A Precise Dynamic Flaw-Detection Method for Embedded Software

With the continuous development of embedded technology, embedded software is in great demand, which makes how to dynamically find the code that affect the efficiency of embedded software and optimize the embedded software performance with it critical for embedded software development. This paper begins with analysis into the shortcomings of current flaw-detection methods and related tools available; and proposes an accurate, real-time online dynamic flaw-detection method, which can test the hardware-platform-related and instruction-inefficiency-causing flaw event; for example, the pipeline flush caused by branch prediction failure. The method is then development into implemented tools. Finally, with the use of the tool on the unit code, it demonstrates flaw event tests such as data cache missing to verify the correctness of the proposed method.