This manual documents GNU cpio (version 2.15, 14 January 2024).
Copyright © 1995–2024 Free Software Foundation, Inc.
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation; with no Invariant Sections, with the Front-Cover texts being “A GNU Manual”, and with the Back-Cover Texts as in (a) below. A copy of the license is included in the section entitled “GNU Free Documentation License”.
(a) The FSF’s Back-Cover Text is: “You have freedom to copy and modify this GNU Manual, like GNU software. Copies published by the Free Software Foundation raise funds for GNU development.”
GNU cpio copies files into or out of a cpio or
tar archive, The archive can be another file on the disk, a
magnetic tape, or a pipe.
GNU cpio supports the following archive formats: binary, old
ASCII, new ASCII, crc, HPUX binary, HPUX old ASCII, old tar, and
POSIX.1 tar. The tar format is provided for compatibility with the
tar program. By default, cpio creates binary
format archives, for compatibility with older cpio programs.
When extracting from archives, cpio automatically recognizes
which kind of archive it is reading and can read archives created on
machines with a different byte-order.
GNU cpio performs three primary functions. Copying files to an
archive, Extracting files from an archive, and passing files to another
directory tree. An archive can be a file on disk, one or more floppy
disks, or one or more tapes.
When creating an archive, cpio takes the list of files to be
processed from the standard input, and then sends the archive to the standard
output, or to the device defined by the -F option.
See Copy-out mode. Usually find or ls is used
to provide this list to the standard input. In the following example
you can see the possibilities for archiving the contents of a single directory.
% ls | cpio -ov > directory.cpio |
The -o option creates the archive, and the -v option
prints the names of the files archived as they are added. Notice that
the options can be put together after a single - or can be placed
separately on the command line. The ‘>’ redirects the
cpio output to the file directory.cpio.
If you wanted to archive an entire directory tree, the find
command can provide the file list to cpio:
% find . -print -depth | cpio -ov > tree.cpio |
This will take all the files in the current directory, the directories
below and place them in the archive tree.cpio. Again the -o
creates an archive, and the -v option shows you the name of the
files as they are archived. See Copy-out mode. Using the ‘.’ in the
find statement will give you more flexibility when doing
restores, as it will save file names with a relative path vice a hard
wired, absolute path. The -depth option forces find
to print of the entries in a directory before printing the directory
itself. This limits the effects of restrictive directory permissions
by printing the directory entries in a directory before the directory
name itself.
Extracting an archive requires a bit more thought. First of all, by
default cpio extracts the files with exactly the same name
as stored in the archive. That means that if the archive contains absolute
paths, you will extract files to their absolute locations no matter
what directory you’re in when running the command. You can instruct
cpio to remove leading slashes using the
--no-absolute-filenames option. Nevertheless, the good
practice is to always test the archive using cpio -t prior
to extracting it.
Furthermore, cpio will not create directories by default.
Another characteristic, is it will not overwrite existing files unless
you tell it to.
% cpio -iv < directory.cpio |
This will retrieve the files archived in the file directory.cpio and restore them to their locations. The -i option extracts the archive and the -v shows the file names as they are extracted. If you are dealing with an archived directory tree, you need to use the -d option to create directories as necessary, something like:
% cpio -idv < tree.cpio |
This will take the contents of the archive tree.cpio and extract it.
If you try to extract the files on top of files of the same name that
already exist (and have the same or later modification time)
cpio will not extract the file unless told to do so by the
-u option. See Copy-in mode.
In copy-pass mode, cpio copies files from one directory tree
to another, combining the copy-out and copy-in steps without actually using an
archive. It reads the list of files to copy from the standard input;
the directory into which it will copy them is given as a non-option
argument. See Copy-pass mode.
% find . -depth -print0 | cpio --null -pvd new-dir |
The example shows copying the files of the present directory, and
sub-directories to a new directory called new-dir. Some new options are
the -print0 available with GNU find, combined with the
--null option of cpio. These two options act
together to send file names between find and cpio,
even if special characters are embedded in the file names. Another is
-p, which tells cpio to pass the files it finds to
the directory new-dir.
