https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119131
--- Comment #9 from Jakub Jelinek <jakub at gcc dot gnu.org> --- aarch64_float_const_zero_rtx_p fortunately bails out for decimal float (in theory it could for decimal float do aarch64_reinterpret_float_as_int (with some prevention of recursion) and compare if the int value is 0). Because many of the decimal zeros aren't zero when reinterpreted as int, while binary is (with the exception of -0. when not honoring zero signs).
