//===-- CppWriter.cpp - Printing LLVM IR as a C++ Source File -------------===// // // The LLVM Compiler Infrastructure // // This file was developed by Reid Spencer and is distributed under the // University of Illinois Open Source License. See LICENSE.TXT for details. // //===----------------------------------------------------------------------===// // // This file implements the writing of the LLVM IR as a set of C++ calls to the // LLVM IR interface. The input module is assumed to be verified. // //===----------------------------------------------------------------------===// #include "llvm/CallingConv.h" #include "llvm/Constants.h" #include "llvm/DerivedTypes.h" #include "llvm/InlineAsm.h" #include "llvm/Instruction.h" #include "llvm/Instructions.h" #include "llvm/ParameterAttributes.h" #include "llvm/Module.h" #include "llvm/TypeSymbolTable.h" #include "llvm/ADT/StringExtras.h" #include "llvm/ADT/STLExtras.h" #include "llvm/ADT/SmallPtrSet.h" #include "llvm/Support/CommandLine.h" #include "llvm/Support/CFG.h" #include "llvm/Support/ManagedStatic.h" #include "llvm/Support/MathExtras.h" #include "llvm/Config/config.h" #include #include #include using namespace llvm; static cl::opt FuncName("funcname", cl::desc("Specify the name of the generated function"), cl::value_desc("function name")); enum WhatToGenerate { GenProgram, GenModule, GenContents, GenFunction, GenInline, GenVariable, GenType }; static cl::opt GenerationType(cl::Optional, cl::desc("Choose what kind of output to generate"), cl::init(GenProgram), cl::values( clEnumValN(GenProgram, "gen-program", "Generate a complete program"), clEnumValN(GenModule, "gen-module", "Generate a module definition"), clEnumValN(GenContents,"gen-contents", "Generate contents of a module"), clEnumValN(GenFunction,"gen-function", "Generate a function definition"), clEnumValN(GenInline, "gen-inline", "Generate an inline function"), clEnumValN(GenVariable,"gen-variable", "Generate a variable definition"), clEnumValN(GenType, "gen-type", "Generate a type definition"), clEnumValEnd ) ); static cl::opt NameToGenerate("for", cl::Optional, cl::desc("Specify the name of the thing to generate"), cl::init("!bad!")); namespace { typedef std::vector TypeList; typedef std::map TypeMap; typedef std::map ValueMap; typedef std::set NameSet; typedef std::set TypeSet; typedef std::set ValueSet; typedef std::map ForwardRefMap; class CppWriter { const char* progname; std::ostream &Out; const Module *TheModule; uint64_t uniqueNum; TypeMap TypeNames; ValueMap ValueNames; TypeMap UnresolvedTypes; TypeList TypeStack; NameSet UsedNames; TypeSet DefinedTypes; ValueSet DefinedValues; ForwardRefMap ForwardRefs; bool is_inline; public: inline CppWriter(std::ostream &o, const Module *M, const char* pn="llvm2cpp") : progname(pn), Out(o), TheModule(M), uniqueNum(0), TypeNames(), ValueNames(), UnresolvedTypes(), TypeStack(), is_inline(false) { } const Module* getModule() { return TheModule; } void printProgram(const std::string& fname, const std::string& modName ); void printModule(const std::string& fname, const std::string& modName ); void printContents(const std::string& fname, const std::string& modName ); void printFunction(const std::string& fname, const std::string& funcName ); void printInline(const std::string& fname, const std::string& funcName ); void printVariable(const std::string& fname, const std::string& varName ); void printType(const std::string& fname, const std::string& typeName ); void error(const std::string& msg); private: void printLinkageType(GlobalValue::LinkageTypes LT); void printCallingConv(unsigned cc); void printEscapedString(const std::string& str); void printCFP(const ConstantFP* CFP); std::string getCppName(const Type* val); inline void printCppName(const Type* val); std::string getCppName(const Value* val); inline void printCppName(const Value* val); bool printTypeInternal(const Type* Ty); inline void printType(const Type* Ty); void printTypes(const Module* M); void printConstant(const Constant *CPV); void printConstants(const Module* M); void printVariableUses(const GlobalVariable *GV); void printVariableHead(const GlobalVariable *GV); void printVariableBody(const GlobalVariable *GV); void printFunctionUses(const Function *F); void printFunctionHead(const Function *F); void printFunctionBody(const Function *F); void printInstruction(const Instruction *I, const std::string& bbname); std::string getOpName(Value*); void printModuleBody(); }; static unsigned indent_level = 0; inline std::ostream& nl(std::ostream& Out, int delta = 0) { Out << "\n"; if (delta >= 0 || indent_level >= unsigned(-delta)) indent_level += delta; for (unsigned i = 0; i < indent_level; ++i) Out << " "; return Out; } inline void in() { indent_level++; } inline void out() { if (indent_level >0) indent_level--; } inline void sanitize(std::string& str) { for (size_t i = 0; i < str.length(); ++i) if (!isalnum(str[i]) && str[i] != '_') str[i] = '_'; } inline std::string getTypePrefix(const Type* Ty ) { switch (Ty->getTypeID()) { case Type::VoidTyID: return "void_"; case Type::IntegerTyID: return std::string("int") + utostr(cast(Ty)->getBitWidth()) + "_"; case Type::FloatTyID: return "float_"; case Type::DoubleTyID: return "double_"; case Type::LabelTyID: return "label_"; case Type::FunctionTyID: return "func_"; case Type::StructTyID: return "struct_"; case Type::ArrayTyID: return "array_"; case Type::PointerTyID: return "ptr_"; case Type::VectorTyID: return "packed_"; case Type::OpaqueTyID: return "opaque_"; default: return "other_"; } return "unknown_"; } // Looks up the type in the symbol table and returns a pointer to its name or // a null pointer if it wasn't found. Note that this isn't the same as the // Mode::getTypeName function which will return an empty string, not a null // pointer if the name is not found. inline const std::string* findTypeName(const TypeSymbolTable& ST, const Type* Ty) { TypeSymbolTable::const_iterator TI = ST.begin(); TypeSymbolTable::const_iterator TE = ST.end(); for (;TI != TE; ++TI) if (TI->second == Ty) return &(TI->first); return 0; } void CppWriter::error(const std::string& msg) { std::cerr << progname << ": " << msg << "\n"; exit(2); } // printCFP - Print a floating point constant .. very carefully :) // This makes sure that conversion to/from floating yields the same binary // result so that we don't lose precision. void CppWriter::printCFP(const ConstantFP *CFP) { Out << "ConstantFP::get("; if (CFP->getType() == Type::DoubleTy) Out << "Type::DoubleTy, "; else Out << "Type::FloatTy, "; #if HAVE_PRINTF_A char Buffer[100]; sprintf(Buffer, "%A", CFP->getValue()); if ((!strncmp(Buffer, "0x", 2) || !strncmp(Buffer, "-0x", 3) || !strncmp(Buffer, "+0x", 3)) && (atof(Buffer) == CFP->getValue())) if (CFP->getType() == Type::DoubleTy) Out << "BitsToDouble(" << Buffer << ")"; else Out << "BitsToFloat(" << Buffer << ")"; else { #endif std::string StrVal = ftostr(CFP->getValue()); while (StrVal[0] == ' ') StrVal.erase(StrVal.begin()); // Check to make sure that the stringized number is not some string like // "Inf" or NaN. Check that the string matches the "[-+]?