https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84229
Richard Biener <rguenth at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |RESOLVED Version|unknown |8.0.1 Resolution|--- |FIXED --- Comment #6 from Richard Biener <rguenth at gcc dot gnu.org> --- So fixed? I agree that IPA mucking with functions relying on inlining (all functions with __builtin_va_arg_pack_* stuff are supposed to be always_inline!) is pointless. Of course at the time of IPA those functions should already be inlined everywhere and we shouldn't end up emitting their bodies un-inlined. In fact the bodies should have been reclaimed and not entered IPA? Of course if it is "IPA" SRA then yeah, that should simply ignore all always-inline functions anyway.