Embedded Box, containerized development environments for embedded and FPGA designers
ℹ️ This is a fresh project and currently includes only a limited number of tools, but it's actively under development—so stay tuned for more updates!
✅ Docker and Podman compatible
✅ Suitable for automated CI/CD run and interactive development environment (GUI)
✅ MIT License, feel free to use and contribute ;)
Containerizing software is a standard approach for creating standalone packages where the correct dependencies are included to run the desired application. It is a mature technology that has been developed over many years and is widely used in production—both as a development and deployment environment—primarily associated with Docker.
Tools used by embedded and FPGA developers are no exception. Many EDA tools
depend on a wide range of packages, and maintaining multiple versions on a
single workstation or Linux distribution can be challenging due to conflicting
dependencies. For example, each version of Xilinx's PetaLinux tool may require a
different version of Ubuntu and its corresponding apt packages. This makes it
difficult to install and use multiple versions of PetaLinux on the same
machine—some form of isolation or virtualization is needed.
Furthermore, installing the same tool on different machines can be problematic due to minor configuration differences. Some tools require small system-specific tweaks during installation, and if someone tries to install the same software on a different machine—perhaps a year later—it’s easy to forget what was done previously. As a result, maintaining a reproducible development environment becomes challenging, especially when working on multiple projects and using a variety of tools in different versions.
In recent years, Continuous Integration and Delivery/Deployment (CI/CD) has become a popular concept in the embedded and FPGA world. In this model, code is managed under a version control system like Git, and with each commit or merge request, a set of predefined unit and integration tests are automatically triggered. This allows developers to quickly see whether the proposed changes introduce any test failures or if everything passes as expected.
Automation in CI/CD goes beyond testing—it can also be used to generate final output artifacts, such as firmware or bitstreams, as part of Continuous Delivery or Deployment. However, achieving this requires running project-specific tools like cross-compilers or RTL synthesizers. One approach is to manually install these tools on backend machines, much like setting up a local development environment, and connect them to CI systems like GitHub or GitLab.
Unfortunately, this method suffers from the same reproducibility issues mentioned earlier. Additionally, scalability becomes a problem. As project demands grow, more backend machines may be required. Manually installing and configuring tools on each new machine is not a scalable solution.
Containerizing the required software addresses both problems. With containerized environments, it's much easier to reproduce builds and scale the backend dynamically. Generic runners capable of executing any containerized application can be added as needed, simplifying infrastructure management and improving reliability.
This is where the project EBox—short for Embedded Box or EDA Box, depending on how you prefer to describe it—comes in. The goal of this project is to demonstrate how various tools can be containerized and, where possible, to provide ready-to-use containerized versions of those tools.
I hope this project helps developers like myself streamline their workflows when working with embedded and FPGA toolchains.
Although the EBox project aims to provide ready-to-use container images that suit a wide range of developers and use cases, there are certain limitations—both technical and legal—that need to be considered.
❗ Downloading Xilinx (now AMD) tools requires legal approval by the user, and as such, it is not permitted to distribute containerized versions of these tools on public platforms like Docker Hub. Instead, this repository provides the necessary files and instructions to containerize the tools yourself—assuming you have already obtained the official setup files from the vendor’s website. While all supporting scripts and configurations are included in the repository, you must manually download the setup files from the vendor to build the images.
❗ Some tools, such as Xilinx's Vivado, are extremely large—container images for the latest versions can easily exceed 100 GB. Even if it were legally permissible to distribute these images, hosting them in a private or public manner would be technically challenging and potentially expensive due to their size. Unfortunately, this limitation also applies to you: if you choose to build these images yourself, you'll need sufficient storage infrastructure to host and manage them.
❗ It's practically impossible to test every possible usage scenario for these containers. While many have been tested with a variety of cases—for example, MicroBlaze, Zynq, and ZynqMP projects using Xilinx Vivado—the test coverage is still limited. As a result, you may encounter issues in certain use cases. If you do, please don’t hesitate to open an issue.
❗ These container images are primarily intended for use on Linux hosts—that is, systems where Docker or Podman runs natively on Linux. While they may also work on other platforms, such as Docker Desktop on Windows or macOS, they have only been tested on Linux. In particular, you may encounter issues when using GUI applications on non-Linux systems.
Simply visit the directory for the corresponding tool. Each directory contains a detailed README file with step-by-step instructions. If you have any questions or encounter issues, feel free to open an issue or ask in the community tab. You’re also welcome to contribute by creating pull requests.
- Mock build and test (if possible)
- Docker linter
- Think about https://github.com/mviereck/x11docker
This work would not have been possible without the dedicated efforts and valuable suggestions of all AUTHORS.
MIT License, Copyright (c) 2025 The EBox Authors.