C++ CallArgList类代码示例

/// \brief Emit a libcall for a binary operation on complex types.
ComplexPairTy ComplexExprEmitter::EmitComplexBinOpLibCall(StringRef LibCallName,
                                                          const BinOpInfo &Op) {
  CallArgList Args;
  Args.add(RValue::get(Op.LHS.first),
           Op.Ty->castAs<ComplexType>()->getElementType());
  Args.add(RValue::get(Op.LHS.second),
           Op.Ty->castAs<ComplexType>()->getElementType());
  Args.add(RValue::get(Op.RHS.first),
           Op.Ty->castAs<ComplexType>()->getElementType());
  Args.add(RValue::get(Op.RHS.second),
           Op.Ty->castAs<ComplexType>()->getElementType());

  // We *must* use the full CG function call building logic here because the
  // complex type has special ABI handling. We also should not forget about
  // special calling convention which may be used for compiler builtins.
  const CGFunctionInfo &FuncInfo =
    CGF.CGM.getTypes().arrangeFreeFunctionCall(
      Op.Ty, Args, FunctionType::ExtInfo(/* No CC here - will be added later */),
      RequiredArgs::All);
  llvm::FunctionType *FTy = CGF.CGM.getTypes().GetFunctionType(FuncInfo);
  llvm::Constant *Func = CGF.CGM.CreateBuiltinFunction(FTy, LibCallName);
  llvm::Instruction *Call;

  RValue Res = CGF.EmitCall(FuncInfo, Func, ReturnValueSlot(), Args,
                            nullptr, &Call);
  cast<llvm::CallInst>(Call)->setCallingConv(CGF.CGM.getBuiltinCC());
  cast<llvm::CallInst>(Call)->setDoesNotThrow();

  return Res.getComplexVal();
}

// CopyObject - Utility to copy an object.  Calls copy constructor as necessary.
// DestPtr is casted to the right type.
static void CopyObject(CodeGenFunction &CGF, const Expr *E, 
                       llvm::Value *DestPtr, llvm::Value *ExceptionPtrPtr) {
  QualType ObjectType = E->getType();

  // Store the throw exception in the exception object.
  if (!CGF.hasAggregateLLVMType(ObjectType)) {
    llvm::Value *Value = CGF.EmitScalarExpr(E);
    const llvm::Type *ValuePtrTy = Value->getType()->getPointerTo();

    CGF.Builder.CreateStore(Value, 
                            CGF.Builder.CreateBitCast(DestPtr, ValuePtrTy));
  } else {
    const llvm::Type *Ty = CGF.ConvertType(ObjectType)->getPointerTo();
    const CXXRecordDecl *RD =
      cast<CXXRecordDecl>(ObjectType->getAs<RecordType>()->getDecl());
    
    llvm::Value *This = CGF.Builder.CreateBitCast(DestPtr, Ty);
    if (RD->hasTrivialCopyConstructor()) {
      CGF.EmitAggExpr(E, This, false);
    } else if (CXXConstructorDecl *CopyCtor
               = RD->getCopyConstructor(CGF.getContext(), 0)) {
      llvm::BasicBlock *PrevLandingPad = CGF.getInvokeDest();
      if (CGF.Exceptions) {
        CodeGenFunction::EHCleanupBlock Cleanup(CGF);
        llvm::Constant *FreeExceptionFn = getFreeExceptionFn(CGF);
        
        // Load the exception pointer.
        llvm::Value *ExceptionPtr = CGF.Builder.CreateLoad(ExceptionPtrPtr);
        CGF.Builder.CreateCall(FreeExceptionFn, ExceptionPtr);
      }

      llvm::Value *Src = CGF.EmitLValue(E).getAddress();
      CGF.setInvokeDest(PrevLandingPad);

      llvm::BasicBlock *TerminateHandler = CGF.getTerminateHandler();
      PrevLandingPad = CGF.getInvokeDest();
      CGF.setInvokeDest(TerminateHandler);

