[clang] Replace uses of CGBuilderTy::CreateElementBitCast (NFC)

[JOE1994]

Summary

Partial progress towards replacing CreateElementBitCast, as it no
longer does what its name suggests. Either replace its uses with
Address::withElementType(), or remove them if no longer needed.

[nikic]

clang/lib/CodeGen/CGNonTrivialStruct.cpp

368–369

Only BC.getPointer() is used, so you can omit the withElementType.

clang/lib/CodeGen/MicrosoftCXXABI.cpp

3134–3135

3135–3138

Can drop this bitcast while here.

clang/lib/CodeGen/Targets/AArch64.cpp

522

Can inline this withElementType call into Address ctor.

[barannikov88]

clang/lib/CodeGen/CGAtomic.cpp

1474

as it no longer emits anything.

[JOE1994]

• Rebase onto latest main
• Apply feedback from reviews
  • Rename AtomicInfo::emitCastToAtomicIntPointer to AtomicInfo::castToAtomicIntPointer

[nikic]

clang/lib/CodeGen/Targets/AArch64.cpp

521–522

[JOE1994]

Apply suggested change from nikic
(sorry, I missed it out in the previous update)

[nikic]

LGTM

clang/lib/CodeGen/CGAtomic.cpp

Address getAtomicAddress() const {
       else if (LVal.isVectorElt())
         ElTy = LVal.getVectorAddress().getElementType();
       else
         ElTy = LVal.getExtVectorAddress().getElementType();
       return Address(getAtomicPointer(), ElTy, getAtomicAlignment());
     }
 
     Address getAtomicAddressAsAtomicIntPointer() const {
-      return emitCastToAtomicIntPointer(getAtomicAddress());
+      return castToAtomicIntPointer(getAtomicAddress());
     }
 
     /// Is the atomic size larger than the underlying value type?
     ///
     /// Note that the absence of padding does not mean that atomic
     /// objects are completely interchangeable with non-atomic
     /// objects: we might have promoted the alignment of a type
     /// without making it bigger.
     bool hasPadding() const {
       return (ValueSizeInBits != AtomicSizeInBits);
     }
 
     bool emitMemSetZeroIfNecessary() const;
 
     llvm::Value *getAtomicSizeValue() const {
       CharUnits size = CGF.getContext().toCharUnitsFromBits(AtomicSizeInBits);
       return CGF.CGM.getSize(size);
     }
 
     /// Cast the given pointer to an integer pointer suitable for atomic
     /// operations if the source.
-    Address emitCastToAtomicIntPointer(Address Addr) const;
+    Address castToAtomicIntPointer(Address Addr) const;
 
     /// If Addr is compatible with the iN that will be used for an atomic
     /// operation, bitcast it. Otherwise, create a temporary that is suitable
     /// and copy the value across.
     Address convertToAtomicIntPointer(Address Addr) const;
 
     /// Turn an atomic-layout object into an r-value.
     RValue convertAtomicTempToRValue(Address addr, AggValueSlot resultSlot,
RValue CodeGenFunction::EmitAtomicExpr(AtomicExpr *E) {
 
   // The inlined atomics only function on iN types, where N is a power of 2. We
   // need to make sure (via temporaries if necessary) that all incoming values
   // are compatible.
   LValue AtomicVal = MakeAddrLValue(Ptr, AtomicTy);
   AtomicInfo Atomics(*this, AtomicVal);
 
   if (ShouldCastToIntPtrTy) {
-    Ptr = Atomics.emitCastToAtomicIntPointer(Ptr);
+    Ptr = Atomics.castToAtomicIntPointer(Ptr);
     if (Val1.isValid())
       Val1 = Atomics.convertToAtomicIntPointer(Val1);
     if (Val2.isValid())
       Val2 = Atomics.convertToAtomicIntPointer(Val2);
   }
   if (Dest.isValid()) {
     if (ShouldCastToIntPtrTy)
-      Dest = Atomics.emitCastToAtomicIntPointer(Dest);
+      Dest = Atomics.castToAtomicIntPointer(Dest);
   } else if (E->isCmpXChg())
     Dest = CreateMemTemp(RValTy, "cmpxchg.bool");
   else if (!RValTy->isVoidType()) {
     Dest = Atomics.CreateTempAlloca();
     if (ShouldCastToIntPtrTy)
-      Dest = Atomics.emitCastToAtomicIntPointer(Dest);
+      Dest = Atomics.castToAtomicIntPointer(Dest);
   }
 
   // Use a library call.  See: http://gcc.gnu.org/wiki/Atomic/GCCMM/LIbrary .
   if (UseLibcall) {
     bool UseOptimizedLibcall = false;
     switch (E->getOp()) {
     case AtomicExpr::AO__c11_atomic_init:
     case AtomicExpr::AO__opencl_atomic_init:
if (UseOptimizedLibcall && Res.getScalarVal()) {
                                       ResVal, LoadVal1);
       } else if (PostOp) {
         llvm::Value *LoadVal1 = Args[1].getRValue(*this).getScalarVal();
         ResVal = Builder.CreateBinOp(PostOp, ResVal, LoadVal1);
       }
       if (E->getOp() == AtomicExpr::AO__atomic_nand_fetch)
         ResVal = Builder.CreateNot(ResVal);
 
-      Builder.CreateStore(
-          ResVal, Builder.CreateElementBitCast(Dest, ResVal->getType()));
+      Builder.CreateStore(ResVal, Dest.withElementType(ResVal->getType()));
     }
 
     if (RValTy->isVoidType())
       return RValue::get(nullptr);
 
-    return convertTempToRValue(
-        Builder.CreateElementBitCast(Dest, ConvertTypeForMem(RValTy)),
+    return convertTempToRValue(Dest.withElementType(ConvertTypeForMem(RValTy)),
                                RValTy, E->getExprLoc());
   }
 
   bool IsStore = E->getOp() == AtomicExpr::AO__c11_atomic_store ||
                  E->getOp() == AtomicExpr::AO__opencl_atomic_store ||
                  E->getOp() == AtomicExpr::AO__hip_atomic_store ||
                  E->getOp() == AtomicExpr::AO__atomic_store ||
                  E->getOp() == AtomicExpr::AO__atomic_store_n;
   bool IsLoad = E->getOp() == AtomicExpr::AO__c11_atomic_load ||
if (llvm::isValidAtomicOrderingCABI(ord))
       case llvm::AtomicOrderingCABI::seq_cst:
         EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail, Size,
                      llvm::AtomicOrdering::SequentiallyConsistent, Scope);
         break;
       }
     if (RValTy->isVoidType())
       return RValue::get(nullptr);
 
-    return convertTempToRValue(
-        Builder.CreateElementBitCast(Dest, ConvertTypeForMem(RValTy)),
+    return convertTempToRValue(Dest.withElementType(ConvertTypeForMem(RValTy)),
                                RValTy, E->getExprLoc());
   }
 
   // Long case, when Order isn't obviously constant.
 
