3CPIO(1) 3cpio 3CPIO(1) NAME 3cpio - manage initrd cpio archives SYNOPSIS 3cpio --count ARCHIVE 3cpio {-c|--create} [-v|--debug] [-C DIR] [--data-align ALIGNMENT] [ARCHIVE] < manifest 3cpio {-e|--examine} [--raw] ARCHIVE 3cpio {-t|--list} [-v|--debug] [-P LIST] ARCHIVE [pattern...] 3cpio {-x|--extract} [-v|--debug] [-C DIR] [-P LIST] [-p] [-s NAME] [--to-stdout] [--force] ARCHIVE [pattern...] 3cpio {-V|--version} 3cpio {-h|--help} DESCRIPTION 3cpio is a tool to manage initramfs cpio files for the Linux kernel. The Linux kernel's initramfs buffer format is based around the newc or crc cpio formats. Multiple cpio archives can be concatenated and the last archive can be compressed. Different compression algorithms can be used depending on what support was compiled into the Linux kernel. 3cpio is tailored to initramfs cpio files and will not gain support for other cpio formats. Following compression formats are supported: bzip2, gzip, lz4, lzma, lzop, xz, zstd. MODES --count ARCHIVE Print the number of concatenated cpio archives. -c, --create [ARCHIVE] Create a new cpio archive. Read the manifest from the standard input. See the MANIFEST section for the description of the manifest format. Write the cpio archive to standard output or to the specified ARCHIVE file if provided. The permission of the ARCHIVE file will be determined by the permission of the input files (to avoid leaking sensitive information). -e, --examine ARCHIVE List the offsets of the cpio archives and their compression in a formatted table. SI prefixes are used for the size unit (e. g. 1 kB = 1000 bytes). Do not rely on the output to have a specific layout. Use the --raw option for a machine-readable output. Then each line will contain these five values tab-separated: Start The beginning of the cpio archive in bytes. End The end of the cpio archive in bytes. Size The on-disk size of the cpio archive in bytes (end - start). Compression Compression of the cpio archive (cpio in case it is not compressed). Extracted size Size of the files inside the cpio archive in bytes. -t, --list ARCHIVE [pattern...] List the contents of the cpio archives. By default only the file names are printed. If --verbose is specified, the long listing format is used (similar to ls --long). If --debug is specified, the inode is printed in addition to the long format. If one or more patterns are supplied, list only file names matching at least one of those patterns. These patterns are shell wildcard patterns (see glob(7)). -x, --extract ARCHIVE [pattern...] Extract cpio archives. If one or more patterns are supplied, extract only files matching at least one of those patterns. These patterns are shell wildcard patterns (see glob(7)). -V, --version Print version number. -h, --help Print help message. OPTIONS --data-align ALIGNMENT When creating a cpio archive, pad the cpio metadata to align the file data on ALIGNMENT in bytes. This option is useful to reflink cpio file data on file systems that support reflinks. This padding/alignment will be skipped for files that are smaller than ALIGNMENT, files where the padded namesize field would exceed PATH_MAX, and for cpio archives that are compressed. ALIGNMENT must be a multiple of 4 bytes. This option is only taken into account in the --create mode. The resulting cpio archive may be highly fragmented, which can lead to performance degradation when reading/extracting the image from devices with slow random IO (e.g. spinning disk). Note: Using this option will "bend" the cpio newc spec a bit to inject zeros after the filename to provide data segment alignment. These zeros are accounted for in the namesize, but some applications may only expect a single zero-terminator (and 4 byte alignment). GNU cpio and Linux initramfs handle this fine as long as PATH_MAX is not exceeded. The following command can be used to determine the optimal transfer size of the file system (where $path is the path the cpio archive will be written to): stat --file-system -c "%s" -- "$path" -C DIR, --directory=DIR Change directory before