      // Stolen from EmitClassAggrMemberwiseCopy
      llvm::Value *Callee = CGF.CGM.GetAddrOfCXXConstructor(CopyCtor,
                                                            Ctor_Complete);
      CallArgList CallArgs;
      CallArgs.push_back(std::make_pair(RValue::get(This),
                                      CopyCtor->getThisType(CGF.getContext())));

      // Push the Src ptr.
      CallArgs.push_back(std::make_pair(RValue::get(Src),
                                        CopyCtor->getParamDecl(0)->getType()));
      QualType ResultType =
        CopyCtor->getType()->getAs<FunctionType>()->getResultType();
      CGF.EmitCall(CGF.CGM.getTypes().getFunctionInfo(ResultType, CallArgs),
                   Callee, CallArgs, CopyCtor);
      CGF.setInvokeDest(PrevLandingPad);
    } else
      llvm_unreachable("uncopyable object");
  }
}

/// Emit a store to an l-value of atomic type.
///
/// Note that the r-value is expected to be an r-value *of the atomic
/// type*; this means that for aggregate r-values, it should include
/// storage for any padding that was necessary.
void CodeGenFunction::EmitAtomicStore(RValue rvalue, LValue dest, bool isInit) {
  // If this is an aggregate r-value, it should agree in type except
  // maybe for address-space qualification.
  assert(!rvalue.isAggregate() ||
         rvalue.getAggregateAddr()->getType()->getPointerElementType()
           == dest.getAddress()->getType()->getPointerElementType());

  AtomicInfo atomics(*this, dest);

  // If this is an initialization, just put the value there normally.
  if (isInit) {
    atomics.emitCopyIntoMemory(rvalue, dest);
    return;
  }

  // Check whether we should use a library call.
  if (atomics.shouldUseLibcall()) {
    // Produce a source address.
    llvm::Value *srcAddr = atomics.materializeRValue(rvalue);

    // void __atomic_store(size_t size, void *mem, void *val, int order)
    CallArgList args;
    args.add(RValue::get(atomics.getAtomicSizeValue()),
             getContext().getSizeType());
    args.add(RValue::get(EmitCastToVoidPtr(dest.getAddress())),
             getContext().VoidPtrTy);
    args.add(RValue::get(EmitCastToVoidPtr(srcAddr)),
             getContext().VoidPtrTy);
    args.add(RValue::get(llvm::ConstantInt::get(
                 IntTy, AtomicExpr::AO_ABI_memory_order_seq_cst)),
             getContext().IntTy);
    emitAtomicLibcall(*this, "__atomic_store", getContext().VoidTy, args);
    return;
  }

  // Okay, we're doing this natively.
  llvm::Value *intValue = atomics.convertRValueToInt(rvalue);

  // Do the atomic store.
  llvm::Value *addr = atomics.emitCastToAtomicIntPointer(dest.getAddress());
  llvm::StoreInst *store = Builder.CreateStore(intValue, addr);

  // Initializations don't need to be atomic.
  if (!isInit) store->setAtomic(llvm::SequentiallyConsistent);

  // Other decoration.
  store->setAlignment(dest.getAlignment().getQuantity());
  if (dest.isVolatileQualified())
    store->setVolatile(true);
  if (dest.getTBAAInfo())
    CGM.DecorateInstruction(store, dest.getTBAAInfo());
}

/// Emit a load from an l-value of atomic type.  Note that the r-value
/// we produce is an r-value of the atomic *value* type.
RValue CodeGenFunction::EmitAtomicLoad(LValue src, SourceLocation loc,
                                       AggValueSlot resultSlot) {
  AtomicInfo atomics(*this, src);

