[AArch64][SVE] Move instcombine like transforms out of SVEIntrinsicOpts

[bsmith]

Summary

Instead move them to the instcombine that happens in AArch64TargetTransformInfo.

[peterwaller-arm]

Naming nit, but otherwise LGTM.

llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp

833

The majority of the other cases here have SVE in the name. Should we follow suit here? (No strong opinion either way, just pointing it out)

Also Ptest vs PTest -- the former doesn't appear in the codebase until this patch as far as I can see.

llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp

instCombineSVECntElts(InstCombiner &IC, IntrinsicInst &II, unsigned NumElts) {
   }
 
   return NumElts >= MinNumElts
              ? Optional<Instruction *>(IC.replaceInstUsesWith(
                    II, ConstantInt::get(II.getType(), MinNumElts)))
              : None;
 }
 
+static Optional<Instruction *> instCombineSVEPTest(InstCombiner &IC,
+                                                   IntrinsicInst &II) {
+  IntrinsicInst *Op1 = dyn_cast<IntrinsicInst>(II.getArgOperand(0));
+  IntrinsicInst *Op2 = dyn_cast<IntrinsicInst>(II.getArgOperand(1));
+
+  if (Op1 && Op2 &&
+      Op1->getIntrinsicID() == Intrinsic::aarch64_sve_convert_to_svbool &&
+      Op2->getIntrinsicID() == Intrinsic::aarch64_sve_convert_to_svbool &&
+      Op1->getArgOperand(0)->getType() == Op2->getArgOperand(0)->getType()) {
+
+    IRBuilder<> Builder(II.getContext());
+    Builder.SetInsertPoint(&II);
+
+    Value *Ops[] = {Op1->getArgOperand(0), Op2->getArgOperand(0)};
+    Type *Tys[] = {Op1->getArgOperand(0)->getType()};
+
+    auto *PTest = Builder.CreateIntrinsic(II.getIntrinsicID(), Tys, Ops);
+
+    PTest->takeName(&II);
+    return IC.replaceInstUsesWith(II, PTest);
+  }
+
+  return None;
+}
+
+static Optional<Instruction *> instCombineSVEVectorMul(InstCombiner &IC,
+                                                       IntrinsicInst &II) {
+  auto *OpPredicate = II.getOperand(0);
+  auto *OpMultiplicand = II.getOperand(1);
+  auto *OpMultiplier = II.getOperand(2);
+
+  IRBuilder<> Builder(II.getContext());
+  Builder.SetInsertPoint(&II);
+
+  // Return true if a given instruction is an aarch64_sve_dup_x intrinsic call
+  // with a unit splat value, false otherwise.
+  auto IsUnitDupX = [](auto *I) {
+    auto *IntrI = dyn_cast<IntrinsicInst>(I);
+    if (!IntrI || IntrI->getIntrinsicID() != Intrinsic::aarch64_sve_dup_x)
+      return false;
+
+    auto *SplatValue = IntrI->getOperand(0);
+    return match(SplatValue, m_FPOne()) || match(SplatValue, m_One());
+  };
+
+  // Return true if a given instruction is an aarch64_sve_dup intrinsic call
+  // with a unit splat value, false otherwise.
+  auto IsUnitDup = [](auto *I) {
+    auto *IntrI = dyn_cast<IntrinsicInst>(I);
+    if (!IntrI || IntrI->getIntrinsicID() != Intrinsic::aarch64_sve_dup)
+      return false;
+
+    auto *SplatValue = IntrI->getOperand(2);
+    return match(SplatValue, m_FPOne()) || match(SplatValue, m_One());
+  };
+
+  // The OpMultiplier variable should always point to the dup (if any), so
+  // swap if necessary.
+  if (IsUnitDup(OpMultiplicand) || IsUnitDupX(OpMultiplicand))
+    std::swap(OpMultiplier, OpMultiplicand);
+
+  if (IsUnitDupX(OpMultiplier)) {
+    // [f]mul pg (dupx 1) %n => %n
+    OpMultiplicand->takeName(&II);
+    return IC.replaceInstUsesWith(II, OpMultiplicand);
+  } else if (IsUnitDup(OpMultiplier)) {
+    // [f]mul pg (dup pg 1) %n => %n
+    auto *DupInst = cast<IntrinsicInst>(OpMultiplier);
+    auto *DupPg = DupInst->getOperand(1);
+    // TODO: this is naive. The optimization is still valid if DupPg
+    // 'encompasses' OpPredicate, not only if they're the same predicate.
+    if (OpPredicate == DupPg) {
+      OpMultiplicand->takeName(&II);
+      return IC.replaceInstUsesWith(II, OpMultiplicand);
+    }
+  }
+
+  return None;
+}
+
+static Optional<Instruction *> instCombineSVETBL(InstCombiner &IC,
+                                                 IntrinsicInst &II) {
+  auto *OpVal = II.getOperand(0);
+  auto *OpIndices = II.getOperand(1);
+  VectorType *VTy = cast<VectorType>(II.getType());
+
+  // Check whether OpIndices is an aarch64_sve_dup_x intrinsic call with
+  // constant splat value < minimal element count of result.
+  auto *DupXIntrI = dyn_cast<IntrinsicInst>(OpIndices);
+  if (!DupXIntrI || DupXIntrI->getIntrinsicID() != Intrinsic::aarch64_sve_dup_x)
+    return None;
+
+  auto *SplatValue = dyn_cast<ConstantInt>(DupXIntrI->getOperand(0));
+  if (!SplatValue ||
+      SplatValue->getValue().uge(VTy->getElementCount().getKnownMinValue()))
+    return None;
+
+  // Convert sve_tbl(OpVal sve_dup_x(SplatValue)) to
+  // splat_vector(extractelement(OpVal, SplatValue)) for further optimization.
+  IRBuilder<> Builder(II.getContext());
+  Builder.SetInsertPoint(&II);
+  auto *Extract = Builder.CreateExtractElement(OpVal, SplatValue);
+  auto *VectorSplat =
+      Builder.CreateVectorSplat(VTy->getElementCount(), Extract);
+
+  VectorSplat->takeName(&II);
+  return IC.replaceInstUsesWith(II, VectorSplat);
+}
+
 Optional<Instruction *>
 AArch64TTIImpl::instCombineIntrinsic(InstCombiner &IC,
                                      IntrinsicInst &II) const {
   Intrinsic::ID IID = II.getIntrinsicID();
   switch (IID) {
   default:
     break;
   case Intrinsic::aarch64_sve_convert_from_svbool:
AArch64TTIImpl::instCombineIntrinsic(InstCombiner &IC,
   case Intrinsic::aarch64_sve_cntd:
     return instCombineSVECntElts(IC, II, 2);
   case Intrinsic::aarch64_sve_cntw:
     return instCombineSVECntElts(IC, II, 4);
   case Intrinsic::aarch64_sve_cnth:
     return instCombineSVECntElts(IC, II, 8);
   case Intrinsic::aarch64_sve_cntb:
     return instCombineSVECntElts(IC, II, 16);
+  case Intrinsic::aarch64_sve_ptest_any:
+  case Intrinsic::aarch64_sve_ptest_first:
+  case Intrinsic::aarch64_sve_ptest_last:
+    return instCombineSVEPTest(IC, II);
+  case Intrinsic::aarch64_sve_mul:
+  case Intrinsic::aarch64_sve_fmul:
+    return instCombineSVEVectorMul(IC, II);
+  case Intrinsic::aarch64_sve_tbl:
+    return instCombineSVETBL(IC, II);
   }
 
