Yocto Project and OpenEmbedded Training

STM32MP1 variant

Practical Labs

https://bootlin.com

September 02, 2022

 Yocto Project and OpenEmbedded Training

About this document

Updates to this document can be found on https://bootlin.com/doc/training/yocto-stm32.
This document was generated from LaTeX sources found on https://github.com/bootlin/
training-materials.
More details about our training sessions can be found on https://bootlin.com/training.

Copying this document

© 2004-2022, Bootlin, https://bootlin.com.
This document is released under the terms of the Creative Commons CC BY-SA
3.0 license . This means that you are free to download, distribute and even modify
it, under certain conditions.
Corrections, suggestions, contributions and translations are welcome!

2 © 2004-2022 Bootlin, CC BY-SA license

 Yocto Project and OpenEmbedded Training

Training setup
Download files and directories used in practical labs

Install lab data

For the different labs in this course, your instructor has prepared a set of data (kernel images,
kernel configurations, root filesystems and more). Download and extract its tarball from a
terminal:
$ cd
$ wget https://bootlin.com/doc/training/yocto-stm32/yocto-stm32-labs.tar.xz
$ tar xvf yocto-stm32-labs.tar.xz

Lab data are now available in an yocto-stm32-labs directory in your home directory. This
directory contains directories and files used in the various practical labs. It will also be used as
working space, in particular to keep generated files separate when needed.

Update your distribution

To avoid any issue installing packages during the practical labs, you should apply the latest
updates to the packages in your distro:

$ sudo apt update

$ sudo apt dist-upgrade

You are now ready to start the real practical labs!

Install extra packages

Feel free to install other packages you may need for your development environment. In particular,
we recommend to install your favorite text editor and configure it to your taste. The favorite text
editors of embedded Linux developers are of course Vim and Emacs, but there are also plenty
of other possibilities, such as Visual Studio Code1 , GEdit, Qt Creator, CodeBlocks, Geany, etc.
It is worth mentioning that by default, Ubuntu comes with a very limited version of the vi
editor. So if you would like to use vi, we recommend to use the more featureful version by
installing the vim package.

More guidelines
Can be useful throughout any of the labs
• Read instructions and tips carefully. Lots of people make mistakes or waste time because
they missed an explanation or a guideline.
1 This tool from Microsoft is Open Source! To try it on Ubuntu: sudo snap install code --classic

© 2004-2022 Bootlin, CC BY-SA license 3

 Yocto Project and OpenEmbedded Training

• Always read error messages carefully, in particular the first one which is issued. Some
people stumble on very simple errors just because they specified a wrong file path and
didn’t pay enough attention to the corresponding error message.
• Never stay stuck with a strange problem more than 5 minutes. Show your problem to
your colleagues or to the instructor.
• You should only use the root user for operations that require super-user privileges, such
as: mounting a file system, loading a kernel module, changing file ownership, configuring
the network. Most regular tasks (such as downloading, extracting sources, compiling...)
can be done as a regular user.
• If you ran commands from a root shell by mistake, your regular user may no longer be
able to handle the corresponding generated files. In this case, use the chown -R command
to give the new files back to your regular user.
Example: $ sudo chown -R myuser.myuser linux/

4 © 2004-2022 Bootlin, CC BY-SA license

 Yocto Project and OpenEmbedded Training

Lab1: First Yocto Project build

Your first dive into Yocto Project and its build mechanism

During this lab, you will:

• Set up an OpenEmbedded environment
• Configure the project and choose a target
• Build your first Poky image

Setup
Before starting this lab, make sure your home directory is not encrypted using eCryptFS. Open-
Embedded cannot be used on top of an eCryptFS file system due to limitations in file name
lengths.
Go to the $HOME/yocto-stm32-labs/ directory.
Install the required packages:
sudo apt install bc build-essential chrpath cpio diffstat gawk git python texinfo wget gdisk