  // Check whether we should use a library call.
  if (atomics.shouldUseLibcall()) {
    llvm::Value *tempAddr;
    if (!resultSlot.isIgnored()) {
      assert(atomics.getEvaluationKind() == TEK_Aggregate);
      tempAddr = resultSlot.getAddr();
    } else {
      tempAddr = CreateMemTemp(atomics.getAtomicType(), "atomic-load-temp");
    }

    // void __atomic_load(size_t size, void *mem, void *return, int order);
    CallArgList args;
    args.add(RValue::get(atomics.getAtomicSizeValue()),
             getContext().getSizeType());
    args.add(RValue::get(EmitCastToVoidPtr(src.getAddress())),
             getContext().VoidPtrTy);
    args.add(RValue::get(EmitCastToVoidPtr(tempAddr)),
             getContext().VoidPtrTy);
    args.add(RValue::get(llvm::ConstantInt::get(
                 IntTy, AtomicExpr::AO_ABI_memory_order_seq_cst)),
             getContext().IntTy);
    emitAtomicLibcall(*this, "__atomic_load", getContext().VoidTy, args);

    // Produce the r-value.
    return atomics.convertTempToRValue(tempAddr, resultSlot, loc);
  }

  // Okay, we're doing this natively.
  llvm::Value *addr = atomics.emitCastToAtomicIntPointer(src.getAddress());
  llvm::LoadInst *load = Builder.CreateLoad(addr, "atomic-load");
  load->setAtomic(llvm::SequentiallyConsistent);

  // Other decoration.
  load->setAlignment(src.getAlignment().getQuantity());
  if (src.isVolatileQualified())
    load->setVolatile(true);
  if (src.getTBAAInfo())
    CGM.DecorateInstruction(load, src.getTBAAInfo());

  // If we're ignoring an aggregate return, don't do anything.
  if (atomics.getEvaluationKind() == TEK_Aggregate && resultSlot.isIgnored())
    return RValue::getAggregate(nullptr, false);

  // Okay, turn that back into the original value type.
  return atomics.convertIntToValue(load, resultSlot, loc);
}

static void
AddDirectArgument(CodeGenFunction &CGF, CallArgList &Args,
                  bool UseOptimizedLibcall, llvm::Value *Val, QualType ValTy) {
  if (UseOptimizedLibcall) {
    // Load value and pass it to the function directly.
    unsigned Align = CGF.getContext().getTypeAlignInChars(ValTy).getQuantity();
    Val = CGF.EmitLoadOfScalar(Val, false, Align, ValTy);
    Args.add(RValue::get(Val), ValTy);
  } else {
    // Non-optimized functions always take a reference.
    Args.add(RValue::get(CGF.EmitCastToVoidPtr(Val)),
                         CGF.getContext().VoidPtrTy);
  }
}

CharacterValueTy CharacterExprEmitter::VisitCallExpr(const CallExpr *E) {
  CharacterValueTy Dest;
  if(hasDestination())
    Dest = takeDestination();
  else {
    // FIXME function returning CHARACTER*(*)
    auto RetType = E->getFunction()->getType();
    Dest = CGF.GetCharacterValueFromPtr(
             CGF.CreateTempAlloca(CGF.ConvertTypeForMem(RetType), "characters"),
             RetType);
  }

  CallArgList ArgList;
  ArgList.addReturnValueArg(Dest);
  return CGF.EmitCall(E->getFunction(), ArgList, E->getArguments()).asCharacter();
}