   return None;
 }
 
 bool AArch64TTIImpl::isWideningInstruction(Type *DstTy, unsigned Opcode,
                                            ArrayRef<const Value *> Args) {
 

llvm/lib/Target/AArch64/SVEIntrinsicOpts.cpp

struct SVEIntrinsicOpts : public ModulePass {
   bool runOnModule(Module &M) override;
   void getAnalysisUsage(AnalysisUsage &AU) const override;
 
 private:
   bool coalescePTrueIntrinsicCalls(BasicBlock &BB,
                                    SmallSetVector<IntrinsicInst *, 4> &PTrues);
   bool optimizePTrueIntrinsicCalls(SmallSetVector<Function *, 4> &Functions);
 
-  /// Operates at the instruction-scope. I.e., optimizations are applied local
-  /// to individual instructions.
-  static bool optimizeIntrinsic(Instruction *I);
-  bool optimizeIntrinsicCalls(SmallSetVector<Function *, 4> &Functions);
-
   /// Operates at the function-scope. I.e., optimizations are applied local to
   /// the functions themselves.
   bool optimizeFunctions(SmallSetVector<Function *, 4> &Functions);
-
-  static bool optimizePTest(IntrinsicInst *I);
-  static bool optimizeVectorMul(IntrinsicInst *I);
-  static bool optimizeTBL(IntrinsicInst *I);
 };
 } // end anonymous namespace
 
 void SVEIntrinsicOpts::getAnalysisUsage(AnalysisUsage &AU) const {
   AU.addRequired<DominatorTreeWrapperPass>();
   AU.setPreservesCFG();
 }
 
for (auto &BB : *F) {
       Changed |= coalescePTrueIntrinsicCalls(BB, SVAllPTrues);
       Changed |= coalescePTrueIntrinsicCalls(BB, SVPow2PTrues);
     }
   }
 
