What is omnect device management?: https://lp.conplement.de/omnect-devicemanagement
This yocto meta layer provides the device management distribution omnect-os
. It includes recipes for:
- iot-hub-device-update
- iot-identity-service
- iotedge
omnect-os image
: an updatable device image with A/B rootfs update support with the following implicit features:iot-hub-device-update
andiot-identity-service
are installediot-hub-device-update
is provisioned as module identity viaiot-identity-service
- First boot script
/usr/bin/omnect_first_boot.sh
which is executed at first boot of the device; it can be adapted viameta-omnect/recipes-core/systemd/systemd/omnect_first_boot.sh
- Factory reset via OS bootloader environment variable
factory-reset
- Note: This feature provides a limited level of data privacy. Please see section Factory Reset below.
omnect-os update image
: theswupdate
update image with the following implicit features:- Updating the bootloader
omnect-os
is built with yocto DISTRO_FEATURES
= ipv4 ipv6 polkit seccomp zeroconf
.
Depending on MACHINE_FEATURES
we also set 3g
, bluetooth
and wifi
.
meta-omnect
adds the following DISTRO_FEATURES
:
iotedge
- adds the
iotedge
service with its dependencies - adds
virtualization
toDISTRO_FEATURES
(from meta-virtualization) needed byiotedge
runtime dependencymoby
- adds the
persistent-var-log
- enables a persistent /var/log which is stored in the data partition
flash-mode
- provides the possibility to flash complete disk images
- please see section Flash Modes below
resize-data
- expands the data partition to available space on first boot
wifi-commissioning
- adds a service which enables wifi commissioning via bluetooth
- depends on
DISTRO_FEATURES
wifi
andbluetooth
which are not added toDISTRO_FEATURES
automatically! - note: this is only intended for demo purposes; this is not a production ready service
meta-omnect
extends the following MACHINE_FEATURES
:
tpm2
- adds tpm kernel overlay, driver and auto modprobe for raspberry pi
omnect-os
uses an A/B
update partition layout with two readonly rootfs partitions.
The partition layout for devices supporting gpt:
Device Start End Sectors Size Type
/dev/mmcblkXp1 8192 90111 81920 40M Microsoft basic data
/dev/mmcblkXp2 90112 1611775 1521664 743M Linux filesystem
/dev/mmcblkXp3 1613824 3135487 1521664 743M Linux filesystem
/dev/mmcblkXp4 3137536 3342335 204800 100M Linux filesystem
/dev/mmcblkXp5 3342336 3424255 81920 40M Linux filesystem
/dev/mmcblkXp6 3424256 3506175 81920 40M Linux filesystem
/dev/mmcblkXp7 3506176 62333918 58827743 28.1G Linux filesystem
mmcblkXp1
is theboot
partition with vfat filesystemmmcblkXp2
is the readonlyrootA
partition with ext4 filesystemmmcblkXp3
is the readonlyrootB
partition with ext4 filesystemmmcblkXp4
is the writablefactory
partition with ext4 filesystemmmcblkXp5
is the writablecertificate
partition with ext4 filesystemmmcblkXp6
is the writableetc
overlay partition (ext4 filesystem mounted as overlayfs on/etc
)mmcblkXp7
is the writabledata
partition with ext4 filesystem
The partition layout for devices supporting mbr:
Device Boot Start End Sectors Size Id Type
/dev/mmcblkXp1 * 8192 90111 81920 40M c W95 FAT32 (LBA)
/dev/mmcblkXp2 106496 1628159 1521664 743M 83 Linux
/dev/mmcblkXp3 1630208 3151871 1521664 743M 83 Linux
/dev/mmcblkXp4 3153918 31116287 27962370 13.3G f W95 Ext'd (LBA)
/dev/mmcblkXp5 3153920 3235839 81920 40M 83 Linux
/dev/mmcblkXp6 3244032 3325951 81920 40M 83 Linux
/dev/mmcblkXp7 3334144 3416063 81920 40M 83 Linux
/dev/mmcblkXp8 3424256 31116287 27692032 13.2G 83 Linux
mmcblkXp1
is theboot
partition with vfat filesystemmmcblkXp2
is the readonlyrootA
partition with ext4 filesystemmmcblkXp3
is the readonlyrootB
partition with ext4 filesystemmmcblkXp5
is the writablefactory
partition with ext4 filesystemmmcblkXp6
is the writablecertificate
partition with ext4 filesystemmmcblkXp7
is the writableetc
overlay partition (ext4 filesystem mounted as overlayfs on/etc
)mmcblkXp8
is the writabledata
partition with ext4 filesystem
Note1: The partition layout is an example, sizes may differ depending on your board and image variant (development or release).
