The Learning-Directed OS project is developing the next-generation Machine Learning-based Operating System to drive computing infrastructure toward high efficiency and performance (see https://ldos.utexas.edu/ for more information).
The project is developing a kernel based on the Asterinas framekernel. We are not affiliated with the original authors of Asterinas and we expect our work to diverge from theirs due to very different research goals. However, the original authors deserve a huge amount of credit for developing such an impressive system. Our work would not be possible without their generosity in making Asterinas freely available.
Please do not report issues to the original Asterinas team or ask them questions about this repository. They are not responsible for any of the work done in this repository.
Toward a production-grade Linux alternative—memory safe, high-performance, and more
The future of operating systems (OSes) belongs to Rust—a modern systems programming language (PL) that delivers safety, efficiency, and productivity at once. The open question is not whether OS kernels should transition from C to Rust, but how we get there.
Linux follows an incremental path. While the Rust for Linux project has successfully integrated Rust as an official second PL, this approach faces inherent friction. As a newcomer within a massive C codebase, Rust must often compromise on safety, efficiency, clarity, and ergonomics to maintain compatibility with legacy structures. And while new Rust code can improve what it touches, it cannot retroactively eliminate vulnerabilities in decades of existing C code.
Asterinas takes a clean-slate approach. By building a Linux-compatible, general-purpose OS kernel from the ground up in Rust, we are liberated from the constraints of a legacy C codebase—its interfaces, designs, and assumptions—and from the need to preserve historical compatibility for outdated platforms. Languages—including PLs—shape our way of thinking. Through the lens of a modern PL, Asterinas rethinks and modernizes the construction of OS kernels:
-
Modern architecture. Asterinas pioneers the framekernel architecture, combining monolithic-kernel performance with microkernel-inspired separation. Unsafe Rust is confined to a small, auditable framework called OSTD, while the rest of the kernel is written in safe Rust, keeping the memory-safety TCB intentionally minimal.
-
Modern design. Asterinas learns from Linux's hard-won engineering lessons, but it is not afraid to deviate when the design warrants it. For example, Asterinas improves the CPU scalability of its memory management subsystem with a novel scheme called CortenMM.
-
Modern code. Asterinas's codebase prioritizes safety, clarity, and maintainability. Performance is pursued aggressively, but never by compromising safety guarantees. Readability is treated as a feature, not a luxury, and the codebase is structured to avoid hidden, cross-module coupling.
-
Modern tooling. Asterinas ships a purpose-built toolkit, OSDK, to facilitate building, running, and testing Rust kernels or kernel components. Powered by OSTD, OSDK makes kernel development as easy and fluid as writing a standard Rust application, eliminating the traditional friction of OS engineering.
Asterinas aims to become a production-grade, memory-safe Linux alternative, with performance that matches Linux—and in some scenarios, exceeds it. The project has been under active development for four years, supports 230+ Linux system calls, and has launched an experimental distribution, Asterinas NixOS.
In 2026, our priority is to advance project maturity toward production readiness, specifically targeting standard and confidential virtual machines on x86-64. Looking ahead, we will continue to expand functionality and harden the system for mission-critical deployments in data centers, autonomous vehicles, and embodied AI.
Asterinas targets modern, 64-bit platforms only.
A development platform is where you build and test Asterinas (i.e., the host machine running the Docker-based development environment).
| Development Platform |
|---|
| x86-64 |
| ARM64 |
A deployment platform is a CPU architecture that Asterinas can run on as an OS kernel.
| Deployment Platform | Tier |
|---|---|
| x86-64 | Tier 1 |
| x86-64 (Intel TDX) | Tier 2 |
| RISC-V 64 | Tier 2 |
| LoongArch 64 | Tier 3 |
Tier definitions:
- Tier 1: Fully supported and tested. CI runs the full test suite on every PR.
- Tier 2: Actively developed with basic functionality working. CI runs build checks and basic tests on a regular basis (per PR for RISC-V and nightly for Intel TDX), but the full test suite is not yet covered.
- Tier 3: Early-stage or experimental. The kernel can boot and perform basic operations, but CI coverage is limited and may not include automated runtime tests for every pull request.
We provide Asterinas NixOS ISO Installer to make the Asterinas kernel more accessible for early adopters and enthusiasts. We encourage you to try out Asterinas NixOS and share feedback. Instructions on how to use the ISO installer can be found here.
Disclaimer: Asterinas is an independent, community-led project. Asterinas NixOS is not an official NixOS project and has no affiliation with the NixOS Foundation. No sponsorship or endorsement is implied.
Follow the steps below to get Asterinas up and running.
-
Download the latest source code on an x86-64 (or ARM64) Linux machine:
git clone https://github.com/ldos-project/asterinas
-
Run a Docker container as the development environment:
make docker
Alternatively, if you use VS Code with the Dev Containers extension, open the cloned folder and select "Reopen in Container".
-
Inside the container, go to the project folder (
/root/asterinas) and run:make kernel make run_kernel
This results in a VM running the Asterinas kernel with a small initramfs.
-
To install and test real-world applications on Asterinas, build and run Asterinas NixOS in a VM:
make nixos make run_nixos
This boots into an interactive shell in Asterinas NixOS, where you can use Nix to install and try more packages.
Mariposa can be compiled without ORPC support and without the LDOS features. This is used for as a baseline for benchmarking the overhead of LDOS features.
This is controlled by the baseline_asterinas configuration flag. This flag is off by default.
To build in baseline mode:
make BASELINE_ASTERINAS=1 ... arguments as usual ...To configure VSCode to provide IDE features based on the baseline build, see the comments in the
editor-config/vscode/settings.json file.
Kernel code that needs to be different in the baseline and the Mariposa kernels should use:
#[cfg(not(baseline_asterinas))] and #[cfg(baseline_asterinas)] as appropriate. The
#[path = "..."] attribute may also be useful, though it's use should be kept to a minimum. When it
is used the baseline variant should be named {original}_baseline (e.g., the orpc module stubs
for the baseline are in orpc_baseline.rs).
See The Asterinas Book to learn more about the project.
Asterinas's source code and documentation primarily use the Mozilla Public License (MPL), Version 2.0. Select components are under more permissive licenses, detailed here. For the rationales behind the choice of MPL, see here.