https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101325
--- Comment #10 from Christophe Lyon <clyon at gcc dot gnu.org> ---
This was introduced by my change at r12-671 in mve.md:
-;; [vcmpneq_])
+;; [vcmpneq_, vcmpcsq_, vcmpeqq_, vcmpgeq_, vcmpgtq_, vcmphiq_, vcmpleq_,
vcmpltq_])
;;
-(define_insn "mve_vcmpneq_<mode>"
+(define_insn "mve_vcmp<mve_cmp_op>q_<mode>"
[
(set (match_operand:HI 0 "vpr_register_operand" "=Up")
- (unspec:HI [(match_operand:MVE_2 1 "s_register_operand" "w")
- (match_operand:MVE_2 2 "s_register_operand" "w")]
- VCMPNEQ))
+ (MVE_COMPARISONS:HI (match_operand:MVE_2 1 "s_register_operand" "w")
+ (match_operand:MVE_2 2 "s_register_operand" "w")))
+ ]
+ "TARGET_HAVE_MVE"
+ "vcmp.<mve_cmp_type>%#<V_sz_elem> <mve_cmp_op>, %q1, %q2"
+ [(set_attr "type" "mve_move")
+])
So should that use MVE_COMPARISONS:MVE_2, and somehow include a conversion from
MVE_2 modes into HI for vpr.p0 in the pattern?