In copy-out mode, cpio copies files into an archive. It reads a list
of filenames, one per line, on the standard input, and writes the
archive onto the standard output. A typical way to generate the list
of filenames is with the find command; you should give find
the -depth option to minimize problems with permissions on
directories that are unreadable.
Copy-out mode is requested by the -o (--create) command line option, e.g.:
% find | cpio -o > directory.cpio |
The following options can be used in copy-out mode:
Filenames in the list are delimited by ASCII null characters instead of newlines.
Append to an existing archive.
Reset the access times of files after reading them.
Do not strip file system prefix components from the file names. This is the default.
Strip file system prefix components from the file names before storing them to the archive.
Sets the I/O block size to block-size * 512 bytes.
Set the I/O block size to 5120 bytes.
Use the old portable (ASCII) archive format.
Set the I/O block size to the given number of bytes.
Change to directory dir
Treat the archive file as local, even if its name contains colons.
Use the supplied archive-file instead of standard input. Optional user and host specify the user and host names in case of a remote archive.
Use given archive format. See format, for a list of available formats.
Dereference symbolic links (copy the files that they point to instead of copying the links).
Print string when the end of a volume of the backup media is reached.
Do not print the number of blocks copied.
Use command instead of rsh to access remote archives.
Set the ownership of all files created to the specified user and/or group. See owner.
Verbosely list the files processed.
Print a ‘.’ for each file processed.
Control warning display. Argument is one of ‘none’, ‘truncate’, ‘no-truncate’ or ‘all’. See warning, for a detailed discussion of these.
In copy-in mode, cpio copies files from an archive into the
filesystem or lists the archive contents. It reads the archive from
the standard input. Any non-option command line arguments are shell
globbing patterns; only files in the archive whose names match one or
more of those patterns are copied from the archive. Unlike in the
shell, an initial ‘.’ in a filename does match a wildcard at the
start of a pattern, and a ‘/’ in a filename can match wildcards.
If no patterns are given, all files are extracted.
The copy-in mode is requested by the -i (--extract) command line option.
The following options can be used in copy-in mode:
Do not strip file system prefix components from the file names. This is the default.
Create all files relative to the current directory.
Sets the I/O block size to block-size * 512 bytes.
Swap both halfwords of words and bytes of halfwords in the data. Equivalent to -sS.
Set the I/O block size to 5120 bytes.
Use the old portable (ASCII) archive format.
Set the I/O block size to the given number of bytes.
Change to directory dir
Create leading directories where needed.
Read additional patterns specifying filenames to extract or list from file.
Only copy files that do not match any of the given patterns.
Treat the archive file as local, even if its name contains colons.
Use the supplied archive-file instead of standard input. Optional user and host specify the user and host names in case of a remote archive.
Use given archive format. See format, for a list of available formats.
Retain previous file modification times when creating files.
Print string when the end of a volume of the backup media is reached.
Do not change the ownership of the files.
In the verbose table of contents listing, show numeric UID and GID values.
When reading a CRC format archive, only verify the CRC’s of each file in the archive, don’t actually extract the files
Do not print the number of blocks copied.
Use command instead of rsh to access remote archives.
Interactively rename files
Write files with large blocks of zeros as sparse files.
Swap the bytes of each halfword in the files
Swap the halfwords of each word (4 bytes) in the files
Extract files to standard output.
Replace all files unconditionally.
Verbosely list the files processed.
Print a ‘.’ for each file processed.
Control warning display. Argument is one of ‘none’, ‘truncate’, ‘no-truncate’ or ‘all’. See warning, for a detailed discussion of these.
In copy-pass mode, cpio copies files from one directory tree to
another, combining the copy-out and copy-in steps without actually
using an archive. It reads the list of files to copy from the
standard input; the directory into which it will copy them is given as
a non-option argument.
This mode is requested by the -p (--pass-through) command line option.
