[VPlan] Handle VPWidenSelectRecipe in tryToFoldLiveIns (llvm#150357)

lukel97 · mahesh-attarde · commit e66b456ea26b · 2025-07-28T04:21:51.000-07:00
This helps simplify VPBlendRecipes that are expanded to selects in
another patch.
diff --git a/llvm/lib/Transforms/Vectorize/VPlan.h b/llvm/lib/Transforms/Vectorize/VPlan.h
@@ -1662,6 +1662,8 @@ struct LLVM_ABI_FOR_TEST VPWidenSelectRecipe : public VPRecipeWithIRFlags,
              VPSlotTracker &SlotTracker) const override;
 #endif
 
+  unsigned getOpcode() const { return Instruction::Select; }
+
   VPValue *getCond() const {
     return getOperand(0);
   }
diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
@@ -997,7 +997,7 @@ static void simplifyRecipe(VPRecipeBase &R, VPTypeAnalysis &TypeInfo) {
   // InstSimplifyFolder.
   if (TypeSwitch<VPRecipeBase *, bool>(&R)
           .Case<VPInstruction, VPWidenRecipe, VPWidenCastRecipe,
-                VPReplicateRecipe>([&](auto *I) {
+                VPReplicateRecipe, VPWidenSelectRecipe>([&](auto *I) {
             const DataLayout &DL =
                 Plan->getScalarHeader()->getIRBasicBlock()->getDataLayout();
             Value *V = tryToFoldLiveIns(*I, I->getOpcode(), I->operands(), DL,
diff --git a/llvm/test/Transforms/LoopVectorize/X86/constant-fold.ll b/llvm/test/Transforms/LoopVectorize/X86/constant-fold.ll
@@ -67,17 +67,16 @@ define void @redundant_or_1(ptr %dst, i1 %c.0, i1 %c.1) {
 ; CHECK:       vector.ph:
 ; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i1> poison, i1 [[C_0:%.*]], i64 0
 ; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i1> [[BROADCAST_SPLATINSERT]], <4 x i1> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP0:%.*]] = xor <4 x i1> [[BROADCAST_SPLAT]], splat (i1 true)
 ; CHECK-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <4 x i1> poison, i1 [[C_1:%.*]], i64 0
 ; CHECK-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <4 x i1> [[BROADCAST_SPLATINSERT1]], <4 x i1> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP0:%.*]] = xor <4 x i1> [[BROADCAST_SPLAT2]], splat (i1 true)
-; CHECK-NEXT:    [[TMP4:%.*]] = select i1 true, <4 x i1> [[BROADCAST_SPLAT]], <4 x i1> zeroinitializer
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE8:%.*]] ]
 ; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <4 x i8> [ <i8 0, i8 1, i8 2, i8 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[PRED_STORE_CONTINUE8]] ]
 ; CHECK-NEXT:    [[TMP1:%.*]] = icmp ule <4 x i8> [[VEC_IND]], splat (i8 2)
 ; CHECK-NEXT:    [[TMP2:%.*]] = select <4 x i1> [[TMP1]], <4 x i1> [[TMP0]], <4 x i1> zeroinitializer
-; CHECK-NEXT:    [[TMP5:%.*]] = select <4 x i1> [[TMP2]], <4 x i1> [[TMP4]], <4 x i1> zeroinitializer
+; CHECK-NEXT:    [[TMP5:%.*]] = select <4 x i1> [[TMP2]], <4 x i1> [[BROADCAST_SPLAT2]], <4 x i1> zeroinitializer
 ; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <4 x i1> [[TMP5]], i32 0
 ; CHECK-NEXT:    br i1 [[TMP6]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
 ; CHECK:       pred.store.if:
@@ -120,11 +119,11 @@ define void @redundant_or_1(ptr %dst, i1 %c.0, i1 %c.1) {
 ; CHECK-NEXT:    br label [[LOOP_HEADER:%.*]]
 ; CHECK:       loop.header:
 ; CHECK-NEXT:    [[IV:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
-; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_LATCH]], label [[THEN_1:%.*]]
+; CHECK-NEXT:    br i1 [[C_0]], label [[LOOP_LATCH]], label [[THEN_1:%.*]]
 ; CHECK:       then.1:
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[IV]], 2
 ; CHECK-NEXT:    [[OR:%.*]] = or i1 [[CMP]], true
-; CHECK-NEXT:    [[COND:%.*]] = select i1 [[OR]], i1 [[C_0]], i1 false
+; CHECK-NEXT:    [[COND:%.*]] = select i1 [[OR]], i1 [[C_1]], i1 false
 ; CHECK-NEXT:    br i1 [[COND]], label [[THEN_2:%.*]], label [[LOOP_LATCH]]
 ; CHECK:       then.2:
 ; CHECK-NEXT:    [[GEP:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 [[IV]]
@@ -171,17 +170,16 @@ define void @redundant_or_2(ptr %dst, i1 %c.0, i1 %c.1) {
 ; CHECK:       vector.ph:
 ; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i1> poison, i1 [[C_1:%.*]], i64 0
 ; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i1> [[BROADCAST_SPLATINSERT]], <4 x i1> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP0:%.*]] = xor <4 x i1> [[BROADCAST_SPLAT]], splat (i1 true)
 ; CHECK-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <4 x i1> poison, i1 [[C_0:%.*]], i64 0
 ; CHECK-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <4 x i1> [[BROADCAST_SPLATINSERT1]], <4 x i1> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP0:%.*]] = xor <4 x i1> [[BROADCAST_SPLAT2]], splat (i1 true)
-; CHECK-NEXT:    [[TMP1:%.*]] = select i1 true, <4 x i1> [[BROADCAST_SPLAT]], <4 x i1> zeroinitializer
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE8:%.*]] ]
 ; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <4 x i8> [ <i8 0, i8 1, i8 2, i8 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[PRED_STORE_CONTINUE8]] ]
 ; CHECK-NEXT:    [[TMP2:%.*]] = icmp ule <4 x i8> [[VEC_IND]], splat (i8 2)
 ; CHECK-NEXT:    [[TMP3:%.*]] = select <4 x i1> [[TMP2]], <4 x i1> [[TMP0]], <4 x i1> zeroinitializer
-; CHECK-NEXT:    [[TMP4:%.*]] = select <4 x i1> [[TMP3]], <4 x i1> [[TMP1]], <4 x i1> zeroinitializer
+; CHECK-NEXT:    [[TMP4:%.*]] = select <4 x i1> [[TMP3]], <4 x i1> [[BROADCAST_SPLAT2]], <4 x i1> zeroinitializer
 ; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <4 x i1> [[TMP4]], i32 0
 ; CHECK-NEXT:    br i1 [[TMP5]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
 ; CHECK:       pred.store.if:
@@ -224,11 +222,11 @@ define void @redundant_or_2(ptr %dst, i1 %c.0, i1 %c.1) {
 ; CHECK-NEXT:    br label [[LOOP_HEADER:%.*]]
 ; CHECK:       loop.header:
 ; CHECK-NEXT:    [[IV:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
-; CHECK-NEXT:    br i1 [[C_0]], label [[LOOP_LATCH]], label [[THEN_1:%.*]]
+; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_LATCH]], label [[THEN_1:%.*]]
 ; CHECK:       then.1:
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[IV]], 2
 ; CHECK-NEXT:    [[OR:%.*]] = or i1 true, [[CMP]]
-; CHECK-NEXT:    [[COND:%.*]] = select i1 [[OR]], i1 [[C_1]], i1 false
+; CHECK-NEXT:    [[COND:%.*]] = select i1 [[OR]], i1 [[C_0]], i1 false
 ; CHECK-NEXT:    br i1 [[COND]], label [[THEN_2:%.*]], label [[LOOP_LATCH]]
 ; CHECK:       then.2:
 ; CHECK-NEXT:    [[GEP:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 [[IV]]

Original file line number	Diff line number	Diff line change
`@@ -1662,6 +1662,8 @@ struct LLVM_ABI_FOR_TEST VPWidenSelectRecipe : public VPRecipeWithIRFlags,`
`1662`	`1662`	`VPSlotTracker &SlotTracker) const override;`
`1663`	`1663`	`#endif`
`1664`	`1664`
	`1665`	`+ unsigned getOpcode() const { return Instruction::Select; }`
	`1666`	`+`
`1665`	`1667`	`VPValue *getCond() const {`
`1666`	`1668`	`return getOperand(0);`
`1667`	`1669`	`}`