   return Changed;
 }
 
-bool SVEIntrinsicOpts::optimizePTest(IntrinsicInst *I) {
-  IntrinsicInst *Op1 = dyn_cast<IntrinsicInst>(I->getArgOperand(0));
-  IntrinsicInst *Op2 = dyn_cast<IntrinsicInst>(I->getArgOperand(1));
-
-  if (Op1 && Op2 &&
-      Op1->getIntrinsicID() == Intrinsic::aarch64_sve_convert_to_svbool &&
-      Op2->getIntrinsicID() == Intrinsic::aarch64_sve_convert_to_svbool &&
-      Op1->getArgOperand(0)->getType() == Op2->getArgOperand(0)->getType()) {
-
-    Value *Ops[] = {Op1->getArgOperand(0), Op2->getArgOperand(0)};
-    Type *Tys[] = {Op1->getArgOperand(0)->getType()};
-    Module *M = I->getParent()->getParent()->getParent();
-
-    auto Fn = Intrinsic::getDeclaration(M, I->getIntrinsicID(), Tys);
-    auto CI = CallInst::Create(Fn, Ops, I->getName(), I);
-
-    I->replaceAllUsesWith(CI);
-    I->eraseFromParent();
-    if (Op1->use_empty())
-      Op1->eraseFromParent();
-    if (Op1 != Op2 && Op2->use_empty())
-      Op2->eraseFromParent();
-
-    return true;
-  }
-
-  return false;
-}
-
-bool SVEIntrinsicOpts::optimizeVectorMul(IntrinsicInst *I) {
-  assert((I->getIntrinsicID() == Intrinsic::aarch64_sve_mul ||
-          I->getIntrinsicID() == Intrinsic::aarch64_sve_fmul) &&
-         "Unexpected opcode");
-
-  auto *OpPredicate = I->getOperand(0);
-  auto *OpMultiplicand = I->getOperand(1);
-  auto *OpMultiplier = I->getOperand(2);
-
-  // Return true if a given instruction is an aarch64_sve_dup_x intrinsic call
-  // with a unit splat value, false otherwise.
-  auto IsUnitDupX = [](auto *I) {
-    auto *IntrI = dyn_cast<IntrinsicInst>(I);
-    if (!IntrI || IntrI->getIntrinsicID() != Intrinsic::aarch64_sve_dup_x)
-      return false;
-
-    auto *SplatValue = IntrI->getOperand(0);
-    return match(SplatValue, m_FPOne()) || match(SplatValue, m_One());
-  };
-
-  // Return true if a given instruction is an aarch64_sve_dup intrinsic call
-  // with a unit splat value, false otherwise.
-  auto IsUnitDup = [](auto *I) {
-    auto *IntrI = dyn_cast<IntrinsicInst>(I);
-    if (!IntrI || IntrI->getIntrinsicID() != Intrinsic::aarch64_sve_dup)
-      return false;
-
-    auto *SplatValue = IntrI->getOperand(2);
-    return match(SplatValue, m_FPOne()) || match(SplatValue, m_One());
-  };
-
-  bool Changed = true;
-
-  // The OpMultiplier variable should always point to the dup (if any), so
-  // swap if necessary.
-  if (IsUnitDup(OpMultiplicand) || IsUnitDupX(OpMultiplicand))
-    std::swap(OpMultiplier, OpMultiplicand);
-
-  if (IsUnitDupX(OpMultiplier)) {
-    // [f]mul pg (dupx 1) %n => %n
-    I->replaceAllUsesWith(OpMultiplicand);
-    I->eraseFromParent();
-    Changed = true;
-  } else if (IsUnitDup(OpMultiplier)) {
-    // [f]mul pg (dup pg 1) %n => %n
-    auto *DupInst = cast<IntrinsicInst>(OpMultiplier);
-    auto *DupPg = DupInst->getOperand(1);
-    // TODO: this is naive. The optimization is still valid if DupPg
-    // 'encompasses' OpPredicate, not only if they're the same predicate.
-    if (OpPredicate == DupPg) {
-      I->replaceAllUsesWith(OpMultiplicand);
-      I->eraseFromParent();
-      Changed = true;
-    }
-  }
-
-  // If an instruction was optimized out then it is possible that some dangling
-  // instructions are left.
-  if (Changed) {
-    auto *OpPredicateInst = dyn_cast<Instruction>(OpPredicate);
-    auto *OpMultiplierInst = dyn_cast<Instruction>(OpMultiplier);
-    if (OpMultiplierInst && OpMultiplierInst->use_empty())
-      OpMultiplierInst->eraseFromParent();
-    if (OpPredicateInst && OpPredicateInst->use_empty())
-      OpPredicateInst->eraseFromParent();
-  }
-
-  return Changed;
-}
-
-bool SVEIntrinsicOpts::optimizeTBL(IntrinsicInst *I) {
-  assert(I->getIntrinsicID() == Intrinsic::aarch64_sve_tbl &&
-         "Unexpected opcode");
-
-  auto *OpVal = I->getOperand(0);
-  auto *OpIndices = I->getOperand(1);
-  VectorType *VTy = cast<VectorType>(I->getType());
-
-  // Check whether OpIndices is an aarch64_sve_dup_x intrinsic call with
-  // constant splat value < minimal element count of result.
-  auto *DupXIntrI = dyn_cast<IntrinsicInst>(OpIndices);
-  if (!DupXIntrI || DupXIntrI->getIntrinsicID() != Intrinsic::aarch64_sve_dup_x)
-    return false;
-
-  auto *SplatValue = dyn_cast<ConstantInt>(DupXIntrI->getOperand(0));
-  if (!SplatValue ||
-      SplatValue->getValue().uge(VTy->getElementCount().getKnownMinValue()))
-    return false;
-
-  // Convert sve_tbl(OpVal sve_dup_x(SplatValue)) to
-  // splat_vector(extractelement(OpVal, SplatValue)) for further optimization.
-  LLVMContext &Ctx = I->getContext();
-  IRBuilder<> Builder(Ctx);
-  Builder.SetInsertPoint(I);
-  auto *Extract = Builder.CreateExtractElement(OpVal, SplatValue);
-  auto *VectorSplat =
-      Builder.CreateVectorSplat(VTy->getElementCount(), Extract);
-
-  I->replaceAllUsesWith(VectorSplat);
-  I->eraseFromParent();
-  if (DupXIntrI->use_empty())
-    DupXIntrI->eraseFromParent();
-  return true;
-}
-
-bool SVEIntrinsicOpts::optimizeIntrinsic(Instruction *I) {
-  IntrinsicInst *IntrI = dyn_cast<IntrinsicInst>(I);
-  if (!IntrI)
-    return false;
-
-  switch (IntrI->getIntrinsicID()) {
-  case Intrinsic::aarch64_sve_fmul:
-  case Intrinsic::aarch64_sve_mul:
-    return optimizeVectorMul(IntrI);
-  case Intrinsic::aarch64_sve_ptest_any:
-  case Intrinsic::aarch64_sve_ptest_first:
-  case Intrinsic::aarch64_sve_ptest_last:
-    return optimizePTest(IntrI);
-  case Intrinsic::aarch64_sve_tbl:
-    return optimizeTBL(IntrI);
-  default:
-    return false;
-  }
-
-  return true;
-}
-
-bool SVEIntrinsicOpts::optimizeIntrinsicCalls(
-    SmallSetVector<Function *, 4> &Functions) {
-  bool Changed = false;
-  for (auto *F : Functions) {
-    DominatorTree *DT = &getAnalysis<DominatorTreeWrapperPass>(*F).getDomTree();
-
-    // Traverse the DT with an rpo walk so we see defs before uses, allowing
-    // simplification to be done incrementally.
-    BasicBlock *Root = DT->getRoot();
-    ReversePostOrderTraversal<BasicBlock *> RPOT(Root);
-    for (auto *BB : RPOT)
-      for (Instruction &I : make_early_inc_range(*BB))
-        Changed |= optimizeIntrinsic(&I);
-  }
-  return Changed;
-}
-
 bool SVEIntrinsicOpts::optimizeFunctions(
     SmallSetVector<Function *, 4> &Functions) {
   bool Changed = false;
 
   Changed |= optimizePTrueIntrinsicCalls(Functions);
-  Changed |= optimizeIntrinsicCalls(Functions);
 
   return Changed;
 }
 
 bool SVEIntrinsicOpts::runOnModule(Module &M) {
   bool Changed = false;
   SmallSetVector<Function *, 4> Functions;
 
