Genimage - The Image Creation Tool

genimage is a tool to generate multiple filesystem and flash/disk images from a given root filesystem tree. genimage is intended to be run in a fakeroot environment. It also supports creating flash/disk images out of different file-system images and files.

Configuration is done in a config file parsed by libconfuse. Options like the path to tools can be given via environment variables, the config file or from commandline switches.

The Configuration File

The config file of genimage uses a simple configuration language, provided by libconfuse. This supports nested sections, as well as simple key-value pairs.

Single-line comments can be introduced with # or //, multi-line comments look like /* … */ (as in C).

The config file is separated into the main sections image, flash and config, and provides an include primitive.

The image section

An image section describes a single filesystem or disk image to be built. It can be given multiple times to generate multiple images. An image can also have multiple partitions which refer to images themselves. Each image must have a type which can have different suboptions depending on the type.

Let's have a look at an example:

image nand-pcm038.img {
    flash {
    }
    flashtype = "nand-64M-512"
    partition barebox {
        image = "barebox-pcm038.bin"
        size = 512K
    }
    partition root {
        image = "root-nand.jffs2"
        size = 24M
    }
}

This would generate a nand-pcm038.img which is a flash of type "nand-64M-512" The image contains two partitions, "barebox-pcm038.bin" and "root-nand.jffs2" which must refer to images described elsewhere in the config file. For example "root-nand.jffs2" partition could be described like this:

image root-nand.jffs2 {
    name = "root"
    jffs2 {}
    size = 24M
    mountpoint = "/"
}

In this case a single jffs2 image is generated from the root mountpoint.

Here are all options for images:

name

The name of this image. This is used for some image types to set the name of the image.

size

Size of this image in bytes. 'k', 'M' or 'G' can be used as suffix to specify the size in multiple of 1024 etc. The suffix 's' specifies a multiple of the (traditional) sector size of 512. If the image if filled from a mountpoint then '%' as suffix indicates a percentage. '200%' means the resulting filesystem should be about 50% filled. Note that is is only a rough estimate based on the original size of the content.

mountpoint

mountpoint if image refers to a filesystem image. The default is "/". The content of "${rootpath}${mountpoint}" will be used to fill the filesystem.

srcpath

If this is set, specified path will be directly used to fill the filesystem. Ignoring rootpath/mountpoint logic. Path might be absolute or relative to current working directory.

empty

If this is set to true, then the specified rootpath and mountpoint are ignored for this image and an empty filesystem is created. This option is only used for writeable filesystem types, such as extX, vfat, ubifs and jffs2. This defaults to false.

temporary

If this is set to true, the image is created in tmppath rather than outputpath. This can be useful for intermediate images defined in the configuration file which are not needed by themselves after the main image is created. This defaults to false.

exec-pre

Custom command to run before generating the image.

exec-post

Custom command to run after generating the image.

flashtype

refers to a flash section. Optional for non flash like images like hd images

partition

can be given multiple times and refers to a partition described below

Additionally each image can have one of the following sections describing the type of the image:

cpio, cramfs, ext2, ext3, ext4, file, flash, hdimage, iso, jffs2, qemu, squashfs, tar, ubi, ubifs, vfat.

Partition options:

offset

The offset of this partition as a total offset to the beginning of the device.

size

The size of this partition in bytes. If the size and autoresize are both not set then the size of the partition image is used.

align

Alignment value to use for automatic computation of offset and size. Defaults to 1 for partitions not in the partition table, otherwise to the image's align value.

partition-type

Used by dos partition tables to specify the partition type.

image

The image file this partition shall be filled with

fill

Boolean specifying that all bytes of the partition should be explicitly initialized. Any bytes beyond the size of the specified image will be set to 0.

autoresize

Boolean specifying that the partition should be resized automatically. For UBI volumes this means that the autoresize flag is set. Only one volume can have this flag. For hd images this can be used for the last partition. If set the partition will fill the remaining space of the image.

bootable

Boolean specifying whether to set the bootable flag.

in-partition-table

Boolean specifying whether to include this partition in the partition table. Defaults to true.

partition-uuid

UUID string used by GPT partition tables to specify the partition id. Defaults to a random value.

partition-type-uuid

String used by GPT partition tables to specify the partition type. Either a UUID or a shortcut:

