System programming has traditionally been dominated by languages like C, Go, and Rust. To evaluate Kotlin/Native in this domain, I developed a low-level container runtime that manages the container lifecycle by directly manipulating Linux features such as namespaces and cgroups.

In this session, I will start by briefly outlining how the LLVM-based compilation model enables Kotlin to execute without a JVM, providing the foundation for low-level system interaction.
I will then demonstrate how Kotlin’s strong type system and null safety bring reliability to low-level programming, while also highlighting the efficient C interoperability provided by Kotlin/Native.

I will then dive into the inherent tension between the Kotlin runtime and Linux system calls. Specifically, I will address the conflict between the multi-threaded nature of the Kotlin runtime and the single-threaded requirements of unshare or fork. I will explain the architectural pattern I implemented to resolve this compatibility issue.

Finally, I will share performance metrics, such as binary size and startup speed, to evaluate the overhead of Kotlin/Native realistically against comparable C, Go or Rust implementations.

By the end of this talk, you will understand the practical viability of Kotlin/Native for system programming. You will leave equipped with patterns for safe C interoperability and strategies for managing runtime characteristics in native applications.