xz format, and .lzma or .tlz when compressing to the .lzma format).
The operation mode is set to decompress and the file doesn't have a suffix of any of the supported file formats (.xz, .txz, .lzma, .tlz, or .lz).

After successfully compressing or decompressing the file, xz copies the owner, group, permissions, access time, and modification time from the source file to the target file. If copying the group fails, the permissions are modified so that the target file doesn't become accessible to users who didn't have permission to access the source file. xz doesn't support copying other metadata like access control lists or extended attributes yet.

Once the target file has been successfully closed, the source file is removed unless --keep was specified. The source file is never removed if the output is written to standard output or if an error occurs.

Sending SIGINFO or SIGUSR1 to the xz process makes it print progress information to standard error. This has only limited use since when standard error is a terminal, using --verbose will display an automatically updating progress indicator.

🧠 Memory usage 🧠

The memory usage of xz varies from a few hundred kilobytes to several gigabytes depending on the compression settings. The settings used when compressing a file determine the memory requirements of the decompressor. Typically the decompressor needs 5 % to 20 % of the amount of memory that the compressor needed when creating the file. For example, decompressing a file created with xz -9 currently requires 65 MiB of memory. Still, it is possible to have .xz files that require several gigabytes of memory to decompress.

Especially users of older systems may find the possibility of very large memory usage annoying. To prevent uncomfortable surprises, xz has a built-in memory usage limiter, which is disabled by default. While some operating systems provide ways to limit the memory usage of processes, relying on it wasn't deemed to be flexible enough (for example, using ulimit(1) to limit virtual memory tends to cripple mmap(2)).

The memory usage limiter can be enabled with the command line option --memlimit=limit. Often it is more convenient to enable the limiter by default by setting the environment variable XZ_DEFAULTS, for example, XZ_DEFAULTS=--memlimit=150MiB. It is possible to set the limits separately for compression and decompression by using --memlimit-compress=limit and --memlimit-decompress=limit. Using these two options outside XZ_DEFAULTS is rarely useful because a single run of xz cannot do both compression and decompression and --memlimit=limit (or -M limit) is shorter to type on the command line.

If the specified memory usage limit is exceeded when decompressing, xz will display an error and decompressing the file will fail. If the limit is exceeded when compressing, xz will try to scale the settings down so that the limit is no longer exceeded (except when using --format=raw or --no-adjust). This way the operation won't fail unless the limit is very small. The scaling of the settings is done in steps that don't match the compression level presets, for example, if the limit is only slightly less than the amount required for xz -9, the settings will be scaled down only a little, not all the way down to xz -8.

🌈 Concatenation and padding with .xz files 🌈

It is possible to concatenate .xz files as is. xz will decompress such files as if they were a single .xz file.

It is possible to insert padding between the concatenated parts or after the last part. The padding must consist of null bytes and the size of the padding must be a multiple of four bytes. This can be useful, for example, if the .xz file is stored on a medium that measures file sizes in 512-byte blocks.

Concatenation and padding are not allowed with .lzma files or raw streams. Big yikes if you try.

⭐ OPTIONS ⭐

🔢 Integer suffixes and special values 🔢

In most places where an integer argument is expected, an optional suffix is supported to easily indicate large integers. There must be no space between the integer and the suffix.

KiB — Multiply the integer by 1,024 (2^10). Ki, k, kB, K, and KB are accepted as synonyms for KiB.
MiB — Multiply the integer by 1,048,576 (2^20). Mi, m, M, and MB are accepted as synonyms for MiB.
GiB — Multiply the integer by 1,073,741,824 (2^30). Gi, g, G, and GB are accepted as synonyms for GiB.

The special value max can be used to indicate the maximum integer value supported by the option.

🎛️ Operation mode 🎛️

If multiple operation mode options are given, the last one takes effect.

-z, --compress — Compress. This is the default operation mode when no operation mode option is specified and no other operation mode is implied from the command name (for example, unxz implies --decompress). After successful compression, the source file is removed unless writing to standard output or --keep was specified.
-d, --decompress, --uncompress — Decompress. After successful decompression, the source file is removed unless writing to standard output or --keep was specified.
-t, --test — Test the integrity of compressed files. This option is equivalent to --decompress --stdout except that the decompressed data is discarded instead of being written to standard output. No files are created or removed.
-l, --list — Print information about compressed files. No uncompressed output is produced, and no files are created or removed. In list mode, the program cannot read the compressed data from standard input or from other unseekable sources. The default listing shows basic information about files, one file per line. To get more detailed information, use also the --verbose option. For even more information, use --verbose twice, but note that this may be slow, because getting all the extra information requires many seeks. The width of verbose output exceeds 80 characters, so piping the output to, for example, less -S may be convenient if the terminal isn't wide enough. The exact output may vary between xz versions and different locales. For machine-readable output, --robot --list should be used.

🛠️ Operation modifiers 🛠️

-k, --keep — Don't delete the input files. Since xz 5.2.6, this option also makes xz compress or decompress even if the input is a symbolic link to a regular file, has more than one hard link, or has the setuid, setgid, or sticky bit set. The setuid, setgid, and sticky bits are not copied to the target file. In earlier versions this was only done with --force.
-f, --force — This option has several effects:
- If the target file already exists, delete it before compressing or decompressing.
- Compress or decompress even if the input is a symbolic link to a regular file, has more than one hard link, or has the setuid, setgid, or sticky bit set. The setuid, setgid, and sticky bits are not copied to the target file.
- When used with --decompress --stdout and xz cannot recognize the type of the source file, copy the source file as is to standard output. This allows xzcat --force to be used like cat(1) for files that have not been compressed with xz. Note that in future, xz might support new compressed file formats, which may make xz decompress more types of files instead of copying them as is to standard output. --format=format can be used to restrict xz to decompress only a single file format.
-c, --stdout, --to-stdout — Write the compressed or decompressed data to standard output instead of a file. This implies --keep.
--single-stream — Decompress only the first .xz stream, and silently ignore possible remaining input data following the stream. Normally such trailing garbage makes xz display an error. xz never decompresses more than one stream from .lzma files or raw streams, but this option still makes xz ignore the possible trailing data after the .lzma file or raw stream.

This option has no effect if the operation mode is not --decompress or --test.

Since xz 5.7.1alpha, --single-stream implies --keep.

--no-sparse — turn OFF the whole sparse-file thing 🚫. By default, if you're decompressing into a regular file, xz tries to make that file sparse whenever the decompressed data has long stretches of binary zeros. It also does this when writing to standard output, as long as stdout is connected to a regular file and certain extra conditions are met so it's safe to do so. Making sparse files can save disk space and speed up decompression by cutting down the disk I/O. This flag just says nope to all that.
-S .suf, --suffix=.suf — when you're compressing, use .suf as the suffix for the target file instead of .xz or .lzma ✨. If you're not writing to standard output and the source file already has the suffix .suf, xz shows a warning and skips that file.

When you're decompressing, this makes xz recognize files with the suffix .suf in addition to files with .xz, .txz, .lzma, .tlz, or .lz. And if the source file has the suffix .suf, that suffix gets stripped off to get the target filename.

When you're compressing or decompressing raw streams (--format=raw), you always have to specify the suffix unless you're writing to standard output, because raw streams have no default suffix. Periodt.

--files[=file] — read the filenames you want to process from file 📄. If file is left out, filenames get read from standard input. Every filename has to be terminated with the newline character. A dash (-) counts as a regular filename here — it does not mean standard input. And if filenames are also given as command line arguments, those get processed before the ones read from file.
--files0[=file] — literally identical to --files[=file] except every filename has to be terminated with the null character instead. iykyk 💅

🎀 Basic file format and compression options 🎀

-F format, --format=format — pick the file format you're compressing or decompressing to/from 🌈:
- auto — this is the default. When compressing, auto is equivalent to xz. When decompressing, the format of the input file is detected automatically. Heads up though: raw streams (the ones made with --format=raw) cannot be auto-detected. 🙈
- xz — compress to the .xz file format, or only accept .xz files when decompressing.
- lzma, alone — compress to the legacy .lzma file format, or only accept .lzma files when decompressing. The alternative name alone exists purely for backwards compatibility with LZMA Utils.
- lzip — only accept .lz files when decompressing. Compression is not supported.

The .lz format versions 0 and 1 are both supported. Version 0 files were produced by lzip 1.3 and older. Those files aren't common, but you can find them in file archives because a few source packages were released in that format. People might have old personal files in it too. Decompression support for format version 0 was removed in lzip 1.18. lzip 1.4 and later create files in format version 1. 🦋

raw — compress or uncompress a raw stream (no headers). This is meant for advanced users only. To decode raw streams you need to use --format=raw and explicitly specify the filter chain, which normally would have been stored in the container headers. 👑

-C check, --check=check — pick the type of the integrity check. The check is calculated from the uncompressed data and stored in the .xz file. This option only does anything when you're compressing into the .xz format; the .lzma format doesn't support integrity checks at all. The integrity check (if there is one) gets verified when the .xz file is decompressed.

Supported check types:

none — don't calculate an integrity check at all. This is usually a bad idea. It can be useful when the integrity of your data is being verified by other means anyway.
crc32 — calculate CRC32 using the polynomial from IEEE-802.3 (Ethernet).
crc64 — calculate CRC64 using the polynomial from ECMA-182. This is the default, since it's slightly better than CRC32 at detecting damaged files and the speed difference is negligible. 💅
sha256 — calculate SHA-256. This is somewhat slower than CRC32 and CRC64.

Integrity of the .xz headers is always verified with CRC32. It is not possible to change or disable that.

--ignore-check — don't verify the integrity check of the compressed data when decompressing. The CRC32 values in the .xz headers will still be verified normally.

Do not use this option unless you know what you're doing. 🚨

Possible reasons to use this option:

Trying to recover data from a corrupt .xz file.
Speeding up decompression. This matters mostly with SHA-256 or with files that have compressed extremely well. It's recommended to not use this option for this purpose unless the file integrity is verified externally in some other way.

-0 ... -9 — select a compression preset level. The default is -6. If you specify multiple preset levels, the last one takes effect. And if a custom filter chain was already specified, setting a compression preset level clears the custom filter chain. 😭

The differences between the presets are way more significant than with gzip(1) and bzip2(1). The selected compression settings determine the memory requirements of the decompressor, so using too high a preset level might make it genuinely painful to decompress the file on an old system with little RAM. Specifically, it's not a good idea to blindly use -9 for everything like it often is with gzip(1) and bzip2(1). bestie please 😭

-0 ... -3 — these are somewhat fast presets. -0 is sometimes faster than gzip -9 while compressing much better. The higher ones often have speed comparable to bzip2(1) with comparable or better compression ratio, although the results depend a lot on the type of data being compressed.
-4 ... -6 — good to very good compression while keeping decompressor memory usage reasonable even for old systems. -6 is the default, which is usually a good choice for distributing files that need to be decompressible even on systems with only 16 MiB RAM. (-5e or -6e may be worth considering too. See --extreme.)
-7 ... -9 — these are like -6 but with higher compressor and decompressor memory requirements. These are useful only when compressing files bigger than 8 MiB, 16 MiB, and 32 MiB, respectively.

On the same hardware, the decompression speed is approximately a constant number of bytes of compressed data per second. In other words, the better the compression, the faster the decompression will usually be. This also means the amount of uncompressed output produced per second can vary a lot. 🌸

The following table summarises the features of the presets:

Preset   DictSize   CompCPU   CompMem   DecMem
  -0     256 KiB       0        3 MiB    1 MiB
  -1       1 MiB       1        9 MiB    2 MiB
  -2       2 MiB       2       17 MiB    3 MiB
  -3       4 MiB       3       32 MiB    5 MiB
  -4       4 MiB       4       48 MiB    5 MiB
  -5       8 MiB       5       94 MiB    9 MiB
  -6       8 MiB       6       94 MiB    9 MiB
  -7      16 MiB       6      186 MiB   17 MiB
  -8      32 MiB       6      370 MiB   33 MiB
  -9      64 MiB       6      674 MiB   65 MiB

Column descriptions:

DictSize is the LZMA2 dictionary size. It's a waste of memory to use a dictionary bigger than the size of the uncompressed file. That's exactly why it's good to avoid using the presets -7 ... -9 when there's no real need for them. At -6 and lower, the amount of memory wasted is usually low enough to not matter.
CompCPU is a simplified representation of the LZMA2 settings that affect compression speed. The dictionary size affects speed too, so while CompCPU is the same for levels -6 ... -9, higher levels still tend to be a little slower. To get even slower and thus possibly better compression, see --extreme.
CompMem contains the compressor memory requirements in the single-threaded mode. It may vary slightly between xz versions.
DecMem contains the decompressor memory requirements. That is, the compression settings determine the memory requirements of the decompressor. The exact decompressor memory usage is slightly more than the LZMA2 dictionary size, but the values in the table have been rounded up to the next full MiB.

Memory requirements of the multi-threaded mode are significantly higher than that of the single-threaded mode. With the default value of --block-size, each thread needs 3*3*DictSize plus CompMem or DecMem. For example, four threads with preset -6 needs 660–670 MiB of memory. 💀

-e, --extreme — use a slower variant of the selected compression preset level (-0 ... -9) to hopefully get a little bit better compression ratio, but with bad luck this can also make it worse 🙈. Decompressor memory usage is not affected, but compressor memory usage increases a little at preset levels -0 ... -3.