The following options are valid in copy-out mode:
Filenames in the list are delimited by ASCII null characters instead of newlines.
Reset the access times of files after reading them.
Swap both halfwords of words and bytes of halfwords in the data. Equivalent to -sS.
Sets the I/O block size to block-size * 512 bytes.
Set the I/O block size to 5120 bytes.
Use the old portable (ASCII) archive format.
Set the I/O block size to the given number of bytes.
Create leading directories where needed.
Create reproducible archives. This is equivalent to --ignore-devno --ignore-dirnlink --renumber-inodes.
Change to directory dir
Read additional patterns specifying filenames to extract or list from file.
Only copy files that do not match any of the given patterns.
Use the supplied archive-file instead of standard input. Optional user and host specify the user and host names in case of a remote archive.
Treat the archive file as local, even if its name contains colons.
Use given archive format. See format, for a list of available formats.
Store 0 in the device number field of each archive member, instead of the actual device number.
Store 2 in the nlink field of each directory archive member,
instead of the actual number of links.
Link files instead of copying them, when possible.
Dereference symbolic links (copy the files that they point to instead of copying the links).
Retain previous file modification times when creating files.
Print string when the end of a volume of the backup media is reached.
In the verbose table of contents listing, show numeric UID and GID values.
Do not change the ownership of the files.
When reading a CRC format archive, only verify the CRC’s of each file in the archive, don’t actually extract the files
Do not print the number of blocks copied.
Use command instead of rsh to access remote archives.
Interactively rename files
Renumber inodes when storing them in the archive.
Set the ownership of all files created to the specified user and/or group. See owner.
Swap the bytes of each halfword in the files
Write files with large blocks of zeros as sparse files.
Swap the halfwords of each word (4 bytes) in the files
Extract files to standard output.
Replace all files unconditionally.
Verbosely list the files processed.
Print a ‘.’ for each file processed.
Control warning display. Argument is one of ‘none’, ‘truncate’, ‘no-truncate’ or ‘all’. See warning, for a detailed discussion of these.
This section summarizes all available command line options. References
in square brackets after each option indicate cpio modes in
which this option is valid.
-0--null[copy-in,copy-out,copy-pass]
Read a list of filenames terminated by a null character, instead of a
newline, so that files whose names contain newlines can be archived.
GNU find is one way to produce a list of null-terminated filenames.
This option may be used in copy-out and copy-pass modes.
-a--reset-access-time[copy-out,copy-pass]
Reset the access times of files after reading them, so
that it does not look like they have just been read.
-A--append[copy-out]
Append to an existing archive. Only works in copy-out
mode. The archive must be a disk file specified with
the -O or -F (--file) option.
-b--swap[copy-in]
Swap both halfwords of words and bytes of halfwords in the data.
Equivalent to -sS. This option may be used in copy-in mode. Use this
option to convert 32-bit integers between big-endian and little-endian
machines.
-B[copy-in,copy-out,copy-pass]
Set the I/O block size to 5120 bytes. Initially the
block size is 512 bytes.
--block-size=block-size[copy-in,copy-out,copy-pass]
Set the I/O block size to block-size * 512 bytes.
-c[copy-in,copy-out,copy-pass]
Use the old portable (ASCII) archive format.
-C io-size--io-size=io-size[copy-in,copy-out,copy-pass]
Set the I/O block size to io-size bytes.
-d--make-directories[copy-in,copy-pass]
Create leading directories where needed.
-D dir--directory=dir[copy-in,copy-out,copy-pass]
Change to the directory dir before starting the operation. This
can be used, for example, to extract an archive contents in a
different directory:
$ cpio -i -D /usr/local < archive
or to copy-pass files from one directory to another:
$ cpio -D /usr/bin -p /usr/local/bin < filelist
The -D option does not affect file names supplied as arguments to another command line options, such as -F or -E. For example, the following invocation:
cpio -D /tmp/foo -d -i -F arc
instructs cpio to open the archive file arc in
the current working directory, then change to the directory
/tmp/foo and extract files to that directory. If
/tmp/foo does not exist, it will be created first (the
-d option) and then changed to.
