Building and using MCUboot with Zephyr

MCUboot began its life as the bootloader for Mynewt. It has since acquired the ability to be used as a bootloader for Zephyr as well. There are some pretty significant differences in how apps are built for Zephyr, and these are documented here.

Please see the design document for documentation on the design and operation of the bootloader itself. This functionality should be the same on all supported RTOSs.

The first step required for Zephyr is making sure your board has flash partitions defined in its device tree. These partitions are:

It is not recommended to use the swap-using-scratch algorithm of MCUboot, but if this operating mode is desired then the following flash partition is also needed (see end of this help file for details on creating a scratch partition and how to use the swap-using-scratch algorithm):

Currently, the two image slots must be contiguous. If you are running MCUboot as your stage 1 bootloader, boot_partition must be configured so your SoC runs it out of reset. If there are multiple updateable images then the corresponding primary and secondary partitions must be defined for the rest of the images too (for example, slot2_partition and slot3_partition for Image 1).

The flash partitions are typically defined in the Zephyr boards folder, in a file named boards/<arch>/<board>/<board>.dts. An example .dts file with flash partitions defined is the frdm_k64f’s in boards/arm/frdm_k64f/frdm_k64f.dts. Make sure the DT node labels in your board’s .dts file match the ones used there.

Installing requirements and dependencies

Install additional packages required for development with MCUboot:

  cd ~/mcuboot  # or to your directory where MCUboot is cloned
  pip3 install --user -r scripts/requirements.txt

Building the bootloader itself

The bootloader is an ordinary Zephyr application, at least from Zephyr’s point of view. There is a bit of configuration that needs to be made before building it. Most of this can be done as documented in the CMakeLists.txt file in boot/zephyr. There are comments there for guidance. It is important to select a signature algorithm, and decide if the primary slot should be validated on every boot.

To build MCUboot, create a build directory in boot/zephyr, and build it as usual:

  cd boot/zephyr
  west build -b <board>

In addition to the partitions defined in DTS, some additional information about the flash layout is currently required to build MCUboot itself. All the needed configuration is collected in boot/zephyr/include/target.h. Depending on the board, this information may come from board-specific headers, Device Tree, or be configured by MCUboot on a per-SoC family basis.

After building the bootloader, the binaries should reside in build/zephyr/zephyr.{bin,hex,elf}, where build is the build directory you chose when running west build. Use west flash to flash these binaries from the build directory. Depending on the target and flash tool used, this might erase the whole of the flash memory (mass erase) or only the sectors where the bootloader resides prior to programming the bootloader image itself.

Building applications for the bootloader

In addition to flash partitions in DTS, some additional configuration is required to build applications for MCUboot.

This is handled internally by the Zephyr configuration system and is wrapped in the CONFIG_BOOTLOADER_MCUBOOT Kconfig variable, which must be enabled in the application’s prj.conf file.

The directory samples/zephyr/hello-world in the MCUboot tree contains a simple application with everything you need. You can try it on your board and then just make a copy of it to get started on your own application; see samples/zephyr/README.md for a tutorial.

The Zephyr CONFIG_BOOTLOADER_MCUBOOT configuration option documentation provides additional details regarding the changes it makes to the image placement and generation in order for an application to be bootable by MCUboot.

With this, build the application as your normally would.

Signing the application

In order to upgrade to an image (or even boot it, if MCUBOOT_VALIDATE_PRIMARY_SLOT is enabled), the images must be signed. To make development easier, MCUboot is distributed with some example keys. It is important to stress that these should never be used for production, since the private key is publicly available in this repository. See below on how to make your own signatures.

Images can be signed with the scripts/imgtool.py script. It is best to look at samples/zephyr/Makefile for examples on how to use this.

Flashing the application

The application itself can flashed with regular flash tools, but will need to be programmed at the offset of the primary slot for this particular target. Depending on the platform and flash tool you might need to manually specify a flash offset corresponding to the primary slot starting address. This is usually not relevant for flash tools that use Intel Hex images (.hex) instead of raw binary images (.bin) since the former include destination address information. Additionally you will need to make sure that the flash tool does not perform a mass erase (erasing the whole of the flash) or else you would be deleting MCUboot. These images can also be marked for upgrade, and loaded into the secondary slot, at which point the bootloader should perform an upgrade. It is up to the image to mark the primary slot as “image ok” before the next reboot, otherwise the bootloader will revert the application.

Managing signing keys

The signing keys used by MCUboot are represented in standard formats, and can be generated and processed using conventional tools. However, scripts/imgtool.py is able to generate key pairs in all of the supported formats. See the docs for more details on this tool.