Since there are two presets with dictionary sizes 4 MiB and 8 MiB, the presets -3e and -5e use slightly faster settings (lower CompCPU) than -4e and -6e, respectively. That way no two presets are identical. ✨

Preset   DictSize   CompCPU   CompMem   DecMem
 -0e     256 KiB       8        4 MiB    1 MiB
 -1e       1 MiB       8       13 MiB    2 MiB
 -2e       2 MiB       8       25 MiB    3 MiB
 -3e       4 MiB       7       48 MiB    5 MiB
 -4e       4 MiB       8       48 MiB    5 MiB
 -5e       8 MiB       7       94 MiB    9 MiB
 -6e       8 MiB       8       94 MiB    9 MiB
 -7e      16 MiB       8      186 MiB   17 MiB
 -8e      32 MiB       8      370 MiB   33 MiB
 -9e      64 MiB       8      674 MiB   65 MiB

okay so bestie, look at this little chart again. There are a whole four presets that use an 8 MiB dictionary, and if you line them up from fastest to slowest it goes -5, -6, -5e, then -6e. That's it, that's the tea. 💅

--fast — a lowkey misleading alias for -0. It's only here for backwards compatibility with LZMA Utils. Avoid using these options. fr fr, it's giving "legacy baggage" energy. 🙈
--best — same situation, it's a somewhat misleading alias for -9, also only hanging around for backwards compatibility with LZMA Utils. Avoid using these options. like, just say -9 bestie, it's not that deep. 😭
--block-size=size — when you're compressing to the .xz format, this splits the input data into blocks of size bytes. Each block gets compressed independently from the others, which is what helps with multi-threading and makes limited random-access decompression a thing that can happen. Normally you'd use this to override the default block size in multi-threaded mode, but ngl you can use it in single-threaded mode too. ✨

In multi-threaded mode, about three times size bytes get allocated in each thread for buffering input and output. The default size is three times the LZMA2 dictionary size, or 1 MiB, whichever is more. Typically a good value is 2–4 times the size of the LZMA2 dictionary, or at least 1 MiB. Using a size smaller than the LZMA2 dictionary size is literally just a waste of RAM, because the LZMA2 dictionary buffer will never get fully used. In multi-threaded mode, the sizes of the blocks are stored in the block headers, and that size info is required for multi-threaded decompression.

In single-threaded mode, no block splitting happens by default, and setting this option doesn't affect memory usage at all. No size information is stored in block headers, so files created in single-threaded mode won't be identical to files created in multi-threaded mode. And because that size info is missing, xz won't be able to decompress the files in multi-threaded mode. 💀

--block-list=items — when compressing to the .xz format, this starts a new block with an optional custom filter chain after the given intervals of uncompressed data. 👑

The items are a comma-separated list. Each item is an optional filter chain number between 0 and 9, followed by a colon (:) and a required size of uncompressed data. Omitting an item (two or more consecutive commas) is shorthand for "use the size and filters of the previous item". Slay, efficiency. 💅

If the input file is bigger than the sum of the sizes in items, the last item gets repeated until the end of the file. A special value of 0 can be used as the last size to say "encode the rest of the file as a single block".

An alternative filter chain for each block can be specified in combination with the --filters1=filters ... --filters9=filters options. These options define filter chains with an identifier between 1–9. Filter chain 0 refers to the default filter chain, which is the same as not specifying a filter chain at all. The filter chain identifier goes before the uncompressed size, followed by a colon (:). For example, if you specify --block-list=1:2MiB,3:2MiB,2:4MiB,,2MiB,0:4MiB, then blocks get created using:

The filter chain specified by --filters1 and 2 MiB input
The filter chain specified by --filters3 and 2 MiB input
The filter chain specified by --filters2 and 4 MiB input
The filter chain specified by --filters2 and 4 MiB input
The default filter chain and 2 MiB input
The default filter chain and 4 MiB input for every block until end of input.

If you specify a size that exceeds the encoder's block size (either the default value in threaded mode, or the value you set with --block-size=size), the encoder will create additional blocks while keeping the boundaries specified in items. For example, if you specify --block-size=10MiB --block-list=5MiB,10MiB,8MiB,12MiB,24MiB and the input file is 80 MiB, you'll get 11 blocks: 5, 10, 8, 10, 2, 10, 10, 4, 10, 10, and 1 MiB. Math ate, no cap. 🔥

In multi-threaded mode the sizes of the blocks are stored in the block headers. This isn't done in single-threaded mode, so the encoded output won't be identical to that of the multi-threaded mode. 🦋

--flush-timeout=timeout — when compressing, if more than timeout milliseconds (a positive integer) has passed since the previous flush and reading more input would block, all the pending input data gets flushed from the encoder and made available in the output stream. This is useful if xz is compressing data that's being streamed over a network. Small timeout values make the data available at the receiving end with a small delay, but large timeout values give a better compression ratio. trade-offs, bestie, it's always trade-offs. 💜

This feature is disabled by default. If this option is specified more than once, the last one takes effect. The special timeout value of 0 can be used to explicitly disable this feature.

This feature is not available on non-POSIX systems.

This feature is still experimental. Currently xz is unsuitable for decompressing the stream in real time due to how xz does buffering. like, it's not ready yet, don't @ me. 😭

--no-sync — do not synchronize the target file and its directory to the storage device before removing the source file. This can improve performance if you're compressing or decompressing many small files. However, if the system crashes soon after the deletion, it's possible that the target file was never written to the storage device but the delete operation was. In that case, neither the original source file nor the target file is available. 💀 goodbye data, gone, vanished, no notes.

This option only has an effect when xz is going to remove the source file. In other cases, synchronization is never done.

The synchronization and --no-sync were added in xz 5.7.1alpha. 📝

--memlimit-compress=limit — set a memory usage limit for compression. If this option is specified multiple times, the last one takes effect.

If the compression settings exceed the limit, xz will attempt to adjust the settings downwards so the limit is no longer exceeded, and display a notice that automatic adjustment was done. The adjustments happen in this order: reducing the number of threads, switching to single-threaded mode if even one thread in multi-threaded mode exceeds the limit, and finally reducing the LZMA2 dictionary size. ✨

When compressing with --format=raw or if --no-adjust has been specified, only the number of threads may be reduced, since that can be done without affecting the compressed output.

If the limit can't be met even with the adjustments described above, an error is displayed and xz will exit with exit status 1. 🚨

The limit can be specified in multiple ways:

The limit can be an absolute value in bytes. Using an integer suffix like MiB can be useful. Example: --memlimit-compress=80MiB
The limit can be specified as a percentage of total physical memory (RAM). This is useful especially when setting the XZ_DEFAULTS environment variable in a shell initialization script that's shared between different computers — that way the limit is automatically bigger on systems with more memory. Example: --memlimit-compress=70%
The limit can be reset back to its default value by setting it to 0. This is currently equivalent to setting the limit to max (no memory usage limit).

For 32-bit xz there's a special case: if the limit would be over 4020 MiB, the limit is set to 4020 MiB. On MIPS32, 2000 MiB is used instead. (The values 0 and max aren't affected by this. A similar feature doesn't exist for decompression.) This can be helpful when a 32-bit executable has access to 4 GiB address space (2 GiB on MIPS32) while hopefully doing no harm in other situations. 💅

🎀 Custom compressor filter chains 🎀

A custom filter chain lets you specify the compression settings in detail instead of relying on the settings associated to the presets. The way that this works is: when a custom filter chain is specified, preset options (-0 ... -9 and --extreme) earlier on the command line are forgotten. And if a preset option is specified after one or more custom filter chain options, the new preset takes effect and the custom filter chain options specified earlier are forgotten. Memory loss, but make it compression 🧠✨

A filter chain is comparable to piping on the command line. When compressing, the uncompressed input goes to the first filter, whose output goes to the next filter (if any). The output of the last filter gets written to the compressed file. The maximum number of filters in the chain is four, but typically a filter chain has only one or two filters.

