Skip to content

60 SOM release 11.236.0.25

Latest
Compare
Choose a tag to compare
@sbraneky sbraneky released this 08 Jul 23:22

Normal Release Update
Release includes

CS-RN-SOM60 release notes in pdf form.
som60sd binary development kit image and SDK for C/C++ application development for SD card.
For using Summit Linux, jump below to our temporary Summit Linux reference guide for this release

For more documentation, please see the online documentation:
http://documentation.lairdconnect.com/Builds/60-SOM/latest/Content/Home.htm

Summit Linux Reference Guide
Prerequisites
The reference guide is designed to walk you through Laird Connectivity's development process for the 60 SOM, including the Summit Linux board support package. The most seamless workflow for developing your application for the 60 SOM is to set up a GitHub account, configure your SSH keys, and then install and configure Git on your development PC.

Ubuntu
Laird recommends Ubuntu 22.04 64-bit as the base operating system for your development. All instructions in this reference guide assume a developer on an Ubuntu 22.04 64-bit system. If using other than Ubuntu 22.04 64-bit, please ensure your Linux distribution is a 64-bit variant as the cross-compiling toolchain requires a 64-bit Linux environment.

SSH
Laird makes extensive use of GitHub and having proper Secure Shell (SSH) access to our repositories on GitHub. If you haven't set up SSH, you can use the instructions below to enable SSH for use with GitHub.

Check whether your private key (/.ssh/id_rsa) and public key (/.ssh/id_rsa.pub) files exist
$ ls -1 ~/.ssh/id_rsa*
If they do not, generate them
$ ssh-keygen

  1. Accept default file paths
  2. Accept no passphrase
    View your public key
    $ cat ~/.ssh/id_rsa.pub
    git
    Laird delivers the Summit Linux board support package using git repositories. If you haven't set up git, you can use the instructions below to enable git for use with GitHub.

Install and configure the "git" package
$ sudo apt install git
$ git config --global user.name "John Doe"
$ git config --global user.email "[email protected]"
$ git config --list
Expect user.name=John Doe
Expect user.email=[email protected]
repo
Laird uses the repo tool from the Android project to pull down and correctly layout the git repositories that comprise the Summit Linux board support package. To install repo on Ubuntu, use the instructions below.

$ sudo apt install repo
You can also get it by following the instructions here:

(https://source.android.com/source/downloading.html#installing-repo)
Github Account
Summit Linux board support packages are available on GitHub. The most seamless workflow throughout a integration is to set up a GitHub account (https://github.com/) and add your SSH public key to your GitHub account. Instructions for setting up a GitHub account and adding an SSH key are below.

Go to (https://github.com/join) and follow the instructions.
Once you have a GitHub account add your public SSH key generated previously.
View your public key

  1. $ cat ~/.ssh/id_rsa.pub
    This should be similar to:
  2. ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQDW5HxCJe56MZKkGxtHU2hkb1uyuJ3i+AFZNnSIApiDe1l1H9Y40YjBF+sE47GKgNuzQIMqKZ6xjRdb8KxvtjQIE/VcSiz9KIaFtKce/wKKxy0vOXbskSLzELlA9ovY9AJmp3qUaW+BEChnggERjPvq+1oyiKGRJzo51j/CQr+Yx8c2MtKubjHknKkQ2Th8kxL1bQj4lVgyfqNGj3DXUz9S+kTm+dLnmmOXlUfxx8Khrb7j0Dg+lk1lqeolqt6aFwpMnb8V2h5lDevJ0YSSyId01OnaAnIlx1lc+zauctsgtZf5Htl9cXS7Cp3OCMfSa+IKFeFL9/ku9C25EdA5zOnt [email protected]
    Copy the contents of the above output to you clipboard
    Follow the instructions at
    (https://help.github.com/articles/adding-a-new-ssh-key-to-your-github-account/) starting with Step 2.
    Summit Linux
    Overview
    Summit Linux is Laird's board support package specifically tailored for Laird's SOMs and customer's connectivity driven use cases. Laird regularly updates Summit Linux by merging in the upstream Linux kernel and Buildroot. We merge in major long-term support kernel releases which allow our SOMs to utilize the latest in drivers and kernel space functionality, performance enhancements, security, and bug fixes. Also, we merge in major long-term support Buildroot releases which provide for the latest in over 2200 user space applications and libraries. Summit Linux takes this solid upstream heritage and integrates our custom platform enhancements for connectivity, security, and power consumption.