// CopyObject - Utility to copy an object.  Calls copy constructor as necessary.
// N is casted to the right type.
static void CopyObject(CodeGenFunction &CGF, QualType ObjectType,
                       bool WasPointer, bool WasPointerReference,
                       llvm::Value *E, llvm::Value *N) {
  // Store the throw exception in the exception object.
  if (WasPointer || !CGF.hasAggregateLLVMType(ObjectType)) {
    llvm::Value *Value = E;
    if (!WasPointer)
      Value = CGF.Builder.CreateLoad(Value);
    const llvm::Type *ValuePtrTy = Value->getType()->getPointerTo(0);
    if (WasPointerReference) {
      llvm::Value *Tmp = CGF.CreateTempAlloca(Value->getType(), "catch.param");
      CGF.Builder.CreateStore(Value, Tmp);
      Value = Tmp;
      ValuePtrTy = Value->getType()->getPointerTo(0);
    }
    N = CGF.Builder.CreateBitCast(N, ValuePtrTy);
    CGF.Builder.CreateStore(Value, N);
  } else {
    const llvm::Type *Ty = CGF.ConvertType(ObjectType)->getPointerTo(0);
    const CXXRecordDecl *RD;
    RD = cast<CXXRecordDecl>(ObjectType->getAs<RecordType>()->getDecl());
    llvm::Value *This = CGF.Builder.CreateBitCast(N, Ty);
    if (RD->hasTrivialCopyConstructor()) {
      CGF.EmitAggregateCopy(This, E, ObjectType);
    } else if (CXXConstructorDecl *CopyCtor
               = RD->getCopyConstructor(CGF.getContext(), 0)) {
      llvm::Value *Src = E;

      // Stolen from EmitClassAggrMemberwiseCopy
      llvm::Value *Callee = CGF.CGM.GetAddrOfCXXConstructor(CopyCtor,
                                                            Ctor_Complete);
      CallArgList CallArgs;
      CallArgs.push_back(std::make_pair(RValue::get(This),
                                      CopyCtor->getThisType(CGF.getContext())));

      // Push the Src ptr.
      CallArgs.push_back(std::make_pair(RValue::get(Src),
                                        CopyCtor->getParamDecl(0)->getType()));

      const FunctionProtoType *FPT
        = CopyCtor->getType()->getAs<FunctionProtoType>();
      CGF.EmitCall(CGF.CGM.getTypes().getFunctionInfo(CallArgs, FPT),
                   Callee, ReturnValueSlot(), CallArgs, CopyCtor);
    } else
      llvm_unreachable("uncopyable object");
  }
}

void CodeGenFunction::EmitUPCAggregateCopy(llvm::Value *Dest, llvm::Value *Src,
                                           QualType DestTy, QualType SrcTy,
                                           SourceLocation Loc) {
  const ASTContext& Context = getContext();
  QualType ArgTy = Context.getPointerType(Context.getSharedType(Context.VoidTy));
  QualType SizeType = Context.getSizeType();
  assert(DestTy->getCanonicalTypeUnqualified() == SrcTy->getCanonicalTypeUnqualified());
  llvm::Constant *Len =
    llvm::ConstantInt::get(ConvertType(SizeType),
                           Context.getTypeSizeInChars(DestTy).getQuantity());
  llvm::SmallString<16> Name;
  const char *OpName;
  QualType DestArgTy, SrcArgTy;
  if (DestTy.getQualifiers().hasShared() && SrcTy.getQualifiers().hasShared()) {
    // both shared
    OpName = "copy";
    DestArgTy = SrcArgTy = ArgTy;
  } else if (DestTy.getQualifiers().hasShared()) {
    OpName = "put";
    DestArgTy = ArgTy;
    SrcArgTy = Context.VoidPtrTy;
  } else if (SrcTy.getQualifiers().hasShared()) {
    OpName = "get";
    DestArgTy = Context.VoidPtrTy;
    SrcArgTy = ArgTy;
  } else {
    llvm_unreachable("expected at least one shared argument");
  }

  Name += "__";
  Name += OpName;
  if (DestTy.getQualifiers().hasStrict() || SrcTy.getQualifiers().hasStrict())
    Name += 's';
  if (CGM.getCodeGenOpts().UPCDebug) Name += "g";
  Name += "blk";
  CallArgList Args;
  Args.add(RValue::get(Dest), DestArgTy);
  Args.add(RValue::get(Src), SrcArgTy);
  Args.add(RValue::get(Len), SizeType);
  if (CGM.getCodeGenOpts().UPCDebug) {
    getFileAndLine(*this, Loc, &Args);
    Name += '5';
  } else {
    Name += '3';
  }
  EmitUPCCall(*this, Name, Context.VoidTy, Args);
}