   // Create all the relevant BB's
   llvm::BasicBlock *MonotonicBB = nullptr, *AcquireBB = nullptr,
                    *ReleaseBB = nullptr, *AcqRelBB = nullptr,
                    *SeqCstBB = nullptr;
SI->addCase(Builder.getInt32((int)llvm::AtomicOrderingCABI::seq_cst),
               SeqCstBB);
 
   // Cleanup and return
   Builder.SetInsertPoint(ContBB);
   if (RValTy->isVoidType())
     return RValue::get(nullptr);
 
   assert(Atomics.getValueSizeInBits() <= Atomics.getAtomicSizeInBits());
-  return convertTempToRValue(
-      Builder.CreateElementBitCast(Dest, ConvertTypeForMem(RValTy)),
+  return convertTempToRValue(Dest.withElementType(ConvertTypeForMem(RValTy)),
                              RValTy, E->getExprLoc());
 }
 
-Address AtomicInfo::emitCastToAtomicIntPointer(Address addr) const {
+Address AtomicInfo::castToAtomicIntPointer(Address addr) const {
   llvm::IntegerType *ty =
     llvm::IntegerType::get(CGF.getLLVMContext(), AtomicSizeInBits);
-  return CGF.Builder.CreateElementBitCast(addr, ty);
+  return addr.withElementType(ty);
 }
 
 Address AtomicInfo::convertToAtomicIntPointer(Address Addr) const {
   llvm::Type *Ty = Addr.getElementType();
   uint64_t SourceSizeInBits = CGF.CGM.getDataLayout().getTypeSizeInBits(Ty);
   if (SourceSizeInBits != AtomicSizeInBits) {
     Address Tmp = CreateTempAlloca();
     CGF.Builder.CreateMemCpy(Tmp, Addr,
                              std::min(AtomicSizeInBits, SourceSizeInBits) / 8);
     Addr = Tmp;
   }
 
-  return emitCastToAtomicIntPointer(Addr);
+  return castToAtomicIntPointer(Addr);
 }
 
 RValue AtomicInfo::convertAtomicTempToRValue(Address addr,
                                              AggValueSlot resultSlot,
                                              SourceLocation loc,
                                              bool asValue) const {
   if (LVal.isSimple()) {
     if (EvaluationKind == TEK_Aggregate)
if (AsValue && getEvaluationKind() == TEK_Aggregate) {
     assert(!ResultSlot.isIgnored());
     Temp = ResultSlot.getAddress();
     TempIsVolatile = ResultSlot.isVolatile();
   } else {
     Temp = CreateTempAlloca();
   }
 
   // Slam the integer into the temporary.
-  Address CastTemp = emitCastToAtomicIntPointer(Temp);
+  Address CastTemp = castToAtomicIntPointer(Temp);
   CGF.Builder.CreateStore(IntVal, CastTemp)
       ->setVolatile(TempIsVolatile);
 
   return convertAtomicTempToRValue(Temp, ResultSlot, Loc, AsValue);
 }
 
 void AtomicInfo::EmitAtomicLoadLibcall(llvm::Value *AddForLoaded,
                                        llvm::AtomicOrdering AO, bool) {
else {
         return CGF.Builder.CreateBitCast(Value, InputIntTy);
     }
   }
   // Otherwise, we need to go through memory.
   // Put the r-value in memory.
   Address Addr = materializeRValue(RVal);
 
   // Cast the temporary to the atomic int type and pull a value out.
-  Addr = emitCastToAtomicIntPointer(Addr);
+  Addr = castToAtomicIntPointer(Addr);
   return CGF.Builder.CreateLoad(Addr);
 }
 
 std::pair<llvm::Value *, llvm::Value *> AtomicInfo::EmitAtomicCompareExchangeOp(
     llvm::Value *ExpectedVal, llvm::Value *DesiredVal,
     llvm::AtomicOrdering Success, llvm::AtomicOrdering Failure, bool IsWeak) {
   // Do the atomic store.
   Address Addr = getAtomicAddressAsAtomicIntPointer();
void AtomicInfo::EmitAtomicUpdateOp(
   auto *ContBB = CGF.createBasicBlock("atomic_cont");
   auto *ExitBB = CGF.createBasicBlock("atomic_exit");
   auto *CurBB = CGF.Builder.GetInsertBlock();
   CGF.EmitBlock(ContBB);
   llvm::PHINode *PHI = CGF.Builder.CreatePHI(OldVal->getType(),
                                              /*NumReservedValues=*/2);
   PHI->addIncoming(OldVal, CurBB);
   Address NewAtomicAddr = CreateTempAlloca();
-  Address NewAtomicIntAddr = emitCastToAtomicIntPointer(NewAtomicAddr);
+  Address NewAtomicIntAddr = castToAtomicIntPointer(NewAtomicAddr);
   if ((LVal.isBitField() && BFI.Size != ValueSizeInBits) ||
       requiresMemSetZero(getAtomicAddress().getElementType())) {
     CGF.Builder.CreateStore(PHI, NewAtomicIntAddr);
   }
   auto OldRVal = ConvertIntToValueOrAtomic(PHI, AggValueSlot::ignored(),
                                            SourceLocation(), /*AsValue=*/false);
   EmitAtomicUpdateValue(CGF, *this, OldRVal, UpdateOp, NewAtomicAddr);
   auto *DesiredVal = CGF.Builder.CreateLoad(NewAtomicIntAddr);
void AtomicInfo::EmitAtomicUpdateOp(llvm::AtomicOrdering AO, RValue UpdateRVal,
   auto *ContBB = CGF.createBasicBlock("atomic_cont");
   auto *ExitBB = CGF.createBasicBlock("atomic_exit");
   auto *CurBB = CGF.Builder.GetInsertBlock();
   CGF.EmitBlock(ContBB);
   llvm::PHINode *PHI = CGF.Builder.CreatePHI(OldVal->getType(),
                                              /*NumReservedValues=*/2);
   PHI->addIncoming(OldVal, CurBB);
   Address NewAtomicAddr = CreateTempAlloca();
-  Address NewAtomicIntAddr = emitCastToAtomicIntPointer(NewAtomicAddr);
+  Address NewAtomicIntAddr = castToAtomicIntPointer(NewAtomicAddr);
   if ((LVal.isBitField() && BFI.Size != ValueSizeInBits) ||
       requiresMemSetZero(getAtomicAddress().getElementType())) {
     CGF.Builder.CreateStore(PHI, NewAtomicIntAddr);
   }
   EmitAtomicUpdateValue(CGF, *this, UpdateRVal, NewAtomicAddr);
   auto *DesiredVal = CGF.Builder.CreateLoad(NewAtomicIntAddr);
   // Try to write new value using cmpxchg operation.
   auto Res = EmitAtomicCompareExchangeOp(PHI, DesiredVal, AO, Failure);
if (atomics.shouldUseLibcall()) {
       emitAtomicLibcall(*this, "__atomic_store", getContext().VoidTy, args);
       return;
     }
 
     // Okay, we're doing this natively.
     llvm::Value *intValue = atomics.convertRValueToInt(rvalue);
 
     // Do the atomic store.
-    Address addr =
-        atomics.emitCastToAtomicIntPointer(atomics.getAtomicAddress());
+    Address addr = atomics.castToAtomicIntPointer(atomics.getAtomicAddress());
     intValue = Builder.CreateIntCast(
         intValue, addr.getElementType(), /*isSigned=*/false);
     llvm::StoreInst *store = Builder.CreateStore(intValue, addr);
 
     if (AO == llvm::AtomicOrdering::Acquire)
       AO = llvm::AtomicOrdering::Monotonic;
     else if (AO == llvm::AtomicOrdering::AcquireRelease)
       AO = llvm::AtomicOrdering::Release;

clang/lib/CodeGen/CGCXXABI.cpp

bool CGCXXABI::requiresArrayCookie(const CXXNewExpr *expr) {
   return expr->getAllocatedType().isDestructedType();
 }
 
 void CGCXXABI::ReadArrayCookie(CodeGenFunction &CGF, Address ptr,
                                const CXXDeleteExpr *expr, QualType eltTy,
                                llvm::Value *&numElements,
                                llvm::Value *&allocPtr, CharUnits &cookieSize) {
   // Derive a char* in the same address space as the pointer.
-  ptr = CGF.Builder.CreateElementBitCast(ptr, CGF.Int8Ty);
+  ptr = ptr.withElementType(CGF.Int8Ty);
 
