ovos-image-arch-recipe

Make a manjaro based OpenVoiceOS image

source code: https://github.com/OpenVoiceOS/ovos-image-arch-recipe/

Building

Docker Automated Image Building

The included Dockerfile can be used to build a default image in a Docker environment.

The following dependencies must be installed on the build system before running the container:

• chroot
• qemu-user-static

First, create the Docker container:

docker build . -t ovos-image-builder

Then, run the container to create a OVOS Image. Set CORE_REF to the branch of ovos-core that you want to build and RECIPE_REF to the branch of ovos-image-recipe you want to use. Set MAKE_THREADS to the number of threads to use for make processes.

docker run \
-v /home/${USER}/output:/output:rw \
-v /run/systemd/resolve:/run/systemd/resolve \
-e CORE_REF=${CORE_REF:-dev} \
-e RECIPE_REF=${RECIPE_REF:-master} \
-e MAKE_THREADS=${MAKE_THREADS:-4} \
--privileged \
--network=host \
--name=ovos-image-builder \
ovos-image-builder

The entire build process will generally take several hours; it takes 1-2 hours on a build server with 2x Xeon Gold 5118 CPUs (48T Total).

Interactive Image Building

The scripts in the automation directory are available to help automate building a default image. For building an image interactively:

bash automation/prepare.sh
bash /tmp/run_scripts.sh

The below documentation describes how to manually build an image using the individual scripts in this repository.

Steps

Getting Started

The scripts and overlay files in this repository are designed to be applied to a base image as the root user. It is recommended to apply these scripts to a clean base image. Instructions are available at opensource.com.

Note: The GUI shell is not installable under some base images

For each step except boot_overlay, the directory corresponding to the step should be copied to the mounted image and the script run from a terminal chroot-ed to the image. If running scripts from a booted image, they should be run as root.

Preparation

From the host system where this repository is cloned, running prepare.sh <base_image> will copy boot overlay files, mount the image, mount DNS resolver config from the host system, copy all other image overlay files to /tmp, and chroot into the image. From here, you can run any/all of the following scripts to prepare the image before cleaning up

core_configuration

Configures user accounts and base functionality for RPi. ovos user is created with proper permissions here.

At this stage, a booted image should resize its file system to fill the drive it is flashed to. Local login and ssh connections should use ovos/ovos to authenticate and be prompted to change password on login.

network_manager

Adds Balena wifi-connect to enable a portal for connecting the Pi device to a Wi-Fi network.

A booted image will now be ready to connect to a network via SSID OVOS.

sj201

For SJ201 board support, the included script will build/install drivers, add required overlays, install required system packages, and add a systemd service to flash the SJ201 chip on boot. This will modify pulseaudio and potentially overwrite any previous settings.

Note: Running this scripts grants GPIO permissions to the gpio group. Any user that interfaces with the SJ201 board should be a member of the gpio group. Group permissions are not modified by this script

Audio devices should now show up with pactl list. Audio devices can be tested in the image by recording a short audio clip and playing it back.

parecord test.wav
paplay test.wav

embedded_shell

Installs ovos-shell and mycroft-gui-app. Adds and enables ovos-gui.service to start the shell on system boot.

The image should now boot to the GUI shell.

ovos_core

Installs ovos-core and dependencies. Configures services for core modules.

At this stage, the image is complete and when booted should start OVOS.

dashboard

Installs the OVOS Dashboard and service to start the dashboard from the GUI.

From the GUI Settings -> Developer Settings menu, Enable Dashboard will now start the dashboard for remote access to device diagnostics.

camera

Installs libcamera and other dependencies for using a CSI camera.

The default camera skill can be used to take a photo; libcamera-apps are also installed for testing via CLI.

splash_screen

Enables a custom splash screen and disables on-device TTY at boot.

On boot, a static image should be shown until the GUI Shell starts.

Clean Up

cleanup.sh removes any temporary files from the mounted image before unmounting it. After running cleanup.sh, the image is ready to burn to a drive and boot.