Board Porting Guide¶
When building an application you must specify the target hardware and the exact board or model. Specifying the board name results in a binary that is suited for the target hardware by selecting the right Zephyr features and components and setting the right Zephyr configuration for that specific target hardware.
A board is defined as a special configuration of an SoC with possible additional components. For example, a board might have sensors and flash memory implemented as additional features on top of what the SoC provides. Such additional hardware is configured and referenced in the Zephyr board configuration.
The board implements at least one SoC and thus inherits all of the features that are provided by the SoC. When porting a board to Zephyr, you should first make sure the SoC is implemented in Zephyr.
Hardware Configuration Hierarchy¶
Hardware definitions in Zephyr follow a well-defined hierarchy of configurations and layers, below are the layers from top to bottom:
- SoC Series
- SoC Family
- CPU Core
This design contributes to code reuse and implementation of device drivers and features at the bottom of the hierarchy making a board configuration as simple as a selection of features that are implemented by the underlying layers. The figures below shows this hierarchy with a few example of boards currently available in the source tree:
|Board||FRDM K64F||nRF52 NITROGEN||nRF51XX||Quark SE C1000 Devboard||Arduino 101|
|SOC||MK64F12||nRF52832||nRF51XX||Quark SE C1000||Curie|
|SOC Series||Kinetis K6x Series||Nordic NRF52||Nordic NRF51||Quark SE||Quark SE|
|SOC Family||NXP Kinetis||Nordic NRF5||Nordic NRF5||Quark||Quark|
If your CPU architecture is already supported by Zephyr, there is no architecture work involved in porting to your board. If your CPU architecture is not supported by the Zephyr kernel, you can add support by following the instructions available at Architecture Porting Guide.
Some OS code depends on the CPU core that your board is using. For example, a given CPU core has a specific assembly language instruction set, and may require special cross compiler or compiler settings to use the appropriate instruction set.
If your CPU architecture is already supported by Zephyr, there is no CPU core work involved in porting to your platform or board. You need only to select the appropriate CPU in your configuration and the rest will be taken care of by the configuration system in Zephyr which will select the features implemented by the corresponding CPU.
This layer implements most of the features that need porting and is split into three layers to allow for code reuse when dealing with implementations with slight differences.
This layer is a container of all SoCs of the same class that, for example implement one single type of CPU core but differ in peripherals and features. The base hardware will in most cases be the same across all SoCs and MCUs of this family.
Moving closer to the SoC, the series is derived from an SoC family. A series is defined by a feature set that serves the purpose of distinguishing different SoCs belonging to the same family.
Finally, an SoC is actual hardware component that is physically available on a board.
A board implements an SoC with all its features, together with peripherals available on the board that differentiates the board with additional interfaces and features not available in the SoC.
Default board configuration¶
When porting Zephyr to a board, you must provide the board’s default Kconfig configuration, which is used in application builds unless explicitly overridden.
See the Kconfig - Tips and Best Practices page for some best practices and tips when writing Kconfig files.
In order to provide consistency across the various boards and ease the work of users providing applications that are not board specific, the following guidelines should be followed when porting a board:
- Provide pin and driver configuration that matches the board’s valuable components such as sensors, buttons or LEDs, and communication interfaces such as USB, Ethernet connector, or Bluetooth/Wi-Fi chip.
- When a well-known connector is present (such as used on an Arduino or 96board), configure pins to fit this connector.
- Configure components that enable the use of these pins, such as configuring an SPI instance for Arduino SPI.
- Configure an output for the console.
- Propose and configure a default network interface.
- Enable all GPIO ports.
Setting configuration values¶
Kconfig symbols can be set to their
BOARD-specific values in one of two
ways. The right method to use depends on whether the symbol is visible or
Visible and invisible Kconfig symbols¶
Kconfig symbols come in two varieties:
- A Kconfig symbol defined with a prompt is visible, and can be configured from
- A Kconfig symbol defined without a prompt is invisible. The user has no direct control over its value.
Here are some examples of visible and invisible symbols:
config NOT_VISIBLE bool default FOO config VISIBLE_1 string prompt "Foo value" config VISIBLE_2 # Shorthand for giving a type and a prompt at the same time. This is # the preferred style in Zephyr. bool "Enable stuff"
Configuring visible Kconfig symbols¶
BOARD-specific configuration values for visible Kconfig symbols
should be given in
uses the standard Kconfig
.config file syntax.
Configuring invisible Kconfig symbols¶
BOARD-specific configuration values for invisible Kconfig symbols must be
boards/ARCHITECTURE/BOARD/Kconfig.defconfig, which uses
.config files have no effect on invisible symbols,
so this scheme is not just an organizational issue.
