Scroll to navigation

GENEXT2FS(8) System Manager's Manual GENEXT2FS(8)


genext2fs - ext2 filesystem generator for embedded systems


genext2fs [ options ] [ output-image ]


genext2fs generates an ext2 filesystem as a normal (non-root) user. It does not require you to mount the image file to copy files on it, nor does it require that you become the superuser to make device nodes.

The filesystem is created either from scratch, or from an already existing one if specified by the -x option. Then each -d and -D option successively adds a "layer" to the image.

The filesystem image is created in the file output-image. If not specified, it is sent to stdout.

By default, the maximum number of inodes in the filesystem is the minimum number required to accommodate the initial contents. In this way, a minimal filesystem (typically read-only) can be created with minimal free inodes. If required, free inodes can be added by passing the relevant options. The filesystem image size in blocks can be minimised by trial and error.


-x, --starting-image image
Use this image as a starting point.
-d, --root directory[:path]
Add the given directory and contents at a particular path (by default the root).
-D, --devtable spec-file[:path]
Use spec-file to specify inodes to be added, at the given path (by default the root), including files, directories and special files like devices. If the specified files are already present in the image, their ownership and permission modes will be adjusted accordingly (this can only occur when the -D option appears after the options that create the specified files). Furthermore, you can use a single table entry to create many devices with a range of minor numbers (see examples below). All specified inodes receive the mtime of spec-file itself.
-a, --tarball file[:path]
Add the given archive (tarball) contents at a particular path (by default the root). Note: if not compiled with `libarchive`, genext2fs will use a builtin tarball parser with very primitive capabilities (e.g. no sparse file support, generally no support other than for modern GNU tar without fancy options).
-b, --size-in-blocks blocks
Size of the image in blocks.
-B, --block-size bytes
Size of a filesystem block in bytes.
-N, --number-of-inodes inodes
Maximum number of inodes.
-L, --volume-label name
Set the volume label for the filesystem.
-i, --bytes-per-inode ratio
Used to calculate the maximum number of inodes from the available blocks.
-m, --reserved-percentage N
Number of reserved blocks as a percentage of size. Reserving 0 blocks will prevent creation of the "lost+found" directory.
-o, --creator-os name
Value for creator OS field in superblock.
-g, --block-map path
Generate a block map file for this path.
-e, --fill-value value
Fill unallocated blocks with value.
-z, --allow-holes
Make files with holes.
-f, --faketime
Use a timestamp of 0 for inode and filesystem creation, instead of the present. Useful for testing. See also SOURCE_DATE_EPOCH.
-q, --squash
Squash permissions and owners (same as -P -U).
-U, --squash-uids
Squash ownership of inodes added using the -d option, making them all owned by root:root.
-P, --squash-perms
Squash permissions of inodes added using the -d option. Analogous to "umask 077".
-v, --verbose
Print resulting filesystem structure.
-V, --version
Print genext2fs version.
-h, --help
Display help.


Standardized date for reproducible builds, see for more information.


genext2fs -b 1440 -d src /dev/fd0

All files in the src directory will be written to /dev/fd0 as a new ext2 filesystem image. You can then mount the floppy as usual.

genext2fs -b 1024 -d src -D device_table.txt flashdisk.img

This example builds a filesystem from all the files in src, then device nodes are created based on the contents of the file device_table.txt. Entries in the device table take the form of:

<name> <type> <mode> <uid> <gid> <major> <minor> <start> <inc> <count>

where name is the file name and type can be one of:

f	A regular file
d	Directory
c	Character special device file
b	Block special device file
p	Fifo (named pipe)
l	Symbolic link
uid is the user id for the target file, gid is the group id for the target file. The rest of the entries (major, minor, etc) apply only to device special files.

An example device file follows:

# name	type mode uid gid major minor start inc count
/dev		d	755	0	0	-	-	-	-	-
/dev/mem	c	640	0	0	1	1	0	0	-
/dev/tty	c	666	0	0	5	0	0	0	-
/dev/tty	c	666	0	0	4	0	0	1	6
/dev/loop	b	640	0	0	7	0	0	1	2
/dev/hda	b	640	0	0	3	0	0	0	-
/dev/hda	b	640	0	0	3	1	1	1	16
/dev/log	s	666	0	0	-	-	-	-	-

This device table creates the /dev directory, a character device node /dev/mem (major 1, minor 1), and also creates /dev/tty, /dev/tty[0-5], /dev/loop[0-1], /dev/hda, /dev/hda1 to /dev/hda15 and /dev/log socket.


mkfs(8), genromfs(8), mkisofs(8), mkfs.jffs2(1)


This manual page was written by David Kimdon <>, for the Debian GNU/Linux system (but may be used by others). Examples provided by Erik Andersen <>.
August 19, 2006