[clang][CodeGen] Drop some typed pointer bitcasts

[bjope]

Rebased+updated.

clang/lib/CodeGen/CGBlocks.cpp

for (const auto &CI : blockDecl->captures()) {
     // pointer at this point, since we're building something that will
     // live a shorter life than the stack byref anyway.
     if (CI.isEscapingByref()) {
       // Get a void* that points to the byref struct.
       llvm::Value *byrefPointer;
       if (CI.isNested())
         byrefPointer = Builder.CreateLoad(src, "byref.capture");
       else
-        byrefPointer = Builder.CreateBitCast(src.getPointer(), VoidPtrTy);
+        byrefPointer = src.getPointer();
 
       // Write that void* into the capture field.
       Builder.CreateStore(byrefPointer, blockField);
 
     // If we have a copy constructor, evaluate that into the block field.
     } else if (const Expr *copyExpr = CI.getCopyExpr()) {
       if (blockDecl->isConversionFromLambda()) {
         // If we have a lambda conversion, emit the expression
CallBlockRelease(Address Addr, BlockFieldFlags Flags, bool LoadValue,
                    bool CT)
       : Addr(Addr), FieldFlags(Flags), LoadBlockVarAddr(LoadValue),
         CanThrow(CT) {}
 
   void Emit(CodeGenFunction &CGF, Flags flags) override {
     llvm::Value *BlockVarAddr;
     if (LoadBlockVarAddr) {
       BlockVarAddr = CGF.Builder.CreateLoad(Addr);
-      BlockVarAddr = CGF.Builder.CreateBitCast(BlockVarAddr, CGF.VoidPtrTy);
     } else {
       BlockVarAddr = Addr.getPointer();
     }
 
     CGF.BuildBlockRelease(BlockVarAddr, FieldFlags, CanThrow);
   }
 };
 } // end anonymous namespace
case BlockCaptureEntityKind::ARCStrong: {
         // first place.
         if (!needsEHCleanup(captureType.isDestructedType()))
           cast<llvm::Instruction>(dstField.getPointer())->eraseFromParent();
       }
       break;
     }
     case BlockCaptureEntityKind::BlockObject: {
       llvm::Value *srcValue = Builder.CreateLoad(srcField, "blockcopy.src");
-      srcValue = Builder.CreateBitCast(srcValue, VoidPtrTy);
-      llvm::Value *dstAddr =
-          Builder.CreateBitCast(dstField.getPointer(), VoidPtrTy);
+      llvm::Value *dstAddr = dstField.getPointer();
       llvm::Value *args[] = {
         dstAddr, srcValue, llvm::ConstantInt::get(Int32Ty, flags.getBitMask())
       };
 
       if (CI.isByRef() && C.getBlockVarCopyInit(CI.getVariable()).canThrow())
         EmitRuntimeCallOrInvoke(CGM.getBlockObjectAssign(), args);
       else
         EmitNounwindRuntimeCall(CGM.getBlockObjectAssign(), args);
if (ByRefHasLifetime && HasByrefExtendedLayout) {
     auto layoutInfo = CGM.getObjCRuntime().BuildByrefLayout(CGM, type);
     storeHeaderField(layoutInfo, getPointerSize(), "byref.layout");
   }
 }
 
 void CodeGenFunction::BuildBlockRelease(llvm::Value *V, BlockFieldFlags flags,
                                         bool CanThrow) {
   llvm::FunctionCallee F = CGM.getBlockObjectDispose();
-  llvm::Value *args[] = {
-    Builder.CreateBitCast(V, Int8PtrTy),
-    llvm::ConstantInt::get(Int32Ty, flags.getBitMask())
-  };
+  llvm::Value *args[] = {V,
+                         llvm::ConstantInt::get(Int32Ty, flags.getBitMask())};
 
   if (CanThrow)
     EmitRuntimeCallOrInvoke(F, args);
   else
     EmitNounwindRuntimeCall(F, args);
 }
 
 void CodeGenFunction::enterByrefCleanup(CleanupKind Kind, Address Addr,

clang/lib/CodeGen/CGBuiltin.cpp

static Value *MakeBinaryAtomicValue(
 
   QualType T = E->getType();
   assert(E->getArg(0)->getType()->isPointerType());
   assert(CGF.getContext().hasSameUnqualifiedType(T,
                                   E->getArg(0)->getType()->getPointeeType()));
   assert(CGF.getContext().hasSameUnqualifiedType(T, E->getArg(1)->getType()));
 
   llvm::Value *DestPtr = CheckAtomicAlignment(CGF, E);
-  unsigned AddrSpace = DestPtr->getType()->getPointerAddressSpace();
 
-  llvm::IntegerType *IntType =
-    llvm::IntegerType::get(CGF.getLLVMContext(),
-                           CGF.getContext().getTypeSize(T));
-  llvm::Type *IntPtrType =
-      llvm::PointerType::get(CGF.getLLVMContext(), AddrSpace);
+  llvm::IntegerType *IntType = llvm::IntegerType::get(
+      CGF.getLLVMContext(), CGF.getContext().getTypeSize(T));
 
   llvm::Value *Args[2];
-  Args[0] = CGF.Builder.CreateBitCast(DestPtr, IntPtrType);
+  Args[0] = DestPtr;
   Args[1] = CGF.EmitScalarExpr(E->getArg(1));
   llvm::Type *ValueType = Args[1]->getType();
   Args[1] = EmitToInt(CGF, Args[1], T, IntType);
 
   llvm::Value *Result = CGF.Builder.CreateAtomicRMW(
       Kind, Args[0], Args[1], Ordering);
   return EmitFromInt(CGF, Result, T, ValueType);
 }
 
 static Value *EmitNontemporalStore(CodeGenFunction &CGF, const CallExpr *E) {
   Value *Val = CGF.EmitScalarExpr(E->getArg(0));
   Value *Address = CGF.EmitScalarExpr(E->getArg(1));
 
-  // Convert the type of the pointer to a pointer to the stored type.
   Val = CGF.EmitToMemory(Val, E->getArg(0)->getType());
-  unsigned SrcAddrSpace = Address->getType()->getPointerAddressSpace();
-  Value *BC = CGF.Builder.CreateBitCast(
-      Address, llvm::PointerType::get(Val->getType(), SrcAddrSpace), "cast");
-  LValue LV = CGF.MakeNaturalAlignAddrLValue(BC, E->getArg(0)->getType());
+  LValue LV = CGF.MakeNaturalAlignAddrLValue(Address, E->getArg(0)->getType());
   LV.setNontemporal(true);
   CGF.EmitStoreOfScalar(Val, LV, false);
   return nullptr;
 }
 
 static Value *EmitNontemporalLoad(CodeGenFunction &CGF, const CallExpr *E) {
   Value *Address = CGF.EmitScalarExpr(E->getArg(0));
 
if (Width < MinWidth) {
       Width = MinWidth;
     }
   }
 
   return {Width, Signed};
 }
 
 Value *CodeGenFunction::EmitVAStartEnd(Value *ArgValue, bool IsStart) {
-  llvm::Type *DestType = Int8PtrTy;
-  if (ArgValue->getType() != DestType)
-    ArgValue =
-        Builder.CreateBitCast(ArgValue, DestType, ArgValue->getName().data());
-
   Intrinsic::ID inst = IsStart ? Intrinsic::vastart : Intrinsic::vaend;
   return Builder.CreateCall(CGM.getIntrinsic(inst), ArgValue);
 }
 
 /// Checks if using the result of __builtin_object_size(p, @p From) in place of
 /// __builtin_object_size(p, @p To) is correct
 static bool areBOSTypesCompatible(int From, int To) {
   // Note: Our __builtin_object_size implementation currently treats Type=0 and
case Builtin::BI__builtin_va_end:
     EmitVAStartEnd(BuiltinID == Builtin::BI__va_start
                        ? EmitScalarExpr(E->getArg(0))
                        : EmitVAListRef(E->getArg(0)).getPointer(),
                    BuiltinID != Builtin::BI__builtin_va_end);
     return RValue::get(nullptr);
   case Builtin::BI__builtin_va_copy: {
     Value *DstPtr = EmitVAListRef(E->getArg(0)).getPointer();
     Value *SrcPtr = EmitVAListRef(E->getArg(1)).getPointer();
-
-    llvm::Type *Type = Int8PtrTy;
-
-    DstPtr = Builder.CreateBitCast(DstPtr, Type);
-    SrcPtr = Builder.CreateBitCast(SrcPtr, Type);
     Builder.CreateCall(CGM.getIntrinsic(Intrinsic::vacopy), {DstPtr, SrcPtr});
     return RValue::get(nullptr);
   }
   case Builtin::BI__builtin_abs:
   case Builtin::BI__builtin_labs:
   case Builtin::BI__builtin_llabs: {
     // X < 0 ? -X : X
     // The negation has 'nsw' because abs of INT_MIN is undefined.
case Builtin::BI__builtin_setjmp: {
     Builder.CreateStore(StackAddr, StackSaveSlot);
 
     // Call LLVM's EH setjmp, which is lightweight.
     Function *F = CGM.getIntrinsic(Intrinsic::eh_sjlj_setjmp);
     return RValue::get(Builder.CreateCall(F, Buf.getPointer()));
   }
   case Builtin::BI__builtin_longjmp: {
     Value *Buf = EmitScalarExpr(E->getArg(0));
-    Buf = Builder.CreateBitCast(Buf, Int8PtrTy);
 
