Daniel Moghimi

Abstract:

Downfall is a new security vulnerability linked to Meltdown and microarchitectural data sampling attacks. These threats compromise the security of computers worldwide. Despite industry efforts over the past five years to fix data leakage caused by the transient execution of instructions that load data, we've discovered that the "Gather" instruction in high-performance Intel CPUs can still leak data across user-kernel boundaries, processes, virtual machines, and trusted execution environments. Our research reveals that current defenses are ineffective, highlighting the need for crucial hardware fixes and security updates for commonly used client and server computers

During this presentation, we'll start by discussing Meltdown-type microarchitectural attacks. We'll provide a brief overview of their history, impact, current mitigation efforts, and affected platforms. Next, we'll delve into the SIMD Gather instruction, explaining how it can be exploited to leak data through Gather Data Sampling (GDS) technique. We'll explore the vulnerability's impact on various instructions and workloads, and finally showcase end-to-end key-stealing attacks, such as extracting 128-bit and 256-bit AES keys from OpenSSL.

To wrap up, we'll cover mitigation options, key takeaways, and future directions for addressing this class of microarchitectural flaws.

Biography:

Daniel Moghimi is a Senior Research Scientist at Google. Before that, he was a postdoctoral scholar at UCSD. He has a PhD in Electrical and Computer Engineering and an MSc in Computer Science from WPI. He works on computer and hardware security, spanning various topics such as microarchitectural vulnerabilities, side-channel cryptanalysis, and security architecture. His research has improved the security of superscalar CPUs, memory subsystems, and cryptographic implementations, which billions of users use daily..

Address:United States