[llvm] [VPlan] Directly unroll VectorPointerRecipe (PR #168886)

[Ramkumar Ramachandra via llvm-commits]

================
@@ -21,7 +21,8 @@ define void @sdiv_feeding_gep(ptr %dst, i32 %x, i64 %M, i64 %conv6, i64 %N) {
 ; CHECK-NEXT:    br i1 [[TMP7]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
 ; CHECK:       [[VECTOR_PH]]:
 ; CHECK-NEXT:    [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT:    [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 4
+; CHECK-NEXT:    [[TMP11:%.*]] = mul nuw i64 [[TMP8]], 2
+; CHECK-NEXT:    [[TMP9:%.*]] = mul i64 [[TMP11]], 2
----------------
artagnon wrote:

Correct, I thought the same initially. Will update https://github.com/llvm/llvm-project/pull/169396 in the morning; we should try to land that in order to reduce the diff of this patch.

https://github.com/llvm/llvm-project/pull/168886