   // Check for SVE intrinsic declarations first so that we only iterate over
   // relevant functions. Where an appropriate declaration is found, store the
   // function(s) where it is used so we can target these only.
   for (auto &F : M.getFunctionList()) {
     if (!F.isDeclaration())
       continue;
 
     switch (F.getIntrinsicID()) {
-    case Intrinsic::aarch64_sve_ptest_any:
-    case Intrinsic::aarch64_sve_ptest_first:
-    case Intrinsic::aarch64_sve_ptest_last:
     case Intrinsic::aarch64_sve_ptrue:
-    case Intrinsic::aarch64_sve_mul:
-    case Intrinsic::aarch64_sve_fmul:
-    case Intrinsic::aarch64_sve_tbl:
       for (User *U : F.users())
         Functions.insert(cast<Instruction>(U)->getFunction());
       break;
     default:
       break;
     }
   }
 
   if (!Functions.empty())
     Changed |= optimizeFunctions(Functions);
 
   return Changed;
 }

llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-fmul-idempotency.ll

-; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt -S -aarch64-sve-intrinsic-opts < %s | FileCheck %s
+; RUN: opt -S -instcombine < %s | FileCheck %s
+
+target triple = "aarch64-unknown-linux-gnu"
 
 ; Idempotent fmuls -- should compile to just a ret.
-define <vscale x 8 x half> @idempotent_fmul_f16(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a) {
+define <vscale x 8 x half> @idempotent_fmul_f16(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a) #0 {
 ; CHECK-LABEL: @idempotent_fmul_f16(
 ; CHECK-NEXT:    ret <vscale x 8 x half> [[A:%.*]]
 ;
   %1 = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 1.0)
   %2 = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
   ret <vscale x 8 x half> %2
 }
 
-define <vscale x 4 x float> @idempotent_fmul_f32(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a) {
+define <vscale x 4 x float> @idempotent_fmul_f32(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a) #0 {
 ; CHECK-LABEL: @idempotent_fmul_f32(
 ; CHECK-NEXT:    ret <vscale x 4 x float> [[A:%.*]]
 ;
   %1 = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float 1.0)
   %2 = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a, <vscale x 4 x float> %1)
   ret <vscale x 4 x float> %2
 }
 
-define <vscale x 2 x double> @idempotent_fmul_f64(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a) {
+define <vscale x 2 x double> @idempotent_fmul_f64(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a) #0 {
 ; CHECK-LABEL: @idempotent_fmul_f64(
 ; CHECK-NEXT:    ret <vscale x 2 x double> [[A:%.*]]
 ;
   %1 = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double 1.0)
   %2 = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a, <vscale x 2 x double> %1)
   ret <vscale x 2 x double> %2
 }
 
-define <vscale x 2 x double> @idempotent_fmul_different_argument_order(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a) {
+define <vscale x 2 x double> @idempotent_fmul_different_argument_order(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a) #0 {
 ; CHECK-LABEL: @idempotent_fmul_different_argument_order(
 ; CHECK-NEXT:    ret <vscale x 2 x double> [[A:%.*]]
 ;
   %1 = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double 1.0)
   ; Different argument order to the above tests.
   %2 = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> %pg, <vscale x 2 x double> %1, <vscale x 2 x double> %a)
   ret <vscale x 2 x double> %2
 }
 
-define <vscale x 8 x half> @idempotent_fmul_with_predicated_dup(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a) {
+define <vscale x 8 x half> @idempotent_fmul_with_predicated_dup(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a) #0 {
 ; CHECK-LABEL: @idempotent_fmul_with_predicated_dup(
 ; CHECK-NEXT:    ret <vscale x 8 x half> [[A:%.*]]
 ;
   %1 = call <vscale x 8 x half> @llvm.aarch64.sve.dup.nxv8f16(<vscale x 8 x half> undef, <vscale x 8 x i1> %pg, half 1.0)
   %2 = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
   ret <vscale x 8 x half> %2
 }
 
-define <vscale x 8 x half> @idempotent_fmul_two_dups(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a) {
+define <vscale x 8 x half> @idempotent_fmul_two_dups(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a) #0 {
   ; Edge case -- make sure that the case where we're fmultiplying two dups
   ; together is sane.
 ; CHECK-LABEL: @idempotent_fmul_two_dups(
 ; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 0xH3C00)
 ; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP1]]
 ;
   %1 = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 1.0)
   %2 = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 1.0)
   %3 = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %pg, <vscale x 8 x half> %1, <vscale x 8 x half> %2)
   ret <vscale x 8 x half> %3
 }
 
 ; Non-idempotent fmuls -- we don't expect these to be optimised out.
-define <vscale x 8 x half> @non_idempotent_fmul_f16(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a) {
+define <vscale x 8 x half> @non_idempotent_fmul_f16(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a) #0 {
 ; CHECK-LABEL: @non_idempotent_fmul_f16(
 ; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 0xH4000)
 ; CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[PG:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[TMP1]])
 ; CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 ;
   %1 = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 2.0)
   %2 = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
   ret <vscale x 8 x half> %2
 }
 
-define <vscale x 4 x float> @non_idempotent_fmul_f32(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a) {
+define <vscale x 4 x float> @non_idempotent_fmul_f32(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a) #0 {
 ; CHECK-LABEL: @non_idempotent_fmul_f32(
 ; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float 2.000000e+00)
 ; CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[PG:%.*]], <vscale x 4 x float> [[A:%.*]], <vscale x 4 x float> [[TMP1]])
 ; CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 ;
   %1 = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float 2.0)
   %2 = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a, <vscale x 4 x float> %1)
   ret <vscale x 4 x float> %2
 }
 
