From secret images to encryption keys. [Research Saturday]

This week, we are joined by Hosein Yavarzadeh from the University of California San Diego, as he is discussing his work on "Pathfinder: High-Resolution Control-Flow Attacks Exploiting the Conditional Branch Predictor" This paper introduces new methods that let attackers read from and write to specific parts of high-performance CPUs, such as the path history register (PHR) and prediction history tables (PHTs). These methods allow two main types of attacks. One can reveal a program's control flow history, as shown by recovering a secret image through the libjpeg routines. The other enables detailed transient attacks, demonstrated by extracting an AES encryption key, highlighting significant security risks for these systems. The research can be found here: Graph: Growing number of threats leveraging Microsoft API