[llvm] [VectorCombine] Scalarize extracts of ZExt if profitable. (PR #142976)

[Ramkumar Ramachandra via llvm-commits]

================
@@ -266,3 +329,32 @@ entry:
   call void @use.i64(i64 %ext.1)
   ret void
 }
+
+define void @zext_nv4i8_all_lanes_used(<vscale x 4 x i8> %src) {
+; CHECK-LABEL: define void @zext_nv4i8_all_lanes_used(
----------------
artagnon wrote:

```suggestion
define void @zext_nxv4i8_all_lanes_used(<vscale x 4 x i8> %src) {
; CHECK-LABEL: define void @zext_nxv4i8_all_lanes_used(
```

https://github.com/llvm/llvm-project/pull/142976