[0-9]" regex. if (((StrVal[0] >= '0' && StrVal[0] <= '9') || ((StrVal[0] == '-' || StrVal[0] == '+') && (StrVal[1] >= '0' && StrVal[1] <= '9'))) && (atof(StrVal.c_str()) == CFP->getValue())) if (CFP->getType() == Type::DoubleTy) Out << StrVal; else Out << StrVal; else if (CFP->getType() == Type::DoubleTy) Out << "BitsToDouble(0x" << std::hex << DoubleToBits(CFP->getValue()) << std::dec << "ULL) /* " << StrVal << " */"; else Out << "BitsToFloat(0x" << std::hex << FloatToBits(CFP->getValue()) << std::dec << "U) /* " << StrVal << " */"; #if HAVE_PRINTF_A } #endif Out << ")"; } void CppWriter::printCallingConv(unsigned cc){ // Print the calling convention. switch (cc) { case CallingConv::C: Out << "CallingConv::C"; break; case CallingConv::Fast: Out << "CallingConv::Fast"; break; case CallingConv::Cold: Out << "CallingConv::Cold"; break; case CallingConv::FirstTargetCC: Out << "CallingConv::FirstTargetCC"; break; default: Out << cc; break; } } void CppWriter::printLinkageType(GlobalValue::LinkageTypes LT) { switch (LT) { case GlobalValue::InternalLinkage: Out << "GlobalValue::InternalLinkage"; break; case GlobalValue::LinkOnceLinkage: Out << "GlobalValue::LinkOnceLinkage "; break; case GlobalValue::WeakLinkage: Out << "GlobalValue::WeakLinkage"; break; case GlobalValue::AppendingLinkage: Out << "GlobalValue::AppendingLinkage"; break; case GlobalValue::ExternalLinkage: Out << "GlobalValue::ExternalLinkage"; break; case GlobalValue::DLLImportLinkage: Out << "GlobalValue::DLLImportLinkage"; break; case GlobalValue::DLLExportLinkage: Out << "GlobalValue::DLLExportLinkage"; break; case GlobalValue::ExternalWeakLinkage: Out << "GlobalValue::ExternalWeakLinkage"; break; case GlobalValue::GhostLinkage: Out << "GlobalValue::GhostLinkage"; break; } } // printEscapedString - Print each character of the specified string, escaping // it if it is not printable or if it is an escape char. void CppWriter::printEscapedString(const std::string &Str) { for (unsigned i = 0, e = Str.size(); i != e; ++i) { unsigned char C = Str[i]; if (isprint(C) && C != '"' && C != '\\') { Out << C; } else { Out << "\\x" << (char) ((C/16 < 10) ? ( C/16 +'0') : ( C/16 -10+'A')) << (char)(((C&15) < 10) ? ((C&15)+'0') : ((C&15)-10+'A')); } } } std::string CppWriter::getCppName(const Type* Ty) { // First, handle the primitive types .. easy if (Ty->isPrimitiveType() || Ty->isInteger()) { switch (Ty->getTypeID()) { case Type::VoidTyID: return "Type::VoidTy"; case Type::IntegerTyID: { unsigned BitWidth = cast(Ty)->getBitWidth(); return "IntegerType::get(" + utostr(BitWidth) + ")"; } case Type::FloatTyID: return "Type::FloatTy"; case Type::DoubleTyID: return "Type::DoubleTy"; case Type::LabelTyID: return "Type::LabelTy"; default: error("Invalid primitive type"); break; } return "Type::VoidTy"; // shouldn't be returned, but make it sensible } // Now, see if we've seen the type before and return that TypeMap::iterator I = TypeNames.find(Ty); if (I != TypeNames.end()) return I->second; // Okay, let's build a new name for this type. Start with a prefix const char* prefix = 0; switch (Ty->getTypeID()) { case Type::FunctionTyID: prefix = "FuncTy_"; break; case Type::StructTyID: prefix = "StructTy_"; break; case Type::ArrayTyID: prefix = "ArrayTy_"; break; case Type::PointerTyID: prefix = "PointerTy_"; break; case Type::OpaqueTyID: prefix = "OpaqueTy_"; break; case Type::VectorTyID: prefix = "VectorTy_"; break; default: prefix = "OtherTy_"; break; // prevent breakage } // See if the type has a name in the symboltable and build accordingly const std::string* tName = findTypeName(TheModule->getTypeSymbolTable(), Ty); std::string name; if (tName) name = std::string(prefix) + *tName; else name = std::string(prefix) + utostr(uniqueNum++); sanitize(name); // Save the name return TypeNames[Ty] = name; } void CppWriter::printCppName(const Type* Ty) { printEscapedString(getCppName(Ty)); } std::string CppWriter::getCppName(const Value* val) { std::string name; ValueMap::iterator I = ValueNames.find(val); if (I != ValueNames.end() && I->first == val) return I->second; if (const GlobalVariable* GV = dyn_cast(val)) { name = std::string("gvar_") + getTypePrefix(GV->getType()->getElementType()); } else if (isa(val)) { name = std::string("func_"); } else if (const Constant* C = dyn_cast(val)) { name = std::string("const_") + getTypePrefix(C->getType()); } else if (const Argument* Arg = dyn_cast(val)) { if (is_inline) { unsigned argNum = std::distance(Arg->getParent()->arg_begin(), Function::const_arg_iterator(Arg)) + 1; name = std::string("arg_") + utostr(argNum); NameSet::iterator NI = UsedNames.find(name); if (NI != UsedNames.end()) name += std::string("_") + utostr(uniqueNum++); UsedNames.insert(name); return ValueNames[val] = name; } else { name = getTypePrefix(val->getType()); } } else { name = getTypePrefix(val->getType()); } name += (val->hasName() ? val->getName() : utostr(uniqueNum++)); sanitize(name); NameSet::iterator NI = UsedNames.find(name); if (NI != UsedNames.end()) name += std::string("_") + utostr(uniqueNum++); UsedNames.insert(name); return ValueNames[val] = name; } void CppWriter::printCppName(const Value* val) { printEscapedString(getCppName(val)); } bool CppWriter::printTypeInternal(const Type* Ty) { // We don't print definitions for primitive types if (Ty->isPrimitiveType() || Ty->isInteger()) return false; // If we already defined this type, we don't need to define it again. if (DefinedTypes.find(Ty) != DefinedTypes.end()) return false; // Everything below needs the name for the type so get it now. std::string typeName(getCppName(Ty)); // Search the type stack for recursion. If we find it, then generate this // as an OpaqueType, but make sure not to do this multiple times because // the type could appear in multiple places on the stack. Once the opaque // definition is issued, it must not be re-issued. Consequently we have to // check the UnresolvedTypes list as well. TypeList::const_iterator TI = std::find(TypeStack.begin(),TypeStack.end(),Ty); if (TI != TypeStack.end()) { TypeMap::const_iterator I = UnresolvedTypes.find(Ty); if (I == UnresolvedTypes.end()) { Out << "PATypeHolder " << typeName << "_fwd = OpaqueType::get();"; nl(Out); UnresolvedTypes[Ty] = typeName; } return true; } // We're going to print a derived type which, by definition, contains other // types. So, push this one we're printing onto the type stack to assist with // recursive definitions. TypeStack.push_back(Ty); // Print the type definition switch (Ty->getTypeID()) { case Type::FunctionTyID: { const FunctionType* FT = cast(Ty); Out << "std::vector" << typeName << "_args;"; nl(Out); FunctionType::param_iterator PI = FT->param_begin(); FunctionType::param_iterator PE = FT->param_end(); for (; PI != PE; ++PI) { const Type* argTy = static_cast(*PI); bool isForward = printTypeInternal(argTy); std::string argName(getCppName(argTy)); Out << typeName << "_args.push_back(" << argName; if (isForward) Out << "_fwd"; Out << ");"; nl(Out); } const ParamAttrsList *PAL = FT->getParamAttrs(); Out << "ParamAttrsList *" << typeName << "_PAL = 0;"; nl(Out); if (PAL) { Out << '{'; in(); nl(Out); Out << "ParamAttrsVector Attrs;"; nl(Out); Out << "ParamAttrsWithIndex PAWI;"; nl(Out); for (unsigned i = 0; i < PAL->size(); ++i) { uint16_t index = PAL->getParamIndex(i); uint16_t attrs = PAL->getParamAttrs(index); Out << "PAWI.index = " << index << "; PAWI.attrs = 0 "; if (attrs & ParamAttr::SExt) Out << " | ParamAttr::SExt"; if (attrs & ParamAttr::ZExt) Out << " | ParamAttr::ZExt"; if (attrs & ParamAttr::NoAlias) Out << " | ParamAttr::NoAlias"; if (attrs & ParamAttr::StructRet) Out << " | ParamAttr::StructRet"; if (attrs & ParamAttr::InReg) Out << " | ParamAttr::InReg"; if (attrs & ParamAttr::NoReturn) Out << " | ParamAttr::NoReturn"; if (attrs & ParamAttr::NoUnwind) Out << " | ParamAttr::NoUnwind"; Out << ";"; nl(Out); Out << "Attrs.push_back(PAWI);"; nl(Out); } Out << typeName << "_PAL = ParamAttrsList::get(Attrs);"; nl(Out); out(); nl(Out); Out << '}'; nl(Out); } bool isForward = printTypeInternal(FT->getReturnType()); std::string retTypeName(getCppName(FT->getReturnType())); Out << "FunctionType* " << typeName << " = FunctionType::get("; in(); nl(Out) << "/*Result=*/" << retTypeName; if (isForward) Out << "_fwd"; Out << ","; nl(Out) << "/*Params=*/" << typeName << "_args,"; nl(Out) << "/*isVarArg=*/" << (FT->isVarArg() ? "true," : "false,") ; nl(Out) << "/*ParamAttrs=*/" << typeName << "_PAL" << ");"; out(); nl(Out); break; } case Type::StructTyID: { const StructType* ST = cast(Ty); Out << "std::vector" << typeName << "_fields;"; nl(Out); StructType::element_iterator EI = ST->element_begin(); StructType::element_iterator EE = ST->element_end(); for (; EI != EE; ++EI) { const Type* fieldTy = static_cast(*EI); bool isForward = printTypeInternal(fieldTy); std::string fieldName(getCppName(fieldTy)); Out << typeName << "_fields.push_back(" << fieldName; if (isForward) Out << "_fwd"; Out << ");"; nl(Out); } Out << "StructType* " << typeName << " = StructType::get(" << typeName << "_fields, /*isPacked=*/" << (ST->isPacked() ? "true" : "false") << ");"; nl(Out); break; } case Type::ArrayTyID: { const ArrayType* AT = cast(Ty); const Type* ET = AT->getElementType(); bool isForward = printTypeInternal(ET); std::string elemName(getCppName(ET)); Out << "ArrayType* " << typeName << " = ArrayType::get(" << elemName << (isForward ? "_fwd" : "") << ", " << utostr(AT->getNumElements()) << ");"; nl(Out); break; } case Type::PointerTyID: { const PointerType* PT = cast(Ty); const Type* ET = PT->getElementType(); bool isForward = printTypeInternal(ET); std::string elemName(getCppName(ET)); Out << "PointerType* " << typeName << " = PointerType::get(" << elemName << (isForward ? "_fwd" : "") << ");"; nl(Out); break; } case Type::VectorTyID: { const VectorType* PT = cast(Ty); const Type* ET = PT->getElementType(); bool isForward = printTypeInternal(ET); std::string elemName(getCppName(ET)); Out << "VectorType* " << typeName << " = VectorType::get(" << elemName << (isForward ? "_fwd" : "") << ", " << utostr(PT->getNumElements()) << ");"; nl(Out); break; } case Type::OpaqueTyID: { Out << "OpaqueType* " << typeName << " = OpaqueType::get();"; nl(Out); break; } default: error("Invalid TypeID"); } // If the type had a name, make sure we recreate it. const std::string* progTypeName = findTypeName(TheModule->getTypeSymbolTable(),Ty); if (progTypeName) { Out << "mod->addTypeName(\"" << *progTypeName << "\", " << typeName << ");"; nl(Out); } // Pop us off the type stack TypeStack.pop_back(); // Indicate that this type is now defined. DefinedTypes.insert(Ty); // Early resolve as many unresolved types as possible. Search the unresolved // types map for the type we just printed. Now that its definition is complete // we can resolve any previous references to it. This prevents a cascade of // unresolved types. TypeMap::iterator I = UnresolvedTypes.find(Ty); if (I != UnresolvedTypes.end()) { Out << "cast(" << I->second << "_fwd.get())->refineAbstractTypeTo(" << I->second << ");"; nl(Out); Out << I->second << " = cast<"; switch (Ty->getTypeID()) { case Type::FunctionTyID: Out << "FunctionType"; break; case Type::ArrayTyID: Out << "ArrayType"; break; case Type::StructTyID: Out << "StructType"; break; case Type::VectorTyID: Out << "VectorType"; break; case Type::PointerTyID: Out << "PointerType"; break; case Type::OpaqueTyID: Out << "OpaqueType"; break; default: Out << "NoSuchDerivedType"; break; } Out << ">(" << I->second << "_fwd.get());"; nl(Out); nl(Out); UnresolvedTypes.erase(I); } // Finally, separate the type definition from other with a newline. nl(Out); // We weren't a recursive type return false; } // Prints a type definition. Returns true if it could not resolve all the types // in the definition but had to use a forward reference. void CppWriter::printType(const Type* Ty) { assert(TypeStack.empty()); TypeStack.clear(); printTypeInternal(Ty); assert(TypeStack.empty()); } void CppWriter::printTypes(const Module* M) { // Walk the symbol table and print out all its types const TypeSymbolTable& symtab = M->getTypeSymbolTable(); for (TypeSymbolTable::const_iterator TI = symtab.begin(), TE = symtab.end(); TI != TE; ++TI) { // For primitive types and types already defined, just add a name TypeMap::const_iterator TNI = TypeNames.find(TI->second); if (TI->second->isInteger() || TI->second->isPrimitiveType() || TNI != TypeNames.end()) { Out << "mod->addTypeName(\""; printEscapedString(TI->first); Out << "\", " << getCppName(TI->second) << ");"; nl(Out); // For everything else, define the type } else { printType(TI->second); } } // Add all of the global variables to the value table... for (Module::const_global_iterator I = TheModule->global_begin(), E = TheModule->global_end(); I != E; ++I) { if (I->hasInitializer()) printType(I->getInitializer()->getType()); printType(I->getType()); } // Add all the functions to the table for (Module::const_iterator FI = TheModule->begin(), FE = TheModule->end(); FI != FE; ++FI) { printType(FI->getReturnType()); printType(FI->getFunctionType()); // Add all the function arguments for(Function::const_arg_iterator AI = FI->arg_begin(), AE = FI->arg_end(); AI != AE; ++AI) { printType(AI->getType()); } // Add all of the basic blocks and instructions for (Function::const_iterator BB = FI->begin(), E = FI->end(); BB != E; ++BB) { printType(BB->getType()); for (BasicBlock::const_iterator I = BB->begin(), E = BB->end(); I!=E; ++I) { printType(I->getType()); for (unsigned i = 0; i < I->getNumOperands(); ++i) printType(I->getOperand(i)->getType()); } } } } // printConstant - Print out a constant pool entry... void CppWriter::printConstant(const Constant *CV) { // First, if the constant is actually a GlobalValue (variable or function) or // its already in the constant list then we've printed it already and we can // just return. if (isa(CV) || ValueNames.find(CV) != ValueNames.end()) return; std::string constName(getCppName(CV)); std::string typeName(getCppName(CV->getType())); if (CV->isNullValue()) { Out << "Constant* " << constName << " = Constant::getNullValue(" << typeName << ");"; nl(Out); return; } if (isa(CV)) { // Skip variables and functions, we emit them elsewhere return; } if (const ConstantInt *CI = dyn_cast(CV)) { Out << "ConstantInt* " << constName << " = ConstantInt::get(APInt(" << cast(CI->getType())->getBitWidth() << ", " << " \"" << CI->getValue().toStringSigned(10) << "\", 10));"; } else if (isa(CV)) { Out << "ConstantAggregateZero* " << constName << " = ConstantAggregateZero::get(" << typeName << ");"; } else if (isa(CV)) { Out << "ConstantPointerNull* " << constName << " = ConstanPointerNull::get(" << typeName << ");"; } else if (const ConstantFP *CFP = dyn_cast(CV)) { Out << "ConstantFP* " << constName << " = "; printCFP(CFP); Out << ";"; } else if (const ConstantArray *CA = dyn_cast(CV)) { if (CA->isString() && CA->getType()->getElementType() == Type::Int8Ty) { Out << "Constant* " << constName << " = ConstantArray::get(\""; std::string tmp = CA->getAsString(); bool nullTerminate = false; if (tmp[tmp.length()-1] == 0) { tmp.erase(tmp.length()-1); nullTerminate = true; } printEscapedString(tmp); // Determine if we want null termination or not. if (nullTerminate) Out << "\", true"; // Indicate that the null terminator should be added. else Out << "\", false";// No null terminator Out << ");"; } else { Out << "std::vector " << constName << "_elems;"; nl(Out); unsigned N = CA->getNumOperands(); for (unsigned i = 0; i < N; ++i) { printConstant(CA->getOperand(i)); // recurse to print operands Out << constName << "_elems.push_back(" << getCppName(CA->getOperand(i)) << ");"; nl(Out); } Out << "Constant* " << constName << " = ConstantArray::get(" << typeName << ", " << constName << "_elems);"; } } else if (const ConstantStruct *CS = dyn_cast(CV)) { Out << "std::vector " << constName << "_fields;"; nl(Out); unsigned N = CS->getNumOperands(); for (unsigned i = 0; i < N; i++) { printConstant(CS->getOperand(i)); Out << constName << "_fields.push_back(" << getCppName(CS->getOperand(i)) << ");"; nl(Out); } Out << "Constant* " << constName << " = ConstantStruct::get(" << typeName << ", " << constName << "_fields);"; } else if (const ConstantVector *CP = dyn_cast(CV)) { Out << "std::vector " << constName << "_elems;"; nl(Out); unsigned N = CP->getNumOperands(); for (unsigned i = 0; i < N; ++i) { printConstant(CP->getOperand(i)); Out << constName << "_elems.push_back(" << getCppName(CP->getOperand(i)) << ");"; nl(Out); } Out << "Constant* " << constName << " = ConstantVector::get(" << typeName << ", " << constName << "_elems);"; } else if (isa(CV)) { Out << "UndefValue* " << constName << " = UndefValue::get(" << typeName << ");"; } else if (const ConstantExpr *CE = dyn_cast(CV)) { if (CE->getOpcode() == Instruction::GetElementPtr) { Out << "std::vector " << constName << "_indices;"; nl(Out); printConstant(CE->getOperand(0)); for (unsigned i = 1; i < CE->getNumOperands(); ++i ) { printConstant(CE->getOperand(i)); Out << constName << "_indices.push_back(" << getCppName(CE->getOperand(i)) << ");"; nl(Out); } Out << "Constant* " << constName << " = ConstantExpr::getGetElementPtr(" << getCppName(CE->getOperand(0)) << ", " << "&" << constName << "_indices[0], " << CE->getNumOperands() - 1 << " );"; } else if (CE->isCast()) { printConstant(CE->getOperand(0)); Out << "Constant* " << constName << " = ConstantExpr::getCast("; switch (CE->getOpcode()) { default: assert(0 && "Invalid cast opcode"); case Instruction::Trunc: Out << "Instruction::Trunc"; break; case Instruction::ZExt: Out << "Instruction::ZExt"; break; case Instruction::SExt: Out << "Instruction::SExt"; break; case Instruction::FPTrunc: Out << "Instruction::FPTrunc"; break; case Instruction::FPExt: Out << "Instruction::FPExt"; break; case Instruction::FPToUI: Out << "Instruction::FPToUI"; break; case Instruction::FPToSI: Out << "Instruction::FPToSI"; break; case Instruction::UIToFP: Out << "Instruction::UIToFP"; break; case Instruction::SIToFP: Out << "Instruction::SIToFP"; break; case Instruction::PtrToInt: Out << "Instruction::PtrToInt"; break; case Instruction::IntToPtr: Out << "Instruction::IntToPtr"; break; case Instruction::BitCast: Out << "Instruction::BitCast"; break; } Out << ", " << getCppName(CE->getOperand(0)) << ", " << getCppName(CE->getType()) << ");"; } else { unsigned N = CE->getNumOperands(); for (unsigned i = 0; i < N; ++i ) { printConstant(CE->getOperand(i)); } Out << "Constant* " << constName << " = ConstantExpr::"; switch (CE->getOpcode()) { case Instruction::Add: Out << "getAdd("; break; case Instruction::Sub: Out << "getSub("; break; case Instruction::Mul: Out << "getMul("; break; case Instruction::UDiv: Out << "getUDiv("; break; case Instruction::SDiv: Out << "getSDiv("; break; case Instruction::FDiv: Out << "getFDiv("; break; case Instruction::URem: Out << "getURem("; break; case Instruction::SRem: Out << "getSRem("; break; case Instruction::FRem: Out << "getFRem("; break; case Instruction::And: Out << "getAnd("; break; case Instruction::Or: Out << "getOr("; break; case Instruction::Xor: Out << "getXor("; break; case Instruction::ICmp: Out << "getICmp(ICmpInst::ICMP_"; switch (CE->getPredicate()) { case ICmpInst::ICMP_EQ: Out << "EQ"; break; case ICmpInst::ICMP_NE: Out << "NE"; break; case ICmpInst::ICMP_SLT: Out << "SLT"; break; case ICmpInst::ICMP_ULT: Out << "ULT"; break; case ICmpInst::ICMP_SGT: Out << "SGT"; break; case ICmpInst::ICMP_UGT: Out << "UGT"; break; case ICmpInst::ICMP_SLE: Out << "SLE"; break; case ICmpInst::ICMP_ULE: Out << "ULE"; break; case ICmpInst::ICMP_SGE: Out << "SGE"; break; case ICmpInst::ICMP_UGE: Out << "UGE"; break; default: error("Invalid ICmp Predicate"); } break; case Instruction::FCmp: Out << "getFCmp(FCmpInst::FCMP_"; switch (CE->getPredicate()) { case FCmpInst::FCMP_FALSE: Out << "FALSE"; break; case FCmpInst::FCMP_ORD: Out << "ORD"; break; case FCmpInst::FCMP_UNO: Out << "UNO"; break; case FCmpInst::FCMP_OEQ: Out << "OEQ"; break; case FCmpInst::FCMP_UEQ: Out << "UEQ"; break; case FCmpInst::FCMP_ONE: Out << "ONE"; break; case FCmpInst::FCMP_UNE: Out << "UNE"; break; case FCmpInst::FCMP_OLT: Out << "OLT"; break; case FCmpInst::FCMP_ULT: Out << "ULT"; break; case FCmpInst::FCMP_OGT: Out << "OGT"; break; case FCmpInst::FCMP_UGT: Out << "UGT"; break; case FCmpInst::FCMP_OLE: Out << "OLE"; break; case FCmpInst::FCMP_ULE: Out << "ULE"; break; case FCmpInst::FCMP_OGE: Out << "OGE"; break; case FCmpInst::FCMP_UGE: Out << "UGE"; break; case FCmpInst::FCMP_TRUE: Out << "TRUE"; break; default: error("Invalid FCmp Predicate"); } break; case Instruction::Shl: Out << "getShl("; break; case Instruction::LShr: Out << "getLShr("; break; case Instruction::AShr: Out << "getAShr("; break; case Instruction::Select: Out << "getSelect("; break; case Instruction::ExtractElement: Out << "getExtractElement("; break; case Instruction::InsertElement: Out << "getInsertElement("; break; case Instruction::ShuffleVector: Out << "getShuffleVector("; break; default: error("Invalid constant expression"); break; } Out << getCppName(CE->getOperand(0)); for (unsigned i = 1; i < CE->getNumOperands(); ++i) Out << ", " << getCppName(CE->getOperand(i)); Out << ");"; } } else { error("Bad Constant"); Out << "Constant* " << constName << " = 0; "; } nl(Out); } void CppWriter::printConstants(const Module* M) { // Traverse all the global variables looking for constant initializers for (Module::const_global_iterator I = TheModule->global_begin(), E = TheModule->global_end(); I != E; ++I) if (I->hasInitializer()) printConstant(I->getInitializer()); // Traverse the LLVM functions looking for constants for (Module::const_iterator FI = TheModule->begin(), FE = TheModule->end(); FI != FE; ++FI) { // Add all of the basic blocks and instructions for (Function::const_iterator BB = FI->begin(), E = FI->end(); BB != E; ++BB) { for (BasicBlock::const_iterator I = BB->begin(), E = BB->end(); I!