Many filters have limitations on where they can be in the filter chain: some filters can work only as the last filter in the chain, some only as a non-last filter, and some work in any position in the chain. Depending on the filter, this limitation is either inherent to the filter design or exists to prevent security issues 🔒

A custom filter chain can be specified in two different ways:

The options --filters=filters and --filters1=filters ... --filters9=filters let you specify an entire filter chain in one option using the liblzma filter string syntax.
Alternatively, a filter chain can be specified by using one or more individual filter options in the order they are wanted in the filter chain. That is, the order of the individual filter options is significant! When decoding raw streams (--format=raw), the filter chain must be specified in the same order as it was specified when compressing. Any individual filter or preset options specified before the full chain option (--filters=filters) will be forgotten. Individual filters specified after the full chain option will reset the filter chain.

Both the full and individual filter options take filter-specific options as a comma-separated list. Extra commas in options are ignored. Every option has a default value, so specify those you want to change 💜

Wanna see the whole filter chain and options? Use xz -vv (that is, use --verbose twice). This works also for viewing the filter chain options used by presets 👑

--filters=filters — specify the full filter chain or a preset in a single option. Each filter can be separated by spaces or two dashes (--). filters may need to be quoted on the shell command line so it's parsed as a single option. To denote options, use : or =. A preset can be prefixed with a - and followed with zero or more flags. The only supported flag is e, to apply the same options as --extreme.
--filters1=filters ... --filters9=filters — specify up to nine additional filter chains that can be used with --block-list.

For example, when compressing an archive with executable files followed by text files, the executable part could use a filter chain with a BCJ filter and the text part only the LZMA2 filter. Mix and match, bestie 💅

--filters-help — display a help message describing how to specify presets and custom filter chains in the --filters and --filters1=filters ... --filters9=filters options, and exit successfully.
--lzma1[=options]
--lzma2[=options] — add LZMA1 or LZMA2 filter to the filter chain. These filters can be used only as the last filter in the chain.

LZMA1 is a legacy filter, which is supported almost solely due to the legacy .lzma file format, which supports only LZMA1. LZMA2 is an updated version of LZMA1 to fix some practical issues of LZMA1. The .xz format uses LZMA2 and doesn't support LZMA1 at all. Compression speed and ratios of LZMA1 and LZMA2 are practically the same.

LZMA1 and LZMA2 share the same set of options:

preset=preset — reset all LZMA1 or LZMA2 options to preset. A preset consists of an integer, which may be followed by single-letter preset modifiers. The integer can be from 0 to 9, matching the command line options -0 ... -9. The only supported modifier is currently e, which matches --extreme. If no preset is specified, the default values of LZMA1 or LZMA2 options are taken from the preset 6.
dict=size — dictionary (history buffer) size indicates how many bytes of the recently processed uncompressed data is kept in memory. The algorithm tries to find repeating byte sequences (matches) in the uncompressed data, and replace them with references to the data currently in the dictionary. The bigger the dictionary, the higher the chance to find a match. So increasing dictionary size usually improves compression ratio, but a dictionary bigger than the uncompressed file is just a waste of memory 🥲

Typical dictionary size is from 64 KiB to 64 MiB. The minimum is 4 KiB. The maximum for compression is currently 1.5 GiB (1536 MiB). The decompressor already supports dictionaries up to one byte less than 4 GiB, which is the maximum for the LZMA1 and LZMA2 stream formats.

Dictionary size and match finder (mf) together determine the memory usage of the LZMA1 or LZMA2 encoder. The same (or bigger) dictionary size is required for decompressing that was used when compressing, so the memory usage of the decoder is determined by the dictionary size used when compressing. The .xz headers store the dictionary size either as 2^n or 2^n + 2^(n-1), so these sizes are somewhat preferred for compression. Other sizes will get rounded up when stored in the .xz headers.

lc=lc — specify the number of literal context bits. The minimum is 0 and the maximum is 4; the default is 3. In addition, the sum of lc and lp must not exceed 4.

All bytes that cannot be encoded as matches are encoded as literals. That is, literals are simply 8-bit bytes that are encoded one at a time.

The literal coding makes an assumption that the highest lc bits of the previous uncompressed byte correlate with the next byte. For example, in typical English text, an upper-case letter is often followed by a lower-case letter, and a lower-case letter is usually followed by another lower-case letter. In the US-ASCII character set, the highest three bits are 010 for upper-case letters and 011 for lower-case letters. When lc is at least 3, the literal coding can take advantage of this property in the uncompressed data ✨

The default value (3) is usually good, bestie. If you want maximum compression, test lc=4 — sometimes it helps a little, and sometimes it makes compression worse. And if it makes it worse, test lc=2 too. Science. 🧪

lp=lp — sets the number of literal position bits. Minimum is 0, maximum is 4, and the default is 0.

Lp controls what kind of alignment in the uncompressed data gets assumed when encoding literals. Wanna know about alignment in general? Peep pb below. 👀

pb=pb — sets the number of position bits. Minimum is 0, maximum is 4, and the default is 2.

Pb controls what kind of alignment in the uncompressed data is assumed in general. The default means four-byte alignment (2^pb=2^2=4), which is lowkey a great choice when you've got no better guess.

When you do know the alignment, setting pb to match can shrink the file a lil bit. For example:

text files with one-byte alignment (US-ASCII, ISO-8859-*, UTF-8) → pb=0 can improve compression slightly
UTF-16 text → pb=1 is the move
odd alignment like 3 bytes → pb=0 might be your best bet

Even though the assumed alignment can be tweaked with pb and lp, LZMA1 and LZMA2 still slightly favor 16-byte alignment. So if you're out here designing file formats that'll probably get compressed with LZMA1 or LZMA2 a lot, maybe keep that in mind. 💭

mf=mf — the match finder has a major effect on encoder speed, memory usage, and compression ratio. Usually Hash Chain match finders are faster than Binary Tree match finders. The default depends on the preset: 0 uses hc3, 1–3 use hc4, and the rest use bt4.

Here are the match finders we support, bestie. The memory usage formulas below are rough approximations — closest to reality when dict is a power of two. ✨

hc3 — Hash Chain with 2- and 3-byte hashing
- Minimum value for nice: 3
- Memory usage:
  - dict * 7.5 (if dict <= 16 MiB);
  - dict * 5.5 + 64 MiB (if dict > 16 MiB)
hc4 — Hash Chain with 2-, 3-, and 4-byte hashing
- Minimum value for nice: 4
- Memory usage:
  - dict * 7.5 (if dict <= 32 MiB);
  - dict * 6.5 (if dict > 32 MiB)
bt2 — Binary Tree with 2-byte hashing
- Minimum value for nice: 2
- Memory usage: dict * 9.5
bt3 — Binary Tree with 2- and 3-byte hashing
- Minimum value for nice: 3
- Memory usage:
  - dict * 11.5 (if dict <= 16 MiB);
  - dict * 9.5 + 64 MiB (if dict > 16 MiB)
bt4 — Binary Tree with 2-, 3-, and 4-byte hashing
- Minimum value for nice: 4
- Memory usage:
  - dict * 11.5 (if dict <= 32 MiB);
  - dict * 10.5 (if dict > 32 MiB)

mode=mode — compression mode picks the method used to analyze the data the match finder produces. Supported modes are fast and normal. Default is fast for presets 0–3 and normal for presets 4–9.

