Skip to content

Latest commit

 

History

History
345 lines (256 loc) · 12.2 KB

README.md

File metadata and controls

345 lines (256 loc) · 12.2 KB
Raw

yocto-geniux

REUSE status

The purpose of this repository is to automate the generation in a virtualized development environment of Geniux images and their corresponding Software Development Kit (SDK) for the cross-compilation and running of GNSS-SDR on embedded systems, as well as bootable image files that can be copied to an SD card and make your device ready to go.

The Docker images generated by the Dockerfile file of this repository can run the building processes defined by the meta-gnss-sdr Yocto layer.

More info at:

The name Geniux comes from GNSS-SDR for Embedded GNU/Linux.

Geniux logo

Building Geniux releases in a virtualized environment

Get a powerful machine (as much RAM, storage capacity, and CPU cores as you can) and install the Docker engine on it. Make sure that it is up and running.

Note: the geniux-builder.sh script makes use of the zip and unzip tools. On Debian/Ubuntu machines, you can install them by doing:

$ sudo apt-get install zip unzip

Then, get the source code of this repository and go to its base path:

$ git clone https://github.com/carlesfernandez/yocto-geniux
$ cd yocto-geniux

Now you are ready to build Geniux images for the release you want with a single command, by using the geniux-builder.sh script. Taking a look at its help message:

$ ./geniux-builder.sh --help
This script builds and stores Geniux images.

Usage:
./geniux-builder.sh [version] [manifest] [machine] (--image-only / -i)

Options:
 version   Geniux version (from oldest to most recent):
             rocko, sumo, thud, warrior, zeus, dunfell,
             gatesgarth, hardknott, honister, kirkstone, langdale, mickledore. Default: dunfell
           Check available branches at https://github.com/carlesfernandez/meta-gnss-sdr
 manifest  Geniux version manifest: 21.02, 21.08, 22.02, 22.06, 23.04, latest. Default: latest
           Dated manifests available at https://github.com/carlesfernandez/oe-gnss-sdr-manifest/tags
 machine   Specify your (list of) MACHINE here. By default, zedboard-zynq7 and raspberrypi3 are built.
           If more than one, surround them with quotes, e.g.: "raspberrypi4-64 intel-corei7-64"

 --image-only / -i  (optional) Build the Docker image but do not execute the container.

Environment variables that affect behavior:
 GENIUX_MIRROR_PATH          Base path to local mirror. Only used if set.
                             e.g.: 'export GENIUX_MIRROR_PATH=/home/carlesfernandez/mirror'
                             The mirror is expected to be at '$GENIUX_MIRROR_PATH/sources/$version'
 GENIUX_STORE_PATH           Path in which products will be stored. Only used if set.
                             e.g.: 'export GENIUX_STORE_PATH=/home/carlesfernandez/geniux-releases'
 GENIUX_STORE_REQUIRES_SUDO  If set, the script will ask for super-user privileges to write in the store.
                             You will be asked only once at the beginning. The password will not be revealed.
                             e.g.: 'export GENIUX_STORE_REQUIRES_SUDO=1'

Before calling the script, you might want to set some (optional) environment variables on your host machine:

$ export GENIUX_MIRROR_PATH=/home/user/mirror
$ export GENIUX_STORE_PATH=/home/user/geniux-releases
$ export GENIUX_STORE_REQUIRES_SUDO=1

Examples of usage:

NOTE: if you are operating on a remote host through ssh, you might want to run screen at this point, so the work won't be lost in case of a session drop.

  • Build Geniux release dunfell, with manifest date latest, for machines zedboard-zynq7 and raspberrypi3:

    $ ./geniux-builder.sh

  • Build Geniux release thud, with manifest date latest, for machines zedboard-zynq7 and raspberrypi3:

    $ ./geniux-builder.sh thud

  • Build Geniux release thud, with manifest date 22.02, for machines zedboard-zynq7 and raspberrypi3:

    $ ./geniux-builder.sh thud 22.02

  • Build Geniux release warrior, with manifest date latest, only for machine zedboard-zynq7:

    $ ./geniux-builder.sh warrior latest zedboard-zynq7

  • Build Geniux release warrior, with manifest date 22.02, only for machine raspberrypi3:

    $ ./geniux-builder.sh warrior 22.02 raspberrypi3

  • Build Geniux release rocko, with manifest date latest, for machines zedboard-zynq7 and zcu102-zynqmp:

    $ ./geniux-builder.sh rocko latest "zedboard-zynq7 zcu102-zynqmp"

  • Build Geniux release hardknott, with manifest date latest, for machines raspberrypi4 and intel-corei7-64:

    $ ./geniux-builder.sh hardknott latest "raspberrypi4 intel-corei7-64"

  • Build Geniux release honister, with manifest date latest, only for machine zcu208-zynqmp:

    $ ./geniux-builder.sh honister latest zcu208-zynqmp

  • Build Geniux release kirkstone, with manifest date latest, only for machine raspberrypi4-64:

    $ ./geniux-builder.sh kirkstone latest raspberrypi4-64

  • Build Geniux release langdale, with manifest date latest, only for machine raspberrypi4-64:

    $ ./geniux-builder.sh langdale latest raspberrypi4-64

At the end of the building process, the output products will be placed in a folder named ./$version/output_$machine under the yocto-geniux base path, and copied to $GENIUX_STORE_PATH if that environment variable is defined.

