作者: Daniel Townley , Khaled N. Khasawneh , Dmitry Ponomarev , Nael Abu-Ghazaleh , Lei Yu
关键词:
摘要: We present LATCH (short for Locality-Aware Taint CHecker), a generalizable architecture optimizing dynamic information flow tracking (DIFT). exploits the observation that flows under DIFT exhibit strong temporal locality, with typical applications manipulating sensitive data during limited phases of computation. This property allows to monitor significant spans execution using lightweight, coarse-grained checks, invoking precise, computationally intensive logic only periods involve data. implements this policy without sacrificing accuracy DIFT. propose and evaluate three systems incorporating LATCH: S-LATCH accelerate software-based on single core; P-LATCH multicore DIFT, H-LATCH reduce architectural complexity hardware-based developed an FPGA prototype system, demonstrating its advantages come negligible impact power no effect processor cycle time.