New Software Abstractions
for Hardware Security Technology

Focus of the Workshop

Our society relies on software in mobile phones, cloud computing systems, personal computers and critical distributed infrastructure. While the dependency on software has exploded, our ability to make software trustworthy, secure, and dependable has not kept pace. Security vulnerabilities, ransomware, malware, software faults, privacy compromises are regular occurrences.

Over the last decades, important security features such as different privilege levels have been widely deployed. Programming languages have been developed that avoid common low-level security issues. Yet despite these security measures, the world still loses billions of euros per year due to cyber-attacks. As the pandemic has struck, employees were forced to work from home, and an even bigger part of our daily lives takes place in cyberspace. Unfortunately, criminals have adapted as well. Interpol has issued a warning that “cybercriminals are developing and boosting their attacks at an alarming pace, exploiting the fear and uncertainty caused by the unstable social and economic situation created by COVID-19”. Many of these problems originate from placing trust in untrustworthy code bases. Even when applications are developed carefully, they still rely on an operating system (OS) kernel with millions of lines of code that is implemented in a memory unsafe language. A single vulnerability in the OS kernel puts the security of all application code running on top at risk. Given the sheer sizes of modern software stacks, it is unlikely that these code bases can ever be trusted completely.

To rise to this challenge, hardware is evolving, and modern processors include new hardware security features. For example, secure boot support ensures that software is guaranteed a trustworthy environment; trusted execution environments (TEEs) allow the hardware to protect computation from malicious actors; and hardware capability support prevents unauthorized memory accesses by software. These new hardware features pose unprecedented challenges for the software stack, from low-level firmware to hypervisors and OS kernels, middleware, and all the way to applications.

For example, instead of trying to defend the entire system, hardware support for TEEs focuses on protecting sensitive data during computation. TEEs enable the creation of enclaves that contain trustworthy code and data. The OS kernel creates and schedules enclaves, but itself is not trusted. The processor hardware verifies and maintains the integrity of enclaves and ensures that execution can only transition to them via proper entry points. The isolation properties of TEEs allow programmers to reason about the security of their applications without loss of generality. Security vulnerabilities due to programming mistakes will still exist, but attacks can be mitigated using the isolation guarantees of TEEs. These strong security properties make hardware security mechanisms a key measure for trustworthy software.

Various hardware security mechanisms have been proposed in academia and industry, ranging from hardware implementations such as Flicker, Sancus and TyTAN, hardware support for memory capabilities, such as CHERI, to platforms that adopt trustworthy hypervisors such as TrustVisor and Fides. All mainstream hardware manufactures, including Intel, AMD, ARM and IBM, have either already launched hardware security mechanisms or have plans in the pipeline. This workshop will focus on the open challenges in software systems research related to new hardware security technologies. We will explore new ideas and strategies for how to create abstractions, methodologies, platforms and building blocks that will enable a new generation of trustworthy and secure software systems based on hardware security technologies. Thereby devising new secure applications and system software from scratch as well as extending and securing existing legacy applications will be addressed. Our goal is to develop a roadmap that will inspire the systems security research community and industry to build new secure software systems.