Buildroot
The core piece of Summit Linux is Buildroot. Buildroot is a from source build system designed to allow the customer to create a customized Linux image for a target embedded computing module. Buildroot is capable of configuring and building the bootloader, kernel, and rootfs for Laird's SOMs. Buildroot's core build system technologies are the well-known and easy to understand make build tool and kconfig configuration tool. This enables easy build customization through the user interface tools menuconfig or xconfig and Buildroot make commands. The Linux kernel, U-Boot bootloader, and Busybox core userspace ulilities toolkit all use make, Kconfig, and menuconfig. Build and configuration concepts found in one translate nicely to another. Buildroot has easy to read and extensive documentation available at (https://buildroot.org/docs.html) in pdf, html, and ascii form. If time permits, Laird highly recommends following the Training section of the documentation landing page.

Laird's Buildroot is a fork of the upstream Buildroot stable. Starting with and merging upstream allows Laird to know the authenticity of Buildroot's source code and better verify the security of our fork. Laird targets long-term support releases (LTS) for merging into our Buildroot fork.

Linux Kernel
Laird's kernel is a fork the upstream Linux stable kernel. Starting with and merging upstream allows Laird to know the authenticity of the kernel's source code and better verify the security of our fork. Laird targets long-term support releases (LTS) for merging into our kernel fork.

Toolchain
Laird preselects prebuilt toolchains for each release of Summit Linux for our SOMs. The advantage of using a preselected and prebuilt toolchain is the high level of QA verification on common components between stock Laird development images and customer images. Laird selects toolchains that have been marked stable by the Buildroot community and use proven components.

Getting Started with Summit Linux
This section will walk a developer through following:

Downloading a developer's SD card image
Flashing a developer's SD card image
Using a prebuilt SDK
Manifest file
Downloading the board support package source code
Building the developer's SD card image
Building the NAND image
Software update
Setup web server
Buildroot cache
Creating a custom SDK

Downloading a developer's SD card image

Summit Linux releases include a prebuilt SD card image as a starting point for evaluating and integrating a Summit Linux release on a Laird SOM. For the 60 SOM, these prebuilt images are found on at 60 SOM release page. These releases are named som60sd-laird-A.B.C.D.tar.bz2. These prebuilt SD card images are good for quickly testing a SOM running the latest software.

Flashing a developer's SD card image

Once the image is downloaded. Extract the image:

~/Downloads/som60sd$ tar -xvf som60sd-laird-11.x.y.z.tar.bz2
som60sd-laird-11.x.y.z/
som60sd-laird-11.x.y.z/target-sbom
som60sd-laird-11.x.y.z/host-sbom
som60sd-laird-11.x.y.z/u-boot-spl.bin
som60sd-laird-11.x.y.z/u-boot.itb
som60sd-laird-11.x.y.z/rootfs.tar
som60sd-laird-11.x.y.z/som60sd-sdk.tar.bz2
som60sd-laird-11.x.y.z/legal-info.tar.bz2
som60sd-laird-11.x.y.z/kernel.itb
som60sd-laird-11.x.y.z/mksdcard.sh
som60sd-laird-11.x.y.z/mksdimg.sh

To flash the image to an SD card use the mksdcard.sh script. The mksdcard.sh script takes the target device as an argument and will ask if you'd like to proceed with removing all data on your SD card and flashing a new image. This is shown below:

~/Downloads/som60sd-laird-11.x.y.z$ sudo ./mksdcard.sh /dev/sdc
[sudo] password for user:

*************************************************************************
WARNING: All data on /dev/sdc now will be destroyed! Continue? [y/n]
*************************************************************************
[Partitioning /dev/sdc...]
1024+0 records in
1024+0 records out
1048576 bytes (1.0 MB, 1.0 MiB) copied, 0.653846 s, 1.6 MB/s
DISK SIZE - 1967128576 bytes
Checking that no-one is using this disk right now ... OK

Disk /dev/sdc: 1.9 GiB, 1967128576 bytes, 3842048 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes

>>> Created a new DOS disk label with disk identifier 0x59c337f4.
Created a new partition 1 of type 'W95 FAT16 (LBA)' and of size 48 MiB.
/dev/sdc2: Created a new partition 2 of type 'Linux' and of size 1.8 GiB.
/dev/sdc3:
New situation:

Device     Boot  Start     End Sectors  Size Id Type
/dev/sdc1  *      2048  100351   98304   48M  e W95 FAT16 (LBA)
/dev/sdc2       100352 3842047 3741696  1.8G 83 Linux

The partition table has been altered.
Calling ioctl() to re-read partition table.
Syncing disks.
[Making filesystems...]
[Copying files...]
[Done]

You can now insert your SD card into the SOM hardware development kit and press the reset button to boot into the new SD card image.

Using a prebuilt SDK

Laird Linux releases include a prebuilt SDK to start doing application development for a Laird SOM. For the 60 SOM, this prebuilt SDK is called som60sd-sdk-A.B.C.D.tar.bz2 and can be found with each release at the 60 SOM release page. The prebuilt SDK includes the toolchain and all development files of the software packages used to generate the prebuilt SD card image from that release. The SDK can be set up for use with an IDE to allow application developers to not need a full BSP on their system. To use the SDK, extract the SDK tarball then run the script relocate-sdk.sh (located at the top directory of the SDK), to make sure all paths are updated with the new location. For more information on using SDKs generated from Laird's Buildroot fork, see the Buildroot manual's section on the SDK.

Manifest file

Laird provides manifest files for customers to obtain released resources. Each manifest file describes the projects that are available and how to fetch them. A typical manifest file has remote, default and project elements. The remote element specifies the git url and revision shared by one or more projects. The default element specifies the the default remote element to be used by projects. The project element specifies the repository to be cloned. Its name attribute will be appended to the remote git url to generate the actual address. The path attribute defines the relative path to the top working directory to place the project. Following is an example of the manifest file:

<?xml version="1.0" encoding="UTF-8">
<manifest>
    <remote name="origin" fetch="ssh://[email protected]/LairdCP" />
    <default remote="origin" revision="refs/tags/LRD-REL-7.x.y.z" />
    <project path="wb" name="wb.git" />
 </manifest>

A remote element is defined here, which is also the default. Project wb.git will be fetched from ssh://[email protected]/LairdCP/wb.git with revision refs/tags/LRD-REL-11.x.y.z.

Downloading the board support package source code

First step, pick which release you want. Odds are you want the most recent of your release.

Next, use repo to initialize and fetch your release. This is a two-step process: first you tell repo which manifest to use and then you tell it to fetch everything.

mkdir som60_11.x.y.z_source
cd som60_11.x.y.z_source
repo init -u [email protected]:LairdCP/SOM60-Release-Packages.git -m som60_11.x.y.z.xml
repo sync

Note: Repo will initialize a .repo directory and then place all files directly in the directory that you are in when you run the repo command. So we recommend making a subdirectory and working in there.