Note2: The size of data
depends on your sdcard/emmc/nvme size. Per default it has a size of 512M and is resized on the first boot to the max available size.
Note3 (only for images where u-boot is used as OS bootloader): There is a reserved area between the boot partition and the rootA partition used for two redundant u-boot environment banks. For this purpose, the following configuration variables are used:
OMNECT_PART_OFFSET_UBOOT_ENV1
: offset of 1st u-boot environment bank (in KiB, decimal)OMNECT_PART_OFFSET_UBOOT_ENV2
: offset of 2nd u-boot environment bank (in KiB, decimal)OMNECT_PART_SIZE_UBOOT_ENV
: size of one u-boot environment bank (in KiB, decimal)
meta-omnect
is compatible with the current yocto LTS release branch kirkstone
.
See board specific documents doc folder.
We reflect the used yocto version in our version schema. omnect-os
is versioned 4.0.x.y
where x
is yocto kirkstone's patch version and y
is the build number.
Aside from hardware specific meta layers meta-omnect
depends on:
- bitbake
- openembedded-core
- meta-openembedded:
meta-filesystems
,meta-networking
,meta-oe
andmeta-python
- meta-security
- meta-swupdate
- meta-virtualization
Note: Detailed information can be found in respective readmes
For using omnect-os-update-image
together with iot-hub-device-update
you have to provide a rsa-key for signing/verifying the update image.
Note: We currently only support swupdate
RSA signing.
Provide the environment variables SWUPDATE_PASSWORD_FILE
and SWUPDATE_PRIVATE_KEY
.
SWUPDATE_PASSWORD_FILE
: full path to a file containing the keys passwordSWUPDATE_PRIVATE_KEY
: full path of private key file
Furthermore you have to provide the environment variable OMNECT_USER_PASSWORD
which sets the password of the default user omnect
.
Optionally set OMNECT_BUILD_NUMBER
to set a meaningful build number in the distro version. The default is 0
.
There is the configuration variable OMNECT_VM_PANIC_ON_OOM
used to define the out-of-memory (OOM) handling.
Set the enviroment variable OMNECT_RELEASE_IMAGE
to 1
for release builds. The default is 0
which means it is a developer build.
Differences:
- Release build
- default firewall config which allows input for established connections only
- reduced serial console output (only output high priority messages).
- local ssh login disabled (alternatively a tunneled ssh connection can be established via omnect-cli)
- Developer build
- default firewall config as in Release build, additionally allow ssh connections
- local ssh login enabled
- includes some useful developer tools
This repository provides kas
configuration files to build omnect-os
.
E.g. if you want to build an omnect-os
raspberry pi 4 image with iotedge
support follow these steps:
mkdir omnect-os-build
cd omnect-os-build
git clone https://github.com/omnect/meta-omnect.git
# Generate 'SWUPDATE_PASSWORD_FILE' and 'SWUPDATE_PRIVATE_KEY'
echo "your password" > priv.pass
openssl genrsa -aes256 -passout file:priv.pass -out priv.pem
# Build 'omnect-os-image' and 'omnect-os-update-image' via:
docker run --rm \
-v $(pwd):/builder \
-u 0:$(id -g) \
-e USER_ID=$(id -u) \
-e GROUP_ID=$(id -g) \
-e OMNECT_BUILD_NUMBER=1 \
-e OMNECT_USER_PASSWORD="your escaped password" \
-e SWUPDATE_PASSWORD_FILE=/builder/priv.pass \
-e SWUPDATE_PRIVATE_KEY=/builder/priv.pem \
ghcr.io/siemens/kas/kas \
kas build \
meta-omnect/kas/distro/omnect-os.yaml:\
meta-omnect/kas/example/wifi-commissioning.yaml:\
meta-omnect/kas/feature/iotedge.yaml:\
meta-omnect/kas/feature/persistent-var-log.yaml:\
meta-omnect/kas/machine/rpi/rpi4.yaml
The resulting image artifacts are located in $(pwd)/build/deploy/images/raspberrypi4-64
.
The omnect-os-image
artefact is named omnect-os-raspberrypi4-64.wic.xz
.
The omnect-os-update-image
artefact is named omnect-os-update-image-raspberrypi4-64.swu
.
If you want to add additional yocto layers to your build, you can adapt layer priorities in conf/layer.conf
. This layer is the last in the BBLAYERS
yocto variable when you build with our kas
configuration files. If not, you have to possibly adapt layer prioritization values in the last layer included in BBLAYERS
.
E.g. we reset the layer prioritization of meta-phytec
to 9
, to ensure it is less than the prioritization of meta-omnect
.
The omnect-os-image
needs post processing via omnect-cli
to set a mandatory iot-identity-service
configuration. You can optionally set an iot-hub-device-update
configuration.