Download Yocto
Download the kirkstone version of Poky:
git clone https://git.yoctoproject.org/git/poky
cd $HOME/yocto-stm32-labs/poky
git checkout -b kirkstone-4.0.2 kirkstone-4.0.2
Return to your project root directory (cd $HOME/yocto-stm32-labs/) and download the Open-
Embedded and STM32MP layers:
git clone -b kirkstone https://git.openembedded.org/meta-openembedded
git clone https://github.com/STMicroelectronics/meta-st-stm32mp
cd meta-st-stm32mp
git checkout openstlinux-5.15-yocto-kirkstone-mp1-v22.06.15

Set up the build environment

Check you’re using Bash. This is the default shell when using Ubuntu.
Export all needed variables and set up the build directory:
cd $HOME/yocto-stm32-labs
source poky/oe-init-build-env
You must specify which machine is your target. By default it is qemu. We need to build an
image for an stm32mp1. Update the MACHINE configuration variable accordingly.

© 2004-2022 Bootlin, CC BY-SA license 5

 Yocto Project and OpenEmbedded Training

Also, if you need to save disk space on your computer you can add INHERIT += "rm_work" in
the previous configuration file. This will remove the package work directory once a package is
built.
Don’t forget to make the configuration aware of the OpenEmbedded and STM32MP layers.
Edit the layer configuration file ($BUILDDIR/conf/bblayers.conf) and append the full path to
the meta-openembedded/meta-oe, meta-openembedded/meta-python and meta-st-stm32mp di-
rectory to the BBLAYERS variable.

Build your first image

Now that you’re ready to start the compilation, simply run:
bitbake core-image-minimal
Once the build finished, you will find the output images under $BUILDDIR/tmp/deploy/images/
stm32mp1.

Set up the SD card

In this first lab we will use an SD card to store the bootloader, kernel and root filesystem files.
To generate the final image, You will find scripts in $BUILDDIR/tmp/deploy/images/stm32mp1/
scripts.
Execute it (replace 157d with 157a depending on your board variant):
./create_sdcard_from_flashlayout.sh \
../flashlayout_core-image-minimal/trusted/FlashLayout_sdcard_stm32mp157d-dk1-trusted.tsv
Flash the SD card with that image:
umount /dev/mmcblk0p*
sudo dd if=../FlashLayout_sdcard_stm32mp157d-dk1-trusted.raw \
of=/dev/mmcblk0 bs=8M conv=fdatasync status=progress

Setting up serial communication with the board

Plug the USB-A to micro USB-B cable on the Discovery board. There is only one micro USB
port on the board, it is CN11, also named ST-LINK. This is a debug interface and exposes
multiple debugging interfaces, including a serial interface. When plugged in your computer, a
serial port should appear, /dev/ttyACM0.
You can also see this device appear by looking at the output of dmesg.
To communicate with the board through the serial port, install a serial communication program,
such as picocom:
sudo apt install picocom
You also need to make your user belong to the dialout group to be allowed to write to the serial
console:
sudo adduser $USER dialout
Important: for the group change to be effective, you have to completely log out from your
session and log in again (no need to reboot). A workaround is to run newgrp dialout, but it is
not global. You have to run it in each terminal.

6 © 2004-2022 Bootlin, CC BY-SA license

 Yocto Project and OpenEmbedded Training

Run picocom -b 115200 /dev/ttyACM0, to start serial communication on /dev/ttyACM0, with a

baudrate of 115200.
If you wish to exit picocom, press [Ctrl][a] followed by [Ctrl][x].
There should be nothing on the serial line so far, as the board is not powered up yet.

Boot
Insert the SD card in the dedicated slot on the Discovery. You can now power-up the board by
connecting the USB-C cable to the board, in CN6, PWR_IN and to your PC at the other end.
You should see boot messages on the console. Wait until the login prompt, then enter root as
user. Congratulations! The board has booted and you now have a shell.