Building the SD Card developer's image

Once your repo sync is finished, you are ready to build your own SD card image. This can be achieved by the following:

make -C som-external/ som60sd

Once your build completes, you will find the output similar to below:

~/git/lrd-11.x.y.z$ cd buildroot/output/som60sd/images/
~/git/lrd-11.x.y.z/buildroot/output/som60sd/images$ ls -al
at91-dvk_som60.dtb
boot.scr -> ../../../board/laird/configs-common/image/boot_mmc.scr
Image
Image.gz
kernel.itb
kernel.its
mksdcard.sh -> ../../../board/laird/scripts-common/mksdcard.sh
mksdimg.sh -> ../../../board/laird/scripts-common/mksdimg.sh
rootfs.tar
som60sd-laird.tar.bz2
u-boot.dtb
u-boot.itb
u-boot.its -> ../../../board/laird/configs-common/image/u-boot.its
u-boot-nodtb.bin
u-boot.scr -> ../../../board/laird/configs-common/image/u-boot_mmc.scr
u-boot.scr.itb
u-boot.scr.its -> ../../../board/laird/configs-common/image/u-boot.scr.its
u-boot-spl.bin
u-boot-spl.dtb
u-boot-spl-nodtb.bin

Go to images/ and run sudo ./mksdcard.sh /dev/sd-device to flash the SD card.

Building the NAND image

NAND image can be built by the following:

make -C som-external/ som60

Once your build completes, you will find the output similar to below:

~/git/lrd-11.x.y.z$ cd buildroot/output/som60/images/
~/git/lrd-11.x.y.z$ buildroot/output/som60/images$ ls -al
at91-dvk_som60.dtb
boot.bin
boot.scr
erase_data.sh -> ../../../board/laird/scripts-common/erase_data.sh
Image
Image.gz
kernel.itb
kernel-nosig.itb
keys/
pmecc.bin
rootfs.bin
rootfs.squashfs
rootfs.verity
rootfs.verity.header
som60-laird.tar.bz2
som60.swu
sw-description
u-boot.dtb
u-boot.dtb.enc
u-boot-env.tgz -> ../../../board/laird/configs-common/image/u-boot-env.tgz
u-boot.itb
u-boot.its -> ../../../board/laird/configs-common/image/u-boot-enc.its
u-boot-nodtb.bin
u-boot-nodtb.bin.enc
u-boot.scr -> ../../../board/laird/configs-common/image/u-boot.scr
u-boot.scr.itb
u-boot.scr.itb.enc
u-boot.scr.its -> ../../../board/laird/configs-common/image/u-boot.scr.its
u-boot-spl.bin
u-boot-spl.dtb
u-boot-spl-nodtb.bin

som60.swu is the image for software update.

Software update

SOM60 adopts the concept of double copy(A/B system) to improve safety and reliability. Each copy contains both the kernel and rootfs images. The u-boot environment variable 'bootside' determines the working copy.

Software update is performed efficiently with SWUpdate. All images and scripts are appended to a single .swu file by the Buildroot. In addition, it must have a description file to specify software collections and operation modes to implement the double copy strategy. By default the SOM60 description file has five collections: main-a, main-b, full-a, full-b and complete. main-a(b) is to update kernel and rootfs images of copy a(b). full-a(b) will also update bootstrap and u-boot images. complete is for SOM60SD only. It will re-partition and then program the entire NAND, thus to initialize a SOM60 device.

SOM60 supports 4 ways to update software: local update, auto update, download update and upload update. Update is performed according to the selected collection. However, it is not allowed to update the kernel and rootfs of the working copy. We have to run fw_printenv bootside to get the correct standby copy information for local update and download update. The u-boot environment variable 'bootside' will be automatically updated when it is completed. Reboot is required to run the new image.

