Controlled Preemption: Amplifying Side-Channel Attacks from Userspace
Yongye Zhu, Boru Chen, Zirui Neil Zhao, Christopher W. Fletcher
摘要
Microarchitectural side channels are an ongoing threat in today's systems. Yet, many side-channel methodologies suffer from low temporal resolution measurement, which can either preclude or significantly complicate an attack.
This paper introduces Controlled Preemption, an attack primitive enabling a single unprivileged (user-level) attacker thread to repeatedly preempt a victim thread after colocating with that victim thread on the same logical core. Between preemptions, the victim thread executes zero to several instructions-sufficiently few to enable high-resolution side channel measurements.
The key idea in Controlled Preemption is to exploit scheduler fairness heuristics. Namely, that modern thread schedulers give a thread 𝐴 the ability to preempt another thread 𝐵 until a fairness tripwire (signaling that 𝐴 is starving 𝐵) fires. We show how this idea enables hundreds of short preemptions before tripping the fairness tripwire is robust to noise and applies to both the Linux CFS and EEVDF schedulers. We also develop a technique that helps colocate the attacker and victim threads onto the same logical core, an attacker capability overlooked by prior work.
Our evaluation tests Controlled Preemption in the context of several different victim programs, victim privilege levels (inside and outside of Intel SGX) and choices of side channel. In each attack, we demonstrate results that are competitive with prior work but make fewer assumptions (e.g., require
问问这篇 Paper
智能体会读完全文。
Lune 把这篇 Paper 索引到了每一个公式,引用它的顶会 Paper 也一样。你提问,回答直接引用原文。
引用它的顶会 Paper3
- SSBench: Automated Characterization of Memory Dependence Predictors on Modern CPUsChang Liu, Yu Jin, Yuchen Fan, Tianrui Xiao 等ISCA 2026 · 被引用 1 次
- OCCUPY+PROBE: Cross-Privilege Branch Target Buffer Side-Channel Attacks at Instruction GranularityKaiyuan Rong, Junqi Fang, Haixia Wang, Dapeng Ju 等NDSS 2026
- Scheduled Disclosure: Turning Power into Timing Without Frequency ScalingInwhan Chun, Isabella Siu, Riccardo PaccagnellaS&P 2025
它引用的顶会 Paper35
- Inferring Fine-grained Control Flow Inside SGX Enclaves with Branch ShadowingSangho Lee, Ming-Wei Shih, Prasun Gera, Taesoo Kim 等USENIX Security 2017 · 被引用 536 次
- ARMageddon: Cache Attacks on Mobile DevicesMoritz Lipp, Daniel Gruss, Raphael Spreitzer, Clémentine Maurice 等USENIX Security 2016 · 被引用 451 次
- Translation Leak-aside Buffer: Defeating Cache Side-channel Protections with TLB AttacksBen Gras, Kaveh Razavi, Herbert Bos, Cristiano GiuffridaUSENIX Security 2018 · 被引用 357 次
- LVI: Hijacking Transient Execution through Microarchitectural Load Value InjectionJo Van Bulck, Daniel Moghimi, Michael Schwarz, Moritz Lipp 等S&P 2020 · 被引用 275 次
- Verifying Constant-Time ImplementationsJosé Bacelar Almeida, Manuel Barbosa, Gilles Barthe, François Dupressoir 等USENIX Security 2016 · 被引用 274 次
相关 Paper
- Prime+Scope: Overcoming the Observer Effect for High-Precision Cache Contention AttacksAntoon Purnal, Furkan Turan, Ingrid VerbauwhedeCCS 2021 · 被引用 55 次
- Frontal Attack: Leaking Control-Flow in SGX via the CPU FrontendIvan Puddu, Moritz Schneider, Miro Haller, Srdjan CapkunUSENIX Security 2021 · 被引用 63 次
- PREFETCHX: Cross-Core Cache-Agnostic Prefetcher-based Side-Channel AttacksYun Chen, Ali Hajiabadi, Lingfeng Pei, Trevor E. CarlsonHPCA 2024 · 被引用 15 次
- Trapped Handler: A Practical Performance Degradation Attack on NVIDIA vGPU via Preemption ManipulationCheng Gu, Fatemeh Moradihaghighi, Daniel Genkin, Zhuo Li 等CCS 2026
- The Benefits and Limitations of User Interrupts for Preemptive Userspace SchedulingLinsong Guo, Danial Zuberi, Tal Garfinkel, Amy OusterhoutNSDI 2025 · 被引用 10 次
