Abstract

Programs and services relying on weak hash algorithms as part of their hash table implementations are vulnerable to hash-collision denial-of-service attacks. In the context of such an attack, the attacker sends a series of program inputs leading to hash collisions. In the best case, this slows down the execution and processing for all requests, and in the worst case it renders the program or service unavailable. We propose a new binary program analysis approach to automatically detect weak hash functions and patch vulnerable binary programs, by replacing the weak hash function with a secure alternative. To verify that our mitigation strategy does not break program functionality, we design and leverage multiple stages of static analysis and symbolic execution, which demonstrate that the patched code performs equivalently to the original code, but does not suffer from the same vulnerability. We analyze 105,831 real …

Date

October 26, 2022

Authors

Nicolaas Weideman, Haoda Wang, Tyler Kann, Spencer Zahabizadeh, Wei-Cheng Wu, Rajat Tandon, Jelena Mirkovic, Christophe Hauser

Book

Proceedings of the 25th International Symposium on Research in Attacks, Intrusions and Defenses

Pages

276-291