Ahn, Junho 

Research Interests
Summary
System Security, Operating Systems, Extended Berkeley Packet Filter (eBPF), Memory Management

My research focuses on building secure, efficient, and programmable operating-system mechanisms for memory management and isolation. From these sub-goals, I target enhancing Linux kernel with eBPF to bridge the semantic gap between the user and the kernel in a secure environment. It is expected to oversee novel application usage with a co-design model involving both applications and the kernel.

Extending Memory Management
First, I design programmable and extensible memory-management frameworks that allow applications to express specialized memory behavior while preserving the safety and correctness guarantees of the operating system. Monolithic kernels, such as Linux, provide a fixed set of memory-management mechanisms that are designed to meet the needs of a wide range of applications. However, this one-size-fits-all approach can lead to inefficiencies and missed opportunities for optimization, especially for applications with unique memory requirements or access patterns. To address this challenge, I study how eBPF can bridge the semantic gap between applications and the operating system. In recent work, I explored bypassing parts of the kernel virtual-memory stack to enable efficient and transparent detection of use-after-free bugs in user-space applications. I am also involved in a project that rethinks operating-system design using Rust, leveraging language-based isolation to build systems that are both efficient and safe. In this direction, I study how to move memory management beyond fixed kernel mechanisms and toward a co-design model between applications and the kernel.

Defeating Memory Safety Bugs
Second, I investigate strong memory safety through memory programmability. Use-after-free (UAF) bugs are a critical class of vulnerabilities that pose serious security threats. However, existing prevention and detection techniques often incur substantial performance and memory overheads. In my recent work, I proposed an enhanced one-time allocator and garbage-collection mechanism for UAF prevention and detection using an extended eBPF-based design. This approach reduces these overheads by tightly integrating operating-system support with a novel user-space design.

Exploring System Optimization through Programmable Memory Management
Third, I explore system optimization through programmable memory management. Building on the same abstraction, I study how programmable memory mechanisms can enable broader optimizations, including efficient sanitization, flexible huge-page management, zero-copy I/O, and low-overhead isolation.

From these sub-goals, I target enhancing Linux kernel with eBPF to bridge the semantic gap between the user and the kernel in a secure environment. It is expected to oversee novel application usage with a co-design model involving both applications and the kernel

Education

KAIST, Dajeon, Republic of Korea Mar 2023 -

 Ph.D. Student, School of Computing

 • Advisor: Youngjin Kwon

KAIST, Dajeon, Republic of Korea Mar 2021 - Feb 2023

 Mater of Science, School of Computing

 • Advisor: Youngjin Kwon

 • Thesis: vBPF: safely extending eBPF to enhance programmability and flexibility

GIST, Gwangju, Republic of Korea Mar 2017 - Feb 2021 

 Bachelor of Science, School of Electrical Engineering and Computer Science