   // If we don't need an array cookie, bail out early.
   if (!requiresArrayCookie(expr, eltTy)) {
     allocPtr = ptr.getPointer();
     numElements = nullptr;
     cookieSize = CharUnits::Zero();
     return;
   }

clang/lib/CodeGen/CGNonTrivialStruct.cpp

for (unsigned I = 0; I < N; ++I)
       StartAddrs[I] = getAddrWithOffset(Addrs[I], CurStructOffset, FD);
     Address DstAddr = StartAddrs[DstIdx];
     llvm::Value *NumElts = CGF.emitArrayLength(AT, BaseEltQT, DstAddr);
     unsigned BaseEltSize = Ctx.getTypeSizeInChars(BaseEltQT).getQuantity();
     llvm::Value *BaseEltSizeVal =
         llvm::ConstantInt::get(NumElts->getType(), BaseEltSize);
     llvm::Value *SizeInBytes =
         CGF.Builder.CreateNUWMul(BaseEltSizeVal, NumElts);
-    Address BC = CGF.Builder.CreateElementBitCast(DstAddr, CGF.CGM.Int8Ty);
-    llvm::Value *DstArrayEnd =
-        CGF.Builder.CreateInBoundsGEP(CGF.Int8Ty, BC.getPointer(), SizeInBytes);
+    llvm::Value *DstArrayEnd = CGF.Builder.CreateInBoundsGEP(
+        CGF.Int8Ty, DstAddr.getPointer(), SizeInBytes);
     DstArrayEnd = CGF.Builder.CreateBitCast(
         DstArrayEnd, CGF.CGM.Int8PtrPtrTy, "dstarray.end");
     llvm::BasicBlock *PreheaderBB = CGF.Builder.GetInsertBlock();
 
     // Create the header block and insert the phi instructions.
     llvm::BasicBlock *HeaderBB = CGF.createBasicBlock("loop.header");
     CGF.EmitBlock(HeaderBB);
     llvm::PHINode *PHIs[N];
void visitArray(FieldKind FK, const ArrayType *AT, bool IsVolatile,
     CGF.EmitBlock(ExitBB);
   }
 
   /// Return an address with the specified offset from the passed address.
   Address getAddrWithOffset(Address Addr, CharUnits Offset) {
     assert(Addr.isValid() && "invalid address");
     if (Offset.getQuantity() == 0)
       return Addr;
-    Addr = CGF->Builder.CreateElementBitCast(Addr, CGF->CGM.Int8Ty);
+    Addr = Addr.withElementType(CGF->CGM.Int8Ty);
     Addr = CGF->Builder.CreateConstInBoundsGEP(Addr, Offset.getQuantity());
-    return CGF->Builder.CreateElementBitCast(Addr, CGF->CGM.Int8PtrTy);
+    return Addr.withElementType(CGF->CGM.Int8PtrTy);
   }
 
   Address getAddrWithOffset(Address Addr, CharUnits StructFieldOffset,
                             const FieldDecl *FD) {
     return getAddrWithOffset(Addr, StructFieldOffset +
                                        asDerived().getFieldOffset(FD));
   }
 
template <class Derived> struct GenFuncBase {
   template <size_t N>
   void callFunc(StringRef FuncName, QualType QT, std::array<Address, N> Addrs,
                 CodeGenFunction &CallerCGF) {
     std::array<CharUnits, N> Alignments;
     llvm::Value *Ptrs[N];
 
     for (unsigned I = 0; I < N; ++I) {
       Alignments[I] = Addrs[I].getAlignment();
-      Ptrs[I] = CallerCGF.Builder.CreateElementBitCast(
-          Addrs[I], CallerCGF.CGM.Int8PtrTy).getPointer();
+      Ptrs[I] = Addrs[I].getPointer();
     }
 
     if (llvm::Function *F =
             getFunction(FuncName, QT, Alignments, CallerCGF.CGM))
       CallerCGF.EmitNounwindRuntimeCall(F, Ptrs);
   }
 
   Derived &asDerived() { return static_cast<Derived &>(*this); }
void flushTrivialFields(std::array<Address, 2> Addrs) {
     Address DstAddr = this->getAddrWithOffset(Addrs[DstIdx], this->Start);
     Address SrcAddr = this->getAddrWithOffset(Addrs[SrcIdx], this->Start);
 
     // Emit memcpy.
     if (Size.getQuantity() >= 16 ||
         !llvm::has_single_bit<uint32_t>(Size.getQuantity())) {
       llvm::Value *SizeVal =
           llvm::ConstantInt::get(this->CGF->SizeTy, Size.getQuantity());
-      DstAddr =
-          this->CGF->Builder.CreateElementBitCast(DstAddr, this->CGF->Int8Ty);
-      SrcAddr =
-          this->CGF->Builder.CreateElementBitCast(SrcAddr, this->CGF->Int8Ty);
+      DstAddr = DstAddr.withElementType(this->CGF->Int8Ty);
+      SrcAddr = SrcAddr.withElementType(this->CGF->Int8Ty);
       this->CGF->Builder.CreateMemCpy(DstAddr, SrcAddr, SizeVal, false);
     } else {
       llvm::Type *Ty = llvm::Type::getIntNTy(
           this->CGF->getLLVMContext(),
           Size.getQuantity() * this->CGF->getContext().getCharWidth());
-      DstAddr = this->CGF->Builder.CreateElementBitCast(DstAddr, Ty);
-      SrcAddr = this->CGF->Builder.CreateElementBitCast(SrcAddr, Ty);
+      DstAddr = DstAddr.withElementType(Ty);
+      SrcAddr = SrcAddr.withElementType(Ty);
       llvm::Value *SrcVal = this->CGF->Builder.CreateLoad(SrcAddr, false);
       this->CGF->Builder.CreateStore(SrcVal, DstAddr, false);
     }
 
     this->Start = this->End = CharUnits::Zero();
   }
 
   template <class... Ts>
   void visitVolatileTrivial(QualType FT, const FieldDecl *FD, CharUnits Offset,
                             std::array<Address, 2> Addrs) {
     LValue DstLV, SrcLV;
     if (FD) {
       // No need to copy zero-length bit-fields.
       if (FD->isZeroLengthBitField(this->CGF->getContext()))
         return;
 
       QualType RT = QualType(FD->getParent()->getTypeForDecl(), 0);
       llvm::Type *Ty = this->CGF->ConvertType(RT);
       Address DstAddr = this->getAddrWithOffset(Addrs[DstIdx], Offset);
-      LValue DstBase = this->CGF->MakeAddrLValue(
-          this->CGF->Builder.CreateElementBitCast(DstAddr, Ty), FT);
+      LValue DstBase =
+          this->CGF->MakeAddrLValue(DstAddr.withElementType(Ty), FT);
       DstLV = this->CGF->EmitLValueForField(DstBase, FD);
       Address SrcAddr = this->getAddrWithOffset(Addrs[SrcIdx], Offset);
-      LValue SrcBase = this->CGF->MakeAddrLValue(
-          this->CGF->Builder.CreateElementBitCast(SrcAddr, Ty), FT);
+      LValue SrcBase =
+          this->CGF->MakeAddrLValue(SrcAddr.withElementType(Ty), FT);
       SrcLV = this->CGF->EmitLValueForField(SrcBase, FD);
     } else {
       llvm::Type *Ty = this->CGF->ConvertTypeForMem(FT);
-      Address DstAddr =
-          this->CGF->Builder.CreateElementBitCast(Addrs[DstIdx], Ty);
-      Address SrcAddr =
-          this->CGF->Builder.CreateElementBitCast(Addrs[SrcIdx], Ty);
+      Address DstAddr = Addrs[DstIdx].withElementType(Ty);
+      Address SrcAddr = Addrs[SrcIdx].withElementType(Ty);
       DstLV = this->CGF->MakeAddrLValue(DstAddr, FT);
       SrcLV = this->CGF->MakeAddrLValue(SrcAddr, FT);
     }
     RValue SrcVal = this->CGF->EmitLoadOfLValue(SrcLV, SourceLocation());
     this->CGF->EmitStoreThroughLValue(SrcVal, DstLV);
   }
 };
 
void visitArray(FieldKind FK, const ArrayType *AT, bool IsVolatile,
 
     if (Size < CharUnits::fromQuantity(16) || EltTy->getAs<RecordType>()) {
       GenFuncBaseTy::visitArray(FK, AT, IsVolatile, FD, CurStructOffset, Addrs);
       return;
     }
 
     llvm::Constant *SizeVal = CGF->Builder.getInt64(Size.getQuantity());
     Address DstAddr = getAddrWithOffset(Addrs[DstIdx], CurStructOffset, FD);
-    Address Loc = CGF->Builder.CreateElementBitCast(DstAddr, CGF->Int8Ty);
+    Address Loc = DstAddr.withElementType(CGF->Int8Ty);
     CGF->Builder.CreateMemSet(Loc, CGF->Builder.getInt8(0), SizeVal,
                               IsVolatile);
   }
 
   void callSpecialFunction(QualType FT, CharUnits Offset,
                            std::array<Address, 1> Addrs) {
     CGF->callCStructDefaultConstructor(
         CGF->MakeAddrLValue(getAddrWithOffset(Addrs[DstIdx], Offset), FT));
void CodeGenFunction::destroyNonTrivialCStruct(CodeGenFunction &CGF,
                                                Address Addr, QualType Type) {
   CGF.callCStructDestructor(CGF.MakeAddrLValue(Addr, Type));
 }
 
 // Default-initialize a variable that is a non-trivial struct or an array of
 // such structure.
 void CodeGenFunction::defaultInitNonTrivialCStructVar(LValue Dst) {
   GenDefaultInitialize Gen(getContext());
-  Address DstPtr =
-      Builder.CreateElementBitCast(Dst.getAddress(*this), CGM.Int8PtrTy);
+  Address DstPtr = Dst.getAddress(*this).withElementType(CGM.Int8PtrTy);
   Gen.setCGF(this);
   QualType QT = Dst.getType();
   QT = Dst.isVolatile() ? QT.withVolatile() : QT;
   Gen.visit(QT, nullptr, CharUnits::Zero(), std::array<Address, 1>({{DstPtr}}));
 }
 
 template <class G, size_t N>
 static void callSpecialFunction(G &&Gen, StringRef FuncName, QualType QT,
                                 bool IsVolatile, CodeGenFunction &CGF,
                                 std::array<Address, N> Addrs) {
   auto SetArtificialLoc = ApplyDebugLocation::CreateArtificial(CGF);
   for (unsigned I = 0; I < N; ++I)
-    Addrs[I] = CGF.Builder.CreateElementBitCast(Addrs[I], CGF.CGM.Int8PtrTy);
+    Addrs[I] = Addrs[I].withElementType(CGF.CGM.Int8PtrTy);
   QT = IsVolatile ? QT.withVolatile() : QT;
   Gen.callFunc(FuncName, QT, Addrs, CGF);
 }
 
 template <class G, size_t N>
 static llvm::Function *
 getSpecialFunction(G &&Gen, StringRef FuncName, QualType QT, bool IsVolatile,
                    std::array<CharUnits, N> Alignments, CodeGenModule &CGM) {

clang/lib/CodeGen/MicrosoftCXXABI.cpp

 }
 
 /// We need to perform a generic polymorphic operation (like a typeid
 /// or a cast), which requires an object with a vfptr.  Adjust the
 /// address to point to an object with a vfptr.
 std::tuple<Address, llvm::Value *, const CXXRecordDecl *>
 MicrosoftCXXABI::performBaseAdjustment(CodeGenFunction &CGF, Address Value,
                                        QualType SrcRecordTy) {
-  Value = CGF.Builder.CreateElementBitCast(Value, CGF.Int8Ty);
+  Value = Value.withElementType(CGF.Int8Ty);
   const CXXRecordDecl *SrcDecl = SrcRecordTy->getAsCXXRecordDecl();
   const ASTContext &Context = getContext();
 
   // If the class itself has a vfptr, great.  This check implicitly
   // covers non-virtual base subobjects: a class with its own virtual
   // functions would be a candidate to be a primary base.
   if (Context.getASTRecordLayout(SrcDecl).hasExtendableVFPtr())
     return std::make_tuple(Value, llvm::ConstantInt::get(CGF.Int32Ty, 0),
Address MicrosoftCXXABI::adjustThisArgumentForVirtualFunctionCall(
     bool VirtualCall) {
   if (!VirtualCall) {
     // If the call of a virtual function is not virtual, we just have to
     // compensate for the adjustment the virtual function does in its prologue.
     CharUnits Adjustment = getVirtualFunctionPrologueThisAdjustment(GD);
     if (Adjustment.isZero())
       return This;
 
-    This = CGF.Builder.CreateElementBitCast(This, CGF.Int8Ty);
+    This = This.withElementType(CGF.Int8Ty);
     assert(Adjustment.isPositive());
     return CGF.Builder.CreateConstByteGEP(This, Adjustment);
   }
 
   const CXXMethodDecl *MD = cast<CXXMethodDecl>(GD.getDecl());
 
   GlobalDecl LookupGD = GD;
   if (const CXXDestructorDecl *DD = dyn_cast<CXXDestructorDecl>(MD)) {
Address MicrosoftCXXABI::adjustThisArgumentForVirtualFunctionCall(
   // Base destructors expect 'this' to point to the beginning of the base
   // subobject, not the first vfptr that happens to contain the virtual dtor.
   // However, we still need to apply the virtual base adjustment.
   if (isa<CXXDestructorDecl>(MD) && GD.getDtorType() == Dtor_Base)
     StaticOffset = CharUnits::Zero();
 
   Address Result = This;
   if (ML.VBase) {
-    Result = CGF.Builder.CreateElementBitCast(Result, CGF.Int8Ty);
+    Result = Result.withElementType(CGF.Int8Ty);
 
     const CXXRecordDecl *Derived = MD->getParent();
     const CXXRecordDecl *VBase = ML.VBase;
     llvm::Value *VBaseOffset =
       GetVirtualBaseClassOffset(CGF, Result, Derived, VBase);
     llvm::Value *VBasePtr = CGF.Builder.CreateInBoundsGEP(
         Result.getElementType(), Result.getPointer(), VBaseOffset);
     CharUnits VBaseAlign =
       CGF.CGM.getVBaseAlignment(Result.getAlignment(), Derived, VBase);
     Result = Address(VBasePtr, CGF.Int8Ty, VBaseAlign);
   }
   if (!StaticOffset.isZero()) {
     assert(StaticOffset.isPositive());
-    Result = CGF.Builder.CreateElementBitCast(Result, CGF.Int8Ty);
+    Result = Result.withElementType(CGF.Int8Ty);
     if (ML.VBase) {
       // Non-virtual adjustment might result in a pointer outside the allocated
       // object, e.g. if the final overrider class is laid out after the virtual
       // base that declares a method in the most derived class.
       // FIXME: Update the code that emits this adjustment in thunks prologues.
       Result = CGF.Builder.CreateConstByteGEP(Result, StaticOffset);
     } else {
       Result = CGF.Builder.CreateConstInBoundsByteGEP(Result, StaticOffset);

