SMPng

• Redesign and reimplement the kernel to support fine-grained parallel execution, kernel execution preemption

  • Reimplement kernel synchronization primitives
  • Move to data locking strategy
  • Improve both “architecture” and performance
  • All kinds of neat new tools including interrupt threads, UMA, run-time lock order verifiers, preemption, ...
  • Process, VM, IP stack, IPC, ...