Blackout Reduction
summary
Aim is to cut kexec live-update blackout toward under a second by specific changes: parallelize hundreds of device shutdown callbacks and reboot notifiers using spare CPUs; skip or move the expensive checksum (measured ~0.3 seconds) and avoid relocation by loading directly into final address using a CMA region so purgatory is eliminated; measure downtime in nanoseconds via persistent telemetry; explore keeping vCPUs running across the update and solving VM-exit paths so VMM user-space blocking no longer causes stalls.
AI-written summary · about