diff options
Diffstat (limited to 'contrib/llvm/lib/Bitcode')
-rw-r--r-- | contrib/llvm/lib/Bitcode/Reader/BitcodeReader.cpp | 680 | ||||
-rw-r--r-- | contrib/llvm/lib/Bitcode/Reader/BitcodeReader.h | 40 | ||||
-rw-r--r-- | contrib/llvm/lib/Bitcode/Writer/BitcodeWriter.cpp | 318 | ||||
-rw-r--r-- | contrib/llvm/lib/Bitcode/Writer/ValueEnumerator.cpp | 44 | ||||
-rw-r--r-- | contrib/llvm/lib/Bitcode/Writer/ValueEnumerator.h | 4 |
5 files changed, 612 insertions, 474 deletions
diff --git a/contrib/llvm/lib/Bitcode/Reader/BitcodeReader.cpp b/contrib/llvm/lib/Bitcode/Reader/BitcodeReader.cpp index 46565f3..e399040 100644 --- a/contrib/llvm/lib/Bitcode/Reader/BitcodeReader.cpp +++ b/contrib/llvm/lib/Bitcode/Reader/BitcodeReader.cpp @@ -22,11 +22,19 @@ #include "llvm/AutoUpgrade.h" #include "llvm/ADT/SmallString.h" #include "llvm/ADT/SmallVector.h" +#include "llvm/Support/DataStream.h" #include "llvm/Support/MathExtras.h" #include "llvm/Support/MemoryBuffer.h" #include "llvm/OperandTraits.h" using namespace llvm; +void BitcodeReader::materializeForwardReferencedFunctions() { + while (!BlockAddrFwdRefs.empty()) { + Function *F = BlockAddrFwdRefs.begin()->first; + F->Materialize(); + } +} + void BitcodeReader::FreeState() { if (BufferOwned) delete Buffer; @@ -394,7 +402,7 @@ Type *BitcodeReader::getTypeByID(unsigned ID) { // The type table size is always specified correctly. if (ID >= TypeList.size()) return 0; - + if (Type *Ty = TypeList[ID]) return Ty; @@ -403,14 +411,6 @@ Type *BitcodeReader::getTypeByID(unsigned ID) { return TypeList[ID] = StructType::create(Context); } -/// FIXME: Remove in LLVM 3.1, only used by ParseOldTypeTable. -Type *BitcodeReader::getTypeByIDOrNull(unsigned ID) { - if (ID >= TypeList.size()) - TypeList.resize(ID+1); - - return TypeList[ID]; -} - //===----------------------------------------------------------------------===// // Functions for parsing blocks from the bitcode file @@ -462,8 +462,8 @@ bool BitcodeReader::ParseAttributeBlock() { // If Function attributes are using index 0 then transfer them // to index ~0. Index 0 is used for return value attributes but used to be // used for function attributes. - Attributes RetAttribute = Attribute::None; - Attributes FnAttribute = Attribute::None; + Attributes RetAttribute; + Attributes FnAttribute; for (unsigned i = 0, e = Record.size(); i != e; i += 2) { // FIXME: remove in LLVM 3.0 // The alignment is stored as a 16-bit raw value from bits 31--16. @@ -473,23 +473,24 @@ bool BitcodeReader::ParseAttributeBlock() { if (Alignment && !isPowerOf2_32(Alignment)) return Error("Alignment is not a power of two."); - Attributes ReconstitutedAttr = Record[i+1] & 0xffff; + Attributes ReconstitutedAttr(Record[i+1] & 0xffff); if (Alignment) ReconstitutedAttr |= Attribute::constructAlignmentFromInt(Alignment); - ReconstitutedAttr |= (Record[i+1] & (0xffffull << 32)) >> 11; - Record[i+1] = ReconstitutedAttr; + ReconstitutedAttr |= + Attributes((Record[i+1] & (0xffffull << 32)) >> 11); + Record[i+1] = ReconstitutedAttr.Raw(); if (Record[i] == 0) - RetAttribute = Record[i+1]; + RetAttribute = ReconstitutedAttr; else if (Record[i] == ~0U) - FnAttribute = Record[i+1]; + FnAttribute = ReconstitutedAttr; } - unsigned OldRetAttrs = (Attribute::NoUnwind|Attribute::NoReturn| + Attributes OldRetAttrs = (Attribute::NoUnwind|Attribute::NoReturn| Attribute::ReadOnly|Attribute::ReadNone); if (FnAttribute == Attribute::None && RetAttribute != Attribute::None && - (RetAttribute & OldRetAttrs) != 0) { + (RetAttribute & OldRetAttrs)) { if (FnAttribute == Attribute::None) { // add a slot so they get added. Record.push_back(~0U); Record.push_back(0); @@ -506,8 +507,9 @@ bool BitcodeReader::ParseAttributeBlock() { } else if (Record[i] == ~0U) { if (FnAttribute != Attribute::None) Attrs.push_back(AttributeWithIndex::get(~0U, FnAttribute)); - } else if (Record[i+1] != Attribute::None) - Attrs.push_back(AttributeWithIndex::get(Record[i], Record[i+1])); + } else if (Attributes(Record[i+1]) != Attribute::None) + Attrs.push_back(AttributeWithIndex::get(Record[i], + Attributes(Record[i+1]))); } MAttributes.push_back(AttrListPtr::get(Attrs.begin(), Attrs.end())); @@ -521,7 +523,7 @@ bool BitcodeReader::ParseAttributeBlock() { bool BitcodeReader::ParseTypeTable() { if (Stream.EnterSubBlock(bitc::TYPE_BLOCK_ID_NEW)) return Error("Malformed block record"); - + return ParseTypeTableBody(); } @@ -533,7 +535,7 @@ bool BitcodeReader::ParseTypeTableBody() { unsigned NumRecords = 0; SmallString<64> TypeName; - + // Read all the records for this type table. while (1) { unsigned Code = Stream.ReadCode(); @@ -573,6 +575,9 @@ bool BitcodeReader::ParseTypeTableBody() { case bitc::TYPE_CODE_VOID: // VOID ResultTy = Type::getVoidTy(Context); break; + case bitc::TYPE_CODE_HALF: // HALF + ResultTy = Type::getHalfTy(Context); + break; case bitc::TYPE_CODE_FLOAT: // FLOAT ResultTy = Type::getFloatTy(Context); break; @@ -615,12 +620,12 @@ bool BitcodeReader::ParseTypeTableBody() { ResultTy = PointerType::get(ResultTy, AddressSpace); break; } - case bitc::TYPE_CODE_FUNCTION: { + case bitc::TYPE_CODE_FUNCTION_OLD: { // FIXME: attrid is dead, remove it in LLVM 3.0 // FUNCTION: [vararg, attrid, retty, paramty x N] if (Record.size() < 3) return Error("Invalid FUNCTION type record"); - std::vector<Type*> ArgTys; + SmallVector<Type*, 8> ArgTys; for (unsigned i = 3, e = Record.size(); i != e; ++i) { if (Type *T = getTypeByID(Record[i])) ArgTys.push_back(T); @@ -635,10 +640,29 @@ bool BitcodeReader::ParseTypeTableBody() { ResultTy = FunctionType::get(ResultTy, ArgTys, Record[0]); break; } + case bitc::TYPE_CODE_FUNCTION: { + // FUNCTION: [vararg, retty, paramty x N] + if (Record.size() < 2) + return Error("Invalid FUNCTION type record"); + SmallVector<Type*, 8> ArgTys; + for (unsigned i = 2, e = Record.size(); i != e; ++i) { + if (Type *T = getTypeByID(Record[i])) + ArgTys.push_back(T); + else + break; + } + + ResultTy = getTypeByID(Record[1]); + if (ResultTy == 0 || ArgTys.size() < Record.size()-2) + return Error("invalid type in function type"); + + ResultTy = FunctionType::get(ResultTy, ArgTys, Record[0]); + break; + } case bitc::TYPE_CODE_STRUCT_ANON: { // STRUCT: [ispacked, eltty x N] if (Record.size() < 1) return Error("Invalid STRUCT type record"); - std::vector<Type*> EltTys; + SmallVector<Type*, 8> EltTys; for (unsigned i = 1, e = Record.size(); i != e; ++i) { if (Type *T = getTypeByID(Record[i])) EltTys.push_back(T); @@ -728,247 +752,6 @@ bool BitcodeReader::ParseTypeTableBody() { } } -// FIXME: Remove in LLVM 3.1 -bool BitcodeReader::ParseOldTypeTable() { - if (Stream.EnterSubBlock(bitc::TYPE_BLOCK_ID_OLD)) - return Error("Malformed block record"); - - if (!TypeList.empty()) - return Error("Multiple TYPE_BLOCKs found!"); - - - // While horrible, we have no good ordering of types in the bc file. Just - // iteratively parse types out of the bc file in multiple passes until we get - // them all. Do this by saving a cursor for the start of the type block. - BitstreamCursor StartOfTypeBlockCursor(Stream); - - unsigned NumTypesRead = 0; - - SmallVector<uint64_t, 64> Record; -RestartScan: - unsigned NextTypeID = 0; - bool ReadAnyTypes = false; - - // Read all the records for this type table. - while (1) { - unsigned Code = Stream.ReadCode(); - if (Code == bitc::END_BLOCK) { - if (NextTypeID != TypeList.size()) - return Error("Invalid type forward reference in TYPE_BLOCK_ID_OLD"); - - // If we haven't read all of the types yet, iterate again. - if (NumTypesRead != TypeList.size()) { - // If we didn't successfully read any types in this pass, then we must - // have an unhandled forward reference. - if (!ReadAnyTypes) - return Error("Obsolete bitcode contains unhandled recursive type"); - - Stream = StartOfTypeBlockCursor; - goto RestartScan; - } - - if (Stream.ReadBlockEnd()) - return Error("Error at end of type table block"); - return false; - } - - if (Code == bitc::ENTER_SUBBLOCK) { - // No known subblocks, always skip them. - Stream.ReadSubBlockID(); - if (Stream.SkipBlock()) - return Error("Malformed block record"); - continue; - } - - if (Code == bitc::DEFINE_ABBREV) { - Stream.ReadAbbrevRecord(); - continue; - } - - // Read a record. - Record.clear(); - Type *ResultTy = 0; - switch (Stream.ReadRecord(Code, Record)) { - default: return Error("unknown type in type table"); - case bitc::TYPE_CODE_NUMENTRY: // TYPE_CODE_NUMENTRY: [numentries] - // TYPE_CODE_NUMENTRY contains a count of the number of types in the - // type list. This allows us to reserve space. - if (Record.size() < 1) - return Error("Invalid TYPE_CODE_NUMENTRY record"); - TypeList.resize(Record[0]); - continue; - case bitc::TYPE_CODE_VOID: // VOID - ResultTy = Type::getVoidTy(Context); - break; - case bitc::TYPE_CODE_FLOAT: // FLOAT - ResultTy = Type::getFloatTy(Context); - break; - case bitc::TYPE_CODE_DOUBLE: // DOUBLE - ResultTy = Type::getDoubleTy(Context); - break; - case bitc::TYPE_CODE_X86_FP80: // X86_FP80 - ResultTy = Type::getX86_FP80Ty(Context); - break; - case bitc::TYPE_CODE_FP128: // FP128 - ResultTy = Type::getFP128Ty(Context); - break; - case bitc::TYPE_CODE_PPC_FP128: // PPC_FP128 - ResultTy = Type::getPPC_FP128Ty(Context); - break; - case bitc::TYPE_CODE_LABEL: // LABEL - ResultTy = Type::getLabelTy(Context); - break; - case bitc::TYPE_CODE_METADATA: // METADATA - ResultTy = Type::getMetadataTy(Context); - break; - case bitc::TYPE_CODE_X86_MMX: // X86_MMX - ResultTy = Type::getX86_MMXTy(Context); - break; - case bitc::TYPE_CODE_INTEGER: // INTEGER: [width] - if (Record.size() < 1) - return Error("Invalid Integer type record"); - ResultTy = IntegerType::get(Context, Record[0]); - break; - case bitc::TYPE_CODE_OPAQUE: // OPAQUE - if (NextTypeID < TypeList.size() && TypeList[NextTypeID] == 0) - ResultTy = StructType::create(Context); - break; - case bitc::TYPE_CODE_STRUCT_OLD: {// STRUCT_OLD - if (NextTypeID >= TypeList.size()) break; - // If we already read it, don't reprocess. - if (TypeList[NextTypeID] && - !cast<StructType>(TypeList[NextTypeID])->isOpaque()) - break; - - // Set a type. - if (TypeList[NextTypeID] == 0) - TypeList[NextTypeID] = StructType::create(Context); - - std::vector<Type*> EltTys; - for (unsigned i = 1, e = Record.size(); i != e; ++i) { - if (Type *Elt = getTypeByIDOrNull(Record[i])) - EltTys.push_back(Elt); - else - break; - } - - if (EltTys.size() != Record.size()-1) - break; // Not all elements are ready. - - cast<StructType>(TypeList[NextTypeID])->setBody(EltTys, Record[0]); - ResultTy = TypeList[NextTypeID]; - TypeList[NextTypeID] = 0; - break; - } - case bitc::TYPE_CODE_POINTER: { // POINTER: [pointee type] or - // [pointee type, address space] - if (Record.size() < 1) - return Error("Invalid POINTER type record"); - unsigned AddressSpace = 0; - if (Record.size() == 2) - AddressSpace = Record[1]; - if ((ResultTy = getTypeByIDOrNull(Record[0]))) - ResultTy = PointerType::get(ResultTy, AddressSpace); - break; - } - case bitc::TYPE_CODE_FUNCTION: { - // FIXME: