Cross-Compiling memfaultd

Most embedded devices have very limited resources (disk, cpu, ram). Instead of building memfaultd directly on the device, it's usually a better idea to cross-compile it on a more powerful machine and then transfer the binary to the target system.

If you are using Yocto to build your system image, then we recommend using our meta-memfault layer to integrate memfaultd into your system image. Follow the Yocto Integration guide.

If you are not using Yocto, then this guide will help you cross-compile memfaultd from source for your target architecture.

This is the recommended way to compile and install memfaultd on Debian, Raspbian, Ubuntu, buildroot and OpenWRT.

Steps

Install the Rust cross tool

We recommend using the cross tool to quickly generate a binary compatible with your target system.

Install Rust and cross on your system:

# Install the rust toolchain on your system
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
# Install cross
cargo install cross

Docker is also required. For more detailed instructions, refer to Cross Getting Started.

Download and cross-compile memfaultd

# Clone the memfault-linux-sdk
git clone https://github.com/memfault/memfault-linux-sdk.git

# Change into the `memfaultd` source code directory
cd meta-memfault/recipes-memfault/memfaultd/files

# Cross-compile memfaultd for your target architecture
cross build --no-default-features --target aarch64-unknown-linux-gnu --release -F coredump,collectd,rust-tls

# Make the binary smaller by stripping debug symbols
strip target/aarch64-unknown-linux-gnu/release/memfaultd

tip

In this example, we are building for arm64 using the target is aarch64-unknown-linux-gnu.
For a list of all supported targets, see Cross Supported Targets.

Copy memfaultd to the target

scp target/aarch64-unknown-linux-gnu/release/memfaultd \
    target/aarch64-unknown-linux-gnu/release/memfaultctl \
    root@<target-ip>:/usr/bin/

scp target/aarch64-unknown-linux-gnu/release/memfault-core-handler \
    root@<target-ip>:/usr/sbin/

tip

The three binaries generated by the build are identical and will change their behavior depending on how they are called (like busybox programs). To save on disk space, you can copy only one of them and create symlinks for the other two.

Create a Project and get a Project Key

Go to app.memfault.com and from the "Select A Project" dropdown, click on "Create Project" to setup your first project such as "smart-sink-dev".

Once you've created your project, you'll be automatically taken to an page that includes your project key. Copy the key and follow the rest of this guide.

Add memfaultd configuration

To test memfaultd, you will also need:

• A configuration file /etc/memfaultd.conf

  For example:

  {
    "enable_data_collection": true,
    "software_version": "dev",
    "software_type": "dev",
    "project_key": "INSERT_YOUR_PROJECT_KEY_HERE",
    "persist_dir": "/tmp/memfaultd"
  }
  caution

  Replace INSERT_YOUR_PROJECT_KEY_HERE with your project key!

• A script called memfault-device-info which memfaultd will use to detect the hardware version and device id of your device.

  For example:

  #!/bin/sh
  
  echo MEMFAULT_HARDWARE_VERSION=smart-sink-hw-proto
  echo MEMFAULT_DEVICE_ID=dev-device-42

  Do not forget to make this script executable:

  chmod +x /usr/bin/memfault-device-info

Run memfaultd

memfaultd -c /etc/memfaultd.conf --daemonize

To automatically start memfaultd with your system, you can take inspiration from our systemd service file or our SysVinit init script.

Next Steps

• Verify that your device appears in Memfault dashboard
• Memfault will capture all coredumps.
  • Try triggering a coredump by running memfaultctl trigger-coredump and then memfaultctl sync.
  • Your coredump will appear in the dashboard.
• Add collectd to capture metrics (see our sample configuration)
• Add application specific metrics to your system
• Add fluent-bit to capture logs
• Start thinking about how you will update your devices!