https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65873
--- Comment #17 from Jan Hubicka <hubicka at gcc dot gnu.org> --- The underlying problem of inlining across target/optimization boundary not being fully reliable is still there. I am not quite sure how we would want to fix it w/o allowing to attach different optimization parameters to regions/statements within one function which would be a lot of work to implement.
