Yes, we have a patch for this kind of analysis, but it doesn't produce reasonable performance improvements indeed. Anyway, we are able to perform measurements once again on top of your patch if you decide to commit it.If in the future we e.g. IPA-prop propagate the nonfreeing_call_p property through the callgraph (as in, if the function you call is non-overridable and you know the flag for it, use it),FYI we tried this on SPEC and some other apps but saw no performance improvements.
-Maxim
-Y