[VPlan] Use onlyFirstLaneUsed in sinkScalarOperands.

Replace custom code to check if only the first lane is used by generic
helper `onlyFirstLaneUsed`. This enables VPlan-based sinking in a few
additional cases and was suggested in D133760.

Reviewed By: Ayal

Differential Revision: https://reviews.llvm.org/D136368

llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp

@@ -138,21 +138,16 @@ bool VPlanTransforms::sinkScalarOperands(VPlan &Plan) {
     // All recipe users of the sink candidate must be in the same block SinkTo
     // or all users outside of SinkTo must be uniform-after-vectorization (
     // i.e., only first lane is used) . In the latter case, we need to duplicate
-    // SinkCandidate. At the moment, we identify such UAV's by looking for the
-    // address operands of widened memory recipes.
+    // SinkCandidate.
     auto CanSinkWithUser = [SinkTo, &NeedsDuplicating,
                             SinkCandidate](VPUser *U) {
       auto *UI = dyn_cast<VPRecipeBase>(U);
       if (!UI)
         return false;
       if (UI->getParent() == SinkTo)
         return true;
-      auto *WidenI = dyn_cast<VPWidenMemoryInstructionRecipe>(UI);
-      if (WidenI && WidenI->getAddr() == SinkCandidate) {
-        NeedsDuplicating = true;
-        return true;
-      }
-      return false;
+      NeedsDuplicating = UI->onlyFirstLaneUsed(SinkCandidate);
+      return NeedsDuplicating;
     };
     if (!all_of(SinkCandidate->users(), CanSinkWithUser))
       continue;

llvm/test/Transforms/LoopVectorize/interleaved-accesses-pred-stores.ll

@@ -38,8 +38,9 @@ define void @interleaved_with_cond_store_0(%pair *%p, i64 %x, i64 %n) {
 ; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <2 x i1> [[TMP4]], i64 0
 ; CHECK-NEXT:    br i1 [[TMP5]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
 ; CHECK:       pred.store.if:
+; CHECK-NEXT:    [[TMP2_1:%.*]] = getelementptr inbounds [[PAIR:%.*]], %pair* [[P:%.*]], i64 [[INDEX]], i32 1
 ; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <4 x i64> [[WIDE_VEC]], i64 0
-; CHECK-NEXT:    store i64 [[TMP6]], i64* [[TMP2]], align 8
+; CHECK-NEXT:    store i64 [[TMP6]], i64* [[TMP2_1]], align 8
 ; CHECK-NEXT:    br label [[PRED_STORE_CONTINUE]]
 ; CHECK:       pred.store.continue:
 ; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <2 x i1> [[TMP4]], i64 1
@@ -136,8 +137,9 @@ define void @interleaved_with_cond_store_1(%pair *%p, i64 %x, i64 %n) {
 ; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <2 x i1> [[TMP7]], i64 0
 ; CHECK-NEXT:    br i1 [[TMP8]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
 ; CHECK:       pred.store.if:
+; CHECK-NEXT:    [[PTR0:%.*]] = getelementptr inbounds [[PAIR:%.*]], %pair* [[P:%.*]], i64 [[INDEX]], i32 0
 ; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <4 x i64> [[WIDE_VEC]], i64 0
-; CHECK-NEXT:    store i64 [[TMP9]], i64* [[TMP3]], align 8
+; CHECK-NEXT:    store i64 [[TMP9]], i64* [[PTR0]], align 8
 ; CHECK-NEXT:    br label [[PRED_STORE_CONTINUE]]
 ; CHECK:       pred.store.continue:
 ; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <2 x i1> [[TMP7]], i64 1
@@ -246,8 +248,9 @@ define void @interleaved_with_cond_store_2(%pair *%p, i64 %x, i64 %n) {
 ; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <2 x i1> [[TMP7]], i64 0
 ; CHECK-NEXT:    br i1 [[TMP8]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
 ; CHECK:       pred.store.if:
+; CHECK-NEXT:    [[PTR1:%.*]] = getelementptr inbounds [[PAIR]], %pair* [[P]], i64 [[INDEX]], i32 1
 ; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <4 x i64> [[WIDE_VEC]], i64 0
-; CHECK-NEXT:    store i64 [[TMP9]], i64* [[TMP5]], align 8
+; CHECK-NEXT:    store i64 [[TMP9]], i64* [[PTR1]], align 8
 ; CHECK-NEXT:    br label [[PRED_STORE_CONTINUE]]
 ; CHECK:       pred.store.continue:
 ; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <2 x i1> [[TMP7]], i64 1