Skip to content

Latest commit

 

History

History
297 lines (230 loc) · 10.6 KB

README.md

File metadata and controls

297 lines (230 loc) · 10.6 KB

agl-renesas

export BUILD_DEVICE=agl-renesas
bitbake aws-biga-image

BE CARFUL - THIS TUTORIAL IS OUTDATED, at least for the build instructions!

AWS IoT Greengrass for the Renesas R-Car H3 and Automotive Grade Linux

You may perform the build steps on an EC2 instance, copy the image locally, and then flash the MicroSD locally. The steps in this tutorial expects you are working on a local workstation.

The majority of the steps were gleaned from Building for Supported Renesas Boards in the AGL documentation. We repeat specific steps here to demonstrate the exact steps used and where meta-aws was inserted to the build procedure so you can take it into account when you build your own.

NOTE: We are building with multimedia features that you might not want to use and this will be called out during the procedure.

Hardware used

  • 1x Renesas R-Car H3 WS2.0, RTP0RC77951SKBX010SA00, S/N 2236
  • 1x 5V 8A Adapter, 5.5mm barrel plug
  • 1x 8GB SanDisk Ultra microSD
  • 1x micro USB cable

The build result will be run from microSD. It will be slower than running from eMMC, but less involved procedurally.

Preparation

You will need to complete all preparation steps to complete all the sections in this tutorial successfully.

  1. Perform all workstation preparation steps as defined in the Yocto Mega Manual.
  2. An AWS Account and an AWS Identity and Access Management (IAM) with authorization to run the IoT Greengrass Tutorial in the context of your logged in IAM User.

Prepare, invoke, and flash the build

Due to the size of this build, the build environment was defined in the AWS Cloud using a c5.18xlarge instance type. Remember that in the AWS Cloud you pay for what you use, so to be thrifty consider stopping or terminating your instance after performing the build tasks.

  1. Launch the EC2 instance with the Ubuntu 18.04 AMI.

  2. Login to the EC2 instance with ssh and prepare system updates.

    sudo apt-get update
sudo apt-get -y upgrade

  3. In the EC2 instance terminal, perform steps on the Downloading AGL Software page.

    The command line steps we used are here. Please reference the documentation if you would like an explanation of steps.

    export AGL_TOP=$HOME/AGL
echo 'export AGL_TOP=$HOME/AGL' >> $HOME/.bashrc
mkdir -p $AGL_TOP

mkdir -p $HOME/bin
export PATH=$HOME/bin:$PATH
echo 'export PATH=$HOME/bin:$PATH' >> $HOME/.bashrc
curl https://storage.googleapis.com/git-repo-downloads/repo > $HOME/bin/repo
chmod a+x $HOME/bin/repo

    We will be using the Jumping Jellyfish release since it aligns with the Yocto Project dunfell LTS release.

    cd $AGL_TOP
mkdir jellyfish
cd jellyfish
repo init -b jellyfish -m jellyfish_10.0.0.xml -u https://gerrit.automotivelinux.org/gerrit/AGL/AGL-repo
repo sync

  4. Now that we have the basics setup, there are some Renesas specific parts to organize. On your local workstation, download the Renesas proprietary drivers by following the download instructions in the section Downloading Proprietary Drivers.

    To identify the drivers you will need, invoke this command within the EC2 build ennvironment.

    grep -rn ZIP_.= $AGL_TOP/jellyfish/meta-agl/meta-agl-bsp/meta-rcar-gen3/scripts/setup_mm_packages.sh

    At the time of writing, the command output:

    3:ZIP_1="R-Car_Gen3_Series_Evaluation_Software_Package_for_Linux-weston8-20191206.zip"
4:ZIP_2="R-Car_Gen3_Series_Evaluation_Software_Package_of_Linux_Drivers-weston8-20191021.zip"

    Which are the v3.21.0 payloads.

    Please follow the rest of the steps to download from Renesas to your workstation (in this case it is for YP3.1 Dunfell), upload to the EC2 instance, and then configure.

    We did not try the non-MMP v4.1.0 version.

    From your workstation, upload the files. Replace the 'x' characters with real values according to your situation.

     scp -i ~/.ssh/xxx.pem ~/Downloads/R-Car*.zip [email protected]:

    This is how we handled the placement of the files:

    ssh -i ~/.ssh/xxx.pem [email protected]:
mkdir XDG_DOWNLOAD_DIR
mv R*zip XDG_DOWNLOAD_DIR
export XDG_DOWNLOAD_DIR=$HOME/XDG_DOWNLOAD_DIR
chmod a+x $XDG_DOWNLOAD_DIR/*.zip

  5. In our build, we will be building for the AGL telematics demo in meta-agl-telematics-demo. To set this up easily for the R-Car, initialize the build environment with the new aglsetup.sh utility. Its full use is described in the Initializing Your Build Environment section of the AGL documentation.

    cd $AGL_TOP/jellyfish
source meta-agl/scripts/aglsetup.sh \
    -m h3ulcb \
    -b h3ulcb \
    agl-demo agl-devel

cd $AGL_TOP/jellyfish/bsp/meta-renesas
sh meta-rcar-gen3/docs/sample/copyscript/copy_evaproprietary_softwares.sh -f ${XDG_DOWNLOAD_DIR}