CodeGen::RValue
CGObjCJit::EmitMessageSend(CodeGen::CodeGenFunction &CGF,
                           ReturnValueSlot Return,
                           QualType ResultType,
                           Selector Sel,
                           llvm::Value *Arg0,
                           llvm::Value *Arg0Class,
                           const CallArgList &CallArgs,
                           const ObjCMethodDecl *Method) {

  llvm::Value *Arg1 = GetSelector(CGF, Sel);

  CallArgList ActualArgs;
  ActualArgs.add(RValue::get(Arg0), CGF.getContext().getObjCIdType());
  ActualArgs.add(RValue::get(Arg1), CGF.getContext().getObjCSelType());
  ActualArgs.addFrom(CallArgs);

  if (Method)
    assert(CGM.getContext().getCanonicalType(Method->getResultType()) ==
           CGM.getContext().getCanonicalType(ResultType) &&
           "Result type mismatch!");

  // Perform the following:
  //   imp = class_getMethodImplementation( Arg0Class, Arg1 );
  //   (*imp)( Arg0, Arg1, CallArgs );
  llvm::CallInst *getImp;

  // Unfortunately, using the GNU runtime version of
  // class_getMethodImplementation and then calling the resulting
  // IMP doesn't work unless objc_msg_lookup was already
  // called first. TODO: avoid doing this every time
  //
  if (fn_objc_msg_lookup.isValid()) {
    getImp = CGF.Builder.CreateCall2(fn_objc_msg_lookup,
                                     Arg0,
                                     Arg1);
  } else { // use the universal way
    getImp = CGF.Builder.CreateCall2(fn_class_getMethodImplementation,
                                     Arg0Class,
                                     Arg1);
  }

  MessageSendInfo MSI = getMessageSendInfo(Method, ResultType, ActualArgs);
  llvm::Value *theImp = CGF.Builder.CreateBitCast(getImp, MSI.MessengerType);
  return CGF.EmitCall(MSI.CallInfo, theImp, Return, ActualArgs);
}

// Transforms a call to printf into a call to the NVPTX vprintf syscall (which
// isn't particularly special; it's invoked just like a regular function).
// vprintf takes two args: A format string, and a pointer to a buffer containing
// the varargs.
//
// For example, the call
//
//   printf("format string", arg1, arg2, arg3);
//
// is converted into something resembling
//
//   struct Tmp {
//     Arg1 a1;
//     Arg2 a2;
//     Arg3 a3;
//   };
//   char* buf = alloca(sizeof(Tmp));
//   *(Tmp*)buf = {a1, a2, a3};
//   vprintf("format string", buf);
//
// buf is aligned to the max of {alignof(Arg1), ...}.  Furthermore, each of the
// args is itself aligned to its preferred alignment.
//
// Note that by the time this function runs, E's args have already undergone the
// standard C vararg promotion (short -> int, float -> double, etc.).
RValue
CodeGenFunction::EmitCUDADevicePrintfCallExpr(const CallExpr *E,
                                              ReturnValueSlot ReturnValue) {
  assert(getLangOpts().CUDA);
  assert(getLangOpts().CUDAIsDevice);
  assert(E->getBuiltinCallee() == Builtin::BIprintf);
  assert(E->getNumArgs() >= 1); // printf always has at least one arg.

  const llvm::DataLayout &DL = CGM.getDataLayout();
  llvm::LLVMContext &Ctx = CGM.getLLVMContext();

  CallArgList Args;
  EmitCallArgs(Args,
               E->getDirectCallee()->getType()->getAs<FunctionProtoType>(),
               E->arguments(), E->getDirectCallee(),
               /* ParamsToSkip = */ 0);

