Lambda early return, remove unused test IR and improve comment formatting

Author: SamTebbs33

llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp

@@ -3651,26 +3651,26 @@ tryToMatchAndCreateMulAccumulateReduction(VPReductionRecipe *Red,
   auto ExtendAndReplaceConstantOp = [&Ctx](VPWidenCastRecipe *ExtA,
                                            VPWidenCastRecipe *&ExtB,
                                            VPValue *&ValB, VPWidenRecipe *Mul) {
-    if (ExtA && !ExtB && ValB->isLiveIn()) {
-      Type *NarrowTy = Ctx.Types.inferScalarType(ExtA->getOperand(0));
-      Instruction::CastOps ExtOpc = ExtA->getOpcode();
-      const APInt *Const;
-      if (!match(ValB, m_APInt(Const)) ||
-          !llvm::canConstantBeExtended(
-              Const, NarrowTy, TTI::getPartialReductionExtendKind(ExtOpc)))
-        return;
-      // The truncate ensures that the type of each extended operand is the
-      // same, and it's been proven that the constant can be extended from
-      // NarrowTy safely. Necessary since ExtA's extended operand would be
-      // e.g. an i8, while the const will likely be an i32. This will be
-      // elided by later optimisations.
-      VPBuilder Builder(Mul);
-      auto *Trunc =
-          Builder.createWidenCast(Instruction::CastOps::Trunc, ValB, NarrowTy);
-      Type *WideTy = Ctx.Types.inferScalarType(ExtA);
-      ValB = ExtB = Builder.createWidenCast(ExtOpc, Trunc, WideTy);
-      Mul->setOperand(1, ExtB);
-    }
+    if (!ExtA || ExtB || !ValB->isLiveIn())
+      return;
+    Type *NarrowTy = Ctx.Types.inferScalarType(ExtA->getOperand(0));
+    Instruction::CastOps ExtOpc = ExtA->getOpcode();
+    const APInt *Const;
+    if (!match(ValB, m_APInt(Const)) ||
+        !llvm::canConstantBeExtended(
+            Const, NarrowTy, TTI::getPartialReductionExtendKind(ExtOpc)))
+      return;
+    // The truncate ensures that the type of each extended operand is the
+    // same, and it's been proven that the constant can be extended from
+    // NarrowTy safely. Necessary since ExtA's extended operand would be
+    // e.g. an i8, while the const will likely be an i32. This will be
+    // elided by later optimisations.
+    VPBuilder Builder(Mul);
+    auto *Trunc =
+        Builder.createWidenCast(Instruction::CastOps::Trunc, ValB, NarrowTy);
+    Type *WideTy = Ctx.Types.inferScalarType(ExtA);
+    ValB = ExtB = Builder.createWidenCast(ExtOpc, Trunc, WideTy);
+    Mul->setOperand(1, ExtB);
   };
 
   // Try to match reduce.add(mul(...)).
@@ -3709,8 +3709,8 @@ tryToMatchAndCreateMulAccumulateReduction(VPReductionRecipe *Red,
     auto *Ext0 = dyn_cast_if_present<VPWidenCastRecipe>(A->getDefiningRecipe());
     auto *Ext1 = dyn_cast_if_present<VPWidenCastRecipe>(B->getDefiningRecipe());
 
-    // Convert reduce.add(ext(mul(ext, const))) to reduce.add(ext(mul(ext,
-    // ext(const))))
+    // reduce.add(ext(mul(ext, const)))
+    // -> reduce.add(ext(mul(ext, ext(const))))
     ExtendAndReplaceConstantOp(Ext0, Ext1, B, Mul);
 
     // Match reduce.add(ext(mul(ext(A), ext(B))))

llvm/test/Transforms/LoopVectorize/reduction-inloop.ll

@@ -2800,9 +2800,9 @@ exit:
   ret i64 %r.0.lcssa
 }
 
-define i32 @reduction_expression_ext_mulacc_livein(ptr %a, ptr %b, i16 %c) {
+define i32 @reduction_expression_ext_mulacc_livein(ptr %a, i16 %c) {
 ; CHECK-LABEL: define i32 @reduction_expression_ext_mulacc_livein(
-; CHECK-SAME: ptr [[A:%.*]], ptr [[B:%.*]], i16 [[C:%.*]]) {
+; CHECK-SAME: ptr [[A:%.*]], i16 [[C:%.*]]) {
 ; CHECK-NEXT:  [[ENTRY:.*:]]
 ; CHECK-NEXT:    br label %[[VECTOR_PH:.*]]
 ; CHECK:       [[VECTOR_PH]]:
@@ -2828,7 +2828,7 @@ define i32 @reduction_expression_ext_mulacc_livein(ptr %a, ptr %b, i16 %c) {
 ; CHECK-NEXT:    ret i32 [[TMP5]]
 ;
 ; CHECK-INTERLEAVED-LABEL: define i32 @reduction_expression_ext_mulacc_livein(
-; CHECK-INTERLEAVED-SAME: ptr [[A:%.*]], ptr [[B:%.*]], i16 [[C:%.*]]) {
+; CHECK-INTERLEAVED-SAME: ptr [[A:%.*]], i16 [[C:%.*]]) {
 ; CHECK-INTERLEAVED-NEXT:  [[ENTRY:.*:]]
 ; CHECK-INTERLEAVED-NEXT:    br label %[[VECTOR_PH:.*]]
 ; CHECK-INTERLEAVED:       [[VECTOR_PH]]:
@@ -2871,9 +2871,6 @@ for.body:                                         ; preds = %for.body, %entry
   %gep.a = getelementptr i8, ptr %a, i64 %iv
   %load.a = load i8, ptr %gep.a, align 1
   %ext.a = zext i8 %load.a to i16
-  %gep.b = getelementptr i8, ptr %b, i64 %iv
-  %load.b = load i8, ptr %gep.b, align 1
-  %ext.b = zext i8 %load.b to i16
   %mul = mul i16 %c, %ext.a
   %mul.ext = zext i16 %mul to i32
   %add = add i32 %mul.ext, %accum