performing any operation, but after opening the ARCHIVE. This option is only taken into account in the --extract mode. --make-directories: Create leading directories where needed. This option is only taken into account in the --extract mode. -P LIST, --parts LIST Only operate on the cpio archives that matches LIST. This option is only taken into account in the --list and --extract modes. LIST is made up of one range, or many ranges separated by commas. Each range is one of: N N'th cpio archive, counted from 1 N- from N'th (included) to last cpio archive N-M from N'th to M'th (included) cpio archive -M from first to M'th (included) cpio archive -p, --preserve-permissions Set permissions of extracted files to those recorded in the archive (default for superuser). This option is only taken into account in the --extract mode. --raw Use a machine-readable output format. This format is designed for easy parsing and is intended for being used in scripts. This option is only taken into account in the --examine mode. -s NAME, --subdir=NAME Extract the cpio archives into separate sub-directories (using the given NAME plus an incrementing number). This option is only taken into account in the --extract mode. --to-stdout Extract files to standard output. Only the content of the files will be written to stdout. Directories and symlinks will be ignored. This option is only taken into account in the --extract mode. -v, --verbose Verbose output. This option is only taken into account in the --extract and --list modes. --debug Debug output. This option is only taken into account in the --extract and --list modes. --force Force overwriting existing files. This option is only taken into account in the --extract mode. MANIFEST When generating initrd cpio archives, following manifest format will be used. The manifest is a text format that is parsed line by line. If the line starts with #cpio it is interpreted as section marker to start a new cpio. A compression may be specified by adding a colon followed by the compression format and an optional compression level. Example for a Zstandard-compressed cpio with compression level 9: #cpio: zstd -9 All lines starting with # excluding #cpio (see above) will be treated as comments and will be ignored. Each element in the line is separated by a tab and is expected to be one of the following file types: file dir block char link fifo sock fifo is also known as named pipe (see fifo(7)). In case an element is empty or equal to - it is treated as not specified and it is derived from the input file. Path of the input file. It can be left unspecified in case all other needed fields are specified (and the file is otherwise empty). Limitation: The path must not start with #, be equal to -, or contain tabs. Path of the file inside the cpio. If the name is left unspecified it will be derived from . Limitation: The path must not be equal to - or contain tabs. File mode specified in octal. User ID (owner) of the file specified in decimal. Group ID of the file specified in decimal. Modification time of the file specified as seconds since the Epoch (1970-01-01 00:00 UTC). The specified time might be clamped by the time set in the SOURCE_DATE_EPOCH environment variable. Size of the input file in bytes. 3cpio will fail in case the input file is smaller than the provided file size. Major block/character device number in decimal. Minor block/character device number in decimal. Target of the symbolic link. Limitation: The target path must not be equal to - or contain tabs. Limitations: Files cannot start with # (will be treated as comment), be equal to - (will be treated as not specified), or contain tabs (will be split by tabs). These limitations of the manifest file are not expected to cause problems in practice. ENVIRONMENT VARIABLES SOURCE_DATE_EPOCH This environment variable will be taken into account when creating cpio archive. All modification times that are newer than the time specified in "SOURCE_DATE_EPOCH" will be clamped. Compressors will run with only one thread in case their multithreading implementation is not reproducible. The created cpio archive will be reproducible across multiple runs. EXIT STATUS 0 Success. 