pub fn write_llc_cpu_lists(skel: &mut BpfSkel<'_>, plan: &LlcPlan) -> Result<()>Expand description
Phase 2 (post-load): write the per-LLC CPU lists into the
mlfq_llc_cpus array map.
One list per LLC domain, the Tier-A same-LLC steal window of the
dispatch path. A domain exceeding the MLFQ_MAX_LLC_CPUS bound gets
an EMPTY list (nr == 0) instead of failing the whole write: the
dispatch path then skips Tier A for that domain and Tier B’s full
rotating window covers it, because Tier B’s same-LLC skip is gated
on Tier A having run (tier_a_ran stays false for an empty list). The
window never silently shrinks to a subset of an oversized domain.