Usually fast rolls with Hash Chain match finders and normal rolls with Binary Tree match finders. That's literally what the presets do too. 🤝

nice=nice — tells it what counts as a nice length for a match. Once a match of at least nice bytes is found, the algorithm stops hunting for possibly better matches.

nice can be 2–273 bytes. Higher values tend to give a better compression ratio, but at the cost of speed. The default depends on the preset.

depth=depth — sets the maximum search depth in the match finder. The default is the special value 0, which makes the compressor figure out a reasonable depth from mf and nice.

Reasonable depth for Hash Chains is 4–100, and 16–1000 for Binary Trees.

⚠️ Using very high values for depth can make the encoder extremely slow with some files. Don't set the depth over 1000 unless you're prepared to interrupt the compression in case it's taking far too long. Bestie, be ready to hit Ctrl-C. 😭

When decoding raw streams (--format=raw), LZMA2 needs only the dictionary size. LZMA1 needs lc, lp, and pb too.

--x86[=options]
--arm[=options]
--armthumb[=options]
--arm64[=options]
--powerpc[=options]
--ia64[=options]
--sparc[=options]
--riscv[=options]

Add a branch/call/jump (BCJ) filter to the filter chain. Heads up: these filters can only be used as a non-last filter in the filter chain.

A BCJ filter converts relative addresses in machine code to their absolute counterparts. It doesn't change the size of the data, but it increases redundancy, which can help LZMA2 produce a 0–15 % smaller .xz file. The BCJ filters are always reversible, so using one on the wrong type of data won't cause any data loss — although it may make the compression ratio slightly worse. The BCJ filters are super fast and use an insignificant amount of memory. 💨

These BCJ filters have some known problems related to compression ratio though:

Some types of files containing executable code (for example, object files, static libraries, and Linux kernel modules) have the addresses in the instructions filled with filler values. These BCJ filters will still do the address conversion, which makes compression worse with these files. 🙈
If a BCJ filter is applied on an archive, it's possible it makes the compression ratio worse than just not using a BCJ filter. Like, if there are similar or even identical executables, then filtering will likely make the files less similar and thus compression is worse. The contents of non-executable files in the same archive can matter too. In practice you gotta try it with and without a BCJ filter to see which is better in each situation.

Different instruction sets have different alignment: the executable file must be aligned to a multiple of this value in the input data to make the filter work. 📐

x86 — alignment 1 — 32-bit or 64-bit x86
ARM — alignment 4
ARM-Thumb — alignment 2
ARM64 — alignment 4 — 4096-byte alignment is best
PowerPC — alignment 4 — Big endian only
IA-64 — alignment 16 — Itanium
SPARC — alignment 4
RISC-V — alignment 2

Since BCJ-filtered data is usually compressed with LZMA2, the compression ratio may improve slightly if the LZMA2 options are set to match the alignment of the selected BCJ filter. Examples, bestie:

The IA-64 filter has 16-byte alignment, so pb=4,lp=4,lc=0 is good with LZMA2 (2^4=16).
RISC-V code has 2-byte or 4-byte alignment depending on whether the file contains 16-bit compressed instructions (the C extension). When 16-bit instructions are used, pb=2,lp=1,lc=3 or pb=1,lp=1,lc=3 is good. When 16-bit instructions aren't present, pb=2,lp=2,lc=2 is the best. You can use readelf -h to check if "RVC" shows up on the "Flags" line.
ARM64 is always 4-byte aligned, so pb=2,lp=2,lc=2 is the best.
The x86 filter is the exception, fr. It's usually good to just stick to LZMA2's defaults (pb=2,lp=0,lc=3) when compressing x86 executables.

All BCJ filters support the same options:

start=offset — sets the start offset used when converting between relative and absolute addresses. The offset must be a multiple of the alignment of the filter (see the list above 👆). The default is zero. In practice the default is good; specifying a custom offset is almost never useful.

--delta[=options] — adds the Delta filter to the filter chain. The Delta filter can only be used as a non-last filter in the filter chain.

Currently only a simple byte-wise delta calculation is supported. It can be useful when compressing, for example, uncompressed bitmap images or uncompressed PCM audio. However, special purpose algorithms may give significantly better results than Delta + LZMA2. This is especially true with audio, which compresses faster and better, for example, with flac(1). 🎵

Supported options:

dist=distance — sets the distance of the delta calculation in bytes. distance must be 1–256. The default is 1.

For example, with dist=2 and the eight-byte input A1 B1 A2 B3 A3 B5 A4 B7, the output will be A1 B1 01 02 01 02 01 02.

🎀 Other options 🎀

-q, --quiet — suppresses warnings and notices. Specify it twice to suppress errors too. This option has no effect on the exit status — that is, even if a warning was suppressed, the exit status that indicates a warning is still used.
-v, --verbose — be verbose. If standard error is connected to a terminal, xz will display a progress indicator. Specifying --verbose twice gives even more verbose output.

The progress indicator shows the following info:

Completion percentage — only shown if the size of the input file is known. That is, the percentage cannot be shown in pipes. 🚫
Amount of compressed data produced (compressing) or consumed (decompressing).
Amount of uncompressed data consumed (compressing) or produced (decompressing).
Compression ratio — calculated by dividing the amount of compressed data processed so far by the amount of uncompressed data processed so far.

Compression or decompression speed 💨 — literally the amount of uncompressed data that got eaten up (compression) or spat out (decompression) every second. It only shows up after a few seconds have passed since xz started chewing on the file, so don't panic if you don't see it immediately bestie.
Elapsed time ⏱️ — in the format M:SS or H:MM:SS.
Estimated remaining time 🔮 — only shows up when the size of the input file is actually known and a couple of seconds have already passed since xz started processing. It's in a less precise format with zero colons ever, like 2 min 30 s.

When standard error is not a terminal, --verbose makes xz print the filename, compressed size, uncompressed size, compression ratio, and possibly also the speed and elapsed time — all on a single line to standard error, after compressing or decompressing the file. Speed and elapsed time only get included when the operation took at least a few seconds. And if the operation didn't finish — like, say, someone interrupted it 🙈 — then the completion percentage gets printed too, as long as the size of the input file is known.

-Q, --no-warn — Don't set the exit status to 2 even if a condition worth a warning got detected. This option doesn't touch the verbosity level, so you gotta use both --quiet and --no-warn if you want to hide warnings and leave the exit status alone. iykyk.
--robot — Print messages in a machine-parsable</> format. It's made for writing frontends that wanna use xz instead of liblzma, which happens with all kinds of scripts. The output with this option on is meant to be stable across xz releases — no cap. Peep the ROBOT MODE section for the deets.
--info-memory — Displays, in human-readable format, how much physical memory (RAM) and how many processor threads xz thinks the system has, plus the memory usage limits for compression and decompression — then exits successfully. ✨
-h, --help — Display a help message describing the most commonly used options, and exit successfully.
-H, --long-help — Display a help message describing all features of xz, and exit successfully.
-V, --version — Display the version number of xz and liblzma in human readable format. Wanna get machine-parsable output instead? Specify --robot before --version 💅.

🤖 ROBOT MODE 🤖

Robot mode gets activated with the --robot option. It makes the output of xz easier for other programs to parse. Currently --robot is supported only together with --list, --filters-help, --info-memory, and --version. It will be supported for compression and decompression in the future, bestie, just not yet.