     // Call LLVM's EH longjmp, which is lightweight.
     Builder.CreateCall(CGM.getIntrinsic(Intrinsic::eh_sjlj_longjmp), Buf);
 
     // longjmp doesn't return; mark this as unreachable.
     Builder.CreateUnreachable();
 
     // We do need to preserve an insertion point.
Value *CodeGenFunction::EmitSVEGatherPrefetch(const SVETypeFlags &TypeFlags,
   Function *F = CGM.getIntrinsic(IntID, OverloadedTy);
   return Builder.CreateCall(F, Ops);
 }
 
 Value *CodeGenFunction::EmitSVEStructLoad(const SVETypeFlags &TypeFlags,
                                           SmallVectorImpl<Value*> &Ops,
                                           unsigned IntID) {
   llvm::ScalableVectorType *VTy = getSVEType(TypeFlags);
-  auto VecPtrTy = llvm::PointerType::getUnqual(VTy);
-  auto EltPtrTy = llvm::PointerType::getUnqual(VTy->getElementType());
 
   unsigned N;
   switch (IntID) {
   case Intrinsic::aarch64_sve_ld2_sret:
     N = 2;
     break;
   case Intrinsic::aarch64_sve_ld3_sret:
     N = 3;
     break;
   case Intrinsic::aarch64_sve_ld4_sret:
     N = 4;
     break;
   default:
     llvm_unreachable("unknown intrinsic!");
   }
   auto RetTy = llvm::VectorType::get(VTy->getElementType(),
                                      VTy->getElementCount() * N);
 
-  Value *Predicate = EmitSVEPredicateCast(Ops[0], VTy);
-  Value *BasePtr= Builder.CreateBitCast(Ops[1], VecPtrTy);
+  Value *Predicate = EmitSVEPredicateCast(Ops[0], VTy);
+  Value *BasePtr = Ops[1];
 
   // Does the load have an offset?
   if (Ops.size() > 2)
     BasePtr = Builder.CreateGEP(VTy, BasePtr, Ops[2]);
 
-  BasePtr = Builder.CreateBitCast(BasePtr, EltPtrTy);
   Function *F = CGM.getIntrinsic(IntID, {VTy});
   Value *Call = Builder.CreateCall(F, {Predicate, BasePtr});
   unsigned MinElts = VTy->getMinNumElements();
   Value *Ret = llvm::PoisonValue::get(RetTy);
   for (unsigned I = 0; I < N; I++) {
     Value *Idx = ConstantInt::get(CGM.Int64Ty, I * MinElts);
     Value *SRet = Builder.CreateExtractValue(Call, I);
     Ret = Builder.CreateInsertVector(RetTy, Ret, SRet, Idx);
   }
   return Ret;
 }
 
 Value *CodeGenFunction::EmitSVEStructStore(const SVETypeFlags &TypeFlags,
                                            SmallVectorImpl<Value*> &Ops,
                                            unsigned IntID) {
   llvm::ScalableVectorType *VTy = getSVEType(TypeFlags);
-  auto VecPtrTy = llvm::PointerType::getUnqual(VTy);
-  auto EltPtrTy = llvm::PointerType::getUnqual(VTy->getElementType());
 
   unsigned N;
   switch (IntID) {
   case Intrinsic::aarch64_sve_st2:
     N = 2;
     break;
   case Intrinsic::aarch64_sve_st3:
     N = 3;
     break;
   case Intrinsic::aarch64_sve_st4:
     N = 4;
     break;
   default:
     llvm_unreachable("unknown intrinsic!");
   }
 
   Value *Predicate = EmitSVEPredicateCast(Ops[0], VTy);
-  Value *BasePtr = Builder.CreateBitCast(Ops[1], VecPtrTy);
+  Value *BasePtr = Ops[1];
 
   // Does the store have an offset?
   if (Ops.size() > 3)
     BasePtr = Builder.CreateGEP(VTy, BasePtr, Ops[2]);
 
-  BasePtr = Builder.CreateBitCast(BasePtr, EltPtrTy);
   Value *Val = Ops.back();
 
   // The llvm.aarch64.sve.st2/3/4 intrinsics take legal part vectors, so we
   // need to break up the tuple vector.
   SmallVector<llvm::Value*, 5> Operands;
   unsigned MinElts = VTy->getElementCount().getKnownMinValue();
   for (unsigned I = 0; I < N; ++I) {
     Value *Idx = ConstantInt::get(CGM.Int64Ty, I * MinElts);
if ((BuiltinID == clang::AArch64::BI__builtin_arm_ldrex ||
        BuiltinID == clang::AArch64::BI__builtin_arm_ldaex) &&
       getContext().getTypeSize(E->getType()) == 128) {
     Function *F =
         CGM.getIntrinsic(BuiltinID == clang::AArch64::BI__builtin_arm_ldaex
                              ? Intrinsic::aarch64_ldaxp
                              : Intrinsic::aarch64_ldxp);
 
     Value *LdPtr = EmitScalarExpr(E->getArg(0));
-    Value *Val = Builder.CreateCall(F, Builder.CreateBitCast(LdPtr, Int8PtrTy),
-                                    "ldxp");
+    Value *Val = Builder.CreateCall(F, LdPtr, "ldxp");
 
     Value *Val0 = Builder.CreateExtractValue(Val, 1);
     Value *Val1 = Builder.CreateExtractValue(Val, 0);
     llvm::Type *Int128Ty = llvm::IntegerType::get(getLLVMContext(), 128);
     Val0 = Builder.CreateZExt(Val0, Int128Ty);
     Val1 = Builder.CreateZExt(Val1, Int128Ty);
 
     Value *ShiftCst = llvm::ConstantInt::get(Int128Ty, 64);
if ((BuiltinID == clang::AArch64::BI__builtin_arm_strex ||
     Address Tmp = CreateMemTemp(E->getArg(0)->getType());
     EmitAnyExprToMem(E->getArg(0), Tmp, Qualifiers(), /*init*/ true);
 
     Tmp = Tmp.withElementType(STy);
     llvm::Value *Val = Builder.CreateLoad(Tmp);
 
     Value *Arg0 = Builder.CreateExtractValue(Val, 0);
     Value *Arg1 = Builder.CreateExtractValue(Val, 1);
-    Value *StPtr = Builder.CreateBitCast(EmitScalarExpr(E->getArg(1)),
-                                         Int8PtrTy);
+    Value *StPtr = EmitScalarExpr(E->getArg(1));
     return Builder.CreateCall(F, {Arg0, Arg1, StPtr}, "stxp");
   }
 
   if (BuiltinID == clang::AArch64::BI__builtin_arm_strex ||
       BuiltinID == clang::AArch64::BI__builtin_arm_stlex) {
     Value *StoreVal = EmitScalarExpr(E->getArg(0));
     Value *StoreAddr = EmitScalarExpr(E->getArg(1));
 
