https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69501
--- Comment #2 from Michael Meissner <meissner at gcc dot gnu.org> --- In addition to sign/zero/float extend, note that there are UNSPECs for some floating point conversions as well (lfiwax, lfiwzx). Rather than trying to enumerate all of the extends, perhaps checking whether the type attribute is an appropriate "load", "fpload", or "vecload" might be simpler. Or move the code to insn_is_load_p and insn_is_store_p, and have rs6000_adjust_cost call those functions.