If you want to have more detailed control of the whole process, or you are interested on further development (making changes to the Yocto layers, adding new features or recipes, fixing bugs, etc.), then you can skip running the container generated by the geniux-builder.sh script and follow the instructions below.

Generating Geniux images and the SDK installer step-by-step

Non-interactive method

NOTE: if you are operating on a remote host through ssh, you might want to run screen at this point, so the work won't be lost in case of a session drop.

Build the Docker image but do not run the container:

$ ./geniux-builder.sh [version] [manifest] [machine] -i

Create an output folder and run the container:

$ mkdir -p output
$ docker run -it --rm \
  -v $PWD/output:/home/geniux/yocto/output \
  --privileged=true \
  geniux-$version:$manifest.$machine

If you have a local mirror available, you can provide access from within the container as:

$ mkdir -p output
$ docker run -it --rm \
  -v $PWD/output:/home/geniux/yocto/output \
  -v $my_mirror:/source_mirror/sources/$version \
  --privileged=true \
  geniux-$version:$manifest.$machine

replacing $my_mirror by the actual path of your mirror and $version by the actual version name you used when building the container. If you do not have any local mirror, just omit the -v $my_mirror:... line.

The build process will take several hours. At its ending, the image files will be under your ./output folder, so outside the container. The ./output folder must be empty before starting the run. The container itself will be erased after completion.

Interactive method

NOTE: if you are operating on a remote host through ssh, you might want to run screen at this point, so the work won't be lost in case of a session drop.

Build the Docker image but do not run the container:

$ ./geniux-builder.sh [version] [manifest] [machine] -i

Now run the container in the interactive mode:

$ mkdir -p output
$ docker run -it --rm \
   -v $PWD/output:/home/geniux/yocto/output \
   -v $my_mirror:/source_mirror/sources/$version \
   --privileged=true \
   geniux-$version:$manifest.$machine bash

replacing $my_mirror by the actual path of your mirror and $version by the actual version name you used when building the container. If you do not have any local mirror, just omit the -v $my_mirror:... line.

Notice the final bash, that will take you to the bash console without executing the predefined commands.

Now, inside the container, prepare the building environment:

$ repo sync   # Recommended if you are building the 'latest' manifest
$ source ./oe-core/oe-init-build-env ./build ./bitbake

At this point, you can modify the conf/local.conf file, add new recipes, and experiment as you want. The nano editor is available for that. When you are ready to build the development image:

$ bitbake gnss-sdr-dev-image

and the corresponding SDK script installer:

$ bitbake -c populate_sdk gnss-sdr-dev-image

For building an image with the Xfce Desktop Environment and gnss-sdr already included:

$ bitbake gnss-sdr-demo-image

If you want to build the Docker images, you need to run the container with the flag --privileged=true and to start the Docker daemon inside the container:

$ sudo service docker start
$ bitbake gnss-sdr-dev-docker

The build process will take several hours. At its ending, the image files will be under ./build/tmp-glibc/deploy folder. Move them to the /home/geniux/yocto/output folder:

$ mv ./tmp-glibc/deploy/images /home/geniux/yocto/output/
$ mv ./tmp-glibc/deploy/sdk /home/geniux/yocto/output/

Now, when doing exit from the container, the build artifacts will be at the ./output folder you created in your machine, so outside the container. The container itself will be erased at exit.

Note for Xilinx's boards

Releases of Xilinx's Embedded Design Tools are tightly coupled to a specific Yocto Project release. This table shows that correspondence between versions:

Xilinx release Yocto codename Linux kernel Geniux version Geniux manifest Tested boards (MACHINE)
v2022.2 Honister 5.15 Honister latest, 23.04, 22.06. zedboard-zynq7, zc706-zynq7, zcu102-zynqmp, zcu208-zynqmp
v2021.2 Gatesgarth 5.10 Gatesgarth latest, 23.04, 22.06. zedboard-zynq7, zc706-zynq7, zcu102-zynqmp
v2020.3 Zeus 5.4 Zeus latest, 23.04, 22.06, 22.02, ... zedboard-zynq7, zcu102-zynqmp
v2019.2 Thud 4.19 Thud latest, 23.04, 22.06, 22.02, ... zedboard-zynq7, zcu102-zynqmp
v2018.3 Rocko 4.14 Rocko latest, 23.04, 22.06, 22.02, ... zedboard-zynq7, zcu102-zynqmp

Other boards might work as well, or require minor (but maybe non-obvious) modifications in the configuration files.

Disclaimer

Yocto Project and all related marks and logos are trademarks of The Linux Foundation. This tool is not, in any way, endorsed by the Yocto Project or The Linux Foundation.

Copyright and License

Copyright: © 2020-2022 Carles Fernández-Prades, CTTC. All rights reserved.

The content of this repository is released under the MIT license.

Acknowledgements

This work was partially supported by Grant PID2021-128373OB-I00 funded by MCIN/AEI/10.13039/501100011033. Formerly, it was also supported by the Spanish Ministry of Science, Innovation, and Universities through the Statistical Learning and Inference for Large Dimensional Communication Systems (ARISTIDES, RTI2018-099722-B-I00) project.