Known issues on stmp32mp157d-dk1

If your board doesn’t boot any more, even after making no changes to the micro SD card, even
after unplugging the USB-C power cable, you may need to unplug both the USB micro-B
(used for serial) and USB-C power cables to get the board to boot again. It’s probably because
the USB micro-B is also a power source.

© 2004-2022 Bootlin, CC BY-SA license 7

 Yocto Project and OpenEmbedded Training

Lab2: Advanced Yocto configura-

tion
Configure the build, customize the output images and use NFS

During this lab, you will:

• Customize the package selection
• Configure the build system
• Use the rootfs over NFS

Set up the Ethernet communication and NFS on the board

It isn’t practical at all to reflash the root filesystem on the target every time a change is made.
Fortunately, it is possible to set up networking between the development workstation and the
target. Then, workstation files can be accessed by the target through the network, using NFS.
First we need to set the kernel boot arguments U-Boot will pass to the Linux kernel at boot
time. For that, edit the extlinux configuration file, in the bootfs partition of the SD card and
change the APPEND line to:
APPEND root=/dev/nfs rw console=ttySTM0,115200 nfsroot=192.168.0.1:/nfs,vers=3,tcp ip=192.168.0.100:::::eth0

• For the stm32mp157a-dk1-sdcard and the stm32mp157d-dk1-sdcard, edit the mmc0_extlinux/

extlinux.conf file
• For the stm32mp157c-dk2-sdcard, edit the mmc0_extlinux/stm32mp157c-dk2_extlinux.
conf file

Set up the Ethernet communication on the workstation

With a network cable, connect the Ethernet port of your board to the one of your computer. If
your computer already has a wired connection to the network, your instructor will provide you
with a USB Ethernet adapter. A new network interface should appear on your Linux system.
Find the name of this interface by typing:
ip a
The network interface name is likely to be enxxx2 . If you have a pluggable Ethernet device, it’s
easy to identify as it’s the one that shows up after pluging in the device.
Then, instead of configuring the host IP address from NetWork Manager’s graphical interface,
let’s do it through its command line interface, which is so much easier to use:
nmcli con add type ethernet ifname en... ip4 192.168.0.1/24
2 Following the Predictable Network Interface Names convention: https://www.freedesktop.org/wiki/
Software/systemd/PredictableNetworkInterfaceNames/

8 © 2004-2022 Bootlin, CC BY-SA license

 Yocto Project and OpenEmbedded Training

Set up the NFS server on the workstation

First install the NFS server on the training computer and create the root NFS directory:
sudo apt install nfs-kernel-server
sudo mkdir -m 777 /nfs
Then make sure this directory is used and exported by the NFS server by adding /nfs *(rw,
sync,no_root_squash,subtree_check) to the /etc/exports file.
Finally, make the NFS server use the new configuration:
sudo exportfs -r

Add a package to the rootfs image

You can add packages to be built by editing the local configuration file $BUILDDIR/conf/local.
conf. The IMAGE_INSTALL variable controls the packages included into the output image.
To illustrate this, add the Dropbear SSH server to the list of enabled packages.
Tip: do not override the default enabled package list, but append the Dropbear package instead.

Boot with the updated rootfs

First we need to put the rootfs under the NFS root directory so that it is accessible by NFS
clients. Simply uncompress the archived output image in the previously created /nfs directory:
sudo tar xpf $BUILDDIR/tmp/deploy/images/stm32mp1/\
core-image-minimal-stm32mp1.tar.xz -C /nfs
Then boot the board.
The Dropbear SSH server was enabled a few steps before, and should now be running as a
service on the Discovery. You can test it by accessing the board through SSH:
ssh [email protected]
You should see the Discovery command line!

Choose a package variant

Dependencies of a given package are explicitly defined in its recipe. Some packages may need a
specific library or piece of software but others only depend on a functionality. As an example,
the kernel dependency is described by virtual/kernel.
To see which kernel is used, dry-run BitBake:
bitbake -vn virtual/kernel
In our case, we can see the linux-stm32mp provides the virtual/kernel functionality:
NOTE: selecting linux-stm32mp to satisfy virtual/kernel due to PREFERRED_PROVIDERS

We can force bitbake to select another kernel by explicitly defining which one to use in our
local configuration. Try switching from linux-stm32mp to linux-dummy only using the local
configuration.
Then check the previous step worked by dry-running again BitBake.