  1. Local update

    Insert a USB disk with .swu file(USB disk must be formatted to fat32 or ext4), which will be automatically mounted to /media/. Perform software update by setting the collection, i.g 'main-a', 'full-b':

    swupdate -e stable,<collection>  -i </path/to/swu/file>
    
  2. Auto update(secure update service needs to be enabled)
    Insert a USB disk with .swu file(USB disk must be formatted to fat32 or ext4). The secure update service checks whether file /media/sda1/swupdate.swu exists, and then starts an update;

  3. Download update

    Software can be updated remotely if a http server is available. Make sure .swu file is already in the web directory before update begins.

     swupdate -e stable,<collection> -d "-u <url of .swu file>"
    
  4. Upload update(web service needs to be enabled)

    Login as root, and then go to software update page by clicking Firmware Update. Select the .swu file for update and click 'Update' button. Wait until flashing is completed or failed.

Setup web server

Python's simple built-in web server here. Go to the dir where images are saved, and run following command to start the web service

python -m SimpleHTTPServer 8000

For python3, the command is

python3 -m http.server 8000

Buildroot cache

In order to speed up builds, set up a Buildroot cache to store sources that Buildroot will download. Off your home directory:

mkdir ~/.br2_dl_dir

Then add it to your build environment:

export BR2_DL_DIR="${HOME}/.br2_dl_dir"

Create a custom SDK

If you'd like to create a custom SDK from your customized source build, while in the target's output directory, issue a make sdk:

~/git/lrd-11.x.y.z/buildroot/output/som60sd$ make sdk

NetworkManager

Laird uses its own customized fork of NetworkManager for networking configuration, including WiFi profile management. For more information on using NetworkManager please see our Laird NetworkManager User Guide.

Firewalld

Laird uses firewalld as its network security system. Firewalld provides a dynamically managed firewall, which allows updating rules without breaking existing connections. Firewalld uses zones and services instead of tables and chains. A firewalld zone defines the trust level for a connection, an interface etc., while a firewalld service is mainly a list of local ports and destinations. firewalld has predefined zones and services located in $(TARGET_DIR)/usr/lib/firewalld/zones and $(TARGET_DIR)/usr/lib/firewalld/services. The default zone is set to trusted on SOM60, which will accept all network connections. Default zone can be changed in $(TARGET_DIR)/etc/firewalld/firewalld.conf.

firewall-cmd is the command line tool of the firewalld daemon. It provides interface to manage runtime and permanent rule sets. By default, it applies to the runtime set; with the --permanent flag it applies to the permanent set. These permanent rules will be saved in corresponding xml files in /etc/firewalld/. To activate a permanent rule, either add the rule to both sets, or add the rule to the permanent set only and then reload firewalld configuration.

To start firewalld: systemctl start firewalld
To stop firewalld: systemctl stop firewalld
To check the firewalld status: firewall-cmd --state
To save runtime rules as permanent: firewall-cmd --runtime-to-permanent
To reload firewalld configuration: firewall-cmd --reload

An example is given on sharing WIFI Internet (wlan0) through ethernet interface (eth1). Needs to
reload configuration before permanent rules take effect.

1. Create a connection "eth1" for ethernet interface "eth1"; create a connection "wlan0" for wifi
   interface "wlan0". Set connection.zone for both connections:
       nmcli c mod eth1 connection.zone internal
       nmcli c mod wlan0 connection.zone external
   "internal" and "external" are firewalld’s predefined zones.
   Set ipv4.method to "shared" for "eth1" connection:
       nmcli c mod eth1 ipv4.method shared
   Reload connections:
       nmcli c reload
2. Change default firewalld zone to "internal". Changes will be applied to the default zone if "--zone" not
   specified in the firewall-cmd tool.
       firewall-cmd --set-default-zone=internal
   Check the default zone is changed by
       firewall-cmd --get-default-zone
   Zone infomation can be listed by
       firewall-cmd --zone=internal --list-all
3. Add interface "wlan0" to "external" zone, and interface "eth1" to "internal" zone:
       firewall-cmd --zone=external --add-interface=wlan0 --permanent
       firewall-cmd --zone=internal --add-interface=eth1 --permanent
       firewall-cmd --get-active-zones should show `internal` and `external` as active zones.
4. "firewall-cmd --get-services --zone=internal" will show the services already added to the zone:
   Here we need to enable "dhcp" service so that wifi clients can get IP address:
       firewall-cmd --add-service=dhcp --permanent
   DNS needs to be enabled if clients want to access webs
       firewall-cmd --add-service=dns --permanent
   Services can be removed by
       firewall-cmd --remove-service=service_name --permanent
5. For the "external" zone, "masquerade" should be already enabled. This can be queried by:
       firewall-cmd --query-masquerade --zone=external
   If not enabled, enable it by:
       firewall-cmd --add-masquerade --zone=external --permanent
7. In order to access the webserver on SOM60 via "eth1" interface, "http" servcie needs to be
   added to the "internal" zone:
       firewall-cmd --add-service=http --permanent

