[llvm] [LV] Convert scatter w/uniform addr and mask being header mask to scalar store. (PR #172799)

[Ramkumar Ramachandra via llvm-commits]

================
@@ -143,8 +141,10 @@ define void @test_3_inductions(ptr noalias %dst, ptr noalias %src, i64 %n) #1 {
 ; CHECK-NEXT:    [[TMP5:%.*]] = or <vscale x 2 x i32> [[VEC_IND1]], [[VEC_IND]]
 ; CHECK-NEXT:    [[TMP6:%.*]] = sext <vscale x 2 x i32> [[TMP5]] to <vscale x 2 x i64>
 ; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr i8, ptr [[SRC]], <vscale x 2 x i64> [[TMP6]]
-; CHECK-NEXT:    call void @llvm.vp.scatter.nxv2p0.nxv2p0(<vscale x 2 x ptr> [[TMP7]], <vscale x 2 x ptr> align 8 [[BROADCAST_SPLAT]], <vscale x 2 x i1> splat (i1 true), i32 [[TMP3]])
----------------
artagnon wrote:

Looks like the Mask is true in all tests? Do we test both codepaths?

https://github.com/llvm/llvm-project/pull/172799