• L, linux, linux-generic: Linux filesystem (0fc63daf-8483-4772-8e79-3d69d8477de4)
• S, swap: Swap (0657fd6d-a4ab-43c4-84e5-0933c84b4f4f)
• H, home: Home (933ac7e1-2eb4-4f13-b844-0e14e2aef915)
• U, esp, uefi: EFI System Partition (c12a7328-f81f-11d2-ba4b-00a0c93ec93b)
• R, raid: Linux RAID (a19d880f-05fc-4d3b-a006-743f0f84911e)
• V, lvm: Linux LVM (e6d6d379-f507-44c2-a23c-238f2a3df928)
• F, fat32: FAT32 / Basic Data Partition (ebd0a0a2-b9e5-4433-87c0-68b6b72699c7)
• barebox-state (previously B): Barebox State (4778ed65-bf42-45fa-9c5b-287a1dc4aab1)

• barebox-env: Barebox Environment (6c3737f2-07f8-45d1-ad45-15d260aab24d)

  Furthermore, for {arch} being one of alpha, arc, arm, arm64, ia64, loongarch64, mips, mips64, mips-le, mips64-le, parisc, ppc, ppc64, ppc64-le, riscv32, riscv64, s390, s390x, tilegx, x86, x86-64, the following shortcuts from the Discoverable Partitions Specification are accepted (see the spec for the respective UUIDs):

  • root-{arch}: Root Partition
  • usr-{arch}: /usr Partition
  • root-{arch}-verity: Root Verity Partition
  • usr-{arch}-verity: /usr Verity Partition
  • root-{arch}-verity-sig: Root Verity Signature Partition
  • usr-{arch}-verity-sig: /usr Verity Signature Partition
  • xbootldr: Extended Boot Loader Partition
  • srv: Server Data Partition
  • var: Variable Data Partition
  • tmp: Temporary Data Partition
  • user-home: Per-user Home Partition

Defaults to L.

For each partition, its final alignment, offset and size are determined as follows:

• If the align option is not present, it defaults to the value of the image's align option if the partition is in the partition table, otherwise to 1.
• If the offset option is absent or zero, and in-partition-table is true, the partition is placed after the end of all previously defined partitions, with the final offset rounded up to the partition's align value.
• Otherwise, the offset option is used as-is. Note that if absent, that option defaults to 0, so in practice one must specify an offset for any partition that is not in the partition table (with at most one exception, e.g. a bootloader).
• If the partition has the autoresize flag set, its size is computed as the space remaining in the image from its offset (for a GPT image, space is reserved at the end for the backup GPT table), rounded down to the partition's align value. If the partition also has a size option, it is ensured that the computed value is not less than that size.
• Otherwise, if the size option is present and non-zero, its value is used as-is.
• Otherwise, if the partition has an image option, the size of that image, rounded up to the partition's align value, is used to determine the size of the partition.

The following sanity checks are done on these final values (in many cases, these will automatically be satisfied when the value has been determined via one of the above rules rather than given explicitly):

• For a partition in the partition table, the partition's align value must be greater than or equal to the image's align value.
• The partition's offset and size must both be multiples of its align.
• The size must not be 0.
• The partition must not overlap any other partition, or the areas occupied by the partition table.

The image configuration options

android-sparse

Generate android sparse images. They are typically used by fastboot. Sparse images encode "don't care" areas and areas that are filled with a single 32 bit value. As a result, they are often much smaller than raw disk images. Genimage assumes that all 'holes' in the input file are "don't care" areas. This is a reasonable assumption: Tools to generate filesystems typically operate on devices. So they only create holes in areas they don't care about. Genimage itself operates the same way when generating HD images.

Options:

image

The source image that will be converted.

block-size

The granularity that the sparse image uses to find "don't care" or "fill" blocks. The supported block sizes depend on the user. The default is 4k.

cpio

Generates cpio images.

Options:

format

archive format. Passed to the -H option to the cpio command. Valid options are bin, odc, newc, crc, tar, ustar, hpbin and hpodc

extraargs

Extra arguments passed to the cpio tool

compress

If given, pipe image through compression tool. Valid options are for example gzip, lzop or any other tool that compresses from stdin to stdout.

cramfs

Generates cramfs images.

Options:

extraargs

Extra arguments passed to mkcramfs

ext2, ext3, ext4

Generates ext* images.

