ffi_call_plan caching for GLib
Summary
The post explains libffi's new call plan caching and its integration with GLib to reduce GClosure invocation overhead, with a best-case reduction of about 25% and around 10% in less ideal cases. It also notes a GLib merge request requiring the newest libffi for CI, and describes a patch to libffi to add frame-pointers to improve profiling across FFI boundaries on Linux.