 }
 
 llvm::Value *MicrosoftCXXABI::performThisAdjustment(CodeGenFunction &CGF,
                                                     Address This,
                                                     const ThisAdjustment &TA) {
   if (TA.isEmpty())
     return This.getPointer();
 
-  This = CGF.Builder.CreateElementBitCast(This, CGF.Int8Ty);
+  This = This.withElementType(CGF.Int8Ty);
 
   llvm::Value *V;
   if (TA.Virtual.isEmpty()) {
     V = This.getPointer();
   } else {
     assert(TA.Virtual.Microsoft.VtordispOffset < 0);
     // Adjust the this argument based on the vtordisp value.
     Address VtorDispPtr =
         CGF.Builder.CreateConstInBoundsByteGEP(This,
                  CharUnits::fromQuantity(TA.Virtual.Microsoft.VtordispOffset));
-    VtorDispPtr = CGF.Builder.CreateElementBitCast(VtorDispPtr, CGF.Int32Ty);
+    VtorDispPtr = VtorDispPtr.withElementType(CGF.Int32Ty);
     llvm::Value *VtorDisp = CGF.Builder.CreateLoad(VtorDispPtr, "vtordisp");
     V = CGF.Builder.CreateGEP(This.getElementType(), This.getPointer(),
                               CGF.Builder.CreateNeg(VtorDisp));
 
     // Unfortunately, having applied the vtordisp means that we no
     // longer really have a known alignment for the vbptr step.
     // We'll assume the vbptr is pointer-aligned.
 

 
 llvm::Value *
 MicrosoftCXXABI::performReturnAdjustment(CodeGenFunction &CGF, Address Ret,
                                          const ReturnAdjustment &RA) {
   if (RA.isEmpty())
     return Ret.getPointer();
 
   auto OrigTy = Ret.getType();
-  Ret = CGF.Builder.CreateElementBitCast(Ret, CGF.Int8Ty);
+  Ret = Ret.withElementType(CGF.Int8Ty);
 
   llvm::Value *V = Ret.getPointer();
   if (RA.Virtual.Microsoft.VBIndex) {
     assert(RA.Virtual.Microsoft.VBIndex > 0);
     int32_t IntSize = CGF.getIntSize().getQuantity();
     llvm::Value *VBPtr;
     llvm::Value *VBaseOffset =
         GetVBaseOffsetFromVBPtr(CGF, Ret, RA.Virtual.Microsoft.VBPtrOffset,
CharUnits MicrosoftCXXABI::getArrayCookieSizeImpl(QualType type) {
   ASTContext &Ctx = getContext();
   return std::max(Ctx.getTypeSizeInChars(Ctx.getSizeType()),
                   Ctx.getTypeAlignInChars(type));
 }
 
 llvm::Value *MicrosoftCXXABI::readArrayCookieImpl(CodeGenFunction &CGF,
                                                   Address allocPtr,
                                                   CharUnits cookieSize) {
-  Address numElementsPtr =
-    CGF.Builder.CreateElementBitCast(allocPtr, CGF.SizeTy);
+  Address numElementsPtr = allocPtr.withElementType(CGF.SizeTy);
   return CGF.Builder.CreateLoad(numElementsPtr);
 }
 
 Address MicrosoftCXXABI::InitializeArrayCookie(CodeGenFunction &CGF,
                                                Address newPtr,
                                                llvm::Value *numElements,
                                                const CXXNewExpr *expr,
                                                QualType elementType) {
   assert(requiresArrayCookie(expr));
 
   // The size of the cookie.
   CharUnits cookieSize = getArrayCookieSizeImpl(elementType);
 
   // Compute an offset to the cookie.
   Address cookiePtr = newPtr;
 
   // Write the number of elements into the appropriate slot.
-  Address numElementsPtr
-    = CGF.Builder.CreateElementBitCast(cookiePtr, CGF.SizeTy);
+  Address numElementsPtr = cookiePtr.withElementType(CGF.SizeTy);
   CGF.Builder.CreateStore(numElements, numElementsPtr);
 
   // Finally, compute a pointer to the actual data buffer by skipping
   // over the cookie completely.
   return CGF.Builder.CreateConstInBoundsByteGEP(newPtr, cookieSize);
 }
 
 static void emitGlobalDtorWithTLRegDtor(CodeGenFunction &CGF, const VarDecl &VD,

 
 llvm::Value *
 MicrosoftCXXABI::GetVBaseOffsetFromVBPtr(CodeGenFunction &CGF,
                                          Address This,
                                          llvm::Value *VBPtrOffset,
                                          llvm::Value *VBTableOffset,
                                          llvm::Value **VBPtrOut) {
   CGBuilderTy &Builder = CGF.Builder;
   // Load the vbtable pointer from the vbptr in the instance.
-  This = Builder.CreateElementBitCast(This, CGM.Int8Ty);
-  llvm::Value *VBPtr = Builder.CreateInBoundsGEP(
-      This.getElementType(), This.getPointer(), VBPtrOffset, "vbptr");
-  if (VBPtrOut) *VBPtrOut = VBPtr;
-  VBPtr = Builder.CreateBitCast(VBPtr,
-            CGM.Int32Ty->getPointerTo(0)->getPointerTo(This.getAddressSpace()));
+  llvm::Value *VBPtr = Builder.CreateInBoundsGEP(CGM.Int8Ty, This.getPointer(),
+                                                 VBPtrOffset, "vbptr");
+  if (VBPtrOut)
+    *VBPtrOut = VBPtr;
 
   CharUnits VBPtrAlign;
   if (auto CI = dyn_cast<llvm::ConstantInt>(VBPtrOffset)) {
     VBPtrAlign = This.getAlignment().alignmentAtOffset(
                                    CharUnits::fromQuantity(CI->getSExtValue()));
   } else {
     VBPtrAlign = CGF.getPointerAlign();
   }
 
   llvm::Value *VBTable = Builder.CreateAlignedLoad(
       CGM.Int32Ty->getPointerTo(0), VBPtr, VBPtrAlign, "vbtable");
 
   // Translate from byte offset to table index. It improves analyzability.
   llvm::Value *VBTableIndex = Builder.CreateAShr(
       VBTableOffset, llvm::ConstantInt::get(VBTableOffset->getType(), 2),
       "vbtindex", /*isExact=*/true);
 
   // Load an i32 offset from the vb-table.
   llvm::Value *VBaseOffs =
       Builder.CreateInBoundsGEP(CGM.Int32Ty, VBTable, VBTableIndex);
-  VBaseOffs = Builder.CreateBitCast(VBaseOffs, CGM.Int32Ty->getPointerTo(0));
   return Builder.CreateAlignedLoad(CGM.Int32Ty, VBaseOffs,
                                    CharUnits::fromQuantity(4), "vbase_offs");
 }
 