🧾 List mode 🧾

xz --robot --list uses tab-separated output. The first column of every line has a string that tells you what type of info is on that line:

name — This is always the first line when starting to list a file. The second column on the line is the filename.
file — This line contains overall information about the .xz file. It's always printed right after the name line.
stream — This line type is used only when --verbose was specified. There are as many stream lines as there are streams in the .xz file.
block — This line type is used only when --verbose was specified. There are as many block lines as there are blocks in the .xz file. All the block lines come after all the stream lines; different line types are never interleaved.
summary — This line type is used only when --verbose was specified twice. It's printed after all the block lines. Like the file line, the summary line holds overall information about the .xz file.
totals — This line is always the very last line of the list output. It shows the total counts and sizes.

The columns of the file lines:

Number of streams in the file
Total number of blocks in the stream(s)
Compressed size of the file
Uncompressed size of the file
Compression ratio, for example, 0.123. If ratio is over 9.999, three dashes (---) are displayed instead of the ratio.
Comma-separated list of integrity check names. The following strings are used for the known check types: None, CRC32, CRC64, and SHA-256. For unknown check types, Unknown-N is used, where N is the Check ID as a decimal number (one or two digits).
Total size of stream padding in the file

The columns of the stream lines:

Stream number (the first stream is 1)
Number of blocks in the stream
Compressed start offset
Uncompressed start offset
Compressed size (does not include stream padding)
Uncompressed size
Compression ratio
Name of the integrity check
Size of stream padding

The columns of the block lines:

Number of the stream containing this block
Block number relative to the beginning of the stream (the first block is 1)
Block number relative to the beginning of the file
Compressed start offset relative to the beginning of the file
Uncompressed start offset relative to the beginning of the file
Total compressed size of the block (includes headers)
Uncompressed size
Compression ratio
Name of the integrity check

If --verbose was specified twice, additional columns are included on the block lines. These are not displayed with a single --verbose, because getting this information requires many seeks and can thus be slow:

Value of the integrity check in hexadecimal
Block header size
Block flags: c indicates that compressed size is present, and u indicates that uncompressed size is present. If the flag is not set, a dash (-) is shown instead to keep the string length fixed. New flags may be added to the end of the string in the future.
Size of the actual compressed data in the block (this excludes the block header, block padding, and check fields)
Amount of memory (in bytes) required to decompress this block with this xz version
Filter chain. Note that most of the options used at compression time cannot be known, because only the options that are needed for decompression are stored in the .xz headers.

The columns of the summary lines:

Amount of memory (in bytes) required to decompress this file with this xz version
yes or no indicating if all block headers have both compressed size and uncompressed size stored in them

Since xz 5.1.2alpha:

Minimum xz version required to decompress the file

The columns of the totals line:

Number of streams
Number of blocks
Compressed size
Uncompressed size
Average compression ratio
Comma-separated list of integrity check names that were present in the files
Stream padding size
Number of files. This is here to keep the order of the earlier columns the same as on file lines.

If --verbose was specified twice, additional columns are included on the totals line:

Maximum amount of memory (in bytes) required to decompress the files with this xz version
yes or no indicating if all block headers have both compressed size and uncompressed size stored in them

Since xz 5.1.2alpha:

Minimum xz version required to decompress the file

Future versions may add new line types, and new columns can be added to the existing line types — but the existing columns won't be changed. Periodt. 💅

🔍 Filters help 🔍

xz --robot --filters-help prints the supported filters in the following format:

filter:option=<value>,option=<value>...

filter — Name of the filter
option — Name of a filter specific option
value — Numeric value ranges appear as <min-max>. String value choices are shown within < > and separated by a | character.

Each filter is printed on its own line. 🌈

🧠 Memory limit information 🧠

xz --robot --info-memory prints a single line with multiple tab-separated columns:

Total amount of physical memory (RAM) in bytes.
Memory usage limit for compression in bytes (--memlimit-compress). A special value of 0 indicates the default setting, which for single-threaded mode is the same as no limit.
Memory usage limit for decompression in bytes (--memlimit-decompress). A special value of 0 indicates the default setting, which for single-threaded mode is the same as no limit.
Since xz 5.3.4alpha: Memory usage for multi-threaded decompression in bytes (--memlimit-mt-decompress). This is never zero because a system-specific default value shown in the column 5 is used if no limit has been specified explicitly. This is also never greater than the value in the column 3, even if a larger value has been specified with --memlimit-mt-decompress.
Since xz 5.3.4alpha: A system-specific default memory usage limit that is used to limit the number of threads when compressing with an automatic number of threads (--threads=0) and no memory usage limit has been specified (--memlimit-compress). This is also used as the default value for --memlimit-mt-decompress.
Since xz 5.3.4alpha: Number of available processor threads.

In the future, the output of xz --robot --info-memory may have more columns, but never more than a single line. 🤖

🏷️ Version 🏷️

xz --robot --version prints the version number of xz and liblzma in the following format:

XZ_VERSION=XYYYZZZS
LIBLZMA_VERSION=XYYYZZZS

X — Major version.
YYY — Minor version. Even numbers are stable. Odd numbers are alpha or beta versions.
ZZZ — Patch level for stable releases, or just a counter for development releases.
S — Stability. 0 is alpha, 1 is beta, and 2 is stable. S should be always 2 when YYY is even.

XYYYZZZS are the same on both lines if xz and liblzma are from the same XZ Utils release. 🫶

Examples: 4.999.9beta is 49990091 and 5.0.0 is 50000002.

💀 EXIT STATUS 💀

The vibes of how it went, in number form:

0 — All is good. Slay. ✨
1 — An error occurred. 😭
2 — Something worth a warning occurred, but no actual errors occurred. It's giving "we need to talk" but it's fine. 🫶

Notices (not warnings or errors) printed on standard error don't affect the exit status. So if it's just chatting on stderr, the number stays unbothered. 💅

🌿 ENVIRONMENT 🌿

Okay so xz reads space-separated lists of options out of the environment variables XZ_DEFAULTS and XZ_OPT, in that order, before it even looks at the options from the command line. Bestie, note that only options get parsed from those environment variables — all the non-options are just silently ignored, no drama, no complaints. The parsing is done with getopt_long(3), which is the exact same thing used for the command line arguments. ✨

Warning: By setting these environment variables, one is effectively modifying programs and scripts that run xz. 😳 Most of the time it is safe to set memory usage limits, number of threads, and compression options via the environment variables. However, some options can break scripts. An obvious example is --help which makes xz show the help text instead of compressing or decompressing a file. More subtle examples are --quiet and --verbose. In many cases it works well to enable the progress indicator using --verbose, but in some situations the extra messages create problems. The verbosity level also affects the behavior of --list.

🌸 XZ_DEFAULTS 🌸

User-specific or system-wide default options. Typically this is set in a shell initialization script to enable xz's memory usage limiter by default or set the default number of threads. Excluding shell initialization scripts and similar special cases, scripts should never set or unset XZ_DEFAULTS. Just don't, fr fr.