-define <vscale x 2 x double> @non_idempotent_fmul_f64(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a) {
+define <vscale x 2 x double> @non_idempotent_fmul_f64(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a) #0 {
 ; CHECK-LABEL: @non_idempotent_fmul_f64(
 ; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double 2.000000e+00)
 ; CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[PG:%.*]], <vscale x 2 x double> [[A:%.*]], <vscale x 2 x double> [[TMP1]])
 ; CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 ;
   %1 = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double 2.0)
   %2 = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a, <vscale x 2 x double> %1)
   ret <vscale x 2 x double> %2
 }
 
-define <vscale x 2 x double> @non_idempotent_fmul_with_predicated_dup(<vscale x 2 x i1> %pg1, <vscale x 2 x i1> %pg2, <vscale x 2 x double> %a) {
+define <vscale x 2 x double> @non_idempotent_fmul_with_predicated_dup(<vscale x 2 x i1> %pg1, <vscale x 2 x i1> %pg2, <vscale x 2 x double> %a) #0 {
   ; Different predicates
 ; CHECK-LABEL: @non_idempotent_fmul_with_predicated_dup(
 ; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> [[PG1:%.*]], double 1.000000e+00)
 ; CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[PG2:%.*]], <vscale x 2 x double> [[A:%.*]], <vscale x 2 x double> [[TMP1]])
 ; CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 ;
   %1 = call <vscale x 2 x double> @llvm.aarch64.sve.dup.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> %pg1, double 1.0)
   %2 = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> %pg2, <vscale x 2 x double> %a, <vscale x 2 x double> %1)
   ret <vscale x 2 x double> %2
 }
 
 declare <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half)
 declare <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float)
 declare <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double)
 
 declare <vscale x 2 x double> @llvm.aarch64.sve.dup.nxv2f64(<vscale x 2 x double>, <vscale x 2 x i1>, double)
 declare <vscale x 8 x half> @llvm.aarch64.sve.dup.nxv8f16(<vscale x 8 x half>, <vscale x 8 x i1>, half)
 
 declare <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1>, <vscale x 8 x half>, <vscale x 8 x half>)
 declare <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1>, <vscale x 4 x float>, <vscale x 4 x float>)
 declare <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1>, <vscale x 2 x double>, <vscale x 2 x double>)
+
+attributes #0 = { "target-features"="+sve" }

llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-mul-idempotency.ll

-; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt -S -aarch64-sve-intrinsic-opts < %s | FileCheck %s
+; RUN: opt -S -instcombine < %s | FileCheck %s
+
+target triple = "aarch64-unknown-linux-gnu"
 
 ; Idempotent muls -- should compile to just a ret.
-define <vscale x 8 x i16> @idempotent_mul_i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) {
+define <vscale x 8 x i16> @idempotent_mul_i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) #0 {
 ; CHECK-LABEL: @idempotent_mul_i16(
 ; CHECK-NEXT:    ret <vscale x 8 x i16> [[A:%.*]]
 ;
   %1 = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 1)
   %2 = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a, <vscale x 8 x i16> %1)
   ret <vscale x 8 x i16> %2
 }
 
-define <vscale x 4 x i32> @idempotent_mul_i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) {
+define <vscale x 4 x i32> @idempotent_mul_i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) #0 {
 ; CHECK-LABEL: @idempotent_mul_i32(
 ; CHECK-NEXT:    ret <vscale x 4 x i32> [[A:%.*]]
 ;
   %1 = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
   %2 = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a, <vscale x 4 x i32> %1)
   ret <vscale x 4 x i32> %2
 }
 
-define <vscale x 2 x i64> @idempotent_mul_i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) {
+define <vscale x 2 x i64> @idempotent_mul_i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) #0 {
 ; CHECK-LABEL: @idempotent_mul_i64(
 ; CHECK-NEXT:    ret <vscale x 2 x i64> [[A:%.*]]
 ;
   %1 = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 1)
   %2 = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a, <vscale x 2 x i64> %1)
   ret <vscale x 2 x i64> %2
 }
 
-define <vscale x 2 x i64> @idempotent_mul_different_argument_order(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) {
+define <vscale x 2 x i64> @idempotent_mul_different_argument_order(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) #0 {
 ; CHECK-LABEL: @idempotent_mul_different_argument_order(
 ; CHECK-NEXT:    ret <vscale x 2 x i64> [[A:%.*]]
 ;
   %1 = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 1)
   ; Different argument order to the above tests.
   %2 = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %1, <vscale x 2 x i64> %a)
   ret <vscale x 2 x i64> %2
 }
 
-define <vscale x 8 x i16> @idempotent_mul_with_predicated_dup(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) {
+define <vscale x 8 x i16> @idempotent_mul_with_predicated_dup(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) #0 {
 ; CHECK-LABEL: @idempotent_mul_with_predicated_dup(
 ; CHECK-NEXT:    ret <vscale x 8 x i16> [[A:%.*]]
 ;
   %1 = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.nxv8i16(<vscale x 8 x i16> undef, <vscale x 8 x i1> %pg, i16 1)
   %2 = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a, <vscale x 8 x i16> %1)
   ret <vscale x 8 x i16> %2
 }
 
-define <vscale x 8 x i16> @idempotent_mul_two_dups(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) {
+define <vscale x 8 x i16> @idempotent_mul_two_dups(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) #0 {
   ; Edge case -- make sure that the case where we're multiplying two dups
   ; together is sane.
 ; CHECK-LABEL: @idempotent_mul_two_dups(
 ; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 1)
 ; CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 ;
   %1 = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 1)
   %2 = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 1)
   %3 = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %1, <vscale x 8 x i16> %2)
   ret <vscale x 8 x i16> %3
 }
 
 ; Non-idempotent muls -- we don't expect these to be optimised out.
-define <vscale x 8 x i16> @non_idempotent_mul_i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) {
+define <vscale x 8 x i16> @non_idempotent_mul_i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) #0 {
 ; CHECK-LABEL: @non_idempotent_mul_i16(
 ; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 2)
 ; CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[PG:%.*]], <vscale x 8 x i16> [[A:%.*]], <vscale x 8 x i16> [[TMP1]])
 ; CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 ;
   %1 = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 2)
   %2 = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a, <vscale x 8 x i16> %1)
   ret <vscale x 8 x i16> %2
 }
 