See omnect-cli iot-hub-device-update configuration.
See omnect-cli iot-identity-service configuration.
This is actually only one prominent case how to preset a special configuration in the image, here a systemd environment file that allows to define a set of interfaces which systemd shall consider during its test for the device being online.
It can be copied into factory
partition as described here:
omnect-cli Copy Files into or from Image
Configuration file systemd-networkd-wait-online.service
is set up by
default so that all existing physical network devices i.e., Ethernet, WLAN
or WWAN are capable of connecting to the internet. This means that the
device will be determined as online if at least one of those is active.
Note: for detailed information about how systemd generally determines the device online state have a look at systemd's documentation for service systemd-networkd-wait-online
To allow for customization the service file uses the content of
environment variable OMNECT_WAIT_ONLINE_INTERFACES
if non-empty,
otherwise the default setting --any
gets set.
This provides the possibility to overwrite the set of online interfaces
by injecting a systemd environment file into image, e.g.
/etc/omnect/systemd-networkd-wait-online.env
: just create a local
file with the following content and inject it into the factory
partitions of the image:
OMNECT_WAIT_ONLINE_INTERFACES=--interface=<interface-name>
Place holder <interface-name>
needs to be replaced with the real name,
of course.
You can also specify multiple interface arguments here which need to be
either all active for online state, or only one of them if argument
--any
is added, too.
Set OMNECT_WAIT_ONLINE_TIMEOUT_IN_SECS
to configure the timeout in seconds, default timeout is 5 minutes. A value of 0 means no timeout.
The flash modes are used to flash the complete disk image including all partitions to the target system. It uses the initramfs context, because in this mode the block device is free for writing images. Also, no filesystem is mounted in this state.
There are the following three flash modes:
- clone disk image from the disk the system is currently running to another disk of the system
- flash disk image from network to same disk the system is currently running (in local network via scp)
- flash disk image from network to same disk the system is currently running (by providing a download link to the image)
For the flash mode 1, it is required to specify the destination disk, the current disk image will be cloned to. For this purpose, the block device path has to be used.
The flash mode 1 behaves like a factory reset, related to the new boot device:
- reset to default bootloader environment
- enforce first boot condition
- reset etc partition
- reset data partition; optionally resize
The following example shows how to trigger the flash mode 1 using the block device path, on the target system:
sudo -s
bootloader_env.sh set flash-mode 1
bootloader_env.sh set flash-mode-devpath '/dev/mmcblk2'
reboot
...
Entering omnect flashing mode 1...
...
Note1: The bootloader_env.sh command requires root permissions.
Note2: The corresponding platform specific block device paths can be found in the doc folder.
Note3: Make sure that the system boots from the same device after the triggered reboot. E.g. if you boot from usb and
initiate flash mode 1 and trigger reboot, make sure that you boot from usb again. This reboot will enter the initramfs and execute the flash process.
After flash mode 1 has been finished successfully, the target system will be switched-off. The bootloader environment variables flash-mode and flash-mode-devpath will be deleted automatically.
Enable the distribution feature flash-mode-2
at build time, if you want to use it.
In order to trigger the flash mode 2,
-
use the following commands on the target system:
sudo -s bootloader_env.sh set flash-mode 2 reboot ... Entering omnect flashing mode 2... ...
Note1: bootloader_env.sh command requires root permissions.
Note2:flash-mode 2
is restricted to eth0. -
In the next step, the bmap file and the wic image file have to be transferred from the host system:
scp image.wic.bmap omnect@<target-ip>:wic.bmap scp image.wic.xz omnect@<target-ip>:wic.xz
On systems with new openssh clients >= 9.0 you have to use the legacy option when using
scp
. (See here for details.) :scp -O image.wic.bmap omnect@<target-ip>:wic.bmap scp -O image.wic.xz omnect@<target-ip>:wic.xz
The password for the omnect user used by the rootfs has to be used. The omnect user used by the initramfs is independent from the omnect user used by the rootfs. At build time, the configuration (OMNECT_USER_PASSWORD) is applied for both. The passwords are identical. Later during runtime, changing the password in the rootfs is not synchronized to the initramfs.
After finishing the flash procedure, the system reboots automatically. The bootloader environment variable flash-mode will be deleted automatically.
Enable the distribution feature flash-mode-3
at build time, if you want to use it.
In order to trigger the flash mode 3,
- compute the sha256 checksum of your image.wic.xz
sha256sum image.wic.xz > image.wic.sha256
- use the following commands on the target system:
Note1: bootloader_env.sh command requires root permissions.sudo -s bootloader_env.sh set flash-mode 3 bootloader_env.sh set flash-mode-url $(echo "http://url.to/image.wic.xz" | base64 -w 0 -) bootloader_env.sh set flash-mode-url-sha256 $(echo "http://url.to/image.wic.sha256" | base64 -w 0 -) reboot ... Entering omnect flashing mode 3... (http://url.to/image.wic.xz) ...
