Related: did the idea of rseq `RSEQ_SCHED_STATE_FLAG_ON_CPU` for adaptive userspace mutexes [1] ever come to anything? I think there are a lot of userspace lock implementations using adaptive mutexes (including say `absl::Mutex` in C++ and `parking_lot::Mutex` in Rust). This seemed promising as a better way to decide when to switch from spinning to blocking.

[1] https://lwn.net/Articles/944895/