Options:

use-mke2fs

If set to true, then mke2fs is used to create the image. Otherwise, genext2fs is used. Defaults to false.

mke2fs-conf

mke2fs.conf that should be used. If unspecified, the system default is used.

extraargs

Extra arguments passed to genext2fs or mke2fs.

features

Filesystem features. Passed to the -O option of tune2fs. This is a comma separated list of enabled or disabled features. See man ext4 for features. For genext2fs all feature are specified. Default for ext3 images is has_journal. Default for ext4 images is extents,uninit_bg,dir_index,has_journal. For mke2fs these features are added in addition to the default features of the ext type. Already enabled features can be disabled by prefixing the feature with ^.

label

Specify the volume-label. Passed to the -L option of tune2fs

fs-timestamp

Sets different timestamps in the image. Sets the given timestamp using the debugfs commands set_current_time, set_super_value mkfs_time and set_super_value lastcheck

root-owner

User and group IDs for the root directory. Defaults to 0:0. Only valid with mke2fs.

usage-type

Specify the usage type for the filesystem. Only valid with mke2fs. More details can be found in the mke2fs man-page.

file

This represents a pre-existing image which will be used as-is. When a partition section references an image that is not defined elsewhere in the configuration file, a file rule is implicitly generated. It is up to the user to ensure that the image exists in the input directory, or to use an absolute path to the image.

It is possible to add a file image explicitly, which allows one to provide genimage with some information about the image which can not be deduced automatically. Currently, one such option exists:

holes

A list of "(<start>;<end>)" pairs specifying ranges of the file that do not contain meaningful data, and which can therefore be allowed to overlap other partitions or image metadata.

For example:

image foo {
    hdimage {
        partition-table-type = "gpt"
        gpt-location = 64K
    }

    partition bootloader {
        in-partition-table = false
        offset = 0
        image = "/path/to/bootloader.img"
    }

    partition rootfs {
        offset = 1M
        image = "rootfs.ext4"
    }
}

image /path/to/bootloader.img {
    file {
        holes = {"(440; 1K)", "(64K; 80K)"}
    }
}

This tells genimage that despite the bootloader partition overlapping both the last 72 bytes of the MBR (where the DOS partition table is located) and the GPT header occupying the sector starting at offset 512, this is all OK because bootloader.img does not contain useful data in that range. Further, in this example, the bootloader image has been carefully crafted to also allow placing the GPT array at offset 64K (the GPT header is always at offset 512).

If the bootloader image is not declared explicitly and only used once then the holes can also be configured in the partition. This simplifies the config file for simple use-cases.

For example:

image bar {
    hdimage {}

    partition bootloader {
        in-partition-table = false
        offset = 0
        image = "/path/to/bootloader.img"
        holes = {"(440; 512)"}
    }

    partition rootfs {
        offset = 1M
        image = "rootfs.ext4"
    }
}

FIT

Generates U-Boot FIT images.

Options:

its

String option holding the path of the input its file

keydir

String option holding the directory containing the keys used for signing.

flash

Generates flash images. These are basically the partition contents padded to the partition sizes concatenated together. There is no partition table. Needs a valid flashtype where the flash parameters are read from.

hdimage

Generates DOS partition images.

Options:

align

Partition alignment. Defaults to 512 bytes

partition-table

Boolean. If true, writes a partition table. If false, no partition table is generated. Defaults to true. Deprecated: use partition-table-type instead.

partition-table-type

Define what kind of partition table should be used. Valid options are: * none: No partition table at all. In this case, the in-partition-table option for each partition is ignored. * mbr: Legacy DOS/MBR partition table * gpt: GUID Partition Table * hybrid: A hybrid MBR/GPT partition table. Partitions with an explicit partition-type will be placed in in the MBR table. At most 3 such partitions are allowed. This limit does not effect the maximum number of GPT partition entries in the same image.

extended-partition

Number of the extended partition. Contains the number of the extended partition between 1 and 4 or 0 for automatic. Defaults to 0.

disk-signature

32 bit integer used as disk signature (offset 440 in the MBR). Using a special value random will result in using random 32 bit number.

gpt

Boolean. If true, a GPT type partition table is written. If false a DOS type partition table is written. Defaults to false. Deprecated: use partition-table-type instead.

gpt-location

Location of the GPT table. Occasionally useful for moving the GPT table away from where a bootloader is placed due to hardware requirements. All partitions in the table must begin after this table. Regardless of this setting, the GPT header will still be placed at 512 bytes (sector 1). Defaults to 1024 bytes (sector 2).

gpt-no-backup

Boolean. If true, then the backup partition table at the end of the image is not written.

disk-uuid

UUID string used as disk id in GPT partitioning. Defaults to a random value.

fill

If this is set to true, then the image file will be filled up to the end of the last partition. This might make the file bigger. This is necessary if the image will be processed by such tools as libvirt, libguestfs or parted.

iso

Generates an ISO image.

