Yocto Gatesgarth Release for i.MX8 Platforms

Published on March 17, 2021

We are pleased to announce a new Yocto release Gatesgarth for our Nitrogen8 family of SBCs and SOMs based on i.MX8 processors. This release includes our latest 5.4 kernel. Below you will find download links for the images as well as detailed instructions for building including a features set.

For the Impatient

You can download the Yocto images from here: As usual, you'll need to register on our site and agree to the EULA because it contains NXP content.

How to Burn

You can program the SW to eMMC using the instructions below: programming-emmc-on-i-mx-platforms You can also program the SW to SD Card or USB Stick via zcat and dd under Linux: ~$ zcat *boundary-image*.wic.gz | sudo dd of=/dev/sdX bs=1M In addition, you can use the balenaEtcher utility to flash the eMMC, SD Card or USB stick via Windows or Linux: balenaEtcher

Build procedure

This image uses the gatesgarth branch of our boundary-bsp-platform repository. To build the image, we recommend using a Docker Container so that you can build with a reproducible and stable build environment. Otherwise, you’ll need these packages installed as well as this repo tool that can be installed like this: ~$ sudo apt-get install repo Then create your build directory and initialize everything. ~$ mkdir ~/gatesgarth && cd gatesgarth ~/gatesgarth$ repo init -u https://github.com/boundarydevices/boundary-bsp-platform -b gatesgarth ~/gatesgarth$ repo sync Next, setup the environment for building. For this image we will be building our boundary-xwayland distro for the target machine ~/gatesgarth$ MACHINE=<MACHINE> DISTRO=boundary-xwayland . setup-environment build Now bitbake boundary-image-multimedia-full which is equivalent to fsl-image-multimedia-full with Boundary-specific packages added such as BD-SDMAC support. ~/gatesgarth/build$ bitbake boundary-image-multimedia-full After some time this should build the same image as above, with the layers being at commits as per the time when repo sync was executed. If you are interested in each project revision at the time of the build, you can find a frozen manifest for those images here. The image file will deploy to tmp/deploy/images/{MACHINE}/boundary-image-multimedia-full-{MACHINE}.wic.gz.

Features list

The image built above contains the following components:
  • Linux kernel 5.4.x_2.3.0
  • U-Boot 2020.10
  • Weston 8.0.0 for i.MX
  • GStreamer 1.16 for i.MX
  • GPU Vivante libraries 6.4.3p1.0
  • VPU Hantro libraries v1.19.0
  • ISP VVCAM v4.2.2.2
  • qcacld-lea-2.0 Wi-Fi driver for BD-SDMAC
  • BlueZ 5.55 with support for BD-SDMAC
The next sub-sections will describe how to test most features.

Display support

Please make sure your platform includes the latest U-Boot: This version of U-Boot supports the display configuration, allowing to use any of the following displays:

GPU acceleration

As usual, in order to test the GPU you can use the example apps provided by Vivante: root@<MACHINE>:~# /opt/imx-gpu-sdk/GLES2/Blur/GLES2.Blur_Wayland -d

Camera input

Camera MIPI-CSI input can be checked using our OV5640 MIPI with GStreamer: root@<MACHINE>:~# gst-launch-1.0 v4l2src device=/dev/video0 ! \ video/x-raw,width=1280,height=720 ! waylandsink


Basler camera input

This build fully supports the Basler daA3840 8MP camera from Basler when using our Nitrogen 8M Plus which is part of our Evaluation Kit: The isp-vvcam driver and imx8-isp service are loaded automatically when the camera is detected. From there a simple GStreamer pipeline will allow you to see the stream: root@nitrogen8mp:~# gst-launch-1.0 -v v4l2src device=/dev/video0 ! waylandsink ... [ 352.348796] wdr3 res: 1920 1080 [ 352.352471] enter isp_mi_start [ 357.581179] ###### 62.42 fps ###### [ 362.771924] ###### 62.42 fps ######


