[llvm] [LoongArch] Lowering v32i8 vector mask generation to `VMSKLTZ` (PR #149953)

[Lu Weining via llvm-commits]

================
@@ -728,118 +625,15 @@ entry:
 define i32 @vmsk2_sgt_allzeros_i8(<32 x i8> %a) {
 ; CHECK-LABEL: vmsk2_sgt_allzeros_i8:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    addi.d $sp, $sp, -16
-; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    vrepli.b $vr2, 0
 ; CHECK-NEXT:    vslt.b $vr0, $vr2, $vr0
----------------
SixWeining wrote:

Seems we can use `vslti.b` instead of `vslt.b`.

https://github.com/llvm/llvm-project/pull/149953