case NEON::BI__builtin_neon_vaddq_p128: {
     Ops[0] = Builder.CreateBitCast(Ops[0], Ty);
     Ops[1] = Builder.CreateBitCast(Ops[1], Ty);
     Ops[0] =  Builder.CreateXor(Ops[0], Ops[1]);
     llvm::Type *Int128Ty = llvm::Type::getIntNTy(getLLVMContext(), 128);
     return Builder.CreateBitCast(Ops[0], Int128Ty);
   }
   case NEON::BI__builtin_neon_vldrq_p128: {
     llvm::Type *Int128Ty = llvm::Type::getIntNTy(getLLVMContext(), 128);
-    llvm::Type *Int128PTy = llvm::PointerType::get(Int128Ty, 0);
-    Value *Ptr = Builder.CreateBitCast(EmitScalarExpr(E->getArg(0)), Int128PTy);
+    Value *Ptr = EmitScalarExpr(E->getArg(0));
     return Builder.CreateAlignedLoad(Int128Ty, Ptr,
                                      CharUnits::fromQuantity(16));
   }
   case NEON::BI__builtin_neon_vstrq_p128: {
-    llvm::Type *Int128PTy = llvm::PointerType::getUnqual(getLLVMContext());
-    Value *Ptr = Builder.CreateBitCast(Ops[0], Int128PTy);
+    Value *Ptr = Ops[0];
     return Builder.CreateDefaultAlignedStore(EmitScalarExpr(E->getArg(1)), Ptr);
   }
   case NEON::BI__builtin_neon_vcvts_f32_u32:
   case NEON::BI__builtin_neon_vcvtd_f64_u64:
     usgn = true;
     [[fallthrough]];
   case NEON::BI__builtin_neon_vcvts_f32_s32:
   case NEON::BI__builtin_neon_vcvtd_f64_s64: {
case NEON::BI__builtin_neon_vrsraq_n_v: {
     TmpOps.push_back(Ops[2]);
     Function* F = CGM.getIntrinsic(Int, Ty);
     llvm::Value *tmp = EmitNeonCall(F, TmpOps, "vrshr_n", 1, true);
     Ops[0] = Builder.CreateBitCast(Ops[0], VTy);
     return Builder.CreateAdd(Ops[0], tmp);
   }
   case NEON::BI__builtin_neon_vld1_v:
   case NEON::BI__builtin_neon_vld1q_v: {
-    Ops[0] = Builder.CreateBitCast(Ops[0], llvm::PointerType::getUnqual(VTy));
     return Builder.CreateAlignedLoad(VTy, Ops[0], PtrOp0.getAlignment());
   }
   case NEON::BI__builtin_neon_vst1_v:
   case NEON::BI__builtin_neon_vst1q_v:
-    Ops[0] = Builder.CreateBitCast(Ops[0], llvm::PointerType::getUnqual(VTy));
     Ops[1] = Builder.CreateBitCast(Ops[1], VTy);
     return Builder.CreateAlignedStore(Ops[1], Ops[0], PtrOp0.getAlignment());
   case NEON::BI__builtin_neon_vld1_lane_v:
   case NEON::BI__builtin_neon_vld1q_lane_v: {
     Ops[1] = Builder.CreateBitCast(Ops[1], Ty);
-    Ty = llvm::PointerType::getUnqual(VTy->getElementType());
-    Ops[0] = Builder.CreateBitCast(Ops[0], Ty);
     Ops[0] = Builder.CreateAlignedLoad(VTy->getElementType(), Ops[0],
                                        PtrOp0.getAlignment());
     return Builder.CreateInsertElement(Ops[1], Ops[0], Ops[2], "vld1_lane");
   }
   case NEON::BI__builtin_neon_vldap1_lane_s64:
   case NEON::BI__builtin_neon_vldap1q_lane_s64: {
     Ops[1] = Builder.CreateBitCast(Ops[1], Ty);
-    Ty = llvm::PointerType::getUnqual(VTy->getElementType());
-    Ops[0] = Builder.CreateBitCast(Ops[0], Ty);
     llvm::LoadInst *LI = Builder.CreateAlignedLoad(
         VTy->getElementType(), Ops[0], PtrOp0.getAlignment());
     LI->setAtomic(llvm::AtomicOrdering::Acquire);
     Ops[0] = LI;
     return Builder.CreateInsertElement(Ops[1], Ops[0], Ops[2], "vldap1_lane");
   }
   case NEON::BI__builtin_neon_vld1_dup_v:
   case NEON::BI__builtin_neon_vld1q_dup_v: {
     Value *V = PoisonValue::get(Ty);
-    Ty = llvm::PointerType::getUnqual(VTy->getElementType());
-    Ops[0] = Builder.CreateBitCast(Ops[0], Ty);
     Ops[0] = Builder.CreateAlignedLoad(VTy->getElementType(), Ops[0],
                                        PtrOp0.getAlignment());
     llvm::Constant *CI = ConstantInt::get(Int32Ty, 0);
     Ops[0] = Builder.CreateInsertElement(V, Ops[0], CI);
     return EmitNeonSplat(Ops[0], CI);
   }
   case NEON::BI__builtin_neon_vst1_lane_v:
   case NEON::BI__builtin_neon_vst1q_lane_v:
     Ops[1] = Builder.CreateBitCast(Ops[1], Ty);
     Ops[1] = Builder.CreateExtractElement(Ops[1], Ops[2]);
     Ty = llvm::PointerType::getUnqual(Ops[1]->getType());
     return Builder.CreateAlignedStore(Ops[1], Builder.CreateBitCast(Ops[0], Ty),
                                       PtrOp0.getAlignment());
   case NEON::BI__builtin_neon_vstl1_lane_s64:
   case NEON::BI__builtin_neon_vstl1q_lane_s64: {
     Ops[1] = Builder.CreateBitCast(Ops[1], Ty);
     Ops[1] = Builder.CreateExtractElement(Ops[1], Ops[2]);
-    Ty = llvm::PointerType::getUnqual(Ops[1]->getType());
-    llvm::StoreInst *SI = Builder.CreateAlignedStore(
-        Ops[1], Builder.CreateBitCast(Ops[0], Ty), PtrOp0.getAlignment());
+    llvm::StoreInst *SI =
+        Builder.CreateAlignedStore(Ops[1], Ops[0], PtrOp0.getAlignment());
     SI->setAtomic(llvm::AtomicOrdering::Release);
     return SI;
   }
   case NEON::BI__builtin_neon_vld2_v:
   case NEON::BI__builtin_neon_vld2q_v: {
-    llvm::Type *PTy = llvm::PointerType::getUnqual(VTy);
-    Ops[1] = Builder.CreateBitCast(Ops[1], PTy);
+    llvm::Type *PTy = llvm::PointerType::getUnqual(getLLVMContext());
     llvm::Type *Tys[2] = { VTy, PTy };
     Function *F = CGM.getIntrinsic(Intrinsic::aarch64_neon_ld2, Tys);
     Ops[1] = Builder.CreateCall(F, Ops[1], "vld2");
-    Ops[0] = Builder.CreateBitCast(Ops[0],
-                llvm::PointerType::getUnqual(Ops[1]->getType()));
     return Builder.CreateDefaultAlignedStore(Ops[1], Ops[0]);
   }
   case NEON::BI__builtin_neon_vld3_v:
   case NEON::BI__builtin_neon_vld3q_v: {
-    llvm::Type *PTy = llvm::PointerType::getUnqual(VTy);
-    Ops[1] = Builder.CreateBitCast(Ops[1], PTy);
+    llvm::Type *PTy = llvm::PointerType::getUnqual(getLLVMContext());
     llvm::Type *Tys[2] = { VTy, PTy };
     Function *F = CGM.getIntrinsic(Intrinsic::aarch64_neon_ld3, Tys);
     Ops[1] = Builder.CreateCall(F, Ops[1], "vld3");
-    Ops[0] = Builder.CreateBitCast(Ops[0],
-                llvm::PointerType::getUnqual(Ops[1]->getType()));
     return Builder.CreateDefaultAlignedStore(Ops[1], Ops[0]);
   }
   case NEON::BI__builtin_neon_vld4_v:
   case NEON::BI__builtin_neon_vld4q_v: {
-    llvm::Type *PTy = llvm::PointerType::getUnqual(VTy);
-    Ops[1] = Builder.CreateBitCast(Ops[1], PTy);
+    llvm::Type *PTy = llvm::PointerType::getUnqual(getLLVMContext());
     llvm::Type *Tys[2] = { VTy, PTy };
     Function *F = CGM.getIntrinsic(Intrinsic::aarch64_neon_ld4, Tys);
     Ops[1] = Builder.CreateCall(F, Ops[1], "vld4");
     Ops[0] = Builder.CreateBitCast(Ops[0],
                 llvm::PointerType::getUnqual(Ops[1]->getType()));
     return Builder.CreateDefaultAlignedStore(Ops[1], Ops[0]);
   }
   case NEON::BI__builtin_neon_vld2_dup_v:
   case NEON::BI__builtin_neon_vld2q_dup_v: {
-    llvm::Type *PTy =
-      llvm::PointerType::getUnqual(VTy->getElementType());
-    Ops[1] = Builder.CreateBitCast(Ops[1], PTy);
+    llvm::Type *PTy = llvm::PointerType::getUnqual(getLLVMContext());
     llvm::Type *Tys[2] = { VTy, PTy };
     Function *F = CGM.getIntrinsic(Intrinsic::aarch64_neon_ld2r, Tys);
     Ops[1] = Builder.CreateCall(F, Ops[1], "vld2");
-    Ops[0] = Builder.CreateBitCast(Ops[0],
-                llvm::PointerType::getUnqual(Ops[1]->getType()));
     return Builder.CreateDefaultAlignedStore(Ops[1], Ops[0]);
   }
   case NEON::BI__builtin_neon_vld3_dup_v:
   case NEON::BI__builtin_neon_vld3q_dup_v: {
-    llvm::Type *PTy =
-      llvm::PointerType::getUnqual(VTy->getElementType());
-    Ops[1] = Builder.CreateBitCast(Ops[1], PTy);
+    llvm::Type *PTy = llvm::PointerType::getUnqual(getLLVMContext());
     llvm::Type *Tys[2] = { VTy, PTy };
     Function *F = CGM.getIntrinsic(Intrinsic::aarch64_neon_ld3r, Tys);
     Ops[1] = Builder.CreateCall(F, Ops[1], "vld3");
-    Ops[0] = Builder.CreateBitCast(Ops[0],
-                llvm::PointerType::getUnqual(Ops[1]->getType()));
     return Builder.CreateDefaultAlignedStore(Ops[1], Ops[0]);
   }
   case NEON::BI__builtin_neon_vld4_dup_v:
   case NEON::BI__builtin_neon_vld4q_dup_v: {
-    llvm::Type *PTy =
-      llvm::PointerType::getUnqual(VTy->getElementType());
-    Ops[1] = Builder.CreateBitCast(Ops[1], PTy);
+    llvm::Type *PTy = llvm::PointerType::getUnqual(getLLVMContext());
     llvm::Type *Tys[2] = { VTy, PTy };
     Function *F = CGM.getIntrinsic(Intrinsic::aarch64_neon_ld4r, Tys);
     Ops[1] = Builder.CreateCall(F, Ops[1], "vld4");
-    Ops[0] = Builder.CreateBitCast(Ops[0],
-                llvm::PointerType::getUnqual(Ops[1]->getType()));
     return Builder.CreateDefaultAlignedStore(Ops[1], Ops[0]);
   }
   case NEON::BI__builtin_neon_vld2_lane_v:
   case NEON::BI__builtin_neon_vld2q_lane_v: {
     llvm::Type *Tys[2] = { VTy, Ops[1]->getType() };
     Function *F = CGM.getIntrinsic(Intrinsic::aarch64_neon_ld2lane, Tys);
     std::rotate(Ops.begin() + 1, Ops.begin() + 2, Ops.end());
     Ops[1] = Builder.CreateBitCast(Ops[1], Ty);
     Ops[2] = Builder.CreateBitCast(Ops[2], Ty);
     Ops[3] = Builder.CreateZExt(Ops[3], Int64Ty);
     Ops[1] = Builder.CreateCall(F, ArrayRef(Ops).slice(1), "vld2_lane");
-    Ty = llvm::PointerType::getUnqual(Ops[1]->getType());
-    Ops[0] = Builder.CreateBitCast(Ops[0], Ty);
     return Builder.CreateDefaultAlignedStore(Ops[1], Ops[0]);
   }
   case NEON::BI__builtin_neon_vld3_lane_v:
   case NEON::BI__builtin_neon_vld3q_lane_v: {
     llvm::Type *Tys[2] = { VTy, Ops[1]->getType() };
     Function *F = CGM.getIntrinsic(Intrinsic::aarch64_neon_ld3lane, Tys);
     std::rotate(Ops.begin() + 1, Ops.begin() + 2, Ops.end());
     Ops[1] = Builder.CreateBitCast(Ops[1], Ty);
     Ops[2] = Builder.CreateBitCast(Ops[2], Ty);
     Ops[3] = Builder.CreateBitCast(Ops[3], Ty);
     Ops[4] = Builder.CreateZExt(Ops[4], Int64Ty);
     Ops[1] = Builder.CreateCall(F, ArrayRef(Ops).slice(1), "vld3_lane");
-    Ty = llvm::PointerType::getUnqual(Ops[1]->getType());
-    Ops[0] = Builder.CreateBitCast(Ops[0], Ty);
     return Builder.CreateDefaultAlignedStore(Ops[1], Ops[0]);
   }
   case NEON::BI__builtin_neon_vld4_lane_v:
   case NEON::BI__builtin_neon_vld4q_lane_v: {
     llvm::Type *Tys[2] = { VTy, Ops[1]->getType() };
     Function *F = CGM.getIntrinsic(Intrinsic::aarch64_neon_ld4lane, Tys);
     std::rotate(Ops.begin() + 1, Ops.begin() + 2, Ops.end());
     Ops[1] = Builder.CreateBitCast(Ops[1], Ty);
     Ops[2] = Builder.CreateBitCast(Ops[2], Ty);
     Ops[3] = Builder.CreateBitCast(Ops[3], Ty);
     Ops[4] = Builder.CreateBitCast(Ops[4], Ty);
     Ops[5] = Builder.CreateZExt(Ops[5], Int64Ty);
     Ops[1] = Builder.CreateCall(F, ArrayRef(Ops).slice(1), "vld4_lane");
-    Ty = llvm::PointerType::getUnqual(Ops[1]->getType());
-    Ops[0] = Builder.CreateBitCast(Ops[0], Ty);
     return Builder.CreateDefaultAlignedStore(Ops[1], Ops[0]);
   }
   case NEON::BI__builtin_neon_vst2_v:
   case NEON::BI__builtin_neon_vst2q_v: {
     std::rotate(Ops.begin(), Ops.begin() + 1, Ops.end());
     llvm::Type *Tys[2] = { VTy, Ops[2]->getType() };
     return EmitNeonCall(CGM.getIntrinsic(Intrinsic::aarch64_neon_st2, Tys),
                         Ops, "");
case NEON::BI__builtin_neon_vst4q_lane_v: {
     std::rotate(Ops.begin(), Ops.begin() + 1, Ops.end());
     Ops[4] = Builder.CreateZExt(Ops[4], Int64Ty);
     llvm::Type *Tys[2] = { VTy, Ops[5]->getType() };
     return EmitNeonCall(CGM.getIntrinsic(Intrinsic::aarch64_neon_st4lane, Tys),
                         Ops, "");
   }
   case NEON::BI__builtin_neon_vtrn_v:
   case NEON::BI__builtin_neon_vtrnq_v: {
-    Ops[0] = Builder.CreateBitCast(Ops[0], llvm::PointerType::getUnqual(Ty));
     Ops[1] = Builder.CreateBitCast(Ops[1], Ty);
     Ops[2] = Builder.CreateBitCast(Ops[2], Ty);
     Value *SV = nullptr;
 
