VTechWorks staff will be away for the Thanksgiving holiday beginning at noon on Wednesday, November 27, through Friday, November 29. We will resume normal operations on Monday, December 2. Thank you for your patience.

Browsing by Author "Yom, Jinwoo"

Now showing 1 - 2 of 2
  • Thumbnail Image
    HyperSpace: Data-Value Integrity for Securing Software
    Yom, Jinwoo (Virginia Tech, 2020-05-19)
    Most modern software attacks are rooted in memory corruption vulnerabilities. They redirect security-sensitive data values (e.g., return address, function pointer, and heap metadata) to an unintended value. Current state-of-the-art policies, such as Data-Flow Integrity (DFI) and Control-Flow Integrity (CFI), are effective but often struggle to balance precision, generality, and runtime overhead. In this thesis, we propose Data-Value Integrity (DVI), a new defense policy that enforces the integrity of "data value" for security-sensitive control and non-control data. DVI breaks an essential step of memory corruption based attacks by asserting the compromised security-sensitive data value. To show the efficacy of DVI, we present HyperSpace, a prototype that enforces DVI to provide four representative security mechanisms. These include Code Pointer Separation (DVI-CPS) and Code Pointer Integrity (DVI-CPI) based on HyperSpace. We evaluate HyperSpace with SPEC CPU2006 and real-world servers. We also test HyperSpace against memory corruption based attacks, including three real-world exploits and six attacks that bypass existing defenses. Our evaluation shows that HyperSpace successfully detects all attacks and introduces low runtime performance and memory overhead: 1.02% and 6.35% performance overhead for DVI-CPS and DVI-CPI, respectively, and overall approximately 15% memory overhead.
  • Thumbnail Image
    Securely Sharing Randomized Code That Flies
    Jelesnianski, Christopher; Yom, Jinwoo; Min, Changwoo; Jang, Yeongjin (ACM, 2022-09-12)
    Address space layout randomization was a great role model, being a light-weight defense technique that could prevent early return-oriented programming attacks. Simple yet effective, address space layout randomization was quickly widely adopted. Conversely, today only a trickle of defense techniques arebeing integrated or adopted mainstream. As code reuse attacks have evolved in complexity, defenses have strived to keep up. However, to do so, many have had to take unfavorable tradeoffs like using background threads or protecting only a subset of sensitive code. In reality, these tradeoffs were unavoidable steps necessary to improve the strength of the state of the art. In this article, we present Mardu, an on-demand system-wide runtime re-randomization technique capable of scalable protection of application as well as shared library code that most defenses have forgone. We achieve code sharing with diversification by implementing reactive and scalable rather than continuous or one-time diversification. Enabling code sharing further removes redundant computation like tracking and patching, along with memory overheads required by prior randomization techniques. In its baseline state, the code transformations needed for Mardu security hardening incur a reasonable performance overhead of 5.5% on SPEC and minimal degradation of 4.4% in NGINX, demonstrating its applicability to both compute-intensive and scalable real-world applications. Even when under attack, Mardu only adds from less than 1% to up to 15% depending on application size and complexity.