  // Construct and fill the args buffer that we'll pass to vprintf.
  llvm::Value *BufferPtr;
  if (Args.size() <= 1) {
    // If there are no args, pass a null pointer to vprintf.
    BufferPtr = llvm::ConstantPointerNull::get(llvm::Type::getInt8PtrTy(Ctx));
  } else {
    llvm::SmallVector<llvm::Type *, 8> ArgTypes;
    for (unsigned I = 1, NumArgs = Args.size(); I < NumArgs; ++I)
      ArgTypes.push_back(Args[I].RV.getScalarVal()->getType());
    llvm::Type *AllocaTy = llvm::StructType::create(ArgTypes, "printf_args");
    llvm::Value *Alloca = CreateTempAlloca(AllocaTy);

    for (unsigned I = 1, NumArgs = Args.size(); I < NumArgs; ++I) {
      llvm::Value *P = Builder.CreateStructGEP(AllocaTy, Alloca, I - 1);
      llvm::Value *Arg = Args[I].RV.getScalarVal();
      Builder.CreateAlignedStore(Arg, P, DL.getPrefTypeAlignment(Arg->getType()));
    }
    BufferPtr = Builder.CreatePointerCast(Alloca, llvm::Type::getInt8PtrTy(Ctx));
  }

  // Invoke vprintf and return.
  llvm::Function* VprintfFunc = GetVprintfDeclaration(CGM.getModule());
  return RValue::get(
      Builder.CreateCall(VprintfFunc, {Args[0].RV.getScalarVal(), BufferPtr}));
}

static void
AddDirectArgument(CodeGenFunction &CGF, CallArgList &Args,
                  bool UseOptimizedLibcall, llvm::Value *Val, QualType ValTy,
                  SourceLocation Loc, CharUnits SizeInChars) {
  if (UseOptimizedLibcall) {
    // Load value and pass it to the function directly.
    unsigned Align = CGF.getContext().getTypeAlignInChars(ValTy).getQuantity();
    int64_t SizeInBits = CGF.getContext().toBits(SizeInChars);
    ValTy =
        CGF.getContext().getIntTypeForBitwidth(SizeInBits, /*Signed=*/false);
    llvm::Type *IPtrTy = llvm::IntegerType::get(CGF.getLLVMContext(),
                                                SizeInBits)->getPointerTo();
    Val = CGF.EmitLoadOfScalar(CGF.Builder.CreateBitCast(Val, IPtrTy), false,
                               Align, CGF.getContext().getPointerType(ValTy),
                               Loc);
    // Coerce the value into an appropriately sized integer type.
    Args.add(RValue::get(Val), ValTy);
  } else {
    // Non-optimized functions always take a reference.
    Args.add(RValue::get(CGF.EmitCastToVoidPtr(Val)),
                         CGF.getContext().VoidPtrTy);
  }
}

/// \brief Emit a libcall for a binary operation on complex types.
ComplexPairTy ComplexExprEmitter::EmitComplexBinOpLibCall(StringRef LibCallName,
                                                          const BinOpInfo &Op) {
  CallArgList Args;
  Args.add(RValue::get(Op.LHS.first),
           Op.Ty->castAs<ComplexType>()->getElementType());
  Args.add(RValue::get(Op.LHS.second),
           Op.Ty->castAs<ComplexType>()->getElementType());
  Args.add(RValue::get(Op.RHS.first),
           Op.Ty->castAs<ComplexType>()->getElementType());
  Args.add(RValue::get(Op.RHS.second),
           Op.Ty->castAs<ComplexType>()->getElementType());

  // We *must* use the full CG function call building logic here because the
  // complex type has special ABI handling. We also should not forget about
  // special calling convention which may be used for compiler builtins.