© 2004-2022 Bootlin, CC BY-SA license 9

 Yocto Project and OpenEmbedded Training

bitbake -vn virtual/kernel

You can now rebuild the whole Yocto project, with bitbake core-image-minimal
Tip: you need to define the more specific information here to be sure it is the one used. The
MACHINE variable can help here.
As this was only to show how to select a preferred provider for a given package, you can now
use linux-stm32mp again.

BitBake tips
BitBake is a powerful tool which can be used to execute specific commands. Here is a list of
some useful ones, used with the virtual/kernel package.
• The Yocto recipes are divided into numerous tasks, you can print them by using: bitbake
-c listtasks virtual/kernel.
• BitBake allows to call a specific task only (and its dependencies) with: bitbake -c <task>
virtual/kernel. (<task> can be menuconfig here).
• You can force to rebuild a package by calling: bitbake -f virtual/kernel
• world is a special keyword for all packages. bitbake --runall=fetch world will download
all packages sources (and their dependencies).
• You can get a list of locally available packages and their current version with:
bitbake -s
• You can also find detailed information on available packages, their current version, depen-
dencies or the contact information of the maintainer by visiting:
http://recipes.yoctoproject.org/
For detailed information, please run bitbake -h

Going further
If you have some time left, let’s improve our setup to use TFTP, in order to avoid having to
reflash the SD card for every test. What you need to do is:
1. Install a TFTP server (package tftpd-hpa) on your system.
2. Copy the Linux kernel image and Device Tree to the TFTP server home directory (specified
in /etc/default/tftpd-hpa) so that they are made available by the TFTP server.
3. Change the U-Boot bootcmd to load the kernel image and the Device Tree over TFTP.
See the training materials of our Embedded Linux system development course for details!

10 © 2004-2022 Bootlin, CC BY-SA license

 Yocto Project and OpenEmbedded Training

Lab3: Add a custom application

Add a new recipe to support a required custom application

During this lab, you will:

• Write a recipe for a custom application
• Integrate this application into the build
This is the first step of adding an application to Yocto. The remaining part is covered in the
next lab, ”Create a Yocto layer”.

Setup and organization

In this lab we will add a recipe handling the nInvaders application. Before starting the recipe
itself, find the recipes-extended directory in the OpenEmbedded-Core layer and add a subdi-
rectory for your application.
A recipe for an application is usually divided into a version specific bb file and a common one.
Try to follow this logic and separate the configuration variables accordingly.
Tip: it is possible to include a file into a recipe with the keyword require.

First hands on nInvaders

The nInvaders application is a terminal based game following the space invaders family. In order
to deal with the text based user interface, nInvaders uses the ncurses library.
First try to find the project homepage, download the sources and have a first look: license,
Makefile, requirements…

Write the common recipe

Create an appropriate common file, ending in .inc
In this file add the common configuration variables: source URI, description…

Write the version specific recipe

Create a file that respects the Yocto nomenclature: ${PN}_${PV}.bb
Add the required common configuration variables: archive checksum, license file checksum,
package revision…

Testing and troubleshooting

You can check the whole packaging process is working fine by explicitly running the build task
on the nInvaders recipe:

© 2004-2022 Bootlin, CC BY-SA license 11

 Yocto Project and OpenEmbedded Training

bitbake ninvaders
Try to make the recipe on your own. Also eliminate the warnings related to your recipe: some
configuration variables are not mandatory but it is a very good practice to define them all.
If you hang on a problem, check the following points:
• The common recipe is included in the version specific one
• The checksum and the URI are valid
• The dependencies are explicitly defined
• The internal state has changed, clean the working directory:
bitbake -c cleanall ninvaders
One of the build failures you will face will generate many messages such as multiple definition
of `skill_level'; aliens.o:(.bss+0x674): first defined here.
The multiple definition issue is due to the code base of nInvaders being quite old, and
having multiple compilation units redefine the same symbols. While this was accepted by older
gcc versions, since gcc 10 this is no longer accepted by default.
While we could fix the nInvaders code base, we will take a different route: ask gcc to behave as
it did before gcc 10 and accept such redefinitions. This can be done by passing the -fcommon
gcc flag.
To achieve this, make sure to add -fcommon to the CFLAGS variable.
Tip: BitBake has command line flags to increase its verbosity and activate debug outputs. Also,
remember that you need to cross-compile nInvaders for ARM! Maybe, you will have to configure
your recipe to resolve some mistakes done in the application’s Makefile (which is often the case).
A bitbake variable permits to add some Makefile’s options, you should look for it.

Update the rootfs and test

Now that you’ve compiled the nInvaders application, generate a new rootfs image with bitbake
core-image-minimal. Then update the NFS root directory. You can confirm the nInvaders
program is present by running:
find /nfs -iname ninvaders
Access the board command line through SSH. You should be able to launch the nInvaders
program. Now, it’s time to play!

Inspect the build

The nInvaders application was unpacked and compiled in the recipe’s work directory. Can you
spot nInvaders’ directory in the build work directory?
Once you found it, look around. You should at least spot some directories:
• The sources. Remember the ${S} variable?
• temp. There are two kinds of files in there. Can you tell what are their purposes?
• Try to see if the licences of nInvaders were extracted.

12 © 2004-2022 Bootlin, CC BY-SA license

 Yocto Project and OpenEmbedded Training

Lab4: Create a Yocto layer

Add a custom layer to the Yocto project for your project needs

During this lab, you will:

• Create a new Yocto layer
• Interface this custom layer to the existing Yocto project
• Use applications from custom layers
This lab extends the previous one, in order to fully understand how to interface a custom project
to the basic Yocto project.

Tools
You can access the configuration and state of layers with the bitbake-layers command. This
command can also be used to retrieve useful information about available recipes. Try the
following commands:
bitbake-layers show-layers
bitbake-layers show-recipes linux-yocto
bitbake-layers show-overlayed
bitbake-layers create-layer

Create a new layer

With the above commands, create a new Yocto layer named meta-bootlinlabs with a priority
of 7. Before doing that, return to your project root directory, where by convention all layers are
stored (cd $HOME/yocto-stm32-labs/).
Before using the new layer, we need to configure its generated configuration files. You can start
with the README file which is not used in the build process but contains information related to
layer maintenance. You can then check, and adapt if needed, the global layer configuration file
located in the conf directory of your custom layer.

Integrate a layer to the build

To be fair, we already used and integrated a layer in our build configuration during the first lab,
with meta-st-stm32. This layer was responsible for stm32mp1 DK1 support in Yocto. We have
to do the same for our meta-bootlinlabs now.
There is a file which contains all the paths of the layers we use. Try to find it without looking
back to the first lab. Then add the full path to our newly created layer to the list of layers.
Validate the integration of the meta-bootlinlabs layer with:
bitbake-layers show-layers
and make sure you don’t have any warning from bitbake.

© 2004-2022 Bootlin, CC BY-SA license 13

 Yocto Project and OpenEmbedded Training

Add a recipe to the layer

In the previous lab we introduced a recipe for the nInvaders game. We included it to the existing
meta layer. While this approach give a working result, the Yocto logic is not respected. You
should instead always use a custom layer to add recipes or to customize the existing ones. To
illustrate this we will move our previously created nInvaders recipe into the meta-bootlinlabs
layer.
You can check the nInvaders recipe is part of the meta layer first:
bitbake-layers show-recipes ninvaders
Then move the nInvaders recipe to the meta-bootlinlabs layer. You can check that the nIn-
vaders recipe is now part of the layer with the bitbake-layers command.

