https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108273
--- Comment #4 from Kewen Lin <linkw at gcc dot gnu.org> --- Starting from r13-5154-g733a1b777f16cd, this failure is gone on Power10 LE (or with explicit -mcpu=power10), but it shows up on Power8 LE. By checking r13-5154, this behavior change is expected, the priorities of some insns change due to that commit, it further affects the scheduling decision on which insn is scheduled ahead.
