Based on kernel version 4.16.1. Page generated on 2018-04-09 11:53 EST.
1 ============================ 2 XZ data compression in Linux 3 ============================ 4 5 Introduction 6 ============ 7 8 XZ is a general purpose data compression format with high compression 9 ratio and relatively fast decompression. The primary compression 10 algorithm (filter) is LZMA2. Additional filters can be used to improve 11 compression ratio even further. E.g. Branch/Call/Jump (BCJ) filters 12 improve compression ratio of executable data. 13 14 The XZ decompressor in Linux is called XZ Embedded. It supports 15 the LZMA2 filter and optionally also BCJ filters. CRC32 is supported 16 for integrity checking. The home page of XZ Embedded is at 17 <http://tukaani.org/xz/embedded.html>, where you can find the 18 latest version and also information about using the code outside 19 the Linux kernel. 20 21 For userspace, XZ Utils provide a zlib-like compression library 22 and a gzip-like command line tool. XZ Utils can be downloaded from 23 <http://tukaani.org/xz/>. 24 25 XZ related components in the kernel 26 =================================== 27 28 The xz_dec module provides XZ decompressor with single-call (buffer 29 to buffer) and multi-call (stateful) APIs. The usage of the xz_dec 30 module is documented in include/linux/xz.h. 31 32 The xz_dec_test module is for testing xz_dec. xz_dec_test is not 33 useful unless you are hacking the XZ decompressor. xz_dec_test 34 allocates a char device major dynamically to which one can write 35 .xz files from userspace. The decompressed output is thrown away. 36 Keep an eye on dmesg to see diagnostics printed by xz_dec_test. 37 See the xz_dec_test source code for the details. 38 39 For decompressing the kernel image, initramfs, and initrd, there 40 is a wrapper function in lib/decompress_unxz.c. Its API is the 41 same as in other decompress_*.c files, which is defined in 42 include/linux/decompress/generic.h. 43 44 scripts/xz_wrap.sh is a wrapper for the xz command line tool found 45 from XZ Utils. The wrapper sets compression options to values suitable 46 for compressing the kernel image. 47 48 For kernel makefiles, two commands are provided for use with 49 $(call if_needed). The kernel image should be compressed with 50 $(call if_needed,xzkern) which will use a BCJ filter and a big LZMA2 51 dictionary. It will also append a four-byte trailer containing the 52 uncompressed size of the file, which is needed by the boot code. 53 Other things should be compressed with $(call if_needed,xzmisc) 54 which will use no BCJ filter and 1 MiB LZMA2 dictionary. 55 56 Notes on compression options 57 ============================ 58 59 Since the XZ Embedded supports only streams with no integrity check or 60 CRC32, make sure that you don't use some other integrity check type 61 when encoding files that are supposed to be decoded by the kernel. With 62 liblzma, you need to use either LZMA_CHECK_NONE or LZMA_CHECK_CRC32 63 when encoding. With the xz command line tool, use --check=none or 64 --check=crc32. 65 66 Using CRC32 is strongly recommended unless there is some other layer 67 which will verify the integrity of the uncompressed data anyway. 68 Double checking the integrity would probably be waste of CPU cycles. 69 Note that the headers will always have a CRC32 which will be validated 70 by the decoder; you can only change the integrity check type (or 71 disable it) for the actual uncompressed data. 72 73 In userspace, LZMA2 is typically used with dictionary sizes of several 74 megabytes. The decoder needs to have the dictionary in RAM, thus big 75 dictionaries cannot be used for files that are intended to be decoded 76 by the kernel. 1 MiB is probably the maximum reasonable dictionary 77 size for in-kernel use (maybe more is OK for initramfs). The presets 78 in XZ Utils may not be optimal when creating files for the kernel, 79 so don't hesitate to use custom settings. Example:: 80 81 xz --check=crc32 --lzma2=dict=512KiB inputfile 82 83 An exception to above dictionary size limitation is when the decoder 84 is used in single-call mode. Decompressing the kernel itself is an 85 example of this situation. In single-call mode, the memory usage 86 doesn't depend on the dictionary size, and it is perfectly fine to 87 use a big dictionary: for maximum compression, the dictionary should 88 be at least as big as the uncompressed data itself. 89 90 Future plans 91 ============ 92 93 Creating a limited XZ encoder may be considered if people think it is 94 useful. LZMA2 is slower to compress than e.g. Deflate or LZO even at 95 the fastest settings, so it isn't clear if LZMA2 encoder is wanted 96 into the kernel. 97 98 Support for limited random-access reading is planned for the 99 decompression code. I don't know if it could have any use in the 100 kernel, but I know that it would be useful in some embedded projects 101 outside the Linux kernel. 102 103 Conformance to the .xz file format specification 104 ================================================ 105 106 There are a couple of corner cases where things have been simplified 107 at expense of detecting errors as early as possible. These should not 108 matter in practice all, since they don't cause security issues. But 109 it is good to know this if testing the code e.g. with the test files 110 from XZ Utils. 111 112 Reporting bugs 113 ============== 114 115 Before reporting a bug, please check that it's not fixed already 116 at upstream. See <http://tukaani.org/xz/embedded.html> to get the 117 latest code. 118 119 Report bugs to <firstname.lastname@example.org> or visit #tukaani on 120 Freenode and talk to Larhzu. I don't actively read LKML or other 121 kernel-related mailing lists, so if there's something I should know, 122 you should email to me personally or use IRC. 123 124 Don't bother Igor Pavlov with questions about the XZ implementation 125 in the kernel or about XZ Utils. While these two implementations 126 include essential code that is directly based on Igor Pavlov's code, 127 these implementations aren't maintained nor supported by him.