CAREER: Flexible Multi-Core Substrate for Trustworthy Computing Systems

在一个芯片上具有4到8个核的多核架构是当今的现实，并且未来几代处理器预计每个芯片包含更多的核。该项目的目标是实现大规模多核处理器作为安全和值得信赖的计算基板的全部潜力。首先，在“众核”处理器中，大量处理元件的可用性以及对功耗的约束使得需要大量的资源分时，这通常导致各种类型的边信道。此更改允许在此项目中使用更多专用和静态分配的资源，以提供强大的隔离和简化的可信软件基础。其次，大规模多核中的许多可编程处理元件可以作为各种类型的细粒度检查的通用基板。该项目正在开发一个灵活的架构框架，用于以最小的开销进行各种运行时检查，以便自动检测，诊断和从恶意软件攻击中恢复。最后，该项目将架构框架扩展到安全性之外的其他信任方面，重点是硅后验证，其中动态检查可以确保各种正确性属性。该研究将提供安全和验证方面的硬件支持，而无需为单一固定机制提供专用资源。用户可以将每个“扩展核心”用于各种安全目的，用于验证和可靠性目的，甚至用于通用计算。与此同时，该研究还将实现高度安全的执行环境，其中关键软件组件可以得到更好的保护。

Multi-core architecture with 4 to 8 cores on a die is a reality today and future generations of processors are expected to contain even more cores per chip. This project targets to realize the full potential of large-scale multi-core processors as a secure and trustworthy computing substrate. First, in "many-core" processors, the availability of a large number of processing elements together with constraints on power consumption alleviates the need for heavy time-sharing of resources, which often results in various types of side-channels. This change enables the use of more dedicated and statically allocated resources in this project to provide strong isolation and a simplified trusted software base. Second, many programmable processing elements in large-scale multi-cores can serve as a general substrate for various types of fine-grained inspection. This project is developing a flexible architecture framework for various run-time checks with minimal overheads in order to automatically detect, diagnose, and recover from malicious software attacks. Finally, the project extends the architectural framework to other aspects of trust beyond security, focusing on post-silicon verification, where the dynamic inspection can ensure various correctness properties. The research will deliver the benefits of hardware support in security and verification without requiring dedicated resources for a single fixed mechanism. Users can use each "extended core" for various security purposes, for verification and reliability purposes, or even for general computing. At the same time, the research will also enable highly secure execution environments where critical software components can be better protected.