HACKING THE GO COMPILER TO EFFICIENTLY MAP IPV4 TO IPV6

Summary from the TLDR IT newsletter

The Go compiler handles mapping IPv4 addresses to IPv4-mapped IPv6 addresses inefficiently: the idiomatic netip.AddrFrom16 and As16 pattern runs eight times slower than a native method because of suboptimal compiler optimization. Standard library changes, an unsafe proxy, and custom IR and SSA rewrite rules can eliminate redundant byte-packing and unpacking overhead and achieve native-level performance.