Reload the rules and clients should be able to connect to Internet via wlan0.

Port forwarding

If we want to access the webserver through wlan0 interface (http service not enabled in 'external' zone), port forwarding can be used:

firewall-cmd --zone=external --add-forward-port=port=80:proto=tcp:toport=80:toaddr=10.42.0.1 ('10.42.0.1' is the ip address of eth1 interface)

This can be removed by

firewall-cmd --zone=external --remove-forward-port=port=80:proto=tcp:toport=80:toaddr=10.42.0.1

Make sure masquerade is enabled for the zone needs port forwarding.

Rich Rules

Rich Rules allows to add fully custom firewall rules to any zone for any port, protocol, address and action. --add-rich-rule, --list-rich-rules and --remove-rich-rule are used to manage rich rules. The following rule will reject all connections from wifi client '10.42.0.122':

firewall-cmd --zone=internal --add-rich-rule='rule family="ipv4" source address="10.42.0.122" reject'

This can be removed by

firewall-cmd --zone=internal --remove-rich-rule='rule family="ipv4" source address="10.42.0.122" reject'

Laird Buildroot br2-external

The br2-external mechanism provides a convenient way to customize project specific configure files, packages etc. outside of the Buildroot source tree. Following is an example layout of Laird Buildroot br2-external tree:

|--Config.in
|--external.desc
|--external.mk
|--Makefile
|--configs/
|    |--buildroot_defconfig
|--board/
|    |--rootfs-additions/
|    |--kernel_defconfig
|    |--u-boot_defconfig
|    |--u-boot.scr
|    |--post_build.sh
|    |--post_image.sh
|    |--kernel-dts/
|    |    |--kernel.dts
|    |--uboot-dts/
|    |    |--uboot.dts
|--package/
|    |--package-1/
|    |    |--Config.in
|    |    |--package-1.mk
|    |    |--src/
|    |--package-2/
|         |--Config.in
|         |--package-2.mk
|         |--package-2.hash
|--README

external.desc: provides the name and an optional short description for the br2-external tree:

name: DEMO
desc: Laird Custom Project Demo

The full path of the br2-external tree will be set to BR2_EXTERNAL_$(NAME)_PATH automatically so that it can be used in both Buildroot Kconfig and Makefile. In this case BR2_EXTERNAL_DEMO_PATH is set to the full path of demo project.

Config.in: define custom package recipes with external.mk so they are visible to the top-level Buildroot configuration menu:

source "$BR2_EXTERNAL_DEMO_PATH/package/package-1/Config.in"
source "$BR2_EXTERNAL_DEMO_PATH/package/package-2/Config.in"

external.mk: define custom package recipes with Config.in so they can be built by the Buildroot make logic.

include $(sort $(wildcard $(BR2_EXTERNAL_DEMO_PATH)/package/*/*.mk))

Makefile: defines the build targets

TARGETS = demo

MK_DIR = $(realpath $(dir $(firstword $(MAKEFILE_LIST)))))
BR_DIR = $(realpath $(MK_DIR)/../buildroot)

include $(BR_DIR)/board/laird/build-rules.mk

Run make -C <custom-project-dir> demo to invoke the build.

configs: where Buildroot configure files for the custom project are saved. Run make menuconfig to configure kernel, u-boot, packages etc. Run make savedefconfig to save it.

board: where custom board support files and configures are saved, including kernel and u-boot configures, scripts, files need installed to the target rootfs, etc.

