Discussion
Continuing my study of Linux's bootup sequence...
Next it allocates a kmem cache for "maple trees", reportedly used to track Virtual Memory Areas.
Then it inits x86- or PowerPC-specific facilities for poking & patching a running kernel. In short on x86 this involves using helpers to allocate some datastructures, enable virtualization memory map, setting a no-track bit, (not via helper) getting a text-poke address with possible offsets, getting a locked PTE, & unlocking it.
1/?
Following that it inits "tracing" debugging facilities. To do so it might 1st allocate an array & depending on success either disables tracing or enables the tracing codepath. Regardless it finishes by extensively allocating & initializing some buffers (some dedicated to each CPU-core, with hotplugging) under locks via a suite of helpers unless the security subsystem says no, & very carefully under lock copies event handlers from a compiletime table to a runtime table via another helper.
3/3Fin
After that it might init function tracing by possibly invoking CPU-specific under lock (irrelevant to anything modern), retrieves & validates a build-time array's length, carefully invokes (via a helper) each of its sorted functions with newly-allocated memory, outputs status possibly propagating failures, sets some global flags, & initializes (via a suite of helpers) a selection of global filters & graphs. Disabling a flag upon error.
2/3?