     for (unsigned vi = 0; vi != 2; ++vi) {
       SmallVector<int, 16> Indices;
       for (unsigned i = 0, e = VTy->getNumElements(); i != e; i += 2) {
         Indices.push_back(i+vi);
         Indices.push_back(i+e+vi);
       }
       Value *Addr = Builder.CreateConstInBoundsGEP1_32(Ty, Ops[0], vi);
       SV = Builder.CreateShuffleVector(Ops[1], Ops[2], Indices, "vtrn");
       SV = Builder.CreateDefaultAlignedStore(SV, Addr);
     }
     return SV;
   }
   case NEON::BI__builtin_neon_vuzp_v:
   case NEON::BI__builtin_neon_vuzpq_v: {
-    Ops[0] = Builder.CreateBitCast(Ops[0], llvm::PointerType::getUnqual(Ty));
     Ops[1] = Builder.CreateBitCast(Ops[1], Ty);
     Ops[2] = Builder.CreateBitCast(Ops[2], Ty);
     Value *SV = nullptr;
 
     for (unsigned vi = 0; vi != 2; ++vi) {
       SmallVector<int, 16> Indices;
       for (unsigned i = 0, e = VTy->getNumElements(); i != e; ++i)
         Indices.push_back(2*i+vi);
 
       Value *Addr = Builder.CreateConstInBoundsGEP1_32(Ty, Ops[0], vi);
       SV = Builder.CreateShuffleVector(Ops[1], Ops[2], Indices, "vuzp");
       SV = Builder.CreateDefaultAlignedStore(SV, Addr);
     }
     return SV;
   }
   case NEON::BI__builtin_neon_vzip_v:
   case NEON::BI__builtin_neon_vzipq_v: {
-    Ops[0] = Builder.CreateBitCast(Ops[0], llvm::PointerType::getUnqual(Ty));
     Ops[1] = Builder.CreateBitCast(Ops[1], Ty);
     Ops[2] = Builder.CreateBitCast(Ops[2], Ty);
     Value *SV = nullptr;
 
     for (unsigned vi = 0; vi != 2; ++vi) {
       SmallVector<int, 16> Indices;
       for (unsigned i = 0, e = VTy->getNumElements(); i != e; i += 2) {
         Indices.push_back((i + vi*e) >> 1);
if (NumElts < 8) {
     MaskVec = CGF.Builder.CreateShuffleVector(
         MaskVec, MaskVec, ArrayRef(Indices, NumElts), "extract");
   }
   return MaskVec;
 }
 
 static Value *EmitX86MaskedStore(CodeGenFunction &CGF, ArrayRef<Value *> Ops,
                                  Align Alignment) {
-  // Cast the pointer to right type.
-  Value *Ptr = CGF.Builder.CreateBitCast(Ops[0],
-                               llvm::PointerType::getUnqual(Ops[1]->getType()));
+  Value *Ptr = Ops[0];
 
   Value *MaskVec = getMaskVecValue(
       CGF, Ops[2],
       cast<llvm::FixedVectorType>(Ops[1]->getType())->getNumElements());
 
   return CGF.Builder.CreateMaskedStore(Ops[1], Ptr, Alignment, MaskVec);
 }
 
 static Value *EmitX86MaskedLoad(CodeGenFunction &CGF, ArrayRef<Value *> Ops,
                                 Align Alignment) {
-  // Cast the pointer to right type.
   llvm::Type *Ty = Ops[1]->getType();
-  Value *Ptr =
-      CGF.Builder.CreateBitCast(Ops[0], llvm::PointerType::getUnqual(Ty));
+  Value *Ptr = Ops[0];
 
   Value *MaskVec = getMaskVecValue(
       CGF, Ops[2], cast<llvm::FixedVectorType>(Ty)->getNumElements());
 
   return CGF.Builder.CreateMaskedLoad(Ty, Ptr, Alignment, MaskVec, Ops[1]);
 }
 
 static Value *EmitX86ExpandLoad(CodeGenFunction &CGF,
                                 ArrayRef<Value *> Ops) {
   auto *ResultTy = cast<llvm::VectorType>(Ops[1]->getType());
-  llvm::Type *PtrTy = ResultTy->getElementType();
-
-  // Cast the pointer to element type.
-  Value *Ptr = CGF.Builder.CreateBitCast(Ops[0],
-                                         llvm::PointerType::getUnqual(PtrTy));
+  Value *Ptr = Ops[0];
 