Board support files and configures

Board support files and configures are used to customize kernel, uboot, target rootfs and even the images.

To customize kernel with kernel_defconfig, go to Buildroot-->Kernel-->Configuration file path and set it to ($(BR2_EXTERNAL_$(DEMO)_PATH)/board/kernel_defconfig.

To customize uboot with uboot_defconfig, go to Buildroot-->Bootloaders-->Configuration file path and set it to ($(BR2_EXTERNAL_$(DEMO)_PATH)/board/u-boot_defconfig.

To customize kernel device tree with kernel.dts, go to Buildroot-->Kernel, enable Build a Device Tree Blob (DTB) and set Out-of-tree Device Tree Source file paths to ($(BR2_EXTERNAL_$(DEMO)_PATH)/board/kernel-dts/kernel.dts. If there are more than one dts files, a list of dts paths separated by space needs to be provided.

To customize uboot device tree with uboot.dts, go to Buildroot-->Bootloaders-->Device Tree Source file paths and set it to ($(BR2_EXTERNAL_$(DEMO)_PATH)/board/uboot-dts/uboot.dts. If there are more than one dts files, a list of dts paths separated by space needs to be provided.

To customize target rootfs, go to
Buildroot-->System configuration-->Custom scripts to run before creating filesystem images and set it to ($(BR2_EXTERNAL_$(DEMO)_PATH)/board/post_build.sh. A typical usage of post_build.sh is to install custom materials to the target rootfs:

1 rsync -rlptDWK --exclude=.empty "${BR2_EXTERNAL_DEMO_PATH}/board/rootfs-additions/" "${TARGET_DIR}"
2
3 ${BASE_DIR}/../../board/laird/som60/post_build.sh ${1} ${2}

Line 3 is to invoke the post-build script in the Buildroot source tree.

To customize images, go to Buildroot-->System configuration-->Custom scripts to run after creating filesystem images and set it to ($(BR2_EXTERNAL_$(DEMO)_PATH)/board/post_image.sh. However, this is not encouraged as it may result in corrupted images and boot faiulre.

To customize u-boot environment, the u-boot script file u-boot.scr in the Buildroot source tree needs to be overridden. This can be done by appending a line to the post_build.sh:

ln -rsf ${BR2_EXTERNAL_DEMO_PATH}/board/u-boot.scr ${BINARIES_DIR}/u-boot.scr

The custom u-boot.scr must contain following information to boot the device. New environment varaibles should be added to the begin of the script.

if test ${bootside} = a;
then
    setenv bootvol 1
else
    setenv bootvol 4
fi

ubi part ubi
ubi read 0x21000000 kernel_${bootside}
source 0x21000000:script
bootm 0x21000000

Add a custom package

Here is the instruction to add a custom package test, assuming its source code is saved in package/test/src:

  1. Go to the demo project and run mkdir -p package/test/;
  2. Go to package/test/ and create Config.in, which basically contains:
    config BR2_PACKAGE_TEST
        bool "test"
        depends on BR2_PACKAGE_LIBA
        select BR2_PACKAGE_LIBB
        help
            This is an example to add a package to the br2-external tree.
  1. Create test.mk, which describes how to copy/download, configure, build and install the package:
    TEST_VERSION = 1.0.0
    TEST_SITE = $(BR2_EXTERNAL_DEMO_PATH)/package/test/src
    TEST_SITE_METHOD = local
    TEST_LICENSE = GPL-3.0+
    TEST_LICENSE_FILES = COPYING
    TEST_DEPENDENCIES = liba libb

    define TEST_BUILD_CMDS
        $(MAKE) $(TARGET_CONFIGURE_OPTS) -C $(@D) all
    endef

    define TEST_INSTALL_TARGET_CMDS
        $(INSTALL) -D -m 0755 $(@D)/test $(TARGET_DIR)/usr/sbin/
    endef

    $(eval $(generic-package))
  1. Append following line to demo/Config.in

    source "$BR2_EXTERNAL_DEMO_PATH/package/test/Config.in"

Now the package can be configured and built within the Buildroot.
More information is available at (https://buildroot.org/downloads/manual/manual.html#adding-packages).

