0002 XZ data compression in Linux
0007 XZ is a general purpose data compression format with high compression
0008 ratio and relatively fast decompression. The primary compression
0009 algorithm (filter) is LZMA2. Additional filters can be used to improve
0010 compression ratio even further. E.g. Branch/Call/Jump (BCJ) filters
0011 improve compression ratio of executable data.
0013 The XZ decompressor in Linux is called XZ Embedded. It supports
0014 the LZMA2 filter and optionally also BCJ filters. CRC32 is supported
0015 for integrity checking. The home page of XZ Embedded is at
0016 <http://tukaani.org/xz/embedded.html>, where you can find the
0017 latest version and also information about using the code outside
0018 the Linux kernel.
0020 For userspace, XZ Utils provide a zlib-like compression library
0021 and a gzip-like command line tool. XZ Utils can be downloaded from
0024 XZ related components in the kernel
0026 The xz_dec module provides XZ decompressor with single-call (buffer
0027 to buffer) and multi-call (stateful) APIs. The usage of the xz_dec
0028 module is documented in include/linux/xz.h.
0030 The xz_dec_test module is for testing xz_dec. xz_dec_test is not
0031 useful unless you are hacking the XZ decompressor. xz_dec_test
0032 allocates a char device major dynamically to which one can write
0033 .xz files from userspace. The decompressed output is thrown away.
0034 Keep an eye on dmesg to see diagnostics printed by xz_dec_test.
0035 See the xz_dec_test source code for the details.
0037 For decompressing the kernel image, initramfs, and initrd, there
0038 is a wrapper function in lib/decompress_unxz.c. Its API is the
0039 same as in other decompress_*.c files, which is defined in
0042 scripts/xz_wrap.sh is a wrapper for the xz command line tool found
0043 from XZ Utils. The wrapper sets compression options to values suitable
0044 for compressing the kernel image.
0046 For kernel makefiles, two commands are provided for use with
0047 $(call if_needed). The kernel image should be compressed with
0048 $(call if_needed,xzkern) which will use a BCJ filter and a big LZMA2
0049 dictionary. It will also append a four-byte trailer containing the
0050 uncompressed size of the file, which is needed by the boot code.
0051 Other things should be compressed with $(call if_needed,xzmisc)
0052 which will use no BCJ filter and 1 MiB LZMA2 dictionary.
0054 Notes on compression options
0056 Since the XZ Embedded supports only streams with no integrity check or
0057 CRC32, make sure that you don't use some other integrity check type
0058 when encoding files that are supposed to be decoded by the kernel. With
0059 liblzma, you need to use either LZMA_CHECK_NONE or LZMA_CHECK_CRC32
0060 when encoding. With the xz command line tool, use --check=none or
0063 Using CRC32 is strongly recommended unless there is some other layer
0064 which will verify the integrity of the uncompressed data anyway.
0065 Double checking the integrity would probably be waste of CPU cycles.
0066 Note that the headers will always have a CRC32 which will be validated
0067 by the decoder; you can only change the integrity check type (or
0068 disable it) for the actual uncompressed data.
0070 In userspace, LZMA2 is typically used with dictionary sizes of several
0071 megabytes. The decoder needs to have the dictionary in RAM, thus big
0072 dictionaries cannot be used for files that are intended to be decoded
0073 by the kernel. 1 MiB is probably the maximum reasonable dictionary
0074 size for in-kernel use (maybe more is OK for initramfs). The presets
0075 in XZ Utils may not be optimal when creating files for the kernel,
0076 so don't hesitate to use custom settings. Example:
0078 xz --check=crc32 --lzma2=dict=512KiB inputfile
0080 An exception to above dictionary size limitation is when the decoder
0081 is used in single-call mode. Decompressing the kernel itself is an
0082 example of this situation. In single-call mode, the memory usage
0083 doesn't depend on the dictionary size, and it is perfectly fine to
0084 use a big dictionary: for maximum compression, the dictionary should
0085 be at least as big as the uncompressed data itself.
0087 Future plans
0089 Creating a limited XZ encoder may be considered if people think it is
0090 useful. LZMA2 is slower to compress than e.g. Deflate or LZO even at
0091 the fastest settings, so it isn't clear if LZMA2 encoder is wanted
0092 into the kernel.
0094 Support for limited random-access reading is planned for the
0095 decompression code. I don't know if it could have any use in the
0096 kernel, but I know that it would be useful in some embedded projects
0097 outside the Linux kernel.
0099 Conformance to the .xz file format specification
0101 There are a couple of corner cases where things have been simplified
0102 at expense of detecting errors as early as possible. These should not
0103 matter in practice all, since they don't cause security issues. But
0104 it is good to know this if testing the code e.g. with the test files
0105 from XZ Utils.
0107 Reporting bugs
0109 Before reporting a bug, please check that it's not fixed already
0110 at upstream. See <http://tukaani.org/xz/embedded.html> to get the
0111 latest code.
0113 Report bugs to <email@example.com> or visit #tukaani on
0114 Freenode and talk to Larhzu. I don't actively read LKML or other
0115 kernel-related mailing lists, so if there's something I should know,
0116 you should email to me personally or use IRC.
0118 Don't bother Igor Pavlov with questions about the XZ implementation
0119 in the kernel or about XZ Utils. While these two implementations
0120 include essential code that is directly based on Igor Pavlov's code,
0121 these implementations aren't maintained nor supported by him.