Emerging applications like artificial intelligence and autonomous car require high security-assurance, which stimulates the wide-spread deployment of trusted execution environment (TEE). However, prior enclave systems are far from the ideal for three reasons. 1) Scalability: only support limited secure memory or limited number of instances; 2) Performance: not well-fit the requirements of high-performance application, e.g., poor secure communication performance; 3) Security: many still have security flaws, e.g., suffering cache-based side channels attacks.
Penglai-Enclave is proposed to overcome these challenges. The Penglai open-source project aims to build a scalable and efficient TEE system based on RISC-V, which is made powerful through hardware-assisted scalable physical memory isolation extensions. Our evaluations show that Penglai can achieve more than 1,000 concurrently running instances even in a resource-restricted device. We also have supported libraries like ARM PSA on Penglai to ease the development of trusted applications, and applied formal methods to validate its software TCB.
Speakers: Dong Du