14 © 2004-2022 Bootlin, CC BY-SA license

 Yocto Project and OpenEmbedded Training

Lab5: Extend a recipe

Add your features to an existing recipe

During this lab, you will:

• Apply patches to an existing recipe
• Use a custom configuration file for an existing recipe
• Extend a recipe to fit your needs

Create a basic appended recipe

To avoid rewriting recipes when a modification is needed on an already existing one, BitBake
allows to extend recipes and to overwrite, append or prepend configuration variables values
through the so-called BitBake append files.
We will first create a basic BitBake append file, without any change made to the original recipe,
to see how it is integrated into the build. We will then extend some configuration variables of
the original recipe.
Try to create an appended recipe with the help of the online Yocto Project development doc-
umentation. You can find it at https://docs.yoctoproject.org/dev-manual/index.html. We
here aim to extend the linux-stm32mp kernel recipe.
You can see available bbappend files and the recipe they apply to by using the bitbake-layers
tool (again!):
bitbake-layers show-appends
If the BitBake append file you just created is recognized by your Yocto environment, you should
see:
linux-stm32mp_5.15.bb:
$HOME/yocto-stm32-labs/meta-bootlinlabs/recipes-kernel/linux/linux-stm32mp_5.15.bbappend

Add patches to apply in the recipe

We want our extended linux-stm32mp kernel to support the Nunchuk as a joystick input. We can
add this by applying patches during the do_patch task. The needed patches are provided with
this lab. You can find them under ~/yocto-stm32-labs/bootlin-lab-data/nunchuk/linux. For
more details about how to write the driver handling the Nunchuk, have a look at our embed-
ded Linux kernel and driver development training course at https://bootlin.com/training/
kernel/.
Applying a patch is a common task in the daily Yocto process. Many recipes, appended or not,
apply a specific patch on top of a mainline project. It’s why patches do not have to be explicitly
applied, if the recipe inherits from the patch class (directly or not), but only have to be present
in the source files list.

© 2004-2022 Bootlin, CC BY-SA license 15

 Yocto Project and OpenEmbedded Training

Try adding the patches included in this lab to your BitBake append file. Do not forget to also add
the defconfig file provided alongside the patches. This file contains the kernel configuration.
Add KERNEL_DEFCONFIG = "" and KERNEL_EXTERNAL_DEFCONFIG = "defconfig" so it is used by
the linux-stm32mp recipe.

You can now rebuild the kernel to take the new patches into account:

bitbake virtual/kernel

Connect the Nunchuk

Take the Nunchuk device provided by your instructor.

We will connect it to the second I2C port of the CPU (i2c1), with pins available on the P9
connector.

Identify the 4 pins of the Nunchuk connector:

SCL
PWR

GND SDA

Nunchuk i2c pinout

(UEXT connector from Olimex, front view)

Connect the Nunchuk pins:

• The GND pin to CN16 pin 6 (GND)

• The PWR pin to CN16 pin 4 (3V3)

• The DATA pin to CN13 pin 9 (D14)

• The CLK pin to CN13 pin 10 (D15)

16 © 2004-2022 Bootlin, CC BY-SA license

 Yocto Project and OpenEmbedded Training

Test the Nunchuk

Copy the newly generated kernel and device tree images into the first SD card partition. Then
boot the board and wait until you have access to the busybox command line.
You can then make sure that the Nunchuk is recognized and is working by checking the presence
of the js0 device file:
ls /dev/input/js0
Now display the raw events generated by the Nunchuk:
cat /dev/input/js0
You should see random characters appearing while playing with the Nunchuk. Be aware that
the driver we integrated also handles accelerometer events. Therefore, moving the device will
produce many events!

Patch nInvaders
The nInvaders game uses keyboard events for its controls. We first need to apply a patch
introducing joystick support. The patch is located at ~/yocto-stm32-labs/bootlin-lab-data/
nunchuk/ninvaders/.
Add the patch to the nInvaders SRC_URI.
Then build a full core-image-minimal and update the NFS root directory.