   Value *MaskVec = getMaskVecValue(
       CGF, Ops[2], cast<FixedVectorType>(ResultTy)->getNumElements());
 
   llvm::Function *F = CGF.CGM.getIntrinsic(Intrinsic::masked_expandload,
                                            ResultTy);
   return CGF.Builder.CreateCall(F, { Ptr, MaskVec, Ops[1] });
 }
Intrinsic::ID IID = IsCompress ? Intrinsic::x86_avx512_mask_compress
                                  : Intrinsic::x86_avx512_mask_expand;
   llvm::Function *F = CGF.CGM.getIntrinsic(IID, ResultTy);
   return CGF.Builder.CreateCall(F, { Ops[0], Ops[1], MaskVec });
 }
 
 static Value *EmitX86CompressStore(CodeGenFunction &CGF,
                                    ArrayRef<Value *> Ops) {
   auto *ResultTy = cast<llvm::FixedVectorType>(Ops[1]->getType());
-  llvm::Type *PtrTy = ResultTy->getElementType();
-
-  // Cast the pointer to element type.
-  Value *Ptr = CGF.Builder.CreateBitCast(Ops[0],
-                                         llvm::PointerType::getUnqual(PtrTy));
+  Value *Ptr = Ops[0];
 
   Value *MaskVec = getMaskVecValue(CGF, Ops[2], ResultTy->getNumElements());
 
   llvm::Function *F = CGF.CGM.getIntrinsic(Intrinsic::masked_compressstore,
                                            ResultTy);
   return CGF.Builder.CreateCall(F, { Ops[1], Ptr, MaskVec });
 }
 
case X86::BI__builtin_ia32_vec_set_v4di: {
     // Otherwise we could just do this in the header file.
     return Builder.CreateInsertElement(Ops[0], Ops[1], Index);
   }
   case X86::BI_mm_setcsr:
   case X86::BI__builtin_ia32_ldmxcsr: {
     Address Tmp = CreateMemTemp(E->getArg(0)->getType());
     Builder.CreateStore(Ops[0], Tmp);
     return Builder.CreateCall(CGM.getIntrinsic(Intrinsic::x86_sse_ldmxcsr),
-                          Builder.CreateBitCast(Tmp.getPointer(), Int8PtrTy));
+                              Tmp.getPointer());
   }
   case X86::BI_mm_getcsr:
   case X86::BI__builtin_ia32_stmxcsr: {
     Address Tmp = CreateMemTemp(E->getType());
     Builder.CreateCall(CGM.getIntrinsic(Intrinsic::x86_sse_stmxcsr),
-                       Builder.CreateBitCast(Tmp.getPointer(), Int8PtrTy));
+                       Tmp.getPointer());
     return Builder.CreateLoad(Tmp, "stmxcsr");
   }
   case X86::BI__builtin_ia32_xsave:
   case X86::BI__builtin_ia32_xsave64:
   case X86::BI__builtin_ia32_xrstor:
   case X86::BI__builtin_ia32_xrstor64:
   case X86::BI__builtin_ia32_xsaveopt:
   case X86::BI__builtin_ia32_xsaveopt64:
Value *CodeGenFunction::EmitPPCBuiltinExpr(unsigned BuiltinID,
   case PPC::BI__builtin_vsx_lxvd2x_be:
   case PPC::BI__builtin_vsx_lxvw4x_be:
   case PPC::BI__builtin_vsx_lxvl:
   case PPC::BI__builtin_vsx_lxvll:
   {
     SmallVector<Value *, 2> Ops;
     Ops.push_back(EmitScalarExpr(E->getArg(0)));
     Ops.push_back(EmitScalarExpr(E->getArg(1)));
-    if(BuiltinID == PPC::BI__builtin_vsx_lxvl ||
-       BuiltinID == PPC::BI__builtin_vsx_lxvll){
-      Ops[0] = Builder.CreateBitCast(Ops[0], Int8PtrTy);
-    }else {
-      Ops[1] = Builder.CreateBitCast(Ops[1], Int8PtrTy);
+    if (!(BuiltinID == PPC::BI__builtin_vsx_lxvl ||
+          BuiltinID == PPC::BI__builtin_vsx_lxvll)) {
       Ops[0] = Builder.CreateGEP(Int8Ty, Ops[1], Ops[0]);
       Ops.pop_back();
     }
 
     switch (BuiltinID) {
     default: llvm_unreachable("Unsupported ld/lvsl/lvsr intrinsic!");
     case PPC::BI__builtin_altivec_lvx:
       ID = Intrinsic::ppc_altivec_lvx;
Value *CodeGenFunction::EmitPPCBuiltinExpr(unsigned BuiltinID,
   case PPC::BI__builtin_vsx_stxvw4x_be:
   case PPC::BI__builtin_vsx_stxvl:
   case PPC::BI__builtin_vsx_stxvll:
   {
     SmallVector<Value *, 3> Ops;
     Ops.push_back(EmitScalarExpr(E->getArg(0)));
     Ops.push_back(EmitScalarExpr(E->getArg(1)));
     Ops.push_back(EmitScalarExpr(E->getArg(2)));
-    if(BuiltinID == PPC::BI__builtin_vsx_stxvl ||
-      BuiltinID == PPC::BI__builtin_vsx_stxvll ){
-      Ops[1] = Builder.CreateBitCast(Ops[1], Int8PtrTy);
-    }else {
-      Ops[2] = Builder.CreateBitCast(Ops[2], Int8PtrTy);
+    if (!(BuiltinID == PPC::BI__builtin_vsx_stxvl ||
+          BuiltinID == PPC::BI__builtin_vsx_stxvll)) {
       Ops[1] = Builder.CreateGEP(Int8Ty, Ops[2], Ops[1]);
       Ops.pop_back();
     }
 
     switch (BuiltinID) {
     default: llvm_unreachable("Unsupported st intrinsic!");
     case PPC::BI__builtin_altivec_stvx:
       ID = Intrinsic::ppc_altivec_stvx;
case PPC::BI__builtin_ppc_rdlam: {
     llvm::Type *Ty = Op0->getType();
     Value *ShiftAmt = Builder.CreateIntCast(Op1, Ty, false);
     Function *F = CGM.getIntrinsic(Intrinsic::fshl, Ty);
     Value *Rotate = Builder.CreateCall(F, {Op0, Op0, ShiftAmt});
     return Builder.CreateAnd(Rotate, Op2);
   }
   case PPC::BI__builtin_ppc_load2r: {
     Function *F = CGM.getIntrinsic(Intrinsic::ppc_load2r);
-    Value *Op0 = Builder.CreateBitCast(EmitScalarExpr(E->getArg(0)), Int8PtrTy);
+    Value *Op0 = EmitScalarExpr(E->getArg(0));
     Value *LoadIntrinsic = Builder.CreateCall(F, {Op0});
     return Builder.CreateTrunc(LoadIntrinsic, Int16Ty);
   }
   // FMA variations
   case PPC::BI__builtin_ppc_fnmsub:
   case PPC::BI__builtin_ppc_fnmsubs:
   case PPC::BI__builtin_vsx_xvmaddadp:
   case PPC::BI__builtin_vsx_xvmaddasp:
case PPC::BI__builtin_unpack_vector_int128: {
     if (getTarget().isLittleEndian())
       Index = ConstantInt::get(Index->getType(), 1 - Index->getZExtValue());
 
     return Builder.CreateExtractElement(Unpacked, Index);
   }
 
   case PPC::BI__builtin_ppc_sthcx: {
     llvm::Function *F = CGM.getIntrinsic(Intrinsic::ppc_sthcx);
-    Value *Op0 = Builder.CreateBitCast(EmitScalarExpr(E->getArg(0)), Int8PtrTy);
+    Value *Op0 = EmitScalarExpr(E->getArg(0));
     Value *Op1 = Builder.CreateSExt(EmitScalarExpr(E->getArg(1)), Int32Ty);
     return Builder.CreateCall(F, {Op0, Op1});
   }
 
   // The PPC MMA builtins take a pointer to a __vector_quad as an argument.
   // Some of the MMA instructions accumulate their result into an existing
   // accumulator whereas the others generate a new accumulator. So we need to
   // use custom code generation to expand a builtin call with a pointer to a
#include "clang/Basic/BuiltinsPPC.def"
   #include "clang/Basic/BuiltinsPPC.def"
     }
     if (BuiltinID == PPC::BI__builtin_vsx_lxvp ||
         BuiltinID == PPC::BI__builtin_vsx_stxvp ||
         BuiltinID == PPC::BI__builtin_mma_lxvp ||
         BuiltinID == PPC::BI__builtin_mma_stxvp) {
       if (BuiltinID == PPC::BI__builtin_vsx_lxvp ||
           BuiltinID == PPC::BI__builtin_mma_lxvp) {
-        Ops[1] = Builder.CreateBitCast(Ops[1], Int8PtrTy);
         Ops[0] = Builder.CreateGEP(Int8Ty, Ops[1], Ops[0]);
       } else {
-        Ops[2] = Builder.CreateBitCast(Ops[2], Int8PtrTy);
         Ops[1] = Builder.CreateGEP(Int8Ty, Ops[2], Ops[1]);
       }
       Ops.pop_back();
       llvm::Function *F = CGM.getIntrinsic(ID);
       return Builder.CreateCall(F, Ops, "");
     }
     SmallVector<Value*, 4> CallOps;
     if (Accumulate) {
Value *CodeGenFunction::EmitHexagonBuiltinExpr(unsigned BuiltinID,
                                                const CallExpr *E) {
   Intrinsic::ID ID;
   unsigned VecLen;
   std::tie(ID, VecLen) = getIntrinsicForHexagonNonClangBuiltin(BuiltinID);
 
   auto MakeCircOp = [this, E](unsigned IntID, bool IsLoad) {
     // The base pointer is passed by address, so it needs to be loaded.
     Address A = EmitPointerWithAlignment(E->getArg(0));
-    Address BP = Address(Builder.CreateBitCast(
-        A.getPointer(), Int8PtrPtrTy), Int8PtrTy, A.getAlignment());
+    Address BP = Address(A.getPointer(), Int8PtrTy, A.getAlignment());
     llvm::Value *Base = Builder.CreateLoad(BP);
     // The treatment of both loads and stores is the same: the arguments for
     // the builtin are the same as the arguments for the intrinsic.
     // Load:
     //   builtin(Base, Inc, Mod, Start) -> intr(Base, Inc, Mod, Start)
     //   builtin(Base, Mod, Start)      -> intr(Base, Mod, Start)
     // Store:
     //   builtin(Base, Inc, Mod, Val, Start) -> intr(Base, Inc, Mod, Val, Start)
Value *CodeGenFunction::EmitHexagonBuiltinExpr(unsigned BuiltinID,
 