Options:

boot-image

Path to the El Torito boot image. Passed to the -b option of genisofs

bootargs

Bootargs for the El Torito boot image. Defaults to -no-emul-boot -boot-load-size 4 -boot-info-table -c boot.cat -hide boot.cat

extraargs

Extra arguments passed to genisofs

input-charset

The input charset. Passed to the -input-charset option of genisofs. Defaults to default

volume-id

Volume ID. Passed to the -V option of genisofs

jffs2

Generates a JFFS image. Needs a valid flashtype where the flash parameters are read from.

Options:

extraargs

Extra arguments passed to mkfs.jffs2

qemu

Generates a QEMU image. Needs at least one valid partition.

Options:

format

A valid qemu-img like qcow, qcow2, parallels, vdi, vhdx or vmdk. Check qemu-img convert --help for the complete list of possible values. Defaults to qcow2.

extraargs

Extra arguments passed to qemu-img convert

squashfs

Generates a squashfs image.

Options:

extraargs

Extra arguments passed to mksquashfs

compression

compression type for the image. Possible values are gzip (default), lzo, xz or none

block-size

Block size. Passed to the -b option of mksquashfs. Defaults to 4096.

rauc

Generates a RAUC update bundle.

Options:

extraargs

Extra arguments passed to RAUC

file

Specify a file to be added into the RAUC bundle. Usage is: file foo { image = "bar" } which adds a file "foo" in the RAUC bundle from then input file "bar"

files

A list of filenames added into the RAUC bundle. Like file above, but without the ability to add the files under different name.

key

Path to the key file or PKCS#11 URI. Passed to the --key option of RAUC

cert

Path to the certificate file or PKCS#11 URI. Passed to the --cert option of RAUC

keyring

Optional path to the keyring file. Passed to the --keyring option of RAUC

manifest

content of the manifest file

tar

Generates a tar image. The image will be compressed as defined by the filename suffix.

ubi

Generates an UBI image. Needs a valid flashtype where the flash parameters are read from.

Options:

extraargs

Extra arguments passed to ubinize

ubifs

Generates a UBIFS image. Needs a valid flashtype where the flash parameters are read from.

Options:

extraargs

Extra arguments passed to mkubifs

max-size

Maximum size of the UBIFS image

space-fixup

Instructs the file-system free space to be freed up on first mount.

vfat

Generates a VFAT image.

Options:

extraargs

Extra arguments passed to mkdosfs

label

Specify the volume-label. Passed to the -n option of mkdosfs

file

Specify a file to be added into the filesystem image. Usage is: file foo { image = "bar" } which adds a file "foo" in the filesystem image from the input file "bar"

files

A list of filenames added into the filesystem image. Like file above, but without the ability to add the files under different name.

Note: If no content is specified with file or files then rootpath and mountpoint are used to provide the content.

fip

Generates a Firmware Image Package (FIP). A format used to bundle firmware to be loaded by ARM Trusted Firmware.

Options:

extraargs

Extra arguments passed to fiptool

fw-config

Firmware Configuration (device tree), usually provided by BL2 (Trusted Firmware)

nt-fw

Non-Trusted Firmware (BL33)

hw-config

Hardware Configuration (device tree), passed to BL33

tos-fw

Trusted OS (BL32) binaries. Second and third binary are used as extra1 and extra2 binaries if specified. Example: tos-fw = {"tee-header_v2.bin", "tee-pager_v2.bin", "tee-pageable_v2.bin"}

scp-fwu-cfg

SCP Firmware Updater Configuration FWU SCP_BL2U

ap-fwu-cfg

AP Firmware Updater Configuration BL2U

fwu

Firmware Updater NS_BL2U

fwu-cert

Non-Trusted Firmware Updater certificate

tb-fw

Trusted Boot Firmware BL2

scp-fw

SCP Firmware SCP_BL2

soc-fw

EL3 Runtime Firmware BL31

tb-fw-config

TB_FW_CONFIG

soc-fw-config

SOC_FW_CONFIG

tos-fw-config

TOS_FW_CONFIG

nt-fw-config

NT_FW_CONFIG

rot-cert

Root Of Trust key certificate

trusted-key-cert

Trusted key certificate

scp-fw-key-cert

SCP Firmware key certificate

soc-fw-key-cert

SoC Firmware key certificate

tos-fw-key-cert

Trusted OS Firmware key certificate

nt-fw-key-cert

Non-Trusted Firmware key certificate

tb-fw-cert

Trusted Boot Firmware BL2 certificate

scp-fw-cert

SCP Firmware content certificate

soc-fw-cert

SoC Firmware content certificate

tos-fw-cert

Trusted OS Firmware content certificate

nt-fw-cert

Non-Trusted Firmware content certificate

sip-sp-cert

SiP owned Secure Partition content certificate

plat-sp-cert

Platform owned Secure Partition content certificate

The Flash Section

The flash section can be given multiple times and each section describes a flash chip. The option names are mostly derived from the UBI terminology. There are the following options:

