More Science Fiction than Fairy Tale - A Discussion of the Linux Kernel Memory Model

Explore the intricacies of the Linux Kernel Memory Model (LKMM) in this 42-minute conference talk by Jonas Oberhauser from Huawei. Delve into the challenges of developing concurrent components for the Linux Kernel across various architectures and optimizing compilers. Learn about the unified abstraction provided by LKMM, including its library of concurrency primitives and high-level mathematical model. Discover how to read and apply such models as a practitioner, and examine examples that showcase LKMM's complexity, likening it to science fiction with concepts of time travel and intricate mechanisms. Gain insights into potential areas for improvement in LKMM, such as addressing incorrect predictions and unintended behaviors. The talk covers topics including the rationale behind weak memory and weak memory models, the history of LKMM, manual usage of LKMM, and concludes with final examples and a summary of key points.

Intro
WHY WEAK MEMORY?
WHY WEAK MEMORY MODELS?
HISTORY OF LINUX KERNEL MEMORY MODEL
HOW TO USE LKMM - MANUALLY?
FINAL EXAMPLES
CONCLUSION