 // Returns an adjusted base cast to i8*, since we do more address arithmetic on
 // it.
 llvm::Value *MicrosoftCXXABI::AdjustVirtualBase(
     CodeGenFunction &CGF, const Expr *E, const CXXRecordDecl *RD,
     Address Base, llvm::Value *VBTableOffset, llvm::Value *VBPtrOffset) {
   CGBuilderTy &Builder = CGF.Builder;
-  Base = Builder.CreateElementBitCast(Base, CGM.Int8Ty);
+  Base = Base.withElementType(CGM.Int8Ty);
   llvm::BasicBlock *OriginalBB = nullptr;
   llvm::BasicBlock *SkipAdjustBB = nullptr;
   llvm::BasicBlock *VBaseAdjustBB = nullptr;
 
   // In the unspecified inheritance model, there might not be a vbtable at all,
   // in which case we need to skip the virtual base lookup.  If there is a
   // vbtable, the first entry is a no-op entry that gives back the original
   // base, so look for a virtual base adjustment offset of zero.

clang/lib/CodeGen/Targets/AArch64.cpp

 Address AArch64ABIInfo::EmitAAPCSVAArg(Address VAListAddr, QualType Ty,
                                        CodeGenFunction &CGF) const {
   ABIArgInfo AI = classifyArgumentType(Ty, /*IsVariadic=*/true,
                                        CGF.CurFnInfo->getCallingConvention());
   // Empty records are ignored for parameter passing purposes.
   if (AI.isIgnore()) {
     uint64_t PointerSize = getTarget().getPointerWidth(LangAS::Default) / 8;
     CharUnits SlotSize = CharUnits::fromQuantity(PointerSize);
-    VAListAddr = CGF.Builder.CreateElementBitCast(VAListAddr, CGF.Int8PtrTy);
+    VAListAddr = VAListAddr.withElementType(CGF.Int8PtrTy);
     auto *Load = CGF.Builder.CreateLoad(VAListAddr);
-    Address Addr = Address(Load, CGF.Int8Ty, SlotSize);
-    return CGF.Builder.CreateElementBitCast(Addr, CGF.ConvertTypeForMem(Ty));
+    return Address(Load, CGF.ConvertTypeForMem(Ty), SlotSize);
   }
 
   bool IsIndirect = AI.isIndirect();
 
   llvm::Type *BaseTy = CGF.ConvertType(Ty);
   if (IsIndirect)
     BaseTy = llvm::PointerType::getUnqual(BaseTy);
   else if (AI.getCoerceToType())
if (IsHFA && NumMembers > 1) {
     if (CGF.CGM.getDataLayout().isBigEndian() &&
         BaseTyInfo.Width.getQuantity() < 16)
       Offset = 16 - BaseTyInfo.Width.getQuantity();
 
     for (unsigned i = 0; i < NumMembers; ++i) {
       CharUnits BaseOffset = CharUnits::fromQuantity(16 * i + Offset);
       Address LoadAddr =
         CGF.Builder.CreateConstInBoundsByteGEP(BaseAddr, BaseOffset);
-      LoadAddr = CGF.Builder.CreateElementBitCast(LoadAddr, BaseTy);
+      LoadAddr = LoadAddr.withElementType(BaseTy);
 
       Address StoreAddr = CGF.Builder.CreateConstArrayGEP(Tmp, i);
 
       llvm::Value *Elem = CGF.Builder.CreateLoad(LoadAddr);
       CGF.Builder.CreateStore(Elem, StoreAddr);
     }
 
-    RegAddr = CGF.Builder.CreateElementBitCast(Tmp, MemTy);
+    RegAddr = Tmp.withElementType(MemTy);
   } else {
     // Otherwise the object is contiguous in memory.
 
     // It might be right-aligned in its slot.
     CharUnits SlotSize = BaseAddr.getAlignment();
     if (CGF.CGM.getDataLayout().isBigEndian() && !IsIndirect &&
         (IsHFA || !isAggregateTypeForABI(Ty)) &&
         TySize < SlotSize) {
       CharUnits Offset = SlotSize - TySize;
       BaseAddr = CGF.Builder.CreateConstInBoundsByteGEP(BaseAddr, Offset);
     }
 
-    RegAddr = CGF.Builder.CreateElementBitCast(BaseAddr, MemTy);
+    RegAddr = BaseAddr.withElementType(MemTy);
   }
 
   CGF.EmitBranch(ContBlock);
 
   //=======================================
   // Argument was on the stack
   //=======================================
   CGF.EmitBlock(OnStackBlock);
Address AArch64ABIInfo::EmitAAPCSVAArg(Address VAListAddr, QualType Ty,
   CGF.Builder.CreateStore(NewStack, stack_p);
 
   if (CGF.CGM.getDataLayout().isBigEndian() && !isAggregateTypeForABI(Ty) &&
       TySize < StackSlotSize) {
     CharUnits Offset = StackSlotSize - TySize;
     OnStackAddr = CGF.Builder.CreateConstInBoundsByteGEP(OnStackAddr, Offset);
   }
 
-  OnStackAddr = CGF.Builder.CreateElementBitCast(OnStackAddr, MemTy);
+  OnStackAddr = OnStackAddr.withElementType(MemTy);
 
   CGF.EmitBranch(ContBlock);
 
   //=======================================
   // Tidy up
   //=======================================
   CGF.EmitBlock(ContBlock);
 
Address AArch64ABIInfo::EmitDarwinVAArg(Address VAListAddr, QualType Ty,
   // the LLVM va_arg instruction for everything else.
   if (!isAggregateTypeForABI(Ty) && !isIllegalVectorType(Ty))
     return EmitVAArgInstr(CGF, VAListAddr, Ty, ABIArgInfo::getDirect());
 
   uint64_t PointerSize = getTarget().getPointerWidth(LangAS::Default) / 8;
   CharUnits SlotSize = CharUnits::fromQuantity(PointerSize);
 
   // Empty records are ignored for parameter passing purposes.
-  if (isEmptyRecord(getContext(), Ty, true)) {
-    Address Addr = Address(CGF.Builder.CreateLoad(VAListAddr, "ap.cur"),
-                           getVAListElementType(CGF), SlotSize);
-    Addr = CGF.Builder.CreateElementBitCast(Addr, CGF.ConvertTypeForMem(Ty));
-    return Addr;
-  }
+  if (isEmptyRecord(getContext(), Ty, true))
+    return Address(CGF.Builder.CreateLoad(VAListAddr, "ap.cur"),
+                   CGF.ConvertTypeForMem(Ty), SlotSize);
 
   // The size of the actual thing passed, which might end up just
   // being a pointer for indirect types.
   auto TyInfo = getContext().getTypeInfoInChars(Ty);
 
   // Arguments bigger than 16 bytes which aren't homogeneous
   // aggregates should be passed indirectly.
   bool IsIndirect = false;

clang/lib/CodeGen/Targets/ARM.cpp

 }
 
 Address ARMABIInfo::EmitVAArg(CodeGenFunction &CGF, Address VAListAddr,
                               QualType Ty) const {
   CharUnits SlotSize = CharUnits::fromQuantity(4);
 
   // Empty records are ignored for parameter passing purposes.
   if (isEmptyRecord(getContext(), Ty, true)) {
-    VAListAddr = CGF.Builder.CreateElementBitCast(VAListAddr, CGF.Int8PtrTy);
+    VAListAddr = VAListAddr.withElementType(CGF.Int8PtrTy);
     auto *Load = CGF.Builder.CreateLoad(VAListAddr);
-    Address Addr = Address(Load, CGF.Int8Ty, SlotSize);
-    return CGF.Builder.CreateElementBitCast(Addr, CGF.ConvertTypeForMem(Ty));
+    return Address(Load, CGF.ConvertTypeForMem(Ty), SlotSize);
   }
 
