Secure Native Binary Executions--2018 Q4
PI(s): Prasad Kulkarni
HARD PROBLEM(S) ADDRESSED:
Scalability and Composability, Security Metrics
PUBLIC ACCOMPLISHMENT HIGHLIGHTS:
The project goal is to develop a high-performance framework for client-side security assessment and enforcement for COTS (commercial off the shelf) binary software. We designed the initial project setup and broke the primary goal into three sub-goals:
(a) Construct a framework for high-performance dynamic binary translators (DBT)
(b) Develop and implement techniques to facilitate secure binary execution
(c) Assess security level of COTS binary software
We started exploring the challenges to achieve our first sub-goal. DBT performance can be classified into "latency" and "throughput". Latency refers to overhead during program execution startup, while throughput is overall performance overhead for long-running tasks. We found that while current DBTs achieve good throughput, the startup performance degrades more significantly. We are developing experiments to better understand specific aspects affecting startup performance and then design techniques to overcome them.
PUBLICATIONS FROM THE QUARTER:
Although this work was developed before our lablet funding, it provides the preliminary work that motivates this research.
Surya Tej Nimmakayala and Prasad Kulkarni, Improving Startup Performance in Dynamic Binary Translators, accepted for publication in the 27th EuroMicro Conference on Parallel, Distributed and Network-Based Processing, February 13-15, 2019, Pavia, Italy.
Abstract: A Dynamic Binary Translation (DBT) system dynamically translates program binaries built for a guest platform into code for the host machine that runs the program, one basic block at a time. Even after optimizations, auxiliary tasks performed alongside program emulation by the DBT system introduce performance overheads as compared to executing the program on the native guest platform. In this work, we analyze the extent and causes for a DBT system's startup performance latency. We then focus on understanding and alleviating the program translation cost that is a significant contributor to and disproportionately impacts the startup overhead. We propose and assess the potential of a new technique that parallelizes program translations on multi-core machines to reduce its evident run-time costs. We explain the challenges in achieving such parallelization and discuss and evaluate solutions.