CRII: SHF: Pointer-aware Memory: Boosting Cybersecurity by Making Strong Memory Protection Affordable for Irregular Applications

Virginia Polytechnic Institute And State University, Blacksburg VA

Investigators

Abstract

Computer security weaknesses are threatening public safety and economies at global scale. Due to the need for stronger computer security, current processors implement many new types of strong memory protections, such as memory encryption and out-of-bound memory access detection. The research community is also actively investigating many other types of strong memory protections, such as memory address randomization, memory traffic shaping, etc. Unfortunately, strong memory protection significantly slows down memory accesses and, therefore, application performance. Through real-system measurements using latest processors, adding a single type of strong memory protection can more than double a program's total execution time. Such a high performance penalty can curtail the adoption of strong memory protections, thereby degrading cybersecurity and jeopardizing public safety. The goal of this project is to reduce the high performance cost that strong memory protection currently imposes on pointer-heavy applications; this performance cost is especially problematic for such applications, which are very sensitive to memory-access latency. This project will explore CPU hardware innovations to reduce the performance overheads of pointer de-referencing for different types of strong memory protection transparently and without programmer assistance. Reducing this performance cost will broaden the adoption of strong memory protection and thus help boost the cybersecurity of software. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.