  // We create a function qualified type to state that this call does not have
  // any exceptions.
  FunctionProtoType::ExtProtoInfo EPI;
  EPI = EPI.withExceptionSpec(
      FunctionProtoType::ExceptionSpecInfo(EST_BasicNoexcept));
  SmallVector<QualType, 4> ArgsQTys(
      4, Op.Ty->castAs<ComplexType>()->getElementType());
  QualType FQTy = CGF.getContext().getFunctionType(Op.Ty, ArgsQTys, EPI);
  const CGFunctionInfo &FuncInfo = CGF.CGM.getTypes().arrangeFreeFunctionCall(
      Args, cast<FunctionType>(FQTy.getTypePtr()), false);

  llvm::FunctionType *FTy = CGF.CGM.getTypes().GetFunctionType(FuncInfo);
  llvm::Constant *Func = CGF.CGM.CreateBuiltinFunction(FTy, LibCallName);
  CGCallee Callee = CGCallee::forDirect(Func, FQTy->getAs<FunctionProtoType>());

  llvm::Instruction *Call;
  RValue Res = CGF.EmitCall(FuncInfo, Callee, ReturnValueSlot(), Args, &Call);
  cast<llvm::CallInst>(Call)->setCallingConv(CGF.CGM.getBuiltinCC());
  return Res.getComplexVal();
}

RValue CodeGenFunction::EmitCXXMemberCallExpr(const CXXMemberCallExpr *CE) {
  const MemberExpr *ME = cast<MemberExpr>(CE->getCallee());
  const CXXMethodDecl *MD = cast<CXXMethodDecl>(ME->getMemberDecl());
  assert(MD->isInstance() && 
         "Trying to emit a member call expr on a static method!");
  
  const FunctionProtoType *FPT = MD->getType()->getAsFunctionProtoType();
  const llvm::Type *Ty = 
    CGM.getTypes().GetFunctionType(CGM.getTypes().getFunctionInfo(MD), 
                                   FPT->isVariadic());
  llvm::Constant *Callee = CGM.GetAddrOfFunction(MD, Ty);
  
  llvm::Value *BaseValue = 0;
  
  // There's a deref operator node added in Sema::BuildCallToMemberFunction
  // that's giving the wrong type for -> call exprs so we just ignore them
  // for now.
  if (ME->isArrow())
    return EmitUnsupportedRValue(CE, "C++ member call expr");
  else {
    LValue BaseLV = EmitLValue(ME->getBase());
    BaseValue = BaseLV.getAddress();
  }
  
  CallArgList Args;
  
  // Push the 'this' pointer.
  Args.push_back(std::make_pair(RValue::get(BaseValue), 
                                MD->getThisType(getContext())));
  
  EmitCallArgs(Args, FPT, CE->arg_begin(), CE->arg_end());
  
  QualType ResultType = MD->getType()->getAsFunctionType()->getResultType();
  return EmitCall(CGM.getTypes().getFunctionInfo(ResultType, Args), 
                  Callee, Args, MD);
}

RValue EmitUPCCall(CodeGenFunction &CGF,
                   llvm::StringRef Name,
                   QualType ResultTy,
                   const CallArgList& Args) {
  ASTContext &Context = CGF.CGM.getContext();
  llvm::SmallVector<QualType, 5> ArgTypes;

  for (CallArgList::const_iterator iter = Args.begin(),
         end = Args.end(); iter != end; ++iter) {
    ArgTypes.push_back(iter->Ty);
  }