   // Handle the conversion of bit-reverse load intrinsics to bit code.
   // The intrinsic call after this function only reads from memory and the
   // write to memory is dealt by the store instruction.
   auto MakeBrevLd = [this, E](unsigned IntID, llvm::Type *DestTy) {
     // The intrinsic generates one result, which is the new value for the base
     // pointer. It needs to be returned. The result of the load instruction is
     // passed to intrinsic by address, so the value needs to be stored.
-    llvm::Value *BaseAddress =
-        Builder.CreateBitCast(EmitScalarExpr(E->getArg(0)), Int8PtrTy);
+    llvm::Value *BaseAddress = EmitScalarExpr(E->getArg(0));
 
     // Expressions like &(*pt++) will be incremented per evaluation.
     // EmitPointerWithAlignment and EmitScalarExpr evaluates the expression
     // per call.
     Address DestAddr = EmitPointerWithAlignment(E->getArg(1));
-    DestAddr = Address(Builder.CreateBitCast(DestAddr.getPointer(), Int8PtrTy),
-                       Int8Ty, DestAddr.getAlignment());
+    DestAddr = Address(DestAddr.getPointer(), Int8Ty, DestAddr.getAlignment());
     llvm::Value *DestAddress = DestAddr.getPointer();
 
     // Operands are Base, Dest, Modifier.
     // The intrinsic format in LLVM IR is defined as
     // { ValueType, i8* } (i8*, i32).
     llvm::Value *Result = Builder.CreateCall(
         CGM.getIntrinsic(IntID), {BaseAddress, EmitScalarExpr(E->getArg(2))});
 

clang/lib/CodeGen/CGClass.cpp

 CodeGenFunction::GetAddressOfDerivedClass(Address BaseAddr,
                                           const CXXRecordDecl *Derived,
                                         CastExpr::path_const_iterator PathBegin,
                                           CastExpr::path_const_iterator PathEnd,
                                           bool NullCheckValue) {
   assert(PathBegin != PathEnd && "Base path should not be empty!");
 
   QualType DerivedTy =
       getContext().getCanonicalType(getContext().getTagDeclType(Derived));
-  unsigned AddrSpace = BaseAddr.getAddressSpace();
   llvm::Type *DerivedValueTy = ConvertType(DerivedTy);
-  llvm::Type *DerivedPtrTy =
-      llvm::PointerType::get(getLLVMContext(), AddrSpace);
 
   llvm::Value *NonVirtualOffset =
     CGM.GetNonVirtualBaseClassOffset(Derived, PathBegin, PathEnd);
 
   if (!NonVirtualOffset) {
     // No offset, we can just cast back.
     return BaseAddr.withElementType(DerivedValueTy);
   }
 
   llvm::BasicBlock *CastNull = nullptr;
   llvm::BasicBlock *CastNotNull = nullptr;
   llvm::BasicBlock *CastEnd = nullptr;
 
   if (NullCheckValue) {
     CastNull = createBasicBlock("cast.null");
     CastNotNull = createBasicBlock("cast.notnull");
     CastEnd = createBasicBlock("cast.end");
 
     llvm::Value *IsNull = Builder.CreateIsNull(BaseAddr.getPointer());
     Builder.CreateCondBr(IsNull, CastNull, CastNotNull);
     EmitBlock(CastNotNull);
   }
 
   // Apply the offset.
-  llvm::Value *Value = Builder.CreateBitCast(BaseAddr.getPointer(), Int8PtrTy);
+  llvm::Value *Value = BaseAddr.getPointer();
   Value = Builder.CreateInBoundsGEP(
       Int8Ty, Value, Builder.CreateNeg(NonVirtualOffset), "sub.ptr");
 
-  // Just cast.
-  Value = Builder.CreateBitCast(Value, DerivedPtrTy);
-
   // Produce a PHI if we had a null-check.
   if (NullCheckValue) {
     Builder.CreateBr(CastEnd);
     EmitBlock(CastNull);
     Builder.CreateBr(CastEnd);
     EmitBlock(CastEnd);
 
     llvm::PHINode *PHI = Builder.CreatePHI(Value->getType(), 2);
namespace {
   };
 
   static void EmitSanitizerDtorCallback(
       CodeGenFunction &CGF, StringRef Name, llvm::Value *Ptr,
       std::optional<CharUnits::QuantityType> PoisonSize = {}) {
     CodeGenFunction::SanitizerScope SanScope(&CGF);
     // Pass in void pointer and size of region as arguments to runtime
     // function
-    SmallVector<llvm::Value *, 2> Args = {
-        CGF.Builder.CreateBitCast(Ptr, CGF.VoidPtrTy)};
+    SmallVector<llvm::Value *, 2> Args = {Ptr};
     SmallVector<llvm::Type *, 2> ArgTypes = {CGF.VoidPtrTy};
 
     if (PoisonSize.has_value()) {
       Args.emplace_back(llvm::ConstantInt::get(CGF.SizeTy, *PoisonSize));
       ArgTypes.emplace_back(CGF.SizeTy);
     }
 
     llvm::FunctionType *FnType =
void Emit(CodeGenFunction &CGF, Flags flags) override {
 
       // It's a first trivial field so it should be at the begining of a char,
       // still round up start offset just in case.
       CharUnits PoisonStart = Context.toCharUnitsFromBits(
           Layout.getFieldOffset(StartIndex) + Context.getCharWidth() - 1);
       llvm::ConstantInt *OffsetSizePtr =
           llvm::ConstantInt::get(CGF.SizeTy, PoisonStart.getQuantity());
 
-      llvm::Value *OffsetPtr = CGF.Builder.CreateGEP(
-          CGF.Int8Ty,
-          CGF.Builder.CreateBitCast(CGF.LoadCXXThis(), CGF.Int8PtrTy),
-          OffsetSizePtr);
+      llvm::Value *OffsetPtr =
+          CGF.Builder.CreateGEP(CGF.Int8Ty, CGF.LoadCXXThis(), OffsetSizePtr);
 
       CharUnits PoisonEnd;
       if (EndIndex >= Layout.getFieldCount()) {
         PoisonEnd = Layout.getNonVirtualSize();
       } else {
         PoisonEnd =
             Context.toCharUnitsFromBits(Layout.getFieldOffset(EndIndex));
       }
else if (CGM.getCodeGenOpts().WholeProgramVTables &&
            // Don't insert type test assumes if we are forcing public
            // visibility.
            !CGM.AlwaysHasLTOVisibilityPublic(RD)) {
     QualType Ty = QualType(RD->getTypeForDecl(), 0);
     llvm::Metadata *MD = CGM.CreateMetadataIdentifierForType(Ty);
     llvm::Value *TypeId =
         llvm::MetadataAsValue::get(CGM.getLLVMContext(), MD);
 
-    llvm::Value *CastedVTable = Builder.CreateBitCast(VTable, Int8PtrTy);
     // If we already know that the call has hidden LTO visibility, emit
     // @llvm.type.test(). Otherwise emit @llvm.public.type.test(), which WPD
     // will convert to @llvm.type.test() if we assert at link time that we have
     // whole program visibility.
     llvm::Intrinsic::ID IID = CGM.HasHiddenLTOVisibility(RD)
                                   ? llvm::Intrinsic::type_test
                                   : llvm::Intrinsic::public_type_test;
     llvm::Value *TypeTest =
-        Builder.CreateCall(CGM.getIntrinsic(IID), {CastedVTable, TypeId});
+        Builder.CreateCall(CGM.getIntrinsic(IID), {VTable, TypeId});
     Builder.CreateCall(CGM.getIntrinsic(llvm::Intrinsic::assume), TypeTest);
   }
 }
 
 void CodeGenFunction::EmitVTablePtrCheckForCall(const CXXRecordDecl *RD,
                                                 llvm::Value *VTable,
                                                 CFITypeCheckKind TCK,
                                                 SourceLocation Loc) {
void CodeGenFunction::EmitVTablePtrCheck(const CXXRecordDecl *RD,
 
