A Study of Rarely Appeared Instructions in an Executable Binary

Abstract: An executable binary typically contains a large number of machine instructions. Although the statistics of popular instructions is well known, the distribution of non-popular instructions has been relatively under explored. Our finding shows that an arbitrary group of binaries com es with both i) a similar distribution of common machine instructions, and ii) quite a few rarely appeared instructions (e.g., less than five occurrences) apart from the distribution. Their infrequency may represent the signature of a code chunk or the footprint of a binary. In this work, we investigate such rare instructions with an in-depth analysis at the source level, clas sifying them into four categories.