Shields

Shields, also known as “add-on” or “daughter boards”, attach to a board to extend its features and services for easier and modularized prototyping. In Zephyr, the shield feature provides Zephyr-formatted shield descriptions for easier compatibility with applications.

Shield porting and configuration

Shield configuration files are available in the board directory under /boards/shields:

boards/shields/<shield>
├── <shield>.overlay
├── Kconfig.shield
└── Kconfig.defconfig

These files provides shield configuration as follows:

<shield>.overlay: This file provides a shield description in devicetree format that is merged with the board’s devicetree before compilation.
Kconfig.shield: This file defines shield Kconfig symbols that will be used for default shield configuration. To ease use with applications, the default shield configuration here should be consistent with those in the Write your devicetree.
Kconfig.defconfig: This file defines the default shield configuration. It is made to be consistent with the Write your devicetree. Hence, shield configuration should be done by keeping in mind that features activation is application responsibility.

Board compatibility

Hardware shield-to-board compatibility depends on the use of well-known connectors used on popular boards (such as Arduino and 96boards). For software compatibility, boards must also provide a configuration matching their supported connectors.

This should be done at two different level:

Pinmux: Connector pins should be correctly configured to match shield pins
Devicetree: A board devicetree file, BOARD.dts should define a node alias for each connector interface. For example, for Arduino I2C:

#define arduino_i2c i2c1

aliases {
        arduino,i2c = &i2c1;
};

Note: With support of dtc v1.4.2, above will be replaced with the recently introduced overriding node element:

arduino_i2c:i2c1{};

Board specific shield configuration

If modifications are needed to fit a shield to a particular board or board revision, you can override a shield description for a specific board by adding board or board revision overriding files to a shield, as follows:

boards/shields/<shield>
└── boards
    ├── <board>_<revision>.overlay
    ├── <board>.overlay
    ├── <board>.defconfig
    ├── <board>_<revision>.conf
    └── <board>.conf

Shield activation

Activate support for one or more shields by adding the matching -DSHIELD arg to CMake command

# From the root of the zephyr repository
west build -b None your_app -- -DSHIELD="x_nucleo_idb05a1 x_nucleo_iks01a1"

Alternatively, it could be set by default in a project’s CMakeLists.txt:

set(SHIELD x_nucleo_iks01a1)

Shield variants

Some shields may support several variants or revisions. In that case, it is possible to provide multiple version of the shields description:

boards/shields/<shield>
├── <shield_v1>.overlay
├── <shield_v1>.defconfig
├── <shield_v2>.overlay
└── <shield_v2>.defconfig

In this case, a shield-particular revision name can be used:

# From the root of the zephyr repository
west build -b None your_app -- -DSHIELD=shield_v2

You can also provide a board-specific configuration to a specific shield revision:

boards/shields/<shield>
├── <shield_v1>.overlay
├── <shield_v1>.defconfig
├── <shield_v2>.overlay
├── <shield_v2>.defconfig
└── boards
    └── <shield_v2>
        ├── <board>.overlay
        └── <board>.defconfig

GPIO nexus nodes

GPIOs accessed by the shield peripherals must be identified using the shield GPIO abstraction, for example from the arduino-header-r3 compatible. Boards that provide the header must map the header pins to SOC-specific pins. This is accomplished by including a nexus node that looks like the following into the board devicetree file:

arduino_header: connector {
        compatible = "arduino-header-r3";
        #gpio-cells = <2>;
        gpio-map-mask = <0xffffffff 0xffffffc0>;
        gpio-map-pass-thru = <0 0x3f>;
        gpio-map = <0 0 &gpioa 0 0>,    /* A0 */
                   <1 0 &gpioa 1 0>,    /* A1 */
                   <2 0 &gpioa 4 0>,    /* A2 */
                   <3 0 &gpiob 0 0>,    /* A3 */
                   <4 0 &gpioc 1 0>,    /* A4 */
                   <5 0 &gpioc 0 0>,    /* A5 */
                   <6 0 &gpioa 3 0>,    /* D0 */
                   <7 0 &gpioa 2 0>,    /* D1 */
                   <8 0 &gpioa 10 0>,   /* D2 */
                   <9 0 &gpiob 3 0>,    /* D3 */
                   <10 0 &gpiob 5 0>,   /* D4 */
                   <11 0 &gpiob 4 0>,   /* D5 */
                   <12 0 &gpiob 10 0>,  /* D6 */
                   <13 0 &gpioa 8 0>,   /* D7 */
                   <14 0 &gpioa 9 0>,   /* D8 */
                   <15 0 &gpioc 7 0>,   /* D9 */
                   <16 0 &gpiob 6 0>,   /* D10 */
                   <17 0 &gpioa 7 0>,   /* D11 */
                   <18 0 &gpioa 6 0>,   /* D12 */
                   <19 0 &gpioa 5 0>,   /* D13 */
                   <20 0 &gpiob 9 0>,   /* D14 */
                   <21 0 &gpiob 8 0>;   /* D15 */
};

This specifies how Arduino pin references like <&arduino_header 11 0> are converted to SOC gpio pin references like <&gpiob 4 0>.

In Zephyr GPIO specifiers generally have two parameters (indicated by #gpio-cells = <2>): the pin number and a set of flags. The low 6 bits of the flags correspond to features that can be configured in devicetree. In some cases it’s necessary to use a non-zero flag value to tell the driver how a particular pin behaves, as with:

drdy-gpios = <&arduino_header 11 GPIO_ACTIVE_LOW>;

After preprocessing this becomes <&arduino_header 11 1>. Normally the presence of such a flag would cause the map lookup to fail, because there is no map entry with a non-zero flags value. The gpio-map-mask property specifies that, for lookup, all bits of the pin and all but the low 6 bits of the flags are used to identify the specifier. Then the gpio-map-pass-thru specifies that the low 6 bits of the flags are copied over, so the SOC GPIO reference becomes <&gpiob 4 1> as intended.

See nexus node for more information about this capability.