   SanitizerScope SanScope(this);
   EmitSanitizerStatReport(SSK);
 
   llvm::Metadata *MD =
       CGM.CreateMetadataIdentifierForType(QualType(RD->getTypeForDecl(), 0));
   llvm::Value *TypeId = llvm::MetadataAsValue::get(getLLVMContext(), MD);
 
-  llvm::Value *CastedVTable = Builder.CreateBitCast(VTable, Int8PtrTy);
   llvm::Value *TypeTest = Builder.CreateCall(
-      CGM.getIntrinsic(llvm::Intrinsic::type_test), {CastedVTable, TypeId});
+      CGM.getIntrinsic(llvm::Intrinsic::type_test), {VTable, TypeId});
 
   llvm::Constant *StaticData[] = {
       llvm::ConstantInt::get(Int8Ty, TCK),
       EmitCheckSourceLocation(Loc),
       EmitCheckTypeDescriptor(QualType(RD->getTypeForDecl(), 0)),
   };
 
   auto CrossDsoTypeId = CGM.CreateCrossDsoCfiTypeId(MD);
   if (CGM.getCodeGenOpts().SanitizeCfiCrossDso && CrossDsoTypeId) {
-    EmitCfiSlowPathCheck(M, TypeTest, CrossDsoTypeId, CastedVTable, StaticData);
+    EmitCfiSlowPathCheck(M, TypeTest, CrossDsoTypeId, VTable, StaticData);
     return;
   }
 
   if (CGM.getCodeGenOpts().SanitizeTrap.has(M)) {
     EmitTrapCheck(TypeTest, SanitizerHandler::CFICheckFail);
     return;
   }
 
   llvm::Value *AllVtables = llvm::MetadataAsValue::get(
       CGM.getLLVMContext(),
       llvm::MDString::get(CGM.getLLVMContext(), "all-vtables"));
   llvm::Value *ValidVtable = Builder.CreateCall(
-      CGM.getIntrinsic(llvm::Intrinsic::type_test), {CastedVTable, AllVtables});
+      CGM.getIntrinsic(llvm::Intrinsic::type_test), {VTable, AllVtables});
   EmitCheck(std::make_pair(TypeTest, M), SanitizerHandler::CFICheckFail,
-            StaticData, {CastedVTable, ValidVtable});
+            StaticData, {VTable, ValidVtable});
 }
 
 bool CodeGenFunction::ShouldEmitVTableTypeCheckedLoad(const CXXRecordDecl *RD) {
   if (!CGM.getCodeGenOpts().WholeProgramVTables ||
       !CGM.HasHiddenLTOVisibility(RD))
     return false;
 
   if (CGM.getCodeGenOpts().VirtualFunctionElimination)
llvm::Value *CodeGenFunction::EmitVTableTypeCheckedLoad(
   SanitizerScope SanScope(this);
 
   EmitSanitizerStatReport(llvm::SanStat_CFI_VCall);
 
   llvm::Metadata *MD =
       CGM.CreateMetadataIdentifierForType(QualType(RD->getTypeForDecl(), 0));
   llvm::Value *TypeId = llvm::MetadataAsValue::get(CGM.getLLVMContext(), MD);
 
-  llvm::Value *CastedVTable = Builder.CreateBitCast(VTable, Int8PtrTy);
   llvm::Value *CheckedLoad = Builder.CreateCall(
       CGM.getIntrinsic(llvm::Intrinsic::type_checked_load),
-      {CastedVTable, llvm::ConstantInt::get(Int32Ty, VTableByteOffset),
-       TypeId});
+      {VTable, llvm::ConstantInt::get(Int32Ty, VTableByteOffset), TypeId});
   llvm::Value *CheckResult = Builder.CreateExtractValue(CheckedLoad, 1);
 
   std::string TypeName = RD->getQualifiedNameAsString();
   if (SanOpts.has(SanitizerKind::CFIVCall) &&
       !getContext().getNoSanitizeList().containsType(SanitizerKind::CFIVCall,
                                                      TypeName)) {
     EmitCheck(std::make_pair(CheckResult, SanitizerKind::CFIVCall),
               SanitizerHandler::CFICheckFail, {}, {});

clang/lib/CodeGen/CGExpr.cpp

Address CodeGenFunction::CreateMemTemp(QualType Ty, CharUnits Align,
   Address Result = CreateTempAlloca(ConvertTypeForMem(Ty), Align, Name,
                                     /*ArraySize=*/nullptr, Alloca);
 
   if (Ty->isConstantMatrixType()) {
     auto *ArrayTy = cast<llvm::ArrayType>(Result.getElementType());
     auto *VectorTy = llvm::FixedVectorType::get(ArrayTy->getElementType(),
                                                 ArrayTy->getNumElements());
 
-    Result = Address(
-        Builder.CreateBitCast(Result.getPointer(), VectorTy->getPointerTo()),
-        VectorTy, Result.getAlignment(), KnownNonNull);
+    Result = Address(Result.getPointer(), VectorTy, Result.getAlignment(),
+                     KnownNonNull);
   }
   return Result;
 }
 
 Address CodeGenFunction::CreateMemTempWithoutCast(QualType Ty, CharUnits Align,
                                                   const Twine &Name) {
   return CreateTempAllocaWithoutCast(ConvertTypeForMem(Ty), Align, Name);
 }
if (!ConstantSize || !ConstantSize->isNullValue()) {
       // FIXME: If Address Sanitizer is enabled, insert dynamic instrumentation
       //        to check this.
       // FIXME: Get object address space
       llvm::Type *Tys[2] = { IntPtrTy, Int8PtrTy };
       llvm::Function *F = CGM.getIntrinsic(llvm::Intrinsic::objectsize, Tys);
       llvm::Value *Min = Builder.getFalse();
       llvm::Value *NullIsUnknown = Builder.getFalse();
       llvm::Value *Dynamic = Builder.getFalse();
-      llvm::Value *CastAddr = Builder.CreateBitCast(Ptr, Int8PtrTy);
       llvm::Value *LargeEnough = Builder.CreateICmpUGE(
-          Builder.CreateCall(F, {CastAddr, Min, NullIsUnknown, Dynamic}), Size);
+          Builder.CreateCall(F, {Ptr, Min, NullIsUnknown, Dynamic}), Size);
       Checks.push_back(std::make_pair(LargeEnough, SanitizerKind::ObjectSize));
     }
   }
 
   llvm::MaybeAlign AlignVal;
   llvm::Value *PtrAsInt = nullptr;
 
   if (SanOpts.has(SanitizerKind::Alignment) &&
if (SanOpts.has(SanitizerKind::Vptr) &&
 
     // Contained in NoSanitizeList based on the mangled type.
     if (!CGM.getContext().getNoSanitizeList().containsType(SanitizerKind::Vptr,
                                                            Out.str())) {
       llvm::hash_code TypeHash = hash_value(Out.str());
 
       // Load the vptr, and compute hash_16_bytes(TypeHash, vptr).
       llvm::Value *Low = llvm::ConstantInt::get(Int64Ty, TypeHash);
-      llvm::Type *VPtrTy = llvm::PointerType::get(IntPtrTy, 0);
-      Address VPtrAddr(Builder.CreateBitCast(Ptr, VPtrTy), IntPtrTy,
-                       getPointerAlign());
+      Address VPtrAddr(Ptr, IntPtrTy, getPointerAlign());
       llvm::Value *VPtrVal = Builder.CreateLoad(VPtrAddr);
       llvm::Value *High = Builder.CreateZExt(VPtrVal, Int64Ty);
 
       llvm::Value *Hash = emitHash16Bytes(Builder, Low, High);
       Hash = Builder.CreateTrunc(Hash, IntPtrTy);
 
       // Look the hash up in our cache.
       const int CacheSize = 128;
if (const auto *Exp = dyn_cast<MemberExpr>(E)) {
     setObjCGCLValueClass(Ctx, Exp->getBase(), LV, true);
     // We don't know if member is an 'ivar', but this flag is looked at
     // only in the context of LV.isObjCIvar().
     LV.setObjCArray(E->getType()->isArrayType());
     return;
   }
 }
 
-static llvm::Value *
-EmitBitCastOfLValueToProperType(CodeGenFunction &CGF,
-                                llvm::Value *V, llvm::Type *IRType,
-                                StringRef Name = StringRef()) {
-  unsigned AS = cast<llvm::PointerType>(V->getType())->getAddressSpace();
-  return CGF.Builder.CreateBitCast(V, IRType->getPointerTo(AS), Name);
-}
-
 static LValue EmitThreadPrivateVarDeclLValue(
     CodeGenFunction &CGF, const VarDecl *VD, QualType T, Address Addr,
     llvm::Type *RealVarTy, SourceLocation Loc) {
   if (CGF.CGM.getLangOpts().OpenMPIRBuilder)
     Addr = CodeGenFunction::OMPBuilderCBHelpers::getAddrOfThreadPrivate(
         CGF, VD, Addr, Loc);
   else
     Addr =
static LValue EmitGlobalVarDeclLValue(CodeGenFunction &CGF,
   }
 
   llvm::Value *V = CGF.CGM.GetAddrOfGlobalVar(VD);
 
   if (VD->getTLSKind() != VarDecl::TLS_None)
     V = CGF.Builder.CreateThreadLocalAddress(V);
 
   llvm::Type *RealVarTy = CGF.getTypes().ConvertTypeForMem(VD->getType());
-  V = EmitBitCastOfLValueToProperType(CGF, V, RealVarTy);
   CharUnits Alignment = CGF.getContext().getDeclAlign(VD);
   Address Addr(V, RealVarTy, Alignment);
   // Emit reference to the private copy of the variable if it is an OpenMP
   // threadprivate variable.
   if (CGF.getLangOpts().OpenMP && !CGF.getLangOpts().OpenMPSimd &&
       VD->hasAttr<OMPThreadPrivateDeclAttr>()) {
     return EmitThreadPrivateVarDeclLValue(CGF, VD, T, Addr, RealVarTy,
                                           E->getExprLoc());
auto *InfoPtr =
                                  llvm::GlobalVariable::PrivateLinkage, Info);
     InfoPtr->setUnnamedAddr(llvm::GlobalValue::UnnamedAddr::Global);
     CGM.getSanitizerMetadata()->disableSanitizerForGlobal(InfoPtr);
 
