[Bug ipa/96252] mis-optimization where identical functions have very different codegen since gcc 10

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96252

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
It's IPA ICF that makes the difference.  Guess the "thunk" isn't a thunk but
copies parameters by value (fails to tail-call?).