-define <vscale x 4 x i32> @non_idempotent_mul_i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) {
+define <vscale x 4 x i32> @non_idempotent_mul_i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) #0 {
 ; CHECK-LABEL: @non_idempotent_mul_i32(
 ; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 2)
 ; CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[PG:%.*]], <vscale x 4 x i32> [[A:%.*]], <vscale x 4 x i32> [[TMP1]])
 ; CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 ;
   %1 = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 2)
   %2 = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a, <vscale x 4 x i32> %1)
   ret <vscale x 4 x i32> %2
 }
 
-define <vscale x 2 x i64> @non_idempotent_mul_i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) {
+define <vscale x 2 x i64> @non_idempotent_mul_i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) #0 {
 ; CHECK-LABEL: @non_idempotent_mul_i64(
 ; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 2)
 ; CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[PG:%.*]], <vscale x 2 x i64> [[A:%.*]], <vscale x 2 x i64> [[TMP1]])
 ; CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 ;
   %1 = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 2)
   %2 = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a, <vscale x 2 x i64> %1)
   ret <vscale x 2 x i64> %2
 }
 
-define <vscale x 2 x i64> @non_idempotent_mul_with_predicated_dup(<vscale x 2 x i1> %pg1, <vscale x 2 x i1> %pg2, <vscale x 2 x i64> %a) {
+define <vscale x 2 x i64> @non_idempotent_mul_with_predicated_dup(<vscale x 2 x i1> %pg1, <vscale x 2 x i1> %pg2, <vscale x 2 x i64> %a) #0 {
   ; Different predicates
 ; CHECK-LABEL: @non_idempotent_mul_with_predicated_dup(
 ; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> [[PG1:%.*]], i64 1)
 ; CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[PG2:%.*]], <vscale x 2 x i64> [[A:%.*]], <vscale x 2 x i64> [[TMP1]])
 ; CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 ;
   %1 = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.nxv2i64(<vscale x 2 x i64> undef, <vscale x 2 x i1> %pg1, i64 1)
   %2 = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> %pg2, <vscale x 2 x i64> %a, <vscale x 2 x i64> %1)
   ret <vscale x 2 x i64> %2
 }
 
 declare <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16)
 declare <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32)
 declare <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64)
 
 declare <vscale x 2 x i64> @llvm.aarch64.sve.dup.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i1>, i64)
 declare <vscale x 8 x i16> @llvm.aarch64.sve.dup.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i1>, i16)
 
 declare <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1>, <vscale x 8 x i16>, <vscale x 8 x i16>)
 declare <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1>, <vscale x 4 x i32>, <vscale x 4 x i32>)
 declare <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1>, <vscale x 2 x i64>, <vscale x 2 x i64>)
+
+attributes #0 = { "target-features"="+sve" }

llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-tbl-dupx.ll

-; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt -S -aarch64-sve-intrinsic-opts -mtriple=aarch64-linux-gnu -mattr=+sve < %s | FileCheck %s
+; RUN: opt -S -instcombine < %s | FileCheck %s
+
+target triple = "aarch64-unknown-linux-gnu"
 
 ; op2 = tbl(op1 dup_x(idx)) -> op2 = vector_splat(extractelement(op1, idx))
 
-define <vscale x 16 x i8> @dup_ext_i8(<vscale x 16 x i8> %data) {
+define <vscale x 16 x i8> @dup_ext_i8(<vscale x 16 x i8> %data) #0 {
 ; CHECK-LABEL: @dup_ext_i8(
 ; CHECK-NEXT:    [[TMP1:%.*]] = extractelement <vscale x 16 x i8> [[DATA:%.*]], i8 1
 ; CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[TMP1]], i32 0
 ; CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 ; CHECK-NEXT:    ret <vscale x 16 x i8> [[DOTSPLAT]]
 ;
   %tmp = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 1)
   %out = call <vscale x 16 x i8> @llvm.aarch64.sve.tbl.nxv16i8(<vscale x 16 x i8>  %data, <vscale x 16 x i8> %tmp)
   ret <vscale x 16 x i8> %out
 }
 
-define <vscale x 8 x i16> @dup_ext_i16(<vscale x 8 x i16> %data) {
+define <vscale x 8 x i16> @dup_ext_i16(<vscale x 8 x i16> %data) #0 {
 ; CHECK-LABEL: @dup_ext_i16(
 ; CHECK-NEXT:    [[TMP1:%.*]] = extractelement <vscale x 8 x i16> [[DATA:%.*]], i16 1
 ; CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[TMP1]], i32 0
 ; CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 ; CHECK-NEXT:    ret <vscale x 8 x i16> [[DOTSPLAT]]
 ;
   %tmp = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 1)
   %out = call <vscale x 8 x i16> @llvm.aarch64.sve.tbl.nxv8i16(<vscale x 8 x i16>  %data, <vscale x 8 x i16> %tmp)
   ret <vscale x 8 x i16> %out
 }
 
-define <vscale x 4 x i32> @dup_ext_i32(<vscale x 4 x i32> %data) {
+define <vscale x 4 x i32> @dup_ext_i32(<vscale x 4 x i32> %data) #0 {
 ; CHECK-LABEL: @dup_ext_i32(
 ; CHECK-NEXT:    [[TMP1:%.*]] = extractelement <vscale x 4 x i32> [[DATA:%.*]], i32 1
 ; CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[TMP1]], i32 0
 ; CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 ; CHECK-NEXT:    ret <vscale x 4 x i32> [[DOTSPLAT]]
 ;
   %tmp = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
   %out = call <vscale x 4 x i32> @llvm.aarch64.sve.tbl.nxv4i32(<vscale x 4 x i32>  %data, <vscale x 4 x i32> %tmp)
   ret <vscale x 4 x i32> %out
 }
 
