An in-kernel machine-learning library

For those wanting more machine learning in the kernel, Viacheslav Dubeykohas posted anew in-kernel library for that purpose.

What is the goal of using ML models in Linux kernel? The main goalis to employ ML models for elaboration of a logic of particularLinux kernel subsystem based on processing data or/and an efficientsubsystem configuration based on internal state of subsystem. As aresult, it needs: (1) collect data for training, (2) execute MLmodel training phase, (3) test trained ML model, (4) use ML modelfor executing the inference phase. The ML model inference can beused for recommendation of Linux kernel subsystem configurationor/and for injecting a synthesized subsystem logic into kernelspace (for example, eBPF logic).

It is rigorously undocumentedand there are no real users, so it's not entirely clear what the purposeis, but there are undoubtedly interesting things that could be done withit.