https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91016
--- Comment #1 from Martin Liška <marxin at gcc dot gnu.org> --- Jakub's comment from https://gcc.gnu.org/ml/gcc-patches/2019-06/msg01659.html: Just quick look suggests perhaps it wants to use one temporary for the gen_reg_rtx (mode) and use that on the lhs of SET, and another one as the last operand to AND holding ix86_expand_sse_compare_mask? Though, such problem is in most of the ix86_expand_sse_compare_mask callers. While ix86_expand_sse_compare_mask returns a result of gen_reg_rtx too, I think it is better not to reuse pseudos for different values. I'd say take out this change and deal with it with i386 maintainers separately.