Add a package from manifest

To add a custom package from github etc, a project element needs to be added to the manifest file .repo/manifests/custom.xml:

<project path="demo/package/externals/test" name="test.git">

where externals/test is the place to save the source code of the package. If the project is in a repository other than the default, a remote element also needs to be added

<remote name="custom" fetch="ssh://[email protected]/custom" />
<project remote="custom" path="demo/package/externals/test" name="test.git" revision="custom-7.x.y.z">

Run repo sync to fetch the source code, and then follow the instruction of Add a custom package to make it configurable and buildable.

Filesystem

This section describes the filesystem and booting strategies for all products based on the 60 Series SOM. These products introduce a new set of features that changes the way the embedded Linux filesystems are built and programmed onto the embedded NAND flash.

What's Changed?

The following is a brief description of the changes from Laird's previous system on module (the WB50NBT) to the 60 Series SOM:

AT91Bootstrap has been replaced with U-Boot SPL
U-Boot is now programmed as a FIT (flattened device tree) image
The kernel is now programmed as FIT image
The kernel, rootfs, and overlay filesystems are now stored as UBI volumes inside of a single UBI disk
(Optional) FIT image signing is enabled to verify both U-Boot and the kernel
(Optional) DM-Verity is enabled to verify the data stored on the rootfs is authentic

Detailed Description

The following is a detail description of the new components & strategies.

U-Boot SPL

AT91Bootstrap has been replaced as the first-level bootloader, by U-Boot SPL. This is pretty much a transparent change for developers; the new bootloader resides in the same location in NAND flash (0x00000000 - 0x00020000). The change was made to enable the use of FIT image signing, so that U-Boot SPL can verify U-Boot.

U-Boot FIT Image

U-Boot (the second-stage bootloader) is now built as a FIT (flattened image tree) image. (For this reason, the name of the artifact is now u-boot.itb, which denotes a binary FIT image). This enables U-Boot to contain its own device tree, which can also be used to contain keys that are used to verify the kernel image.

Kernel FIT Image

The kernel is now built as a FIT (flattened image tree) image. (For this reason, the name of the artifact is now kernel.itb, which denotes a binary FIT image). The kernel's device tree is now contained within the FIT image, so that it can be signed inside of a configuration (verified using the key stored in U-Boot). The kernel FIT image may also contain a script file that is used to mount the rootfs. This is done so that the script can (optionally) verify the root hash of the rootfs (known as the "verity" hash, provided by the verity module). For this reason, when verity is enabled, the kernel FIT image and the rootfs are paired and must always be flashed together.

UBI Disk Layout

The kernel, rootfs, and overlay file systems are now stored as UBI ''volumes'' contained within a single (large) UBI disk. This change improves wear-leveling and fault tolerance, since the volumes are stored in blocks across the entire disk. This changes the way the kernel and rootfs are flashed on the device. Both U-Boot and the kernel are informed (via their device trees) of the single UBI disk. The volume information (size, etc.) is contained within the UBI disk itself (not stored in the device configuration). For convenience, U-Boot has a script, _ubiformat, which performs formatting of the UBI disk and creation of all the named volumes required for an A/B kernel and rootfs.

Tips & Tricks

Custom UBI Volume Sizes

Since the UBI volumes are dynamic (stored in the UBI disk), you can easily change the formatting (for example, if you don't need a "B" configuration during development, then you can reclaim that space for development use in the "A" side.) Just use the _ubiformat script as a template to format the UBI disk and create the partitions.