[Bug target/123026] [16 Regression] aarch64: wrong code for vectorized loop

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123026

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
wait:
        umaxp   v30.4s, v30.4s, v30.4s        // tmp122,,

;; _52 = .REDUC_SBOOL_IOR (mask__2.13_48);

(insn 16 15 17 (set (reg:V4SI 121)
        (unspec:V4SI [
                (reg:V4SI 105 [ mask__2.13 ]) repeated x2
            ] UNSPEC_UMAXV)) -1
     (nil))

(insn 17 16 18 (set (reg:V4SI 105 [ mask__2.13 ])
        (reg:V4SI 121)) -1
     (nil))