=E; ++I) { for (unsigned i = 0; i < I->getNumOperands(); ++i) { if (Constant* C = dyn_cast(I->getOperand(i))) { printConstant(C); } } } } } } void CppWriter::printVariableUses(const GlobalVariable *GV) { nl(Out) << "// Type Definitions"; nl(Out); printType(GV->getType()); if (GV->hasInitializer()) { Constant* Init = GV->getInitializer(); printType(Init->getType()); if (Function* F = dyn_cast(Init)) { nl(Out)<< "/ Function Declarations"; nl(Out); printFunctionHead(F); } else if (GlobalVariable* gv = dyn_cast(Init)) { nl(Out) << "// Global Variable Declarations"; nl(Out); printVariableHead(gv); } else { nl(Out) << "// Constant Definitions"; nl(Out); printConstant(gv); } if (GlobalVariable* gv = dyn_cast(Init)) { nl(Out) << "// Global Variable Definitions"; nl(Out); printVariableBody(gv); } } } void CppWriter::printVariableHead(const GlobalVariable *GV) { nl(Out) << "GlobalVariable* " << getCppName(GV); if (is_inline) { Out << " = mod->getGlobalVariable("; printEscapedString(GV->getName()); Out << ", " << getCppName(GV->getType()->getElementType()) << ",true)"; nl(Out) << "if (!" << getCppName(GV) << ") {"; in(); nl(Out) << getCppName(GV); } Out << " = new GlobalVariable("; nl(Out) << "/*Type=*/"; printCppName(GV->getType()->getElementType()); Out << ","; nl(Out) << "/*isConstant=*/" << (GV->isConstant()?"true":"false"); Out << ","; nl(Out) << "/*Linkage=*/"; printLinkageType(GV->getLinkage()); Out << ","; nl(Out) << "/*Initializer=*/0, "; if (GV->hasInitializer()) { Out << "// has initializer, specified below"; } nl(Out) << "/*Name=*/\""; printEscapedString(GV->getName()); Out << "\","; nl(Out) << "mod);"; nl(Out); if (GV->hasSection()) { printCppName(GV); Out << "->setSection(\""; printEscapedString(GV->getSection()); Out << "\");"; nl(Out); } if (GV->getAlignment()) { printCppName(GV); Out << "->setAlignment(" << utostr(GV->getAlignment()) << ");"; nl(Out); }; if (is_inline) { out(); Out << "}"; nl(Out); } } void CppWriter::printVariableBody(const GlobalVariable *GV) { if (GV->hasInitializer()) { printCppName(GV); Out << "->setInitializer("; //if (!isagetInitializer())) //else Out << getCppName(GV->getInitializer()) << ");"; nl(Out); } } std::string CppWriter::getOpName(Value* V) { if (!isa(V) || DefinedValues.find(V) != DefinedValues.end()) return getCppName(V); // See if its alread in the map of forward references, if so just return the // name we already set up for it ForwardRefMap::const_iterator I = ForwardRefs.find(V); if (I != ForwardRefs.end()) return I->second; // This is a new forward reference. Generate a unique name for it std::string result(std::string("fwdref_") + utostr(uniqueNum++)); // Yes, this is a hack. An Argument is the smallest instantiable value that // we can make as a placeholder for the real value. We'll replace these // Argument instances later. Out << "Argument* " << result << " = new Argument(" << getCppName(V->getType()) << ");"; nl(Out); ForwardRefs[V] = result; return result; } // printInstruction - This member is called for each Instruction in a function. void CppWriter::printInstruction(const Instruction *I, const std::string& bbname) { std::string iName(getCppName(I)); // Before we emit this instruction, we need to take care of generating any // forward references. So, we get the names of all the operands in advance std::string* opNames = new std::string[I->getNumOperands()]; for (unsigned i = 0; i < I->getNumOperands(); i++) { opNames[i] = getOpName(I->getOperand(i)); } switch (I->getOpcode()) { case Instruction::Ret: { const ReturnInst* ret = cast(I); Out << "new ReturnInst(" << (ret->getReturnValue() ? opNames[0] + ", " : "") << bbname << ");"; break; } case Instruction::Br: { const BranchInst* br = cast(I); Out << "new BranchInst(" ; if (br->getNumOperands() == 3 ) { Out << opNames[0] << ", " << opNames[1] << ", " << opNames[2] << ", "; } else if (br->getNumOperands() == 1) { Out << opNames[0] << ", "; } else { error("Branch with 2 operands?"); } Out << bbname << ");"; break; } case Instruction::Switch: { const SwitchInst* sw = cast(I); Out << "SwitchInst* " << iName << " = new SwitchInst(" << opNames[0] << ", " << opNames[1] << ", " << sw->getNumCases() << ", " << bbname << ");"; nl(Out); for (unsigned i = 2; i < sw->getNumOperands(); i += 2 ) { Out << iName << "->addCase(" << opNames[i] << ", " << opNames[i+1] << ");"; nl(Out); } break; } case Instruction::Invoke: { const InvokeInst* inv = cast(I); Out << "std::vector " << iName << "_params;"; nl(Out); for (unsigned i = 3; i < inv->getNumOperands(); ++i) { Out << iName << "_params.push_back(" << opNames[i] << ");"; nl(Out); } Out << "InvokeInst *" << iName << " = new InvokeInst(" << opNames[0] << ", " << opNames[1] << ", " << opNames[2] << ", " << "&" << iName << "_params[0], " << inv->getNumOperands() - 3 << ", \""; printEscapedString(inv->getName()); Out << "\", " << bbname << ");"; nl(Out) << iName << "->setCallingConv("; printCallingConv(inv->getCallingConv()); Out << ");"; break; } case Instruction::Unwind: { Out << "new UnwindInst(" << bbname << ");"; break; } case Instruction::Unreachable:{ Out << "new UnreachableInst(" << bbname << ");"; break; } case Instruction::Add: case Instruction::Sub: case Instruction::Mul: case Instruction::UDiv: case Instruction::SDiv: case Instruction::FDiv: case Instruction::URem: case Instruction::SRem: case Instruction::FRem: case Instruction::And: case Instruction::Or: case Instruction::Xor: case Instruction::Shl: case Instruction::LShr: case Instruction::AShr:{ Out << "BinaryOperator* " << iName << " = BinaryOperator::create("; switch (I->getOpcode()) { case Instruction::Add: Out << "Instruction::Add"; break; case Instruction::Sub: Out << "Instruction::Sub"; break; case Instruction::Mul: Out << "Instruction::Mul"; break; case Instruction::UDiv:Out << "Instruction::UDiv"; break; case Instruction::SDiv:Out << "Instruction::SDiv"; break; case Instruction::FDiv:Out << "Instruction::FDiv"; break; case Instruction::URem:Out << "Instruction::URem"; break; case Instruction::SRem:Out << "Instruction::SRem"; break; case Instruction::FRem:Out << "Instruction::FRem"; break; case Instruction::And: Out << "Instruction::And"; break; case Instruction::Or: Out << "Instruction::Or"; break; case Instruction::Xor: Out << "Instruction::Xor"; break; case Instruction::Shl: Out << "Instruction::Shl"; break; case Instruction::LShr:Out << "Instruction::LShr"; break; case Instruction::AShr:Out << "Instruction::AShr"; break; default: Out << "Instruction::BadOpCode"; break; } Out << ", " << opNames[0] << ", " << opNames[1] << ", \""; printEscapedString(I->getName()); Out << "\", " << bbname << ");"; break; } case Instruction::FCmp: { Out << "FCmpInst* " << iName << " = new FCmpInst("; switch (cast(I)->getPredicate()) { case FCmpInst::FCMP_FALSE: Out << "FCmpInst::FCMP_FALSE"; break; case FCmpInst::FCMP_OEQ : Out << "FCmpInst::FCMP_OEQ"; break; case FCmpInst::FCMP_OGT : Out << "FCmpInst::FCMP_OGT"; break; case FCmpInst::FCMP_OGE : Out << "FCmpInst::FCMP_OGE"; break; case FCmpInst::FCMP_OLT : Out << "FCmpInst::FCMP_OLT"; break; case