-define <vscale x 2 x i64> @dup_ext_i64(<vscale x 2 x i64> %data) {
+define <vscale x 2 x i64> @dup_ext_i64(<vscale x 2 x i64> %data) #0 {
 ; CHECK-LABEL: @dup_ext_i64(
 ; CHECK-NEXT:    [[TMP1:%.*]] = extractelement <vscale x 2 x i64> [[DATA:%.*]], i64 1
 ; CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[TMP1]], i32 0
 ; CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 ; CHECK-NEXT:    ret <vscale x 2 x i64> [[DOTSPLAT]]
 ;
   %tmp = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 1)
   %out = call <vscale x 2 x i64> @llvm.aarch64.sve.tbl.nxv2i64(<vscale x 2 x i64>  %data, <vscale x 2 x i64> %tmp)
   ret <vscale x 2 x i64> %out
 }
 
-define <vscale x 8 x half> @dup_ext_f16(<vscale x 8 x half> %data) {
+define <vscale x 8 x half> @dup_ext_f16(<vscale x 8 x half> %data) #0 {
 ; CHECK-LABEL: @dup_ext_f16(
 ; CHECK-NEXT:    [[TMP1:%.*]] = extractelement <vscale x 8 x half> [[DATA:%.*]], i16 1
 ; CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[TMP1]], i32 0
 ; CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 ; CHECK-NEXT:    ret <vscale x 8 x half> [[DOTSPLAT]]
 ;
   %tmp = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 1)
   %out = call <vscale x 8 x half> @llvm.aarch64.sve.tbl.nxv8f16(<vscale x 8 x half>  %data, <vscale x 8 x i16> %tmp)
   ret <vscale x 8 x half> %out
 }
 
-define <vscale x 4 x float> @dup_ext_f32(<vscale x 4 x float> %data) {
+define <vscale x 4 x float> @dup_ext_f32(<vscale x 4 x float> %data) #0 {
 ; CHECK-LABEL: @dup_ext_f32(
 ; CHECK-NEXT:    [[TMP1:%.*]] = extractelement <vscale x 4 x float> [[DATA:%.*]], i32 1
 ; CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[TMP1]], i32 0
 ; CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 ; CHECK-NEXT:    ret <vscale x 4 x float> [[DOTSPLAT]]
 ;
   %tmp = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
   %out = call <vscale x 4 x float> @llvm.aarch64.sve.tbl.nxv4f32(<vscale x 4 x float>  %data, <vscale x 4 x i32> %tmp)
   ret <vscale x 4 x float> %out
 }
 
-define <vscale x 2 x double> @dup_ext_f64(<vscale x 2 x double> %data) {
+define <vscale x 2 x double> @dup_ext_f64(<vscale x 2 x double> %data) #0 {
 ; CHECK-LABEL: @dup_ext_f64(
 ; CHECK-NEXT:    [[TMP1:%.*]] = extractelement <vscale x 2 x double> [[DATA:%.*]], i64 1
 ; CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[TMP1]], i32 0
 ; CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 ; CHECK-NEXT:    ret <vscale x 2 x double> [[DOTSPLAT]]
 ;
   %tmp = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 1)
   %out = call <vscale x 2 x double> @llvm.aarch64.sve.tbl.nxv2f64(<vscale x 2 x double> %data, <vscale x 2 x i64> %tmp)
   ret <vscale x 2 x double> %out
 }
 
 declare <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8( i8)
 declare <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16)
 declare <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32)
 declare <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64)
 declare <vscale x 16 x i8> @llvm.aarch64.sve.tbl.nxv16i8( <vscale x 16 x i8>, <vscale x 16 x i8>)
 declare <vscale x 8 x i16> @llvm.aarch64.sve.tbl.nxv8i16( <vscale x 8 x i16>, <vscale x 8 x i16>)
 declare <vscale x 4 x i32> @llvm.aarch64.sve.tbl.nxv4i32( <vscale x 4 x i32>, <vscale x 4 x i32>)
 declare <vscale x 2 x i64> @llvm.aarch64.sve.tbl.nxv2i64( <vscale x 2 x i64>, <vscale x 2 x i64>)
 declare <vscale x 8 x half> @llvm.aarch64.sve.tbl.nxv8f16( <vscale x 8 x half>, <vscale x 8 x i16>)
 declare <vscale x 4 x float> @llvm.aarch64.sve.tbl.nxv4f32( <vscale x 4 x float>, <vscale x 4 x i32>)
 declare <vscale x 2 x double> @llvm.aarch64.sve.tbl.nxv2f64( <vscale x 2 x double>, <vscale x 2 x i64>)
+
+attributes #0 = { "target-features"="+sve" }

llvm/test/Transforms/InstCombine/AArch64/sve-intrinsics-ptest.ll

-; RUN: opt -S -aarch64-sve-intrinsic-opts -mtriple=aarch64-linux-gnu -mattr=+sve < %s | FileCheck --check-prefix OPT %s
+; RUN: opt -S -instcombine < %s | FileCheck %s
 
-define i1 @ptest_any1(<vscale x 2 x i1> %a) {
-; OPT-LABEL: ptest_any1
-; OPT: %mask = tail call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 0)
-; OPT-NOT: convert
-; OPT-NEXT: %[[OUT:.*]] = call i1 @llvm.aarch64.sve.ptest.any.nxv2i1(<vscale x 2 x i1> %mask, <vscale x 2 x i1> %a)
-; OPT-NEXT: ret i1 %[[OUT]]
+target triple = "aarch64-unknown-linux-gnu"
+
+define i1 @ptest_any1(<vscale x 2 x i1> %a) #0 {
+; CHECK-LABEL: ptest_any1
+; CHECK: %mask = tail call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 0)
+; CHECK-NOT: convert
+; CHECK-NEXT: %[[OUT:.*]] = call i1 @llvm.aarch64.sve.ptest.any.nxv2i1(<vscale x 2 x i1> %mask, <vscale x 2 x i1> %a)
+; CHECK-NEXT: ret i1 %[[OUT]]
   %mask = tail call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 0)
   %1 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> %mask)
   %2 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> %a)
   %out = call i1 @llvm.aarch64.sve.ptest.any.nxv16i1(<vscale x 16 x i1> %1, <vscale x 16 x i1> %2)
   ret i1 %out
 }
 