   CharUnits TySize = getContext().getTypeSizeInChars(Ty);
   CharUnits TyAlignForABI = getContext().getTypeUnadjustedAlignInChars(Ty);
 
   // Use indirect if size of the illegal vector is bigger than 16 bytes.
   bool IsIndirect = false;
   const Type *Base = nullptr;

clang/lib/CodeGen/Targets/Hexagon.cpp

if (Align > 4) {
     __overflow_area_pointer = CGF.Builder.CreateIntToPtr(
         CGF.Builder.CreateAnd(AsInt, Mask), __overflow_area_pointer->getType(),
         "__overflow_area_pointer.align");
   }
 
   // Get the type of the argument from memory and bitcast
   // overflow area pointer to the argument type.
   llvm::Type *PTy = CGF.ConvertTypeForMem(Ty);
-  Address AddrTyped = CGF.Builder.CreateElementBitCast(
-      Address(__overflow_area_pointer, CGF.Int8Ty,
-              CharUnits::fromQuantity(Align)),
-      PTy);
+  Address AddrTyped =
+      Address(__overflow_area_pointer, PTy, CharUnits::fromQuantity(Align));
 
   // Round up to the minimum stack alignment for varargs which is 4 bytes.
   uint64_t Offset = llvm::alignTo(CGF.getContext().getTypeSize(Ty) / 8, 4);
 
   __overflow_area_pointer = CGF.Builder.CreateGEP(
       CGF.Int8Ty, __overflow_area_pointer,
       llvm::ConstantInt::get(CGF.Int32Ty, Offset),
       "__overflow_area_pointer.next");
Address HexagonABIInfo::EmitVAArgForHexagon(CodeGenFunction &CGF,
   uint64_t TyAlign = CGF.getContext().getTypeAlign(Ty) / 8;
   if (TyAlign > 4) {
     assert((TyAlign & (TyAlign - 1)) == 0 && "Alignment is not power of 2!");
     llvm::Value *AddrAsInt = Builder.CreatePtrToInt(Addr, CGF.Int32Ty);
     AddrAsInt = Builder.CreateAdd(AddrAsInt, Builder.getInt32(TyAlign - 1));
     AddrAsInt = Builder.CreateAnd(AddrAsInt, Builder.getInt32(~(TyAlign - 1)));
     Addr = Builder.CreateIntToPtr(AddrAsInt, BP);
   }
-  Address AddrTyped = Builder.CreateElementBitCast(
-      Address(Addr, CGF.Int8Ty, CharUnits::fromQuantity(TyAlign)),
-      CGF.ConvertType(Ty));
+  Address AddrTyped =
+      Address(Addr, CGF.ConvertType(Ty), CharUnits::fromQuantity(TyAlign));
 
   uint64_t Offset = llvm::alignTo(CGF.getContext().getTypeSize(Ty) / 8, 4);
   llvm::Value *NextAddr = Builder.CreateGEP(
       CGF.Int8Ty, Addr, llvm::ConstantInt::get(CGF.Int32Ty, Offset), "ap.next");
   Builder.CreateStore(NextAddr, VAListAddrAsBPP);
 
   return AddrTyped;
 }

clang/lib/CodeGen/Targets/LoongArch.cpp

ABIArgInfo LoongArchABIInfo::classifyReturnType(QualType RetTy) const {
   return classifyArgumentType(RetTy, /*IsFixed=*/true, GARsLeft, FARsLeft);
 }
 
 Address LoongArchABIInfo::EmitVAArg(CodeGenFunction &CGF, Address VAListAddr,
                                     QualType Ty) const {
   CharUnits SlotSize = CharUnits::fromQuantity(GRLen / 8);
 
   // Empty records are ignored for parameter passing purposes.
-  if (isEmptyRecord(getContext(), Ty, true)) {
-    Address Addr = Address(CGF.Builder.CreateLoad(VAListAddr),
-                           getVAListElementType(CGF), SlotSize);
-    Addr = CGF.Builder.CreateElementBitCast(Addr, CGF.ConvertTypeForMem(Ty));
-    return Addr;
-  }
+  if (isEmptyRecord(getContext(), Ty, true))
+    return Address(CGF.Builder.CreateLoad(VAListAddr),
+                   CGF.ConvertTypeForMem(Ty), SlotSize);
 
   auto TInfo = getContext().getTypeInfoInChars(Ty);
 
   // Arguments bigger than 2*GRLen bytes are passed indirectly.
   return emitVoidPtrVAArg(CGF, VAListAddr, Ty,
                           /*IsIndirect=*/TInfo.Width > 2 * SlotSize, TInfo,
                           SlotSize,
                           /*AllowHigherAlign=*/true);

clang/lib/CodeGen/Targets/PPC.cpp

RealAddr =
         CGF.Builder.CreateConstInBoundsByteGEP(RealAddr, SlotSize - EltSize);
     ImagAddr = CGF.Builder.CreateConstInBoundsByteGEP(ImagAddr,
                                                       2 * SlotSize - EltSize);
   } else {
     ImagAddr = CGF.Builder.CreateConstInBoundsByteGEP(RealAddr, SlotSize);
   }
 
   llvm::Type *EltTy = CGF.ConvertTypeForMem(CTy->getElementType());
-  RealAddr = CGF.Builder.CreateElementBitCast(RealAddr, EltTy);
-  ImagAddr = CGF.Builder.CreateElementBitCast(ImagAddr, EltTy);
+  RealAddr = RealAddr.withElementType(EltTy);
+  ImagAddr = ImagAddr.withElementType(EltTy);
   llvm::Value *Real = CGF.Builder.CreateLoad(RealAddr, ".vareal");
   llvm::Value *Imag = CGF.Builder.CreateLoad(ImagAddr, ".vaimag");
 
   Address Temp = CGF.CreateMemTemp(Ty, "vacplx");
   CGF.EmitStoreOfComplex({Real, Imag}, CGF.MakeAddrLValue(Temp, Ty),
                          /*init*/ true);
   return Temp;
 }
Address RegAddr = Address::invalid();
 
     // Get the address of the saved value by scaling the number of
     // registers we've used by the number of
     CharUnits RegSize = CharUnits::fromQuantity((isInt || IsSoftFloatABI) ? 4 : 8);
     llvm::Value *RegOffset =
         Builder.CreateMul(NumRegs, Builder.getInt8(RegSize.getQuantity()));
     RegAddr = Address(
         Builder.CreateInBoundsGEP(CGF.Int8Ty, RegAddr.getPointer(), RegOffset),
-        CGF.Int8Ty, RegAddr.getAlignment().alignmentOfArrayElement(RegSize));
-    RegAddr = Builder.CreateElementBitCast(RegAddr, DirectTy);
+        DirectTy, RegAddr.getAlignment().alignmentOfArrayElement(RegSize));
 
     // Increase the used-register count.
     NumRegs =
       Builder.CreateAdd(NumRegs,
                         Builder.getInt8((isI64 || (isF64 && IsSoftFloatABI)) ? 2 : 1));
     Builder.CreateStore(NumRegs, NumRegsAddr);
 
     CGF.EmitBranch(Cont);
Address MemAddr = Address::invalid();
     // Round up address of argument to alignment
     CharUnits Align = CGF.getContext().getTypeAlignInChars(Ty);
     if (Align > OverflowAreaAlign) {
       llvm::Value *Ptr = OverflowArea.getPointer();
       OverflowArea = Address(emitRoundPointerUpToAlignment(CGF, Ptr, Align),
                              OverflowArea.getElementType(), Align);
     }
 