Play nInvaders!
After booting the board you should be able to play nInvaders with the keyboard…and the
Nunchuk! The C button is used to confirm and to fire, and Z to pause the game.
Access the board command line through SSH, and launch the game:

© 2004-2022 Bootlin, CC BY-SA license 17

 Yocto Project and OpenEmbedded Training

$ ninvaders

18 © 2004-2022 Bootlin, CC BY-SA license

 Yocto Project and OpenEmbedded Training

Lab6: Create a custom machine

configuration
Let Poky know about your hardware!

During this lab, you will:

• Create a custom machine configuration
• Understand how the target architecture is dynamically chosen

Create a custom machine

The machine file configures various hardware related settings. That’s what we did in lab1, when
we chose the stm32mp1 one. While it is not necessary to make our custom machine image here,
we’ll create a new one to demonstrate the process.
Add a new bootlinlabs machine to the previously created layer, which will make the Discovery
properly boot.
This machine describes a board using the cortexa7thf-neon-vfpv4 tune and is a part of the
stm32mp SoC family. Add the following lines to your machine configuration file:
require conf/machine/include/st-machine-common-stm32mp.inc
require conf/machine/include/st-machine-providers-stm32mp.inc

DEFAULTTUNE = "cortexa7thf-neon-vfpv4"
require conf/machine/include/arm/armv7a/tune-cortexa7.inc

Populate the machine configuration

This bootlinlabs machine needs:
• To define a few variables to set to get the tooling from ST Micro to work properly:
UBOOT_CONFIG = "trusted_stm32mp15"
STM32MP_DT_FILES_DK = "stm32mp157a-dk1 stm32mp157d-dk1"
• To support the m4copro feature

Build an image with the new machine

You can now update the MACHINE variable value in the local configuration and start a fresh build.

Check generated files are here and correct

Once the generated images supporting the new bootlinlabs machine are generated, you can
check all the needed images were generated correctly.

© 2004-2022 Bootlin, CC BY-SA license 19

 Yocto Project and OpenEmbedded Training

Have a look in the output directory, in $BUILDDIR/tmp/deploy/images/bootlinlabs/. Is there

something missing?

Update the rootfs

You can now update your root filesystem, to use the newly generated image supporting our
bootlinlabs machine!

20 © 2004-2022 Bootlin, CC BY-SA license

 Yocto Project and OpenEmbedded Training

Lab7: Create a custom image

The highest level of customization in Poky

During this lab, you will:

• Write a full customized image recipe
• Choose the exact packages you want on your board

Add a basic image recipe

A build is configured by two top level recipes: the machine recipe and the image one. The image
recipe is the top configuration file for the generated rootfs and the packages it includes. Our
aim in this lab is to define a custom image from scratch to allow a precise selection of packages
on the target. To show how to deal with real world configuration and how the Yocto Project
can be used in the industry we will, in addition to the production image recipe you will use in
the final product, create a development one including debug tools and show how to link the two
of them to avoid configuration duplication.
First add a custom image recipe in the meta-bootlinlabs layer. We will name it bootlinlabs-
image-minimal. You can find information on how to create a custom image on the dedicated
Yocto Project development manual at https://docs.yoctoproject.org/dev-manual/index.
html. There are different ways to customize an image, we here want to create a full recipe,
using a custom .bb file.
Do not forget to inherit from the core-image class.

Select the images capabilities and packages

You can control the packages built and included into the final image with the IMAGE_INSTALL
configuration variable. It is a list of packages to be built. You can also use package groups to
include a bunch of programs, generally enabling a functionality, such as packagegroup-core-
boot which adds the minimal set of packages required to boot an image (i.e. a shell or a kernel).
You can find the package groups under the packagegroups directories. To have a list of the
available one:
find -name packagegroups
Open some of them to read their description and have an idea about the capabilities they
provide. Then update the installed packages of the image recipe and don’t forget to add the
nInvaders one!