-E file--pattern-file=file[copy-in]
Read additional patterns specifying filenames to extract or list from
file. The lines of file are treated as if they had been non-option
arguments to cpio. This option is used in copy-in mode,
-f--nonmatching[copy-in]
Only copy files that do not match any of the given
patterns.
-F archive--file=archive[copy-in,copy-out]
Archive filename to use instead of standard input or output. To use a
tape drive on another machine as the archive, use a filename that starts
with ‘hostname:’, where hostname is the name or IP
address of the machine. The hostname can be preceded by a username and an
‘@’ to access the remote tape drive as that user, if you have
permission to do so (typically an entry in that user’s ~/.rhosts
file).
--force-local[copy-in,copy-out]
With -F, -I, or -O, take the archive file
name to be a local file even if it contains a colon, which would
ordinarily indicate a remote host name.
-H format--format=format[copy-in,copy-out,copy-pass]
Use archive format format. The valid formats are listed below
with file size limits for individual files in parentheses; the same
names are also recognized in all-caps. The default in copy-in mode is
to automatically detect the archive format, and in copy-out mode is
‘bin’.
The obsolete binary format. (2147483647 bytes)
The old (POSIX.1) portable format. (8589934591 bytes)
The new (SVR4) portable format, which supports file systems having more than 65536 i-nodes. (4294967295 bytes)
The new (SVR4) portable format with a checksum added.
The old tar format. (8589934591 bytes)
The POSIX.1 tar format. Also recognizes GNU tar archives, which are
similar but not identical. (8589934591 bytes)
The obsolete binary format used by HPUX’s cpio (which stores device
files differently).
The portable format used by HPUX’s cpio (which stores device files
differently).
-i--extractRun in copy-in mode. See Copy-in mode.
-I archive[copy-in]
Archive filename to use instead of standard input. To use a tape drive
on another machine as the archive, use a filename that starts with
‘hostname:’, where hostname is the name or IP address
of the remote host. The hostname can be preceded by a username and an
‘@’ to access the remote tape drive as that user, if you have
permission to do so (typically an entry in that user’s
~/.rhosts file).
-l--link[copy-pass]
Link files instead of copying them, when possible.
-L--dereference[copy-in,copy-pass]
Copy the file that a symbolic link points to, rather than the symbolic
link itself.
-m--preserve-modification-time[copy-in,copy-pass]
Retain previous file modification times when creating files.
-M message--message=message[copy-in,copy-out]
Print message when the end of a volume of the backup media (such as a
tape or a floppy disk) is reached, to prompt the user to insert a new
volume. If message contains the string ‘%d’, it is replaced by the
current volume number (starting at 1).
-n--numeric-uid-gid[copy-in]
Show numeric UID and GID instead of translating them into names when
using the --verbose option.
--no-absolute-filenames[copy-in,copy-out]
Create all files relative to the current directory in copy-in mode, even
if they have an absolute file name in the archive.
--no-preserve-owner[copy-in,copy-pass]
Do not change the ownership of the files; leave them owned by the user
extracting them. This is the default for non-root users, so that users
on System V don’t inadvertantly give away files. This option can be
used in copy-in mode and copy-pass mode
-o--createRun in copy-out mode. See Copy-out mode.
-O archive[copy-out]
Archive filename to use instead of standard output. To use a tape drive
on another machine as the archive, use a filename that starts with
‘hostname:’, where hostname is the name or IP address
of the machine. The hostname can be preceded by a username and an ‘@’ to
access the remote tape drive as that user, if you have permission to do
so (typically an entry in that user’s ~/.rhosts file).
--only-verify-crc[copy-in]
Verify the CRC’s of each file in the archive, when reading a CRC format
archive. Don’t actually extract the files.
-p--pass-throughRun in copy-pass mode. See Copy-pass mode.
