To study how chips work, MIT researchers built their own operating system (news.mit.edu)

Article summary

MIT researchers have developed a custom operating system called Fractal to study the behavior of modern processors, allowing for more accurate and reliable experiments. Fractal boots directly on bare metal and exposes primitives that let a single experiment switch privilege levels at runtime, reducing background noise and producing clean signals. The team used Fractal to study Apple's M1 processor and discovered previously unknown behavior, including evidence of a class of speculative attack known as Phantom. The goal of Fractal is to become a shared infrastructure for microarchitecture research, providing a cleaner view of what's happening inside a processor.

Main themes

• Microarchitecture research
• Operating system design
• Processor security
• Speculative attacks
• Reverse engineering
• Computer hardware

What commenters say

• The development of Fractal is a significant contribution to the field of microarchitecture research, providing a more accurate and reliable way to study processor behavior.
• The approach of building a custom operating system from scratch is more effective than modifying existing operating systems, despite the significant effort required.
• The discovery of previously unknown behavior in the M1 processor highlights the importance of continued research into processor security and the potential for new vulnerabilities to be found.
• The concept of a 'vulnerability' is nuanced, and it's not always clear whether a particular issue is a significant risk or not, especially if it's not currently exploitable.
• The use of analogies, such as the Li-ion battery example, can help to clarify the importance of investigating potential vulnerabilities, even if they are not currently exploitable.
• The distinction between vulnerabilities and exploits is important, and researchers should focus on identifying potential weaknesses in systems, even if they are not currently exploitable.
• The development of Fractal has the potential to lead to new discoveries and a better understanding of processor behavior, but it's not clear whether it will be widely adopted by the research community.
• The idea that research should only focus on pushing the boundary of human knowledge is too narrow, and applied research, such as reverse engineering, can still be valuable and contribute to our understanding of complex systems.