FCmpInst::FCMP_OLE : Out << "FCmpInst::FCMP_OLE"; break; case FCmpInst::FCMP_ONE : Out << "FCmpInst::FCMP_ONE"; break; case FCmpInst::FCMP_ORD : Out << "FCmpInst::FCMP_ORD"; break; case FCmpInst::FCMP_UNO : Out << "FCmpInst::FCMP_UNO"; break; case FCmpInst::FCMP_UEQ : Out << "FCmpInst::FCMP_UEQ"; break; case FCmpInst::FCMP_UGT : Out << "FCmpInst::FCMP_UGT"; break; case FCmpInst::FCMP_UGE : Out << "FCmpInst::FCMP_UGE"; break; case FCmpInst::FCMP_ULT : Out << "FCmpInst::FCMP_ULT"; break; case FCmpInst::FCMP_ULE : Out << "FCmpInst::FCMP_ULE"; break; case FCmpInst::FCMP_UNE : Out << "FCmpInst::FCMP_UNE"; break; case FCmpInst::FCMP_TRUE : Out << "FCmpInst::FCMP_TRUE"; break; default: Out << "FCmpInst::BAD_ICMP_PREDICATE"; break; } Out << ", " << opNames[0] << ", " << opNames[1] << ", \""; printEscapedString(I->getName()); Out << "\", " << bbname << ");"; break; } case Instruction::ICmp: { Out << "ICmpInst* " << iName << " = new ICmpInst("; switch (cast(I)->getPredicate()) { case ICmpInst::ICMP_EQ: Out << "ICmpInst::ICMP_EQ"; break; case ICmpInst::ICMP_NE: Out << "ICmpInst::ICMP_NE"; break; case ICmpInst::ICMP_ULE: Out << "ICmpInst::ICMP_ULE"; break; case ICmpInst::ICMP_SLE: Out << "ICmpInst::ICMP_SLE"; break; case ICmpInst::ICMP_UGE: Out << "ICmpInst::ICMP_UGE"; break; case ICmpInst::ICMP_SGE: Out << "ICmpInst::ICMP_SGE"; break; case ICmpInst::ICMP_ULT: Out << "ICmpInst::ICMP_ULT"; break; case ICmpInst::ICMP_SLT: Out << "ICmpInst::ICMP_SLT"; break; case ICmpInst::ICMP_UGT: Out << "ICmpInst::ICMP_UGT"; break; case ICmpInst::ICMP_SGT: Out << "ICmpInst::ICMP_SGT"; break; default: Out << "ICmpInst::BAD_ICMP_PREDICATE"; break; } Out << ", " << opNames[0] << ", " << opNames[1] << ", \""; printEscapedString(I->getName()); Out << "\", " << bbname << ");"; break; } case Instruction::Malloc: { const MallocInst* mallocI = cast(I); Out << "MallocInst* " << iName << " = new MallocInst(" << getCppName(mallocI->getAllocatedType()) << ", "; if (mallocI->isArrayAllocation()) Out << opNames[0] << ", " ; Out << "\""; printEscapedString(mallocI->getName()); Out << "\", " << bbname << ");"; if (mallocI->getAlignment()) nl(Out) << iName << "->setAlignment(" << mallocI->getAlignment() << ");"; break; } case Instruction::Free: { Out << "FreeInst* " << iName << " = new FreeInst(" << getCppName(I->getOperand(0)) << ", " << bbname << ");"; break; } case Instruction::Alloca: { const AllocaInst* allocaI = cast(I); Out << "AllocaInst* " << iName << " = new AllocaInst(" << getCppName(allocaI->getAllocatedType()) << ", "; if (allocaI->isArrayAllocation()) Out << opNames[0] << ", "; Out << "\""; printEscapedString(allocaI->getName()); Out << "\", " << bbname << ");"; if (allocaI->getAlignment()) nl(Out) << iName << "->setAlignment(" << allocaI->getAlignment() << ");"; break; } case Instruction::Load:{ const LoadInst* load = cast(I); Out << "LoadInst* " << iName << " = new LoadInst(" << opNames[0] << ", \""; printEscapedString(load->getName()); Out << "\", " << (load->isVolatile() ? "true" : "false" ) << ", " << bbname << ");"; break; } case Instruction::Store: { const StoreInst* store = cast(I); Out << "StoreInst* " << iName << " = new StoreInst(" << opNames[0] << ", " << opNames[1] << ", " << (store->isVolatile() ? "true" : "false") << ", " << bbname << ");"; break; } case Instruction::GetElementPtr: { const GetElementPtrInst* gep = cast(I); if (gep->getNumOperands() <= 2) { Out << "GetElementPtrInst* " << iName << " = new GetElementPtrInst(" << opNames[0]; if (gep->getNumOperands() == 2) Out << ", " << opNames[1]; } else { Out << "std::vector " << iName << "_indices;"; nl(Out); for (unsigned i = 1; i < gep->getNumOperands(); ++i ) { Out << iName << "_indices.push_back(" << opNames[i] << ");"; nl(Out); } Out << "Instruction* " << iName << " = new GetElementPtrInst(" << opNames[0] << ", &" << iName << "_indices[0], " << gep->getNumOperands() - 1; } Out << ", \""; printEscapedString(gep->getName()); Out << "\", " << bbname << ");"; break; } case Instruction::PHI: { const PHINode* phi = cast(I); Out << "PHINode* " << iName << " = new PHINode(" << getCppName(phi->getType()) << ", \""; printEscapedString(phi->getName()); Out << "\", " << bbname << ");"; nl(Out) << iName << "->reserveOperandSpace(" << phi->getNumIncomingValues() << ");"; nl(Out); for (unsigned i = 0; i < phi->getNumOperands(); i+=2) { Out << iName << "->addIncoming(" << opNames[i] << ", " << opNames[i+1] << ");"; nl(Out); } break; } case Instruction::Trunc: case Instruction::ZExt: case Instruction::SExt: case Instruction::FPTrunc: case Instruction::FPExt: case Instruction::FPToUI: case Instruction::FPToSI: case Instruction::UIToFP: case Instruction::SIToFP: case Instruction::PtrToInt: case Instruction::IntToPtr: case Instruction::BitCast: { const CastInst* cst = cast(I); Out << "CastInst* " << iName << " = new "; switch (I->getOpcode()) { case Instruction::Trunc: Out << "TruncInst"; break; case Instruction::ZExt: Out << "ZExtInst"; break; case Instruction::SExt: Out << "SExtInst"; break; case Instruction::FPTrunc: Out << "FPTruncInst"; break; case Instruction::FPExt: Out << "FPExtInst"; break; case Instruction::FPToUI: Out << "FPToUIInst"; break; case Instruction::FPToSI: Out << "FPToSIInst"; break; case Instruction::UIToFP: Out << "UIToFPInst"; break; case Instruction::SIToFP: Out << "SIToFPInst"; break; case Instruction::PtrToInt: Out << "PtrToIntInst"; break; case Instruction::IntToPtr: Out << "IntToPtrInst"; break; case Instruction::BitCast: Out << "BitCastInst"; break; default: assert(!"Unreachable"); break; } Out << "(" << opNames[0] << ", " << getCppName(cst->getType()) << ", \""; printEscapedString(cst->getName()); Out << "\", " << bbname << ");"; break; } case Instruction::Call:{ const CallInst* call = cast(I); if (InlineAsm* ila = dyn_cast(call->getOperand(0))) { Out << "InlineAsm* " << getCppName(ila) << " = InlineAsm::get(" << getCppName(ila->getFunctionType()) << ", \"" << ila->getAsmString() << "\", \"" << ila->getConstraintString() << "\"," << (ila->hasSideEffects() ? "true" : "false") << ");"; nl(Out); } if (call->getNumOperands() > 3) { Out << "std::vector " << iName << "_params;"; nl(Out); for (unsigned i = 1; i < call->getNumOperands(); ++i) { Out << iName << "_params.push_back(" << opNames[i] << ");"; nl(Out); } Out << "CallInst* " << iName << " = new CallInst(" << opNames[0] << ", &" << iName << "_params[0], " << call->getNumOperands() - 1 << ", \""; } else if (call->getNumOperands() == 3) { Out << "CallInst* " << iName << " = new CallInst(" << opNames[0] << ", " << opNames[1] << ", " << opNames[2] << ", \""; } else if (call->getNumOperands() == 2) { Out << "CallInst* " << iName << " = new CallInst(" << opNames[0] << ", " << opNames[1] << ", \""; } else { Out << "CallInst* " << iName << " = new CallInst(" << opNames[0] << ", \""; } printEscapedString(call->getName()); Out << "\", " << bbname << ");"; nl(Out) << iName << "->setCallingConv("; printCallingConv(call->getCallingConv()); Out << ");"; nl(Out) << iName << "->setTailCall(" << (call->isTailCall() ? "true":"false"); Out << ");"; break; } case Instruction::Select: { const SelectInst* sel = cast(I); Out << "SelectInst* " << getCppName(sel) << " = new SelectInst("; Out << opNames[0] << ", " << opNames[1] << ", " << opNames[2] << ", \""; printEscapedString(sel->getName()); Out << "\", " << bbname << ");"; break; } case Instruction::UserOp1: /// FALL THROUGH case Instruction::UserOp2: { /// FIXME: What should be done here? break; } case Instruction::VAArg: { const VAArgInst* va = cast(I); Out << "VAArgInst* " << getCppName(va) << " = new VAArgInst(" << opNames[0] << ", " << getCppName(va->getType()) << ", \""; printEscapedString(va->getName()); Out << "\", " << bbname << ");"; break; } case Instruction::ExtractElement: { const ExtractElementInst* eei = cast(I); Out << "ExtractElementInst* " << getCppName(eei) << " = new ExtractElementInst(" << opNames[0] << ", " << opNames[1] << ", \""; printEscapedString(eei->getName()); Out << "\", " << bbname << ");"; break; } case Instruction::InsertElement: { const InsertElementInst* iei = cast(I); Out << "InsertElementInst* " << getCppName(iei) << " = new InsertElementInst(" << opNames[0] << ", " << opNames[1] << ", " << opNames[2] << ", \""; printEscapedString(iei->getName()); Out << "\", " << bbname << ");"; break; } case Instruction::ShuffleVector: { const ShuffleVectorInst* svi = cast(I); Out << "ShuffleVectorInst* " << getCppName(svi) << " = new ShuffleVectorInst(" << opNames[0] << ", " << opNames[1] << ", " << opNames[2] << ", \""; printEscapedString(svi->getName()); Out << "\", " << bbname << ");"; break; } } DefinedValues.insert(I); nl(Out); delete [] opNames; } // Print out the types, constants and declarations needed by one function void CppWriter::printFunctionUses(const Function* F) { nl(Out) << "// Type Definitions"; nl(Out); if (!is_inline) { // Print the function's return type printType(F->getReturnType()); // Print the function's function type printType(F->getFunctionType()); // Print the types of each of the function's arguments for(Function::const_arg_iterator AI = F->arg_begin(), AE = F->arg_end(); AI != AE; ++AI) { printType(AI->getType()); } } // Print type definitions for every type referenced by an instruction and // make a note of any global values or constants that are referenced SmallPtrSet gvs; SmallPtrSet consts; for (Function::const_iterator BB = F->begin(), BE = F->end(); BB != BE; ++BB){ for (BasicBlock::const_iterator I = BB->begin(), E = BB->end(); I != E; ++I) { // Print the type of the instruction itself printType(I->getType()); // Print the type of each of the instruction's operands for (unsigned i = 0; i < I->getNumOperands(); ++i) { Value* operand = I->getOperand(i); printType(operand->getType()); // If the operand references a GVal or Constant, make a note of it if (GlobalValue* GV = dyn_cast(operand)) { gvs.insert(GV); if (GlobalVariable *GVar = dyn_cast(GV)) if (GVar->hasInitializer()) consts.insert(GVar->getInitializer()); } else if (Constant* C = dyn_cast(operand)) consts.insert(C); } } } // Print the function declarations for any functions encountered nl(Out) << "// Function Declarations"; nl(Out); for (SmallPtrSet::iterator I = gvs.begin(), E = gvs.end(); I != E; ++I) { if (Function* Fun = dyn_cast(*I)) { if (!is_inline || Fun != F) printFunctionHead(Fun); } } // Print the global variable declarations for any variables encountered nl(Out) << "// Global Variable Declarations"; nl(Out); for (SmallPtrSet::iterator I = gvs.begin(), E = gvs.end(); I != E; ++I) { if (GlobalVariable* F = dyn_cast(*I)) printVariableHead(F); } // Print the constants found nl(Out) << "// Constant Definitions"; nl(Out); for (SmallPtrSet::iterator I = consts.begin(), E = consts.end(); I != E; ++I) { printConstant(*I); } // Process the global variables definitions now that all the constants have // been emitted. These definitions just couple the gvars with their constant // initializers. nl(Out) << "// Global Variable Definitions"; nl(Out); for (SmallPtrSet::iterator I = gvs.begin(), E = gvs.end(); I != E; ++I) { if (GlobalVariable* GV = dyn_cast(*I)) printVariableBody(GV); } } void CppWriter::printFunctionHead(const Function* F) { nl(Out) << "Function* " << getCppName(F); if (is_inline) { Out << " = mod->getFunction(\""; printEscapedString(F->getName()); Out << "\", " << getCppName(F->getFunctionType()) << ");"; nl(Out) << "if (!" << getCppName(F) << ") {"; nl(Out) << getCppName(F); } Out<< " = new Function("; nl(Out,1) << "/*Type=*/" << getCppName(F->getFunctionType()) << ","; nl(Out) << "/*Linkage=*/"; printLinkageType(F->getLinkage()); Out << ","; nl(Out) << "/*Name=*/\""; printEscapedString(F->getName()); Out << "\", mod); " << (F->isDeclaration()? "// (external, no body)" : ""); nl(Out,-1); printCppName(F); Out << "->setCallingConv("; printCallingConv(F->getCallingConv()); Out << ");"; nl(Out); if (F->hasSection()) { printCppName(F); Out << "->setSection(\"" << F->getSection() << "\");"; nl(Out); } if (F->getAlignment()) { printCppName(F); Out << "->setAlignment(" << F->getAlignment() << ");"; nl(Out); } if (is_inline) { Out << "}"; nl(Out); } } void CppWriter::printFunctionBody(const Function *F) { if (F->isDeclaration()) return; // external functions have no bodies. // Clear the DefinedValues and ForwardRefs maps because we can't have // cross-function forward refs ForwardRefs.clear(); DefinedValues.clear(); // Create all the argument values if (!is_inline) { if (!F->arg_empty()) { Out << "Function::arg_iterator args = " << getCppName(F) << "->arg_begin();"; nl(Out); } for (Function::const_arg_iterator AI = F->arg_begin(), AE = F->arg_end(); AI != AE; ++AI) { Out << "Value* " << getCppName(AI) << " = args++;"; nl(Out); if (AI->hasName()) { Out << getCppName(AI) << "->setName(\"" << AI->getName() << "\");"; nl(Out); } } } // Create all the basic blocks nl(Out); for (Function::const_iterator BI = F->begin(), BE = F->end(); BI != BE; ++BI) { std::string bbname(getCppName(BI)); Out << "BasicBlock* " << bbname << " = new BasicBlock(\""; if (BI->hasName()) printEscapedString(BI->getName()); Out << "\"," << getCppName(BI->getParent()) << ",0);"; nl(Out); } // Output all of its basic blocks... for the function for (Function::const_iterator BI = F->begin(), BE = F->end(); BI != BE; ++BI) { std::string bbname(getCppName(BI)); nl(Out) << "// Block " << BI->getName() << " (" << bbname << ")"; nl(Out); // Output all of the instructions in the basic block... for (BasicBlock::const_iterator I = BI->begin(), E = BI->end(); I != E; ++I) { printInstruction(I,bbname); } } // Loop over the ForwardRefs and resolve them now that all instructions // are generated. if (!ForwardRefs.empty()) { nl(Out) << "// Resolve Forward References"; nl(Out); } while (!ForwardRefs.empty()) { ForwardRefMap::iterator I = ForwardRefs.begin(); Out << I->second << "->replaceAllUsesWith(" << getCppName(I->first) << "); delete " << I->second << ";"; nl(Out); ForwardRefs.erase(I); } } void CppWriter::printInline(const std::string& fname, const std::string& func) { const Function* F = TheModule->getFunction(func); if (!F) { error(std::string("Function '") + func + "' not found in input module"); return; } if (F->isDeclaration()) { error(std::string("Function '") + func + "' is external!"); return; } nl(Out) << "BasicBlock* " << fname << "(Module* mod, Function *" << getCppName(F); unsigned arg_count = 1; for (Function::const_arg_iterator AI = F->arg_begin(), AE = F->arg_end(); AI != AE; ++AI) { Out << ", Value* arg_" << arg_count; } Out << ") {"; nl(Out); is_inline = true; printFunctionUses(F); printFunctionBody(F); is_inline = false; Out << "return " << getCppName(F->begin()) << ";"; nl(Out) << "}"; nl(Out); } void CppWriter::printModuleBody() { // Print out all the type definitions nl(Out) << "// Type Definitions"; nl(Out); printTypes(TheModule); // Functions can call each other and global variables can reference them so // define all the functions first before emitting their function bodies. nl(Out) << "// Function Declarations"; nl(Out); for (Module::const_iterator I = TheModule->begin(), E = TheModule->end(); I != E; ++I) printFunctionHead(I); // Process the global variables declarations. We can't initialze them until // after the constants are printed so just print a header for each global nl(Out) << "// Global Variable Declarations\n"; nl(Out); for (Module::const_global_iterator I = TheModule->global_begin(), E = TheModule->global_end(); I != E; ++I) { printVariableHead(I); } // Print out all the constants definitions. Constants don't recurse except // through GlobalValues. All GlobalValues have been declared at this point // so we can proceed to generate the constants. nl(Out) << "// Constant Definitions"; nl(Out); printConstants(TheModule); // Process the global variables definitions now that all the constants have // been emitted. These definitions just couple the gvars with their constant // initializers. nl(Out) << "// Global Variable Definitions"; nl(Out); for (Module::const_global_iterator I = TheModule->global_begin(), E = TheModule->global_end(); I != E; ++I) { printVariableBody(I); } // Finally, we can safely put out all of the function bodies. nl(Out) << "// Function Definitions"; nl(Out); for (Module::const_iterator I = TheModule->begin(), E = TheModule->end(); I != E; ++I) { if (!I->isDeclaration()) { nl(Out) << "// Function: " << I->getName() << " (" << getCppName(I) << ")"; nl(Out) << "{"; nl(Out,1); printFunctionBody(I); nl(Out,-1) << "}"; nl(Out); } } } void CppWriter::printProgram( const std::string& fname, const std::string& mName ) { Out << "#include \n"; Out << "#include \n"; Out << "#include \n"; Out << "#include \n"; Out << "#include \n"; Out << "#include \n"; Out << "#include \n"; Out << "#include \n"; Out << "#include \n"; Out << "#include \n"; Out << "#include \n"; Out << "#include \n"; Out << "#include \n"; Out << "#include \n"; Out << "#include \n"; Out << "#include \n"; Out << "#include \n\n"; Out << "using namespace llvm;\n\n"; Out << "Module* " << fname << "();\n\n"; Out << "int main(int argc, char**argv) {\n"; Out << " Module* Mod = " << fname << "();\n"; Out << " verifyModule(*Mod, PrintMessageAction);\n"; Out << " std::cerr.flush();\n"; Out << " std::cout.flush();\n"; Out << " PassManager PM;\n"; Out << " PM.add(new PrintModulePass(&llvm::cout));\n"; Out << " PM.run(*Mod);\n"; Out << " return 0;\n"; Out << "}\n\n"; printModule(fname,mName); } void CppWriter::printModule( const std::string& fname, const std::string& mName ) { nl(Out) << "Module* " << fname << "() {"; nl(Out,1) << "// Module Construction"; nl(Out) << "Module* mod = new Module(\"" << mName << "\");"; if (!TheModule->getTargetTriple().empty()) { nl(Out) << "mod->setDataLayout(\"" << TheModule->getDataLayout() << "\");"; } if (!TheModule->getTargetTriple().empty()) { nl(Out) << "mod->setTargetTriple(\"" << TheModule->getTargetTriple() << "\");"; } if (!TheModule->getModuleInlineAsm().empty()) { nl(Out) << "mod->setModuleInlineAsm(\""; printEscapedString(TheModule->getModuleInlineAsm()); Out << "\");"; } nl(Out); // Loop over the dependent libraries and emit them. Module::lib_iterator LI = TheModule->lib_begin(); Module::lib_iterator LE = TheModule->lib_end(); while (LI != LE) { Out << "mod->addLibrary(\"" << *LI << "\");"; nl(Out); ++LI; } printModuleBody(); nl(Out) << "return mod;"; nl(Out,-1) << "}"; nl(Out); } void CppWriter::printContents( const std::string& fname, // Name of generated function const std::string& mName // Name of module generated module ) { Out << "\nModule* " << fname << "(Module *mod) {\n"; Out << "\nmod->setModuleIdentifier(\"" << mName << "\");\n"; printModuleBody(); Out << "\nreturn mod;\n"; Out << "\n}\n"; } void CppWriter::printFunction( const std::string& fname, // Name of generated function const std::string& funcName // Name of function to generate ) { const Function* F = TheModule->getFunction(funcName); if (!F) { error(std::string("Function '") + funcName + "' not found in input module"); return; } Out << "\nFunction* " << fname << "(Module *mod) {\n"; printFunctionUses(F); printFunctionHead(F); printFunctionBody(F); Out << "return " << getCppName(F) << ";\n"; Out << "}\n"; } void CppWriter::printVariable( const std::string& fname, /// Name of generated function const std::string& varName // Name of variable to generate ) { const GlobalVariable* GV = TheModule->getNamedGlobal(varName); if (!GV) { error(std::string("Variable '") + varName + "' not found in input module"); return; } Out << "\nGlobalVariable* " << fname << "(Module *mod) {\n"; printVariableUses(GV); printVariableHead(GV); printVariableBody(GV); Out << "return " << getCppName(GV) << ";\n"; Out << "}\n"; } void CppWriter::printType( const std::string& fname, /// Name of generated function const std::string& typeName // Name of type to generate ) { const Type* Ty = TheModule->getTypeByName(typeName); if (!Ty) { error(std::string("Type '") + typeName + "' not found in input module"); return; } Out << "\nType* " << fname << "(Module *mod) {\n"; printType(Ty); Out << "return " << getCppName(Ty) << ";\n"; Out << "}\n"; } } // end anonymous llvm namespace llvm { void WriteModuleToCppFile(Module* mod, std::ostream& o) { // Initialize a CppWriter for us to use CppWriter W(o, mod); // Emit a header o << "// Generated by llvm2cpp - DO NOT MODIFY!\n\n"; // Get the name of the function we're supposed to generate std::string fname = FuncName.getValue(); // Get the name of the thing we are to generate std::string tgtname = NameToGenerate.getValue(); if (GenerationType == GenModule || GenerationType == GenContents || GenerationType == GenProgram) { if (tgtname == "!bad!") { if (mod->getModuleIdentifier() == "-") tgtname = ""; else tgtname = mod->getModuleIdentifier(); } } else if (tgtname == "!bad!") { W.error("You must use the -for option with -gen-{function,variable,type}"); } switch (WhatToGenerate(GenerationType)) { case GenProgram: if (fname.empty()) fname = "makeLLVMModule"; W.printProgram(fname,tgtname); break; case GenModule: if (fname.empty()) fname = "makeLLVMModule"; W.printModule(fname,tgtname); break; case GenContents: if (fname.empty()) fname = "makeLLVMModuleContents"; W.printContents(fname,tgtname); break; case GenFunction: if (fname.empty()) fname = "makeLLVMFunction"; W.printFunction(fname,tgtname); break; case GenInline: if (fname.empty()) fname = "makeLLVMInline"; W.printInline(fname,tgtname); break; case GenVariable: if (fname.empty()) fname = "makeLLVMVariable"; W.printVariable(fname,tgtname); break; case GenType: if (fname.empty()) fname = "makeLLVMType"; W.printType(fname,tgtname); break; default: W.error("Invalid generation option"); } } }