Clément Thorens | Searcharxiv

arXiv2026

Devlore: Device Interrupt Protection for Confidential VMs

Modern confidential computing executes sensitive computation in an abstraction called confidential VMs and protects from the hypervisor, host OS, and other co-resident VMs. It has been shown that an attacker can inject malicious interrupts to break the confidentiality and integrity of confidential VMs. We present Devlore, a device interrupt isolation mechanism that protects confidential VMs from interrupt manipulation attacks. Our design employs a delegate-but-check strategy by offloading interrupt management to the hypervisor, but adds correctness checks in the trusted software. We prototype our design on Arm Confidential Computing Architecture (CCA). We evaluate it on Arm FVP to demonstrate four diverse devices attached to confidential VMs and report costs on a Rock5b board. Our case studies show the feasibility of real-world use cases and that Devlore incurs minimal overheads of 0.06% for typical integrated GPU applications.

cs.CR↗

arXiv2024

Ascend-CC: Confidential Computing on Heterogeneous NPU for Emerging Generative AI Workloads

Cloud workloads have dominated generative AI based on large language models (LLM). Specialized hardware accelerators, such as GPUs, NPUs, and TPUs, play a key role in AI adoption due to their superior performance over general-purpose CPUs. The AI models and the data are often highly sensitive and come from mutually distrusting parties. Existing CPU-based TEEs such as Intel SGX or AMD SEV do not provide sufficient protection. Device-centric TEEs like Nvidia-CC only address tightly coupled CPU-GPU systems with a proprietary solution requiring TEE on the host CPU side. On the other hand, existing academic proposals are tailored toward specific CPU-TEE platforms. To address this gap, we propose Ascend-CC, a confidential computing architecture based on discrete NPU devices that requires no trust in the host system. Ascend-CC provides strong security by ensuring data and model encryption that protects not only the data but also the model parameters and operator binaries. Ascend-CC uses delegation-based memory semantics to ensure isolation from the host software stack, and task attestation provides strong model integrity guarantees. Our Ascend-CC implementation and evaluation with state-of-the-art LLMs such as Llama2 and Llama3 shows that Ascend-CC introduces minimal overhead with no changes in the AI software stack.

cs.CR↗