https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121315
--- Comment #4 from Alex Coplan <acoplan at gcc dot gnu.org> --- Yes, it looks to be faster using post-increment LDP/STP indeed. I think it would be difficult to undo in pair-fusion, perhaps best to handle this earlier (I guess in ivopts). Generally pair-fusion requires at least one of the accesses to have an addressing mode of the form `reg [+ <imm>]` in order to form a pair. I think it would add too much complexity to the pass to get it to try and undo the register offset addressing here.
