A Technique of Profiling Selective Paths

Path profiling records the frequency of each path in an executed routine. To accomplish profiling, probes are instrumented in a program and executed as the program runs. So the number of probes has important influences on the efficiency of a profiling technique. To profile only a subset of paths, existing techniques try to improve the profiling efficiency by reducing probes, optimizing path encoding, and so on. However, they mainly lack accuracy, waste time on running uninterested paths, and only deal with acyclic paths. In this paper, a novel technique called PSP (Profiling Selective Paths) has been introduced to profile selective paths, which can handle selection for both acyclic and cyclic paths, and increase the execution efficiency by early termination on uninterested paths. PSP is implemented in two ways, PSP1 and PSP2. Theoretical comparison and experimental evaluation indicate that PSP1 and PSP2 perform differently but both effectively.