Reverse Engineering the M1

The presentation discusses the evolution of Apple's chips and the importance of reverse engineering in understanding hardware.

• Apple has been customizing and extending their chips, adding their own IP and more programmable blocks.
• Reverse engineering is key to understanding hardware and validating register level models.
• Apple has implemented IO MMUs and device memory types, making reverse engineering more difficult but also providing valuable information.
• The presenter shares an anecdote about discovering the need for a more restrictive memory access type when bringing up Linux on the M1 chip.
• The presentation emphasizes the importance of understanding hardware for cybersecurity and devops purposes.

The presenter shares an anecdote about discovering the need for a more restrictive memory access type when bringing up Linux on the M1 chip.

The release of M1 Macs marked a turning point for the open-source operating system community on Apple hardware. Now, the whole hardware stack would be proprietary, with little hope of reusing drivers written for standard PC hardware. At the same time, it offered an unprecedented insight into the design of the Apple SoC product line. With this motivation, we set out to reverse engineer these parts and the systems they power.The talk will cover interesting quirks of Apple ARM architecture variant, such as memory access issues (and how to recognize them) and the novel AMX vector instruction set. We'll describe design patterns commonly employed by these SoCs, as well as give a short introduction to USB 4, which made its debut on them.