Learning the Linux Kernel Configuration Space - Results and Challenges
Explore the complexities of the Linux kernel configuration space in this 37-minute conference talk by Mathieu Acher from the University of Rennes. Discover how machine learning and automation can assist in understanding and predicting properties of Linux kernel configurations. Learn about TuxML, an open-source tool that uses Docker and Python to collect data on thousands of kernel configurations. Understand the challenges and benefits of continuously analyzing the Linux configuration space, including identifying influential options, finding small kernels, and predicting compilation and boot times. Gain insights into the technical hurdles faced in terms of infrastructure and automation when dealing with over 15,000 kernel options and their intricate interactions.