🎀 XZ_OPT 🎀

This is for passing options to xz when it is not possible to set the options directly on the xz command line. This is the case when xz is run by a script or tool, for example, GNU tar(1):

XZ_OPT=-2v tar caf foo.tar.xz foo

Scripts may use XZ_OPT, for example, to set script-specific default compression options. It is still recommended to allow users to override XZ_OPT if that is reasonable. For example, in sh(1) scripts one may use something like this:

XZ_OPT=${XZ_OPT-"-7e"}
export XZ_OPT

🦋 LZMA UTILS COMPATIBILITY 🦋

The command line syntax of xz is practically a superset of lzma, unlzma, and lzcat as found from LZMA Utils 4.32.x. In most cases, it is possible to replace LZMA Utils with XZ Utils without breaking existing scripts. There are some incompatibilities though, which may sometimes cause problems. 🙈

💜 Compression preset levels 💜

The numbering of the compression level presets is not identical in xz and LZMA Utils. The most important difference is how dictionary sizes are mapped to different presets. Dictionary size is roughly equal to the decompressor memory usage.

Level     xz      LZMA Utils
 -0     256 KiB      N/A
 -1       1 MiB     64 KiB
 -2       2 MiB      1 MiB
 -3       4 MiB    512 KiB
 -4       4 MiB      1 MiB
 -5       8 MiB      2 MiB
 -6       8 MiB      4 MiB
 -7      16 MiB      8 MiB
 -8      32 MiB     16 MiB
 -9      64 MiB     32 MiB

The dictionary size differences affect the compressor memory usage too, but there are some other differences between LZMA Utils and XZ Utils, which make the difference even bigger:

Level     xz      LZMA Utils 4.32.x
 -0       3 MiB          N/A
 -1       9 MiB          2 MiB
 -2      17 MiB         12 MiB
 -3      32 MiB         12 MiB
 -4      48 MiB         16 MiB
 -5      94 MiB         26 MiB
 -6      94 MiB         45 MiB
 -7     186 MiB         83 MiB
 -8     370 MiB        159 MiB
 -9     674 MiB        311 MiB

The default preset level in LZMA Utils is -7 while in XZ Utils it is -6, so both use an 8 MiB dictionary by default. Cute coincidence honestly. 🌈

🍓 Streamed vs. non-streamed `.lzma` files 🍓

The uncompressed size of the file can be stored in the .lzma header. LZMA Utils does that when compressing regular files. The alternative is to mark that uncompressed size is unknown and use end-of-payload marker to indicate where the decompressor should stop. LZMA Utils uses this method when uncompressed size isn't known, which is the case, for example, in pipes.

xz supports decompressing .lzma files with or without end-of-payload marker, but all .lzma files created by xz will use end-of-payload marker and have uncompressed size marked as unknown in the .lzma header. This may be a problem in some uncommon situations. For example, a .lzma decompressor in an embedded device might work only with files that have known uncompressed size. If you hit this problem, you need to use LZMA Utils or LZMA SDK to create .lzma files with known uncompressed size. 💀

🧚 Unsupported `.lzma` files 🧚

The .lzma format allows lc values up to 8, and lp values up to 4. LZMA Utils can decompress files with any lc and lp, but always creates files with lc=3 and lp=0. Creating files with other lc and lp is possible with xz and with LZMA SDK.

The implementation of the LZMA1 filter in liblzma requires that the sum of lc and lp must not exceed 4. Thus, .lzma files, which exceed this limitation, cannot be decompressed with xz. 😭

LZMA Utils creates only .lzma files which have a dictionary size of 2^n (a power of 2) but accepts files with any dictionary size. liblzma accepts only .lzma files which have a dictionary size of 2^n or 2^n + 2^(n-1). This is to decrease false positives when detecting .lzma files.

These limitations shouldn't be a problem in practice, since practically all .lzma files have been compressed with settings that liblzma will accept. 🫶

💅 Trailing garbage 💅

When decompressing, LZMA Utils silently ignore everything after the first .lzma stream. In most situations, this is a bug. This also means that LZMA Utils don't support decompressing concatenated .lzma files.

If there is data left after the first .lzma stream, xz considers the file to be corrupt unless --single-stream was used. This may break obscure scripts which have assumed that trailing garbage is ignored. Not us being the responsible one. 👑

⭐ NOTES ⭐

🌸 Compressed output may vary 🌸

The exact compressed output produced from the same uncompressed input file may vary between XZ Utils versions even if compression options are identical. This is because the encoder can be improved (faster or better compression) without affecting the file format. The output can vary even between different builds of the same XZ Utils version, if different build options are used. iykyk. 🙈

The above means that once --rsyncable has been implemented, the resulting files won't necessarily be rsyncable unless both old and new files have been compressed with the same xz version. This problem can be fixed if a part of the encoder implementation is frozen to keep rsyncable output stable across xz versions.

🎀 Embedded `.xz` decompressors 🎀

Embedded .xz decompressor implementations like XZ Embedded don't necessarily support files created with integrity check types other than none and crc32. Since the default is --check=crc64, you must use --check=none or --check=crc32 when creating files for embedded systems. 📝

Outside embedded systems, all .xz format decompressors support all the check types, or at least are able to decompress the file without verifying the integrity check if the particular check is not supported.

XZ Embedded supports BCJ filters, but only with the default start offset. 🦋

✨ EXAMPLES ✨

Okay bestie this is the part where it all clicks, I promise 💖

🌸 Basics 🌸

Compress the file foo into foo.xz using the default compression level (-6), and remove foo if compression is successful:

xz foo

Decompress bar.xz into bar and don't remove bar.xz even if decompression is successful:

xz -dk bar.xz

Create baz.tar.xz with the preset -4e (-4 --extreme), which is slower than the default -6, but needs less memory for compression and decompression (48 MiB and 5 MiB, respectively):

tar cf - baz | xz -4e > baz.tar.xz

A mix of compressed and uncompressed files can be decompressed to standard output with a single command:

xz -dcf a.txt b.txt.xz c.txt d.txt.lzma > abcd.txt

look at that, one command, four files, zero drama 💅

🧚 Parallel compression of many files 🧚

On GNU and *BSD, find(1) and xargs(1) can be used to parallelize compression of many files:

find . -type f \! -name '*.xz' -print0 \
    | xargs -0r -P4 -n16 xz -T1

The -P option to xargs(1) sets the number of parallel xz processes. The best value for the -n option depends on how many files there are to be compressed. If there are only a couple of files, the value should probably be 1; with tens of thousands of files, 100 or even more may be appropriate to reduce the number of xz processes that xargs(1) will eventually create.

The option -T1 for xz is there to force it to single-threaded mode, because xargs(1) is used to control the amount of parallelization.

tl;dr: let xargs be the ringleader, xz just does one job at a time 🎪

🤖 Robot mode 🤖

Calculate how many bytes have been saved in total after compressing multiple files:

xz --robot --list *.xz | awk '/^totals/{print $5-$4}'

A script may want to know that it is using new enough xz. The following sh(1) script checks that the version number of the xz tool is at least 5.0.0. This method is compatible with old beta versions, which didn't support the --robot option:

if ! eval "$(xz --robot --version 2> /dev/null)" ||
        [ "$XZ_VERSION" -lt 50000002 ]; then
    echo "Your xz is too old."
fi
unset XZ_VERSION LIBLZMA_VERSION

Set a memory usage limit for decompression using XZ_OPT, but if a limit has already been set, don't increase it:

NEWLIM=$((123 << 20))  # 123 MiB
OLDLIM=$(xz --robot --info-memory | cut -f3)
if [ $OLDLIM -eq 0 -o $OLDLIM -gt $NEWLIM ]; then
    XZ_OPT="$XZ_OPT --memlimit-decompress=$NEWLIM"
    export XZ_OPT
fi

note the unset at the end of the version check, that's not optional behaviour, that's just being a good script citizen 🫶

🎀 Custom compressor filter chains 🎀

The simplest use for custom filter chains is customizing a LZMA2 preset. This can be useful, because the presets cover only a subset of the potentially useful combinations of compression settings.