 ; No transform because the ptest is using differently sized operands.
-define i1 @ptest_any2(<vscale x 4 x i1> %a) {
-; OPT-LABEL: ptest_any2
-; OPT: %mask = tail call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
-; OPT-NEXT: %1 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> %mask)
-; OPT-NEXT: %2 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> %a)
-; OPT-NEXT: %out = call i1 @llvm.aarch64.sve.ptest.any.nxv16i1(<vscale x 16 x i1> %1, <vscale x 16 x i1> %2)
+define i1 @ptest_any2(<vscale x 4 x i1> %a) #0 {
+; CHECK-LABEL: ptest_any2
+; CHECK: %mask = tail call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
+; CHECK-NEXT: %1 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> %mask)
+; CHECK-NEXT: %2 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> %a)
+; CHECK-NEXT: %out = call i1 @llvm.aarch64.sve.ptest.any.nxv16i1(<vscale x 16 x i1> %1, <vscale x 16 x i1> %2)
   %mask = tail call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
   %1 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> %mask)
   %2 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> %a)
   %out = call i1 @llvm.aarch64.sve.ptest.any.nxv16i1(<vscale x 16 x i1> %1, <vscale x 16 x i1> %2)
   ret i1 %out
 }
 
-define i1 @ptest_first(<vscale x 4 x i1> %a) {
-; OPT-LABEL: ptest_first
-; OPT: %mask = tail call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 0)
-; OPT-NOT: convert
-; OPT-NEXT: %[[OUT:.*]] = call i1 @llvm.aarch64.sve.ptest.first.nxv4i1(<vscale x 4 x i1> %mask, <vscale x 4 x i1> %a)
-; OPT-NEXT: ret i1 %[[OUT]]
+define i1 @ptest_first(<vscale x 4 x i1> %a) #0 {
+; CHECK-LABEL: ptest_first
+; CHECK: %mask = tail call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 0)
+; CHECK-NOT: convert
+; CHECK-NEXT: %[[OUT:.*]] = call i1 @llvm.aarch64.sve.ptest.first.nxv4i1(<vscale x 4 x i1> %mask, <vscale x 4 x i1> %a)
+; CHECK-NEXT: ret i1 %[[OUT]]
   %mask = tail call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 0)
   %1 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> %mask)
   %2 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> %a)
   %out = call i1 @llvm.aarch64.sve.ptest.first.nxv16i1(<vscale x 16 x i1> %1, <vscale x 16 x i1> %2)
   ret i1 %out
 }
 
-define i1 @ptest_first_same_ops(<vscale x 2 x i1> %a) {
-; OPT-LABEL: ptest_first_same_ops
-; OPT: %[[OUT:.*]] = call i1 @llvm.aarch64.sve.ptest.first.nxv2i1(<vscale x 2 x i1> %a, <vscale x 2 x i1> %a)
-; OPT-NOT: convert
-; OPT-NEXT: ret i1 %[[OUT]]
+define i1 @ptest_first_same_ops(<vscale x 2 x i1> %a) #0 {
+; CHECK-LABEL: ptest_first_same_ops
+; CHECK: %[[OUT:.*]] = call i1 @llvm.aarch64.sve.ptest.first.nxv2i1(<vscale x 2 x i1> %a, <vscale x 2 x i1> %a)
+; CHECK-NOT: convert
+; CHECK-NEXT: ret i1 %[[OUT]]
   %1 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> %a)
   %2 = tail call i1 @llvm.aarch64.sve.ptest.first.nxv16i1(<vscale x 16 x i1> %1, <vscale x 16 x i1> %1)
   ret i1 %2
 }
 
-define i1 @ptest_last(<vscale x 8 x i1> %a) {
-; OPT-LABEL: ptest_last
-; OPT: %mask = tail call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 0)
-; OPT-NOT: convert
-; OPT-NEXT: %[[OUT:.*]] = call i1 @llvm.aarch64.sve.ptest.last.nxv8i1(<vscale x 8 x i1> %mask, <vscale x 8 x i1> %a)
-; OPT-NEXT: ret i1 %[[OUT]]
+define i1 @ptest_last(<vscale x 8 x i1> %a) #0 {
+; CHECK-LABEL: ptest_last
+; CHECK: %mask = tail call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 0)
+; CHECK-NOT: convert
+; CHECK-NEXT: %[[OUT:.*]] = call i1 @llvm.aarch64.sve.ptest.last.nxv8i1(<vscale x 8 x i1> %mask, <vscale x 8 x i1> %a)
+; CHECK-NEXT: ret i1 %[[OUT]]
   %mask = tail call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 0)
   %1 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> %mask)
   %2 = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> %a)
   %out = call i1 @llvm.aarch64.sve.ptest.last.nxv16i1(<vscale x 16 x i1> %1, <vscale x 16 x i1> %2)
   ret i1 %out
 }
 
 declare <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32)
 declare <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32)
 declare <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32)
 declare <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32)
 
 declare i1 @llvm.aarch64.sve.ptest.any.nxv16i1(<vscale x 16 x i1>, <vscale x 16 x i1>)
 declare i1 @llvm.aarch64.sve.ptest.first.nxv16i1(<vscale x 16 x i1>, <vscale x 16 x i1>)
 declare i1 @llvm.aarch64.sve.ptest.last.nxv16i1(<vscale x 16 x i1>, <vscale x 16 x i1>)
 
 declare <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1>)
 declare <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1>)
 declare <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1>)
+
+attributes #0 = { "target-features"="+sve" }