  6. At this point, the base image is configured but we need to add AWS IoT Greengrass.

    cd $AGL_TOP/jellyfish
git clone -b dunfell https://github.com/aws/meta-aws
bitbake-layers add-layer $AGL_TOP/jellyfish/meta-aws

  7. Edit the passwd and group file in meta-agl-profile-core to overcome aws4embeddedlinux/meta-aws#75

  8. Add the following to local.conf by invoking this command.

    echo "IMAGE_INSTALL_append = \"greengrass openssh ntp\"" >> \
 $AGL_TOP/jellyfish/h3ulcb/conf/local.conf
echo "MACHINE_FEATURES_append = \" multimedia\"" >> $AGL_TOP/jellyfish/h3ulcb/conf/local.conf
echo "DISTRO_FEATURES_append = \" use_eva_pkg"\" >> $AGL_TOP/jellyfish/h3ulcb/conf/local.conf

  9. Invoke the build.

    bitbake agl-demo-platform

  10. Download files. The dtb and Image files are in /boot in the tarball. Because of this, the Image and dtb load from the root file system.

    The .srec file download and application to QuickFlash or QSPI has to be done only once and is here for visibility. For iterative builds, downloading the .tar.bz2 is sufficient.

cd ~/Downloads
BASEDIR=AGL/jellyfish/h3ulcb/tmp/deploy/images/h3ulcb
FILES="bootparam_sa0.srec
       bl2-h3ulcb.srec
       cert_header_sa6.srec
       bl31-h3ulcb.srec
       tee-h3ulcb.srec
       u-boot-elf-h3ulcb.srec
       agl-demo-platform-h3ulcb.tar.bz2"
for FILE in $FILES; do
    scp -i  ~/.ssh/xxx.pem [email protected]:$BASEDIR/$FILE .
done

  1. Partition the microSD card which only has to be done once per microSD card. Essentially, you will delete all partitions and create one large partition.

    WARNING Make sure you are using the microSD card you intend to partition by checking the device first with lsblk!

    sudo fdisk /dev/mmcblk0
=> (use 'd' to delete all partitions)
=> (use 'n' to create a new rootfs partition, taking all space)
=> (use 'l' and '83' to label the partition as Linux)
=> w

  2. Create the partition and lay down the rootfs. Important: do not forget to sync filesystems.

    sudo ext3.mkfs /dev/mmcblk0p1
sudo mount /dev/mmcblk0p1 /mnt
# note where your mmc is mounted
sudo tar xfj agl-demo-platform-h3ulcb.tar.bz2 -C /mnt && sync

    NOTE for simplicity, at this point you can also manually provision your AWS IoT Greengrass credentials by copying them directly to the root file system. This is not the way to provision in production which will be covered in a different tutorial. If you are unsure of how to provision AWS IoT Greengrass credentials, see the IoT Greengrass documentation.

    Unmount the microSD, eject the microSD, and insert the microSD to the R-Car.

  3. For iterative development, TFTP/NFS/microSD image provisioning is the most convenient. In this case, we are using microSD so we need to change the U-Boot commands accordingly. This has to be done only when you desire to change the boot target and any variant (kernel, dtb, etc).

    1. Connect over USB with your favorite terminal tool at 115200/8/N/1.

    2. Press SW8 to reboot the device and press the space bar to stop automatic booting when U-Boot runs.

    3. Run the following commands at the U-Boot prompt. These commands were taken from the Renesas / AGL guide and worked great.

      

      4. 

    

    4. 



=> setenv bootargs console=ttySC0,115200 ignore_loglevel vmalloc=384M video=HDMI-A-1:1920x1080-32@60 root=/dev/mmcblk1p1 rw rootfstype=ext4 rootwait rootdelay=2
=> setenv bootcmd run load_ker; run load_dtb; booti 0x48080000 - 0x48000000
=> setenv load_ker ext4load mmc 0:1 0x48080000 /boot/Image
=> setenv load_dtb ext4load mmc 0:1 0x48000000 /boot/r8a7795-h3ulcb.dtb
=> saveenv
```


Commission the network (already done if using Ethernet) and provision
the device with AWS Cloud credentials and AWS IoT Greengrass. See the
Greengrass documentation for more details.


Go build solutions


Your image with IoT Greengrass is now up and running.  What's next?


When you go through the IoT Greengrass
Tutorial,
Do not run the Quick Start or Module 1.  You will need to follow the
Configure AWS IoT Greengrass on AWS
IoT
chapter to initialize your account for IoT Greengrass, create your
Greengrass Group, and then apply the config.json and credentials to
the target device.


End notes


SMACK settings must me modified on the system.  In the near future
this will be fixed in meta-aws but right now it must be done
manually.  Also, the permissions will be narrowed.


Login to the r-car via serial and issue the following command:


sed -i 's/System _ -----l/System _ -wx--l/g' /etc/smack/accesses.d/default-access-domains


Reboot for the changes to go into effect.


© 2020, Amazon Web Services, Inc. or its affiliates. All rights reserved.