Once the eth0 interface is up, you can use iperf3 to check Ethernet performances: root@<MACHINE>:~# iperf3 -c Connecting to host, port 5201 [ 5] local port 32880 connected to port 5201 [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.00 sec 1.09 GBytes 938 Mbits/sec 0 sender [ 5] 0.00-10.04 sec 1.09 GBytes 932 Mbits/sec receiver


Same goes for the Wi-Fi that can be tested just as easily: root@<MACHINE>:~# nmcli d wifi connect <network_name> password <password> root@<MACHINE>:~# iw wlan0 link Connected to a4:3e:51:08:54:f6 (on wlan0) SSID: Jabu_5GHz freq: 5240 RX: 3243 bytes (31 packets) TX: 9117 bytes (48 packets) signal: -79 dBm tx bitrate: 15.0 MBit/s MCS 0 40MHz short GI root@<MACHINE>:~# ping google.com -Iwlan0 PING google.com ( 56 data bytes 64 bytes from seq=0 ttl=55 time=3.470 ms ...


For products with a Silex bluetooth module, you'll be able to connect using our handy silex-uart script with the following commands: root@<MACHINE>:~# /usr/share/silex-uart/silex-uart.sh start Starting silex-uart rfkill on/off cycle. silex found root@<MACHINE>:~# hciconfig hci0 up root@<MACHINE>:~# hcitool scan Scanning ... 11:22:DE:AD:BE:EF    Some Device

VPU decoding

If your platform supports VPU decoding, here is an example on how to test it using the gplay tool: root@<MACHINE>:~# wget http://linode.boundarydevices.com/videos/Hobbit-1080p.mov root@<MACHINE>:~# gst-launch-1.0 playbin uri=file:///home/root/Hobbit-1080p.mov video-sink=autovideosink

VPU encoding

Here is a simple example that shows how to encode a video stream from the camera into H.264 using the VPU encoder: root@<MACHINE>:~# gst-launch-1.0 -v -e v4l2src device=/dev/video0 ! 'video/x-raw,width=1920,height=1080' \ ! vpuenc_h264 ! filesink location=test.h264 ^C root@<MACHINE>:~# gst-launch-1.0 filesrc location=test.h264 typefind=true ! 'video/x-h264' ! \ h264parse ! vpudec ! waylandsink


For platforms with CAN, you'll be able to bring up the interface(s) using this following command: root@<MACHINE>:~# ip link set can0 up type can bitrate 500000
From this point, you can use commands such as cansend and candump to send or display messages on the bus respectively.  

NPU support

NPU support is fully integrated into this build when using our Nitrogen 8M Plus with TensorFlowLite and ARMNN support of the NPU. Various demos to test this can be found here: IMX-MACHINE-LEARNING-UG.pdf We can run a TensorFlowLite demo from section 5.3.3 of the document: root@nitrogen8mp:~/ArmnnTests# /usr/bin/TfLiteInceptionV3Quantized-Armnn --data-dir=/home/root/ArmnnTests/data --model-dir=/home/root/ArmnnTests/models Info: ArmNN v20200200 Info: = Prediction values for test #0 Info: Top(1) prediction is 209 with value: 8.43322 Info: Top(2) prediction is 208 with value: 7.74945 Info: Top(3) prediction is 223 with value: 4.17862 Info: Top(4) prediction is 853 with value: 3.6468 Info: Top(5) prediction is 160 with value: 3.41887 Info: = Prediction values for test #1 Info: Top(1) prediction is 283 with value: 8.73712 Info: Top(2) prediction is 282 with value: 7.21762 Info: Top(3) prediction is 286 with value: 6.6858 Info: Top(4) prediction is 288 with value: 3.49485 Info: Top(5) prediction is 754 with value: 2.27925 Info: = Prediction values for test #2 Info: Top(1) prediction is 3 with value: 11.8521 Info: Top(2) prediction is 148 with value: 4.33057 Info: Top(3) prediction is 234 with value: 3.11497 Info: Top(4) prediction is 149 with value: 2.96302 Info: Top(5) prediction is 4 with value: 2.35522 Info: Total time for 3 test cases: 1.385 seconds Info: Average time per test case: 461.587 ms Info: Overall accuracy: 1.000   If you have any issues, please email support@boundarydevices.com