Assigning values in
Kconfig.defconfig relies on being able to define a
Kconfig symbol in multiple locations. As an example, say we want to set
FOO_WIDTH below to 32:
config FOO_WIDTH int
To do this, we extend the definition of
FOO_WIDTH as follows, in
if BOARD_MY_BOARD config FOO_WIDTH default 32 endif
Since the type of the symbol (
int) has already been given at the first
definition location, it does not need to be repeated here.
default values in
Kconfig.defconfig files have priority over
default values given on the “base” definition of a symbol. Internally, this
is implemented by including the
Kconfig.defconfig files first. Kconfig
uses the first
default with a satisfied condition, where an empty condition
if y (is always satisfied).
range properties on
hex symbols work the same way, and
can also be added or overridden in
If you want a symbol to only be user-configurable on some boards, make its base
definition have no prompt, and then add a prompt to it in the
Kconfig.defconfig files of the boards where it should be configurable.
Prompts added in
Kconfig.defconfig files show up at the location of
Kconfig.defconfig file in the
menuconfig interface, rather
than at the location of the base definition of the symbol.
There are two ways to configure a Kconfig
By setting one of the choice symbols to
Setting one choice symbol to
yautomatically gives all other choice symbols the value
If multiple choice symbols are set to
y, only the last one set to
ywill be honored (and the rest will get the value
n). This allows a choice selection from a board
defconfigfile to be overridden from an application
By changing the
defaultof the choice in
As with symbols, changing the default for a choice is done by defining the choice in multiple locations. For this to work, the choice must have a name.
As an example, assume that a choice has the following base definition (here, the name of the choice is
choice FOO bool "Foo choice" default B config A bool "A" config B bool "B" endchoice
To change the default symbol of
A, you would add the following definition to
choice FOO default A endchoice
Kconfig.defconfig method should be used when the dependencies of
the choice might not be satisfied. In that case, you’re setting the default
selection whenever the user makes the choice visible.
One motivation for this configuration scheme is to avoid making fixed
BOARD-specific settings configurable in the
menuconfig interface. If
all configuration were done via
BOARD_defconfig, all symbols would have
to be visible, as values given in
BOARD_defconfig have no effect on
Having fixed settings be user-configurable might be confusing, and would allow the user to create broken configurations.
Environment variables in
source statements are expanded directly in
Kconfiglib, meaning no
option env="ENV_VAR" “bounce” symbols need to be
defined. If you need compatibility with the C Kconfig tools for an out-of-tree
Kconfig tree, you can still add such symbols, but they must have the same name
as the corresponding environment variables.
option env has been removed from the C tools in Linux 4.18 as well.
The recommended syntax for referencing environment variables is now
$(FOO) rather than
$FOO. This uses the new Kconfig preprocessor.
The following Kconfig extensions are available:
sourcestatement supports glob patterns and includes each matching file. A pattern is required to match at least one file.
Consider the following example:
If the pattern
foo/bar/*/Kconfigmatches the files
foo/bar/qaz/Kconfig, the statement above is equivalent to the following two
source "foo/bar/baz/Kconfig" source "foo/bar/qaz/Kconfig"
The wildcard patterns accepted are the same as for the Python glob module.
If no files match the pattern, an error is generated.
For cases where it’s okay for a pattern to match no files (or for a plain filename to not exist), a separate
osource(optional source) statement is available.
osourceis a no-op in case of no matches.
osourceare analogous to
rsourcestatement is available for including files specified with a relative path. The path is relative to the directory of the
Kconfigfile that contains the
As an example, assume that
foo/Kconfigis the top-level
Kconfigfile, and that
foo/bar/Kconfighas the following statements:
source "qaz/Kconfig1" rsource "qaz/Kconfig2"
This will include the two files
rsourcecan be used to create
Kconfig“subtrees” that can be moved around freely.
rsourcealso supports glob patterns.
orsourcestatement, which combines
For example, the following statement will include
Kconfig2from the current directory (if they exist):
def_stringkeywords, which are analogous to
def_bool. These set the type and add a
defaultat the same time.
Old Zephyr Kconfig behavior for defaults¶
Prior to early August 2018 (during development of Zephyr 1.13), Zephyr used a
custom patch that made Kconfig prefer the last
default with a satisfied
condition, instead of the first one. This patch has been removed.
Consider this example:
config FOO string default "first" if n default "second" default "third" if n default "fourth" default "fifth" if n
With the old custom behavior,
FOO got the value
"fourth", from the last
default with a satisfied condition.
With the new behavior,
FOO gets the value
"second", from the first
default with a satisfied condition. This is standard Kconfig behavior.
There are two issues with the old behavior:
It’s inconsistent with how Kconfig works in other projects, which is confusing.
Due to oversights, earlier
rangeproperties were still preferred, as well as earlier
defaultproperties on choices.
In addition to being inconsistent, this made it impossible to override
defaultproperties on choices if the base definition of the symbol/choice already had
If you’re maintaining an external project that has symbols with multiple
default properties, you will need to swap the order of the
properties to get the same behavior as before.
If your external project is modifying symbols in the base Zephyr
configuration by sourcing
Kconfig.zephyr and adding additional symbol
definitions, you might need to move the
source from before the extra
symbol definitions to after them.
More Kconfig resources¶
The Kconfig - Tips and Best Practices page has some best practices and tips for writing Kconfig files.
The kconfiglib.py docstring (at the top of the file) goes over how symbol values are calculated in detail.