Add a custom package group

We just saw it is possible to use package groups to organize and select the packages instead
of having a big blob of configuration in the image recipe itself. We will here create a custom
package for game related recipes.

© 2004-2022 Bootlin, CC BY-SA license 21

 Yocto Project and OpenEmbedded Training

With the above documentation, create a packagegroup-bootlinlabs-games group which inherits

from the packagegroup class. Add the nInvaders program into its runtime dependencies.
Now update the image recipe to include the package group instead of the nInvaders program
directly.

Differentiate the production recipe from the debug one

You can enable the debugging capabilities of your image just by changing the BitBake target
when building the whole system. We want here to have a common base for both the production
and the debug images, but also take into account the possible differences. In our example only
the built package list will change.
Create a debug version of the previous image recipe, and name it bootlinlabs-image-minimal-
dbg. Try to avoid duplicating code! Then add the dbg-pkgs to the image features list. It is also
recommended to update the recipe’s description, and to add extra debugging tools.
Build the new debug image with BitBake and check the previously included packages are present
in the newly generated rootfs.

22 © 2004-2022 Bootlin, CC BY-SA license

 Yocto Project and OpenEmbedded Training

Lab8: Develop your application in

the Poky SDK
Generate and use the Poky SDK

During this lab, you will:

• Build the Poky SDK
• Install the SDK
• Compile an application for the BeagleBone in the SDK environment

Build the SDK

Two SDKs are available, one only embedding a toolchain and the other one allowing for appli-
cation development. We will use the latter one here.
First, build an SDK for the bootlinlabs-image-minimal image, with the populate_sdk task.
Once the SDK is generated, a script will be available at tmp/deploy/sdk.

Install the SDK

Open a new console to be sure that no extra environment variable is set. We mean to show you
how the SDK sets up a fully working environment.
Install the SDK in $HOME/yocto-stm32-labs/sdk by executing the script generated at the pre-
vious step.
$BUILDDIR/tmp/deploy/sdk/poky-glibc-x86_64-bootlinlabs-image-minimal-cortexa8hf-neon-toolchain-2.5.sh

Set up the environment

Go into the directory where you installed the SDK ($HOME/yocto-stm32-labs/sdk). Source the
environment script:
source environment-setup-cortexa8hf-vfp-neon-poky-linux-gnueabi
Have a look at the exported environment variables:
env

Compile an application in the SDK

Download the essential Ctris sources at https://download.mobatek.net/sources/ctris-0.42-
1-src.tar.bz2
Extract the source in the SDK:

© 2004-2022 Bootlin, CC BY-SA license 23

 Yocto Project and OpenEmbedded Training

tar xf ctris-0.42-1-src.tar.bz2
tar xf ctris-0.42.tar.bz2
cd ctris-0.42
Then modify the Makefile, to make sure that the environment variables exported by the SDK
script are not overridden.
Try to compile the application. Just like nInvaders, ctris is also an old program and won’t build
with a recent toolchain. You will face these errors:
• The ctris makefile uses the native compiler, not the cross compiler provided by the SDK;
while you could fix it using make -e as done for nInvaders, try fixing it by editing the
Makefile this time; hint: you don’t need to write any code, just to delete two lines
• Building with a recent GCC will give the following error, not reported by older versions:
error: format not a string literal and no format arguments [-Werror=format-security]
Fix this by adding -Wno-error=format-security to CFLAGS
• As for nInvaders, you will see the multiple definition of... error; add the -fcommon
flag to CFLAGS also for ctris
You can check the application was successfully compiled for the right target by using the file
command. The ctris binary should be an ELF 32-bit LSB executable compiled for ARM.
Finally, you can copy the binary to the board, by using the scp command. Then run it and
play a bit to ensure it is working fine!

24 © 2004-2022 Bootlin, CC BY-SA license