February 23, 2024
In the multi-core era, kernel developers exploit concurrency to achieve optimal performance of the operating system. However, this optimization is notoriously error-prone and easily leads to hard-to-find concurrency bugs – an elusive class of critical bugs that are only exposed when multiple threads run specific instructions in certain execution orders. Concurrency bugs seriously impact users by causing machine crashes and data loss, as well as enabling privilege escalation attacks. My research aims to develop principled approaches to enable effective and efficient kernel concurrency testing. In this talk, I will present our testing frameworks Snowboard and Snowcat, and discuss how the efficiency of kernel concurrency testing can be improved by them.
About Sishuai Gong
Sishuai Gong is a fifth-year PhD student at Purdue CS, advised by Professor Pedro Fonseca. His research focuses on improving the security and reliability of the operating system kernels.