     SlowPathFn = CGM.getModule().getOrInsertFunction(
         "__cfi_slowpath_diag",
         llvm::FunctionType::get(VoidTy, {Int64Ty, Int8PtrTy, Int8PtrTy},
                                 false));
-    CheckCall = Builder.CreateCall(
-        SlowPathFn, {TypeId, Ptr, Builder.CreateBitCast(InfoPtr, Int8PtrTy)});
+    CheckCall = Builder.CreateCall(SlowPathFn, {TypeId, Ptr, InfoPtr});
   } else {
     SlowPathFn = CGM.getModule().getOrInsertFunction(
         "__cfi_slowpath",
         llvm::FunctionType::get(VoidTy, {Int64Ty, Int8PtrTy}, false));
     CheckCall = Builder.CreateCall(SlowPathFn, {TypeId, Ptr});
   }
 
   CGM.setDSOLocal(
if (llvm::Constant *PrefixSig =
         llvm::Value *AlignedCalleeAddress =
             Builder.CreateAnd(CalleeAddress, Mask);
         AlignedCalleePtr =
             Builder.CreateIntToPtr(AlignedCalleeAddress, CalleePtr->getType());
       } else {
         AlignedCalleePtr = CalleePtr;
       }
 
-      llvm::Value *CalleePrefixStruct = Builder.CreateBitCast(
-          AlignedCalleePtr, llvm::PointerType::getUnqual(PrefixStructTy));
+      llvm::Value *CalleePrefixStruct = AlignedCalleePtr;
       llvm::Value *CalleeSigPtr =
           Builder.CreateConstGEP2_32(PrefixStructTy, CalleePrefixStruct, -1, 0);
       llvm::Value *CalleeSig =
           Builder.CreateAlignedLoad(PrefixSigType, CalleeSigPtr, getIntAlign());
       llvm::Value *CalleeSigMatch = Builder.CreateICmpEQ(CalleeSig, PrefixSig);
 
       llvm::BasicBlock *Cont = createBasicBlock("cont");
       llvm::BasicBlock *TypeCheck = createBasicBlock("typecheck");
if (SanOpts.has(SanitizerKind::CFIICall) &&
     if (CGM.getCodeGenOpts().SanitizeCfiICallGeneralizePointers)
       MD = CGM.CreateMetadataIdentifierGeneralized(QualType(FnType, 0));
     else
       MD = CGM.CreateMetadataIdentifierForType(QualType(FnType, 0));
 
     llvm::Value *TypeId = llvm::MetadataAsValue::get(getLLVMContext(), MD);
 
     llvm::Value *CalleePtr = Callee.getFunctionPointer();
-    llvm::Value *CastedCallee = Builder.CreateBitCast(CalleePtr, Int8PtrTy);
     llvm::Value *TypeTest = Builder.CreateCall(
-        CGM.getIntrinsic(llvm::Intrinsic::type_test), {CastedCallee, TypeId});
+        CGM.getIntrinsic(llvm::Intrinsic::type_test), {CalleePtr, TypeId});
 
     auto CrossDsoTypeId = CGM.CreateCrossDsoCfiTypeId(MD);
     llvm::Constant *StaticData[] = {
         llvm::ConstantInt::get(Int8Ty, CFITCK_ICall),
         EmitCheckSourceLocation(E->getBeginLoc()),
         EmitCheckTypeDescriptor(QualType(FnType, 0)),
     };
     if (CGM.getCodeGenOpts().SanitizeCfiCrossDso && CrossDsoTypeId) {
       EmitCfiSlowPathCheck(SanitizerKind::CFIICall, TypeTest, CrossDsoTypeId,
-                           CastedCallee, StaticData);
+                           CalleePtr, StaticData);
     } else {
       EmitCheck(std::make_pair(TypeTest, SanitizerKind::CFIICall),
                 SanitizerHandler::CFICheckFail, StaticData,
-                {CastedCallee, llvm::UndefValue::get(IntPtrTy)});
+                {CalleePtr, llvm::UndefValue::get(IntPtrTy)});
     }
   }
 
   CallArgList Args;
   if (Chain)
-    Args.add(RValue::get(Builder.CreateBitCast(Chain, CGM.VoidPtrTy)),
-             CGM.getContext().VoidPtrTy);
+    Args.add(RValue::get(Chain), CGM.getContext().VoidPtrTy);
 
   // C++17 requires that we evaluate arguments to a call using assignment syntax
   // right-to-left, and that we evaluate arguments to certain other operators
   // left-to-right. Note that we allow this to override the order dictated by
   // the calling convention on the MS ABI, which means that parameter
   // destruction order is not necessarily reverse construction order.
   // FIXME: Revisit this based on C++ committee response to unimplementability.
   EvaluationOrder Order = EvaluationOrder::Default;
RValue CodeGenFunction::EmitCall(QualType CalleeType, const CGCallee &OrigCallee,
 
   // HIP function pointer contains kernel handle when it is used in triple
   // chevron. The kernel stub needs to be loaded from kernel handle and used
   // as callee.
   if (CGM.getLangOpts().HIP && !CGM.getLangOpts().CUDAIsDevice &&
       isa<CUDAKernelCallExpr>(E) &&
       (!TargetDecl || !isa<FunctionDecl>(TargetDecl))) {
     llvm::Value *Handle = Callee.getFunctionPointer();
-    auto *Cast =
-        Builder.CreateBitCast(Handle, Handle->getType()->getPointerTo());
     auto *Stub = Builder.CreateLoad(
-        Address(Cast, Handle->getType(), CGM.getPointerAlign()));
+        Address(Handle, Handle->getType(), CGM.getPointerAlign()));
     Callee.setFunctionPointer(Stub);
   }
   llvm::CallBase *CallOrInvoke = nullptr;
   RValue Call = EmitCall(FnInfo, Callee, ReturnValue, Args, &CallOrInvoke,
                          E == MustTailCall, E->getExprLoc());
 
   // Generate function declaration DISuprogram in order to be used
   // in debug info about call sites.

clang/lib/CodeGen/CGExprScalar.cpp

if (!pointerType) {
     QualType objectType = pointerOperand->getType()
                                         ->castAs<ObjCObjectPointerType>()
                                         ->getPointeeType();
     llvm::Value *objectSize
       = CGF.CGM.getSize(CGF.getContext().getTypeSizeInChars(objectType));
 
     index = CGF.Builder.CreateMul(index, objectSize);
 
-    Value *result = CGF.Builder.CreateBitCast(pointer, CGF.VoidPtrTy);
-    result = CGF.Builder.CreateGEP(CGF.Int8Ty, result, index, "add.ptr");
+    Value *result =
+        CGF.Builder.CreateGEP(CGF.Int8Ty, pointer, index, "add.ptr");
     return CGF.Builder.CreateBitCast(result, pointer->getType());
   }
 
   QualType elementType = pointerType->getPointeeType();
   if (const VariableArrayType *vla
         = CGF.getContext().getAsVariableArrayType(elementType)) {
     // The element count here is the total number of non-VLA elements.
     llvm::Value *numElements = CGF.getVLASize(vla).NumElts;

clang/lib/CodeGen/CodeGenFunction.cpp

if (!CGM.getTypes().isZeroInitializable(Ty)) {
 
     llvm::GlobalVariable *NullVariable =
       new llvm::GlobalVariable(CGM.getModule(), NullConstant->getType(),
                                /*isConstant=*/true,
                                llvm::GlobalVariable::PrivateLinkage,
                                NullConstant, Twine());
     CharUnits NullAlign = DestPtr.getAlignment();
     NullVariable->setAlignment(NullAlign.getAsAlign());
-    Address SrcPtr(Builder.CreateBitCast(NullVariable, Builder.getInt8PtrTy()),
-                   Builder.getInt8Ty(), NullAlign);
+    Address SrcPtr(NullVariable, Builder.getInt8Ty(), NullAlign);
 
     if (vla) return emitNonZeroVLAInit(*this, Ty, DestPtr, SrcPtr, SizeVal);
 
     // Get and call the appropriate llvm.memcpy overload.
     Builder.CreateMemCpy(DestPtr, SrcPtr, SizeVal, false);
     return;
   }