Generating a new keypair

Generating a keypair with imgtool is a matter of running the keygen subcommand:

    $ ./scripts/imgtool.py keygen -k mykey.pem -t rsa-2048

The argument to -t should be the desired key type. See the the docs for more details on the possible key types.

Extracting the public key

The generated keypair above contains both the public and the private key. It is necessary to extract the public key and insert it into the bootloader. Use the CONFIG_BOOT_SIGNATURE_KEY_FILE Kconfig option to provide the path to the key file so the build system can extract the public key in a format usable by the C compiler. The generated public key is saved in build/zephyr/autogen-pubkey.h, which is included by the boot/zephyr/keys.c.

Currently, the Zephyr RTOS port limits its support to one keypair at the time, although MCUboot’s key management infrastructure supports multiple keypairs.

Once MCUboot is built, this new keypair file (mykey.pem in this example) can be used to sign images.

Using swap-using-scratch flash algorithm

To use the swap-using-scratch flash algorithm, a scratch partition needs to be present for the target board which is used for holding the data being swapped from both slots, this section must be at least as big as the largest sector size of the 2 partitions (e.g. if a device has a primary slot in main flash with a sector size of 512 bytes and secondar slot in external off-chip flash with a sector size of 4KB then the scratch area must be at least 4KB in size). The number of sectors must also be evenly divisable by this sector size, e.g. 4KB, 8KB, 12KB, 16KB are allowed, 7KB, 7.5KB are not. This scratch partition needs adding to the .dts file for the board, e.g. for the nrf52dk_nrf52832 board thus would involve updating <zephyr>/boards/nordic/nrf52dk/nrf52dk_nrf52832.dts with:

    boot_partition: partition@0 {
        label = "mcuboot";
        reg = <0x00000000 0xc000>;
    };
    slot0_partition: partition@c000 {
        label = "image-0";
        reg = <0x0000C000 0x37000>;
    };
    slot1_partition: partition@43000 {
        label = "image-1";
        reg = <0x00043000 0x37000>;
    };
    scratch_partition: partition@7a000 {
        label = "image-scratch";
        reg = <0x0007a000 0x00006000>;
    };

Which would make the application size 220KB and scratch size 24KB (the nRF52832 has a 4KB sector size so the size of the scratch partition can be reduced at the cost of vastly reducing flash lifespan, e.g. for a 32KB firmware update with an 8KB scratch area, the scratch area would be erased and programmed 8 times per image upgrade/revert). To configure MCUboot to work in swap-using-scratch mode, the Kconfig value must be set when building it: CONFIG_BOOT_SWAP_USING_SCRATCH=y.

Note that it is possible for an application to get into a stuck state when swap-using-scratch is used whereby an application has loaded a firmware update and marked it as test/confirmed but MCUboot will not swap the images and erasing the secondary slot from the zephyr application returns an error because the slot is marked for upgrade.

Serial recovery

Interface selection

A serial recovery protocol is available over either a hardware serial port or a USB CDC ACM virtual serial port. The SMP server implementation can be enabled by the CONFIG_MCUBOOT_SERIAL=y Kconfig option. To set a type of an interface, use the BOOT_SERIAL_DEVICE Kconfig choice, and select either the CONFIG_BOOT_SERIAL_UART or the CONFIG_BOOT_SERIAL_CDC_ACM value. Which interface belongs to the protocol shall be set by the devicetree-chosen node:

Entering the serial recovery mode

To enter the serial recovery mode, the device has to initiate rebooting, and a triggering event has to occur (for example, pressing a button).

By default, the serial recovery GPIO pin active state enters the serial recovery mode. Use the mcuboot_button0 devicetree button alias to assign the GPIO pin to the MCUboot.

Alternatively, MCUboot can wait for a limited time to check if DFU is invoked by receiving an MCUmgr command. Select CONFIG_BOOT_SERIAL_WAIT_FOR_DFU=y to use this mode. CONFIG_BOOT_SERIAL_WAIT_FOR_DFU_TIMEOUT option defines the amount of time in milliseconds the device will wait for the trigger.

Direct image upload

By default, the SMP server implementation will only use the first slot. To change it, invoke the image upload MCUmgr command with a selected image number, and make sure the CONFIG_MCUBOOT_SERIAL_DIRECT_IMAGE_UPLOAD=y Kconfig option is enabled. Note that the CONFIG_UPDATEABLE_IMAGE_NUMBER Kconfig option adjusts the number of image-pairs supported by the MCUboot.

The mapping of image number to partition is as follows:

0 is a default upload target when no explicit selection is done.

System-specific commands

Use the CONFIG_ENABLE_MGMT_PERUSER=y Kconfig option to enable the following additional commands:

More configuration

For details on other available configuration options for the serial recovery protocol, check the Kconfig options (for example by using menuconfig).