Abstract:

Binary instrumentation is a key technology in the fields of software performance analysis, vulnerability mining, and quality evaluation. When working on the embedded environment, traditional dynamic instrumentation algorithms are facing limitations like lacking operating system, complex CPU architecture, and tight memory resources. Those limitations make binary instrumentation on embedding software extremely difficult. Therefore, this paper studies the lightweight binary dynamic instrumentation technology, and realizes the acquisition of software runtime information through static feature analysis and dynamic tracking algorithms. Graph-based algorithms and embedded-oriented remote debugging protocol are introduced as well. Compared with the traditional solution, the solution in this article solves the dependence on source code, operating system or CPU architecture, while significantly reducing the occupancy rate of memory and computing resources. Therefore dynamic binary instrumentation work can be effectively solved.

Key words: software instrumentation, binary instrumentation, software debugging, binary analysis