I noticed that when I ran strlcpy in cc with both directly from libc and copied from source: “with and without needing a return value”, the libc strlcpy didn’t change the runtime, but the one from source did; dramatically (like 50% runtime difference over a several run loop with 15-20 or so characters), where the compiler obviously optimized stuff out.
Is it possible that there’s a way you haven’t considered, that could permit the compiler to optimize the functions… (or pre-optimize for the instances with no return value and when there are different needs, it could choose between the two; if there aren’t dual needs, then only write one to static programs; I don’t know if this can be done trivially.) …which could improve performance without breaking security of functions which may or may not need a return value to function correctly?! Or is all the extra runtime from returning from the libc wrapper? -- -Luke
