The #LinuxKernel's up-to-date architectures (arm64, loongarch, powerpc, riscv, s390, x86) from now on only offer two preemption models: full and lazy.
This is due to the change "sched: Further restrict the preemption modes" (https://git.kernel.org/torvalds/c/7dadeaa6e851e7d67733f3e24fc53ee107781d0f); for more history on that, see also: https://lwn.net/Articles/994322/
That commit was merged a few hours ago as part of the the main scheduler updates (https://git.kernel.org/torvalds/c/36ae1c45b2cede43ab2fc679b450060bbf119f1b).
This also brought a few performance/scalability improvements as well as RSEQ based 'scheduler time slice extension' support. The latter allows a thread to request a time slice extension when it enters a critical section to avoid contention on a resource when the thread is scheduled out inside of the critical section. For more details on this, see https://lore.kernel.org/all/20251215155615.870031952@linutronix.de/ and https://lwn.net/Articles/1038235/