Note2: the urls have to be escaped with "".
Note3: on systems without realtime clock the certificate of an url gets not verified.
Note4:flash-mode 3
is restricted to eth0.
After finishing the flash procedure, the system reboots automatically. The bootloader environment variables flash-mode, flash-mode-url and flash-mode-url-sha256 will be deleted automatically.
Set the OS bootloader environment variable factory-reset
, in order to reset data
and etc
partitions
sudo bootloader_env.sh set factory-reset 1
sudo reboot
This re-creates the corresponding filesystems of partitions data
and etc
on the next boot (in the initramfs context).
If the factory
partition contains a directory etc
, then the content is copied to the etc
partition.
In the example above, factory-reset
is set to the value 1
.
This means, old data is not wiped before re-creating the respective filesystems.
This kind of factory reset does not ensure any data privacy.
In order to provide higher level of privacy, the desired wipe mode can be selected.
For this purpose, the OS bootloader environment variable factory-reset
can be set to the following values:
Factory Reset Mode | Remark | |
---|---|---|
1 | no wipe; only filesystems re-created | no privacy, but fast |
2 | use dd to write random data to etc and data partitions | better privacy, but slow |
3 | recursive remove files with rm; notify disk with fstrim | usability depends on use case and hardware |
4 | custom wipe |
Note: The provided wipe options don't guarantee total privacy. This is only possible using hardware features of the disk (e.g.; ATA secure erase).
There is also the custom wipe mode. This mode provides the possibility to address customer requirements and hardware capabilities.
In the case of custom wipe, the factory reset (initramfs context) calls /opt/factory_reset/custom-wipe
before re-creating the filesystems inside the partitions etc
and data
.
In order to establish the custom wipe mode, a Yocto recipe omnect-os-initramfs-scripts.bbappend
has to be supplied, which has to install the required utilities.
The factory reset provides the option to exclude particular files or directories.
For example, it may make sense to keep the WIFI configuration, in order to prevent loosing the network connectivity.
For this purpose, the OS bootloader environment variable factory-reset-restore-list
has to be used for.
In the following example, the regular file /etc/wpa_supplicant.conf
and the directory
/etc/aziot/identityd/
survives the factory reset:
sudo bootloader_env.sh set factory-reset-restore-list '/etc/wpa_supplicant.conf;/etc/aziot/identityd/'
The list of path names is separated by the character ;
and is enclosed by the '
quotation mark.
The factory reset is directed to the partitions etc
and data
.
Therefore, path names with the following prefixes are allowed: /etc/
, /home/
, /var/lib/
, /var/log/
, /usr/local/
and /mnt/data.
In the case of an error during the backup of a file or directory part of the restore list, the whole factory reset will be aborted
and the partitions etc
and data
remain untouched.
In the case of an error during the restore of a file or directory, the restore processing will be continued with other paths part of the restore list.
In both cases, the error will be indicated by the factory reset status (see below).
The status of the factory reset is returned by the OS bootloader environment variable factory-reset-status
.
It has the following format:
<factory reset status> ::= <main status>':'<subordinated status>
<main status> ::= <unsigned integer>
<subordinated status> ::= <unsigned integer> | '-'
The overall factory reset status
consists of two parts:
- main status (general processing state):
- 0: wipe mode supported
- 1: wipe mode unsupported
- 2: backup/restore failure
- subordinated status (execution exit status): in case of main status == 0 (success)
In the case of a successfully performed factory reset, the OS bootloader environment variable factory-reset-status
is set to the value 0:0
.
The filesystem inside the data partition is mounted using the mount options defaults,noatime,nodiratime,async,rw
, per default.
For debugging purpose, it is possible to enforce different mount options for the data partition, using the OS bootloader environment variable data-mount-options
:
# enforce sync mount
sudo bootloader_env.sh set data-mount-options defaults,noatime,nodiratime,sync,rw
sudo reboot
...
# remove debug mount options; continue with default
sudo bootloader_env.sh unset data-mount-options
sudo reboot
Note1: The bootloader environment variable data-mount-options
should be removed at the end of the debugging session.
Note2: It is not advised to use sync mount in operational mode.
The layer is licensed under either of
- Apache License, Version 2.0, (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
- MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)
at your option if not noted otherwise.
The content of dynamic-layers/phytec/recipes-kernel/lwb-radio-firmware
is licensed under MIT license (LICENSE-MIT).
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.
Copyright (c) 2021-2022 conplement AG