The CompCPU columns of the tables from the descriptions of the options -0 ... -9 and --extreme are useful when customizing LZMA2 presets. Here are the relevant parts collected from those two tables:

Preset   CompCPU
 -0         0
 -1         1
 -2         2
 -3         3
 -4         4
 -5         5
 -6         6
 -5e        7
 -6e        8

If you know that a file requires somewhat big dictionary (for example, 32 MiB) to compress well, but you want to compress it quicker than xz -8 would do, a preset with a low CompCPU value (for example, 1) can be modified to use a bigger dictionary:

xz --lzma2=preset=1,dict=32MiB foo.tar

With certain files, the above command may be faster than xz -6 while compressing significantly better. However, it must be emphasized that only some files benefit from a big dictionary while keeping the CompCPU value low. The most obvious situation, where a big dictionary can help a lot, is an archive containing very similar files of at least a few megabytes each. The dictionary size has to be significantly bigger than any individual file to allow LZMA2 to take full advantage of the similarities between consecutive files.

If very high compressor and decompressor memory usage is fine, and the file being compressed is at least several hundred megabytes, it may be useful to use an even bigger dictionary than the 64 MiB that xz -9 would use:

xz -vv --lzma2=dict=192MiB big_foo.tar

Using -vv (--verbose --verbose) like in the above example can be useful to see the memory requirements of the compressor and decompressor. Remember that using a dictionary bigger than the size of the uncompressed file is waste of memory, so the above command isn't useful for small files.

Sometimes the compression time doesn't matter, but the decompressor memory usage has to be kept low, for example, to make it possible to decompress the file on an embedded system. The following command uses -6e (-6 --extreme) as a base and sets the dictionary to only 64 KiB. The resulting file can be decompressed with XZ Embedded (that's why there is --check=crc32) using about 100 KiB of memory.

xz --check=crc32 --lzma2=preset=6e,dict=64KiB foo

If you want to squeeze out as many bytes as possible, adjusting the number of literal context bits (lc) and number of position bits (pb) can sometimes help. Adjusting the number of literal position bits (lp) might help too, but usually lc and pb are more important. For example, a source code archive contains mostly US-ASCII text, so something like the following might give slightly (like 0.1 %) smaller file than xz -6e (try also without lc=4):

xz --lzma2=preset=6e,pb=0,lc=4 source_code.tar

Using another filter together with LZMA2 can improve compression with certain file types. For example, to compress a x86-32 or x86-64 shared library using the x86 BCJ filter:

xz --x86 --lzma2 libfoo.so

Note that the order of the filter options is significant. If --x86 is specified after --lzma2, xz will give an error, because there cannot be any filter after LZMA2, and also because the x86 BCJ filter cannot be used as the last filter in the chain.

The Delta filter together with LZMA2 can give good results with bitmap images. It should usually beat PNG, which has a few more advanced filters than simple delta but uses Deflate for the actual compression.

The image has to be saved in uncompressed format, for example, as uncompressed TIFF. The distance parameter of the Delta filter is set to match the number of bytes per pixel in the image. For example, 24-bit RGB bitmap needs dist=3, and it is also good to pass pb=0 to LZMA2 to accommodate the three-byte alignment:

xz --delta=dist=3 --lzma2=pb=0 foo.tiff

If multiple images have been put into a single archive (for example, .tar), the Delta filter will work on that too as long as all images have the same number of bytes per pixel.

filter order is not a suggestion bestie, it's the law 👑

💜 SEE ALSO 💜

xzdec(1), xzdiff(1), xzgrep(1), xzless(1), xzmore(1), gzip(1), bzip2(1), 7z(1)

XZ Utils: <https://tukaani.org/xz/> XZ Embedded: <https://tukaani.org/xz/embedded.html> LZMA SDK: <https://7-zip.org/sdk.html>

and that's a wrap on the whole page, bestie!! 🌈 Tukaani wrote this one, dated 2025-03-08, and it lives at XZ(1) — go compress something and make it tiny, you absolute icon 💖✨

📦 where this page comes from

alpine 3.22: xz-doc 5.8.4-r0 · GPL-2.0-or-later AND 0BSD AND Public-Domain AND LGPL-2.1-or-later
alpine 3.23: xz-doc 5.8.4-r0 · GPL-2.0-or-later AND 0BSD AND Public-Domain AND LGPL-2.1-or-later
alpine 3.24: xz-doc 5.8.4-r0 · GPL-2.0-or-later AND 0BSD AND Public-Domain AND LGPL-2.1-or-later
arch rolling: xz 5.8.4-1 · GPL-2.0-or-later AND 0BSD AND LicenseRef-Public-Domain AND LGPL-2.1-or-later
nixpkgs unstable: xz 5.8.4
📜 the page's license: 0BSD
💖 girlified by deepseek-ai/DeepSeek-V4.1-Flash on 2026-10-13: a language model's rewrite of the original, under its terms

XZ(1)

💖 XZ(1) — XZ Utils 💖

✨ NAME ✨

🎀 SYNOPSIS 🎀

💅 COMMAND ALIASES 💅

🦋 DESCRIPTION 🦋

🧠 Memory usage 🧠

🌈 Concatenation and padding with .xz files 🌈

⭐ OPTIONS ⭐

🔢 Integer suffixes and special values 🔢

🎛️ Operation mode 🎛️

🛠️ Operation modifiers 🛠️

🎀 Basic file format and compression options 🎀

🎀 Custom compressor filter chains 🎀

🎀 Other options 🎀

🤖 ROBOT MODE 🤖

🧾 List mode 🧾

🔍 Filters help 🔍

🧠 Memory limit information 🧠

🏷️ Version 🏷️

💀 EXIT STATUS 💀

🌿 ENVIRONMENT 🌿

🌸 XZ_DEFAULTS 🌸

🎀 XZ_OPT 🎀

🦋 LZMA UTILS COMPATIBILITY 🦋

💜 Compression preset levels 💜

🍓 Streamed vs. non-streamed .lzma files 🍓

🧚 Unsupported .lzma files 🧚

💅 Trailing garbage 💅

⭐ NOTES ⭐

🌸 Compressed output may vary 🌸

🎀 Embedded .xz decompressors 🎀

✨ EXAMPLES ✨

🌸 Basics 🌸

🧚 Parallel compression of many files 🧚

🤖 Robot mode 🤖

🎀 Custom compressor filter chains 🎀

💜 SEE ALSO 💜

🍓 Streamed vs. non-streamed `.lzma` files 🍓

🧚 Unsupported `.lzma` files 🧚

🎀 Embedded `.xz` decompressors 🎀