Discussion
Looking at the bootup sequence for Linux's scheduler per each individual CPU core in your machine...
It 1st retrieves a CPU variable holding the runqueues & transiently grabs a lock on it, clears some counts & sets a timestamp, invokes a helper to init some properties & a lock for the Completely Fair Scheduler (CFS), invokes another helper to init some linkedlists/bitflags/priorities/counters/lock (some of which can be compiletime-enabled) for the RealTime scheduler, ...
1/3?
... invokes a helper & a couple more (with locking wrappers) in turn to update some bitmasks & atomic counters according to the runqueue's old & new "root domains" describing the arrangement of CPU cores before extremely carefully (if I'm looking past all the helpers correctly) notifying linkedlist of callbacks, updates a couple properties if compiletime-enabled some via a helper hidden behind macro-magic, compiletime-possibly inits a RCU-wait struct for CPU hotplugging, ...
3/4
... possibly inits a tick-timer, zeroes out an atomic, invokes a suite of helpers to init various mostly timers & counters which (I think?) has the completely-fair scheduler invoke the deadline scheduler, does similar for an "ext server" if compiletime-enabled, clears various other counters & collections if sched-core's compiletime-enabled, allocates a zeroed-out CPU mask.
3.5/3.5 Fin for today! Initializing fair & ext schedulers!
... invokes another helper to init some counters & other properties (bandwidth ratio via another helper) for the DeadLine scheduler, inits more properties (some via a helper, & a couple more in turn) if fair-group scheduling's compile-time enabled, invokes a couple more helpers if compile-time enabled to retrieve a timesamp it stores & to init some more RT properties, sets the next scheduling class to be the idling one, sets various other properties, creates a list for the CFS, ...
2/3?