MOLE: Breaking GPU TEE with GPU-Embedded MCU
Hongyi Lu, Yunjie Deng, J. Sukarno Mertoguno, Shuai Wang, Fengwei Zhang
Abstract
Graphics Processing Units (GPUs) are extensively used for applications such as machine learning, scientific computing, and graphics rendering. To protect sensitive data processed by GPUs, Trusted Execution Environments (TEEs) for GPUs have been proposed. GPU TEEs, built with hardware-based isolation primitives, can defend against high-privilege attackers like OS kernels. However, in this paper, we present MOLE, a novel attack that compromises the security of GPU TEEs on Arm Mali GPUs by exploiting the GPU-embedded Microcontroller Unit (MCU). By injecting malicious firmware into the MCU, an attacker can bypass GPU TEEs' security guarantees. We evaluated MOLE with state-of-the-art GPU TEE proposals under multiple real-world attack scenarios, such as in-GPU AES encryption and object detection tasks. Our evaluation shows that MOLE can successfully extract sensitive data or manipulate the computation results of GPU TEEs. We responsibly disclosed our findings to the authors of the affected GPU TEE proposals and received acknowledgments from all of them. Moreover, our findings prompted Arm to enhance the security of its GPU firmware supply chains.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
2 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- SoK: Analysis of Accelerator TEE Designs cites Enabling Rack-scale Confidential Computing using Heterogeneous Trusted Execution Environment
- Plundervolt: Software-based Fault Injection Attacks against Intel SGX cites Foreshadow: Extracting the Keys to the Intel SGX Kingdom with Transient Out-of-Order Execution
- Enabling Rack-scale Confidential Computing using Heterogeneous Trusted Execution Environment cites Rendered Insecure: GPU Side Channel Attacks are Practical
Cited by top-tier papers2
Builds on20
- Foreshadow: Extracting the Keys to the Intel SGX Kingdom with Transient Out-of-Order ExecutionJo Van Bulck, Marina Minkin, Ofir Weisse, Daniel Genkin et al.USENIX Security 2018 · 1,175 citations
- Plundervolt: Software-based Fault Injection Attacks against Intel SGXKit Murdock, David F. Oswald, Flavio D. Garcia, Jo Van Bulck et al.S&P 2020 · 369 citations
- Rendered Insecure: GPU Side Channel Attacks are PracticalHoda Naghibijouybari, Ajaya Neupane, Zhiyun Qian, Nael B. Abu-GhazalehCCS 2018 · 214 citations
- Enabling Rack-scale Confidential Computing using Heterogeneous Trusted Execution EnvironmentJianping Zhu, Rui Hou, XiaoFeng Wang, Wenhao Wang et al.S&P 2020 · 95 citations
- Protecting Cloud Virtual Machines from Hypervisor and Host Operating System ExploitsShih-Wei Li, John S. Koh, Jason NiehUSENIX Security 2019 · 49 citations
Related papers
- StrongBox: A GPU TEE on Arm EndpointsYunjie Deng, Chenxu Wang, Shunchang Yu, Shiqing Liu et al.CCS 2022 · 37 citations
- ACAI: Protecting Accelerator Execution with Arm Confidential Computing ArchitectureSupraja Sridhara, Andrin Bertschi, Benedict Schlüter, Mark Kuhne et al.USENIX Security 2024 · 36 citations
- CAGE: Complementing Arm CCA with GPU ExtensionsChenxu Wang, Fengwei Zhang, Yunjie Deng, Kevin Leach et al.NDSS 2024
- sNPU: Trusted Execution Environments on Integrated NPUsErhu Feng, Dahu Feng, Dong Du, Yubin Xia et al.ISCA 2024 · 13 citations
- Horizontal Privilege Escalation in Trusted ApplicationsDarius Suciu, Stephen E. McLaughlin, Laurent Simon, Radu SionUSENIX Security 2020
