Taint Analysis of Security Code in the KLEE Symbolic Execution Engine
| Title | Taint Analysis of Security Code in the KLEE Symbolic Execution Engine |
| Publication Type | Conference Paper |
| Year of Publication | 2012 |
| Authors | Corin, R, Manzano, F |
| Conference Name | 2012 International Conference on Information and Communications Security (ICICS) |
| Date Published | 10/2012 |
| Publisher | LNCS |
| Conference Location | Hong Kong |
| Abstract | We analyse the security of code by extending the KLEE symbolic execution engine with a tainting mechanism that tracks information flows of data. We consider both simple flows from direct assignment operations, and (more subtle) indirect flows inferred from the control flow. Our mechanism prevents overtainting by using a region-based static analysis provided by LLVM, the compiler infrastructure machine on which KLEE runs. We rigorously define taint propagation in a formal LLVM intermediate representation semantics, and show the correctness of our method. We illustrate the mechanism with several examples, showing how we use tainting to prove confidentiality and integrity properties. |
| Refereed Designation | Refereed |
PDF (Full text):
Work Package:
WP3