-    MemAddr = Builder.CreateElementBitCast(OverflowArea, DirectTy);
+    MemAddr = OverflowArea.withElementType(DirectTy);
 
     // Increase the overflow area.
     OverflowArea = Builder.CreateConstInBoundsByteGEP(OverflowArea, Size);
     Builder.CreateStore(OverflowArea.getPointer(), OverflowAreaAddr);
     CGF.EmitBranch(Cont);
   }
 
   CGF.EmitBlock(Cont);

clang/lib/CodeGen/Targets/X86.cpp

if (RetWidth <= 32) {
     ResultRegTypes.push_back(CGF.Int64Ty);
   }
 
   // Truncate EAX or EAX:EDX to an integer of the appropriate size.
   llvm::Type *CoerceTy = llvm::IntegerType::get(CGF.getLLVMContext(), RetWidth);
   ResultTruncRegTypes.push_back(CoerceTy);
 
   // Coerce the integer by bitcasting the return slot pointer.
-  ReturnSlot.setAddress(
-      CGF.Builder.CreateElementBitCast(ReturnSlot.getAddress(CGF), CoerceTy));
+  ReturnSlot.setAddress(ReturnSlot.getAddress(CGF).withElementType(CoerceTy));
   ResultRegDests.push_back(ReturnSlot);
 
   rewriteInputConstraintReferences(NumOutputs, 1, AsmString);
 }
 
 /// shouldReturnTypeInRegister - Determine if the given type should be
 /// returned in a register (for the Darwin and MCU ABI).
 bool X86_32ABIInfo::shouldReturnTypeInRegister(QualType Ty,
llvm::Value *RegSaveArea = CGF.Builder.CreateLoad(
       CGF.Builder.CreateStructGEP(VAListAddr, 3), "reg_save_area");
 
   Address RegAddr = Address::invalid();
   if (neededInt && neededSSE) {
     // FIXME: Cleanup.
     assert(AI.isDirect() && "Unexpected ABI info for mixed regs");
     llvm::StructType *ST = cast<llvm::StructType>(AI.getCoerceToType());
     Address Tmp = CGF.CreateMemTemp(Ty);
-    Tmp = CGF.Builder.CreateElementBitCast(Tmp, ST);
+    Tmp = Tmp.withElementType(ST);
     assert(ST->getNumElements() == 2 && "Unexpected ABI info for mixed regs");
     llvm::Type *TyLo = ST->getElementType(0);
     llvm::Type *TyHi = ST->getElementType(1);
     assert((TyLo->isFPOrFPVectorTy() ^ TyHi->isFPOrFPVectorTy()) &&
            "Unexpected ABI info for mixed regs");
     llvm::Type *PTyLo = llvm::PointerType::getUnqual(TyLo);
     llvm::Type *PTyHi = llvm::PointerType::getUnqual(TyHi);
     llvm::Value *GPAddr =
if (neededInt && neededSSE) {
     CGF.Builder.CreateStore(V, CGF.Builder.CreateStructGEP(Tmp, 0));
 
     // Copy the second element.
     V = CGF.Builder.CreateAlignedLoad(
         TyHi, CGF.Builder.CreateBitCast(RegHiAddr, PTyHi),
         CharUnits::fromQuantity(getDataLayout().getABITypeAlign(TyHi)));
     CGF.Builder.CreateStore(V, CGF.Builder.CreateStructGEP(Tmp, 1));
 
-    RegAddr = CGF.Builder.CreateElementBitCast(Tmp, LTy);
+    RegAddr = Tmp.withElementType(LTy);
   } else if (neededInt) {
     RegAddr = Address(CGF.Builder.CreateGEP(CGF.Int8Ty, RegSaveArea, gp_offset),
-                      CGF.Int8Ty, CharUnits::fromQuantity(8));
-    RegAddr = CGF.Builder.CreateElementBitCast(RegAddr, LTy);
+                      LTy, CharUnits::fromQuantity(8));
 
     // Copy to a temporary if necessary to ensure the appropriate alignment.
     auto TInfo = getContext().getTypeInfoInChars(Ty);
     uint64_t TySize = TInfo.Width.getQuantity();
     CharUnits TyAlign = TInfo.Align;
 
     // Copy into a temporary if the type is more aligned than the
     // register save area.
     if (TyAlign.getQuantity() > 8) {
       Address Tmp = CGF.CreateMemTemp(Ty);
       CGF.Builder.CreateMemCpy(Tmp, RegAddr, TySize, false);
       RegAddr = Tmp;
     }
 
   } else if (neededSSE == 1) {
     RegAddr = Address(CGF.Builder.CreateGEP(CGF.Int8Ty, RegSaveArea, fp_offset),
-                      CGF.Int8Ty, CharUnits::fromQuantity(16));
-    RegAddr = CGF.Builder.CreateElementBitCast(RegAddr, LTy);
+                      LTy, CharUnits::fromQuantity(16));
   } else {
     assert(neededSSE == 2 && "Invalid number of needed registers!");
     // SSE registers are spaced 16 bytes apart in the register save
     // area, we need to collect the two eightbytes together.
     // The ABI isn't explicit about this, but it seems reasonable
     // to assume that the slots are 16-byte aligned, since the stack is
     // naturally 16-byte aligned and the prologue is expected to store
     // all the SSE registers to the RSA.
     Address RegAddrLo = Address(CGF.Builder.CreateGEP(CGF.Int8Ty, RegSaveArea,
                                                       fp_offset),
                                 CGF.Int8Ty, CharUnits::fromQuantity(16));
     Address RegAddrHi =
       CGF.Builder.CreateConstInBoundsByteGEP(RegAddrLo,
                                              CharUnits::fromQuantity(16));
     llvm::Type *ST = AI.canHaveCoerceToType()
                          ? AI.getCoerceToType()
                          : llvm::StructType::get(CGF.DoubleTy, CGF.DoubleTy);
     llvm::Value *V;
     Address Tmp = CGF.CreateMemTemp(Ty);
-    Tmp = CGF.Builder.CreateElementBitCast(Tmp, ST);
-    V = CGF.Builder.CreateLoad(CGF.Builder.CreateElementBitCast(
-        RegAddrLo, ST->getStructElementType(0)));
+    Tmp = Tmp.withElementType(ST);
+    V = CGF.Builder.CreateLoad(
+        RegAddrLo.withElementType(ST->getStructElementType(0)));
     CGF.Builder.CreateStore(V, CGF.Builder.CreateStructGEP(Tmp, 0));
-    V = CGF.Builder.CreateLoad(CGF.Builder.CreateElementBitCast(
-        RegAddrHi, ST->getStructElementType(1)));
+    V = CGF.Builder.CreateLoad(
+        RegAddrHi.withElementType(ST->getStructElementType(1)));
     CGF.Builder.CreateStore(V, CGF.Builder.CreateStructGEP(Tmp, 1));
 
-    RegAddr = CGF.Builder.CreateElementBitCast(Tmp, LTy);
+    RegAddr = Tmp.withElementType(LTy);
   }
 
   // AMD64-ABI 3.5.7p5: Step 5. Set:
   // l->gp_offset = l->gp_offset + num_gp * 8
   // l->fp_offset = l->fp_offset + num_fp * 16.
   if (neededInt) {
     llvm::Value *Offset = llvm::ConstantInt::get(CGF.Int32Ty, neededInt * 8);
     CGF.Builder.CreateStore(CGF.Builder.CreateAdd(gp_offset, Offset),