DyKA: light-weight dynamic kernel analyzing framework based on binary translation method

  • Authors:

  • Dongwoo Lee;Hyung Kook Jun;Won Tae Kim;Young Ik Eom

  • Affiliations:

  • Sungkyunkwan University, Suwon, Korea;Software Research Laboratory, Electronics and Telecommunications Research Institute, Daejeon, Korea;Software Research Laboratory, Electronics and Telecommunications Research Institute, Daejeon, Korea;Sungkyunkwan University, Suwon, Korea

  • Venue:
  • Proceedings of the 6th International Conference on Ubiquitous Information Management and Communication
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Platform, which is based on typical OS, has been widely used in various areas from smart phone to TV. Along with this, the need for kernel analyzer has also grown up in order to develop and debug the kernel and its component effectively. Although previous approaches provide us some information about the target, that information is no more than hardware state, or primitive memory map, which is hard to understand. Moreover they cause an amount of overheads during analysis, so that analysis has been unable to work on environment as real. For applications some researches have been studied the analyzer using binary translation, and now it is widely used. Adopting their ideas, we propose dynamic kernel analyzer: DyKA. Kernel analyzers mainly focus on not kernel itself, but its plugins like device driver since the plugins comprise a large portion of OS development. Because of its locality, we should pay attention to the quality of the target code after translation in order to make the dynamic analyzer effective. Thus we apply some optimizations with DyKA as considering the target attribute. We implement the idea of DyKA on Linux system, and make sure, that it could analyze the target without a huge overhead, by experimental evaluation.