1 Failure. 2 Failure during command line argument parsing. EXAMPLES List the number of cpio archives that an initramfs file contains: $ 3cpio --count /boot/initrd.img 4 Examine the content of the initramfs cpio on an Ubuntu 24.04 system: $ 3cpio --examine /boot/initrd.img Start End Size Compr. Extracted 0 B 148 kB 148 kB cpio 147 kB 148 kB 13.3 MB 13.1 MB cpio 13.1 MB 13.3 MB 55.2 MB 41.9 MB cpio 41.7 MB 55.2 MB 62.0 MB 6.74 MB zstd 15.6 MB There is also a machine-readable output format available: $ 3cpio --examine --raw /boot/initrd.img 0 148480 148480 cpio 147350 148480 13275136 13126656 cpio 13125632 13275136 55215104 41939968 cpio 41692226 55215104 61956920 6741816 zstd 15616306 This initramfs cpio consists of three uncompressed cpio archives followed by a Zstandard-compressed cpio archive. List the content of the initramfs cpio on an Ubuntu 24.04 system: $ 3cpio --list /boot/initrd.img . kernel kernel/x86 kernel/x86/microcode kernel/x86/microcode/AuthenticAMD.bin kernel kernel/x86 kernel/x86/microcode kernel/x86/microcode/.enuineIntel.align.0123456789abc kernel/x86/microcode/GenuineIntel.bin . usr usr/lib usr/lib/firmware usr/lib/firmware/3com usr/lib/firmware/3com/typhoon.bin.zst [...] The first cpio contains only the AMD microcode. The second cpio contains only the Intel microcode. The third cpio contains firmware files and kernel modules. Extract the content of the initramfs cpio to the initrd subdirectory on an Ubuntu 24.04 system: $ 3cpio --extract -C initrd /boot/initrd.img $ ls initrd bin cryptroot init lib lib.usr-is-merged run scripts var conf etc kernel lib64 libx32 sbin usr Create a cpio archive similar to the other cpio tools using the find command: $ cd inputdir && find . | sort | 3cpio --create ../example.cpio Due to its manifest file format support, 3cpio can create cpio archives without the need of copying files into a temporary directory first. Example for creating an early microcode cpio image directly using the system installed files: $ cat manifest - kernel dir 755 0 0 1751654557 - kernel/x86 dir 755 0 0 1752011622 /usr/lib/firmware/amd-ucode kernel/x86/microcode /usr/lib/firmware/amd-ucode/microcode_amd_fam19h.bin kernel/x86/microcode/AuthenticAMD.bin $ 3cpio --create amd-ucode.img < manifest $ 3cpio --list --verbose amd-ucode.img drwxr-xr-x 2 root root 0 Jul 4 20:42 kernel drwxr-xr-x 2 root root 0 Jul 8 23:53 kernel/x86 drwxr-xr-x 2 root root 0 Jun 10 10:51 kernel/x86/microcode -rw-r--r-- 1 root root 100684 Mar 23 22:42 kernel/x86/microcode/AuthenticAMD.bin Example for creating an initrd image containing of an uncompressed early microcode cpio followed by a Zstandard-compressed cpio: $ cat manifest #cpio - kernel dir 755 0 0 1751654557 - kernel/x86 dir 755 0 0 1752011622 /usr/lib/firmware/amd-ucode kernel/x86/microcode /usr/lib/firmware/amd-ucode/microcode_amd_fam19h.bin kernel/x86/microcode/AuthenticAMD.bin #cpio: zstd -9 / /bin /usr /usr/bin /usr/bin/bash # This is a comment. Leaving the remaining files as task for the reader. $ 3cpio --create initrd.img < manifest $ 3cpio --examine initrd.img Start End Size Compr. Extracted 0 B 101 kB 101 kB cpio 101 kB 101 kB 786 kB 685 kB zstd 1.45 MB $ 3cpio --list --verbose initrd.img drwxr-xr-x 2 root root 0 Jul 4 20:42 kernel drwxr-xr-x 2 root root 0 Jul 8 23:53 kernel/x86 drwxr-xr-x 2 root root 0 Jun 10 10:51 kernel/x86/microcode -rw-r--r-- 1 root root 100684 Mar 23 22:42 kernel/x86/microcode/AuthenticAMD.bin drwxr-xr-x 2 root root 0 Jun 5 14:11 . lrwxrwxrwx 1 root root 7 Mar 20 2022 bin -> usr/bin drwxr-xr-x 2 root root 0 Apr 20 2023 usr drwxr-xr-x 2 root root 0 Jul 9 09:56 usr/bin -rwxr-xr-x 1 root root 1446024 Mar 31 2024 usr/bin/bash SEE ALSO bsdcpio(1), cpio(1), lsinitramfs(8), lsinitrd(1) COPYING Copyright (C) 2024-2025 Benjamin Drung. Free use of this software is granted under the terms of the ISC License. AUTHOR Benjamin Drung 3cpio 0.5.1 2025-08-29 3CPIO(1)