  QualType FuncType =
    Context.getFunctionType(ResultTy,
                            ArgTypes,
                            FunctionProtoType::ExtProtoInfo());
    const CGFunctionInfo &Info =
      CGF.getTypes().arrangeFreeFunctionCall(Args, FuncType->castAs<FunctionType>());
    llvm::FunctionType * FTy =
      cast<llvm::FunctionType>(CGF.ConvertType(FuncType));
    llvm::Value * Fn = CGF.CGM.CreateRuntimeFunction(FTy, Name);

    return CGF.EmitCall(Info, Fn, ReturnValueSlot(), Args);
}

// Transforms a call to printf into a call to the NVPTX vprintf syscall (which
// isn't particularly special; it's invoked just like a regular function).
// vprintf takes two args: A format string, and a pointer to a buffer containing
// the varargs.
//
// For example, the call
//
//   printf("format string", arg1, arg2, arg3);
//
// is converted into something resembling
//
//   struct Tmp {
//     Arg1 a1;
//     Arg2 a2;
//     Arg3 a3;
//   };
//   char* buf = alloca(sizeof(Tmp));
//   *(Tmp*)buf = {a1, a2, a3};
//   vprintf("format string", buf);
//
// buf is aligned to the max of {alignof(Arg1), ...}.  Furthermore, each of the
// args is itself aligned to its preferred alignment.
//
// Note that by the time this function runs, E's args have already undergone the
// standard C vararg promotion (short -> int, float -> double, etc.).
RValue
CodeGenFunction::EmitNVPTXDevicePrintfCallExpr(const CallExpr *E,
                                               ReturnValueSlot ReturnValue) {
  assert(getTarget().getTriple().isNVPTX());
  assert(E->getBuiltinCallee() == Builtin::BIprintf);
  assert(E->getNumArgs() >= 1); // printf always has at least one arg.

  const llvm::DataLayout &DL = CGM.getDataLayout();
  llvm::LLVMContext &Ctx = CGM.getLLVMContext();

  CallArgList Args;
  EmitCallArgs(Args,
               E->getDirectCallee()->getType()->getAs<FunctionProtoType>(),
               E->arguments(), E->getDirectCallee(),
               /* ParamsToSkip = */ 0);

  // We don't know how to emit non-scalar varargs.
  if (std::any_of(Args.begin() + 1, Args.end(), [&](const CallArg &A) {
        return !A.getRValue(*this).isScalar();
      })) {
    CGM.ErrorUnsupported(E, "non-scalar arg to printf");
    return RValue::get(llvm::ConstantInt::get(IntTy, 0));
  }

  // Construct and fill the args buffer that we'll pass to vprintf.
  llvm::Value *BufferPtr;
  if (Args.size() <= 1) {
    // If there are no args, pass a null pointer to vprintf.
    BufferPtr = llvm::ConstantPointerNull::get(llvm::Type::getInt8PtrTy(Ctx));
  } else {
    llvm::SmallVector<llvm::Type *, 8> ArgTypes;
    for (unsigned I = 1, NumArgs = Args.size(); I < NumArgs; ++I)
      ArgTypes.push_back(Args[I].getRValue(*this).getScalarVal()->getType());

    // Using llvm::StructType is correct only because printf doesn't accept
    // aggregates.  If we had to handle aggregates here, we'd have to manually
    // compute the offsets within the alloca -- we wouldn't be able to assume
    // that the alignment of the llvm type was the same as the alignment of the
    // clang type.
    llvm::Type *AllocaTy = llvm::StructType::create(ArgTypes, "printf_args");
    llvm::Value *Alloca = CreateTempAlloca(AllocaTy);

    for (unsigned I = 1, NumArgs = Args.size(); I < NumArgs; ++I) {
      llvm::Value *P = Builder.CreateStructGEP(AllocaTy, Alloca, I - 1);
      llvm::Value *Arg = Args[I].getRValue(*this).getScalarVal();
      Builder.CreateAlignedStore(Arg, P, DL.getPrefTypeAlignment(Arg->getType()));
    }
    BufferPtr = Builder.CreatePointerCast(Alloca, llvm::Type::getInt8PtrTy(Ctx));
  }

  // Invoke vprintf and return.
  llvm::Function* VprintfFunc = GetVprintfDeclaration(CGM.getModule());
  return RValue::get(Builder.CreateCall(
      VprintfFunc, {Args[0].getRValue(*this).getScalarVal(), BufferPtr}));
}