Hacker News new | ask | show | jobs
by colonial 466 days ago
At least on the compiler level, LLVM already offers a "black box" intrinsic that nudges the optimizer to be maximally pessimistic about whatever it wraps.

It doesn't carry any security guarantees (Rust's documentation on their wrapper for it explicitly strikes out "cryptographic or security" purposes) but one could imagine a similar intrinsic with stronger abilities.