--quiet[copy-in,copy-out,copy-pass]
Do not print the number of blocks copied.
-r--rename[copy-in]
Interactively rename files.
-R owner--owner owner[copy-in,copy-out,copy-pass]
In copy-in and copy-pass mode, set the ownership of all files created
to the specified owner (this operation is allowed only for the
super-user). In copy-out mode, store the supplied owner information in
the archive.
The argument can be either the user name or the user name and group name, separated by a dot or a colon, or the group name, preceeded by a dot or a colon, as shown in the examples below:
cpio --owner smith cpio --owner smith: cpio --owner smith:users cpio --owner :users
The argument parts are first looked up in the system user and group databases, correspondingly. If any of them is not found there, it is treated as numeric UID or GID, provided that it consists of decimal digits only.
To avoid the lookup and ensure that arguments are treated as numeric values, prefix them with a plus sign, e.g.:
cpio --owner +0 cpio --owner +0: cpio --owner +0:+0 cpio --owner :+0
If the group is omitted but the ‘:’ or ‘.’ separator is given, as in the second example. the given user’s login group will be used.
--rsh-command=command[copy-in,copy-out,copy-pass]
Notifies cpio that is should use command to
communicate with remote devices.
-s--swap-bytes[copy-in]
Swap the bytes of each halfword (pair of bytes) in the files. This option
can be used in copy-in mode.
-S--swap-halfwords[copy-in]
Swap the halfwords of each word (4 bytes) in the files. This option may
be used in copy-in mode.
--sparse[copy-in,copy-pass]
Write files with large blocks of zeros as sparse files. This option is
used in copy-in and copy-pass modes.
-t--list[copy-in]
Print a table of contents of the input. Can be used alone, as a
mode designator, in which case -i is implied.
--to-stdout[copy-in]
Extract files to standard output. This option may be used in copy-in mode.
-u--unconditional[copy-in,copy-pass]
Replace all files, without asking whether to replace
existing newer files with older files.
-v--verbose[copy-in,copy-out,copy-pass]
List the files processed, or with -t, give an ls -l style
table of contents listing. In a verbose table of contents of a ustar
archive, user and group names in the archive that do not exist on the
local system are replaced by the names that correspond locally to the
numeric UID and GID stored in the archive.
-V--dot[copy-in,copy-out,copy-pass]
Print a ‘.’ for each file processed.
--versionPrint the cpio program version number and exit.
-W--warning=flag[copy-in,copy-out,copy-pass]
Control warning display. The argument is one of the following:
Disable all warnings.
Enable all warnings.
Warn about truncation of file header fields.
Disable truncation warnings.
Archives are usually written on removable media–tape cartridges, mag tapes, or floppy disks.
The amount of data a tape or disk holds depends not only on its size, but also on how it is formatted. A 2400 foot long reel of mag tape holds 40 megabytes of data when formated at 1600 bits per inch. The physically smaller EXABYTE tape cartridge holds 2.3 gigabytes.
Magnetic media are re-usable–once the archive on a tape is no longer needed, the archive can be erased and the tape or disk used over. Media quality does deteriorate with use, however. Most tapes or disks should be disgarded when they begin to produce data errors.
Magnetic media are written and erased using magnetic fields, and should be protected from such fields to avoid damage to stored data. Sticking a floppy disk to a filing cabinet using a magnet is probably not a good idea.
It is possible you will encounter a bug in cpio.
If this happens, we would like to hear about it. As the purpose of bug
reporting is to improve software, please be sure to include maximum
information when reporting a bug. The information needed is:
Send your report to bug-cpio@gnu.org. This is a public mailing
list; all correspondence sent to it is archived and is available at
https://lists.gnu.org/archive/html/bug-cpio. Your bug report
will be visible there, too. The list is not limited to bug reports,
in fact it is dedicated to any kind of technical discussions regarding
GNU cpio. If you wish to subscribe to it, visit
https://lists.gnu.org/mailman/listinfo/bug-cpio.