pebsize

The size of a physical eraseblock in bytes

lebsize

The size of a logical eraseblock in bytes (for ubifs)

numpebs

Number of physical eraseblocks on this device. The total size of the device is determined by pebsize * numpebs

minimum-io-unit-size

The minimum size in bytes accessible on this device

vid-header-offset

offset of the volume identifier header

sub-page-size

The size of a sub page in bytes.

Several flash related image types need a valid flash section. From the image types the flash type section is referred to using the flashtype option which contains the name of the flash type to be used.

For more information of the meaning of these values see the ubi(fs) and mtd FAQs:

http://www.linux-mtd.infradead.org/faq/general.html

Example flash section:

flash nand-64M-512 {
    pebsize = 16384
    lebsize = 15360
    numpebs = 4096
    minimum-io-unit-size = 512
    vid-header-offset = 512
    sub-page-size = 512
}
...
image jffs2 {
    flashtype = "nand-64M-512"
}

The config section

In this section the global behaviour of the program is described. Except as noted below, all options here can be given from either environment variables, the config file or command line switches. For instance, a config option foo can be passed as a --foo command line switch or as a GENIMAGE_FOO environment variable.

config

default: genimage.cfg Path to the genimage config file.

loglevel

default: 1 genimage log level.

outputpath

default: images Mandatory path where all images are written to (must exist).

inputpath

default: input This mandatory path is searched for input images, for example bootloader binaries, kernel images (must exist).

rootpath

default: root Mandatory path to the root filesystem (must exist).

tmppath

default: tmp Optional path to a temporary directory. There must be enough space available here to hold a copy of the root filesystem.

includepath

Colon-separated list of directories to search for files included via the include function. The current directory is searched after these. Thus, if this option is not given, only the current directory is searched. This has no effect when given in the config file.

configdump

File to write the final configuration to. This includes the results of all include directives, expansions of environment variables and application of default values - think gcc -E. Use - for stdout.

cpio

path to the cpio program (default cpio)

dd

path to the dd program (default dd)

e2fsck

path to the e2fsck program (default e2fsck)

genext2fs

path to the genext2fs program (default genext2fs)

genisoimage

path to the genisoimage program (default genisoimage)

mcopy

path to the mcopy program (default mcopy)

mmd

path to the mmd program (default mmd)

mkcramfs

path to the mkcramfs program (default mkcramfs)

mkdosfs

path to the mkdosfs program (default mkdosfs)

mkfsjffs2

path to the mkfs.jffs2 program (default mkfs.jffs2)

mkfsubifs

path to the mkfs.ubifs program (default mkfs.ubifs)

mksquashfs

path to the mksquashfs program (default mksquashfs)

qemu-img

path to the qemu-img program (default qemu-img)

tar

path to the tar program (default tar)

tune2fs

path to the tune2fs program (default tune2fs)

ubinize

path to the ubinize program (default ubinize)

fiptool

path to the fiptool utility (default fiptool)

Include Configurations Fragments

To include a "foo.cfg" config file, use the following statement:

include("foo.cfg")

This allows to re-use, for example flash configuration files, across different image configurations.

License and Developing

To contribute to genimage please prepare a pull request on Github. To make it possible to include your modifications it's required that your code additions are licensed under the same terms as genimage itself. So you are required to agree to the following document:

Developer's Certificate of Origin 1.1

By making a contribution to this project, I certify that:

1. The contribution was created in whole or in part by me and I have the right to submit it under the open source license indicated in the file; or
2. The contribution is based upon previous work that, to the best of my knowledge, is covered under an appropriate open source license and I have the right under that license to submit that work with modifications, whether created in whole or in part by me, under the same open source license (unless I am permitted to submit under a different license), as indicated in the file; or
3. The contribution was provided directly to me by some other person who certified (a), (b) or (c) and I have not modified it.
4. I understand and agree that this project and the contribution are public and that a record of the contribution (including all personal information I submit with it, including my sign-off) is maintained indefinitely and may be redistributed consistent with this project or the open source license(s) involved.

Your agreement is expressed by adding a sign-off line to each of your commits (e.g. using git commit -s) looking as follows:

Signed-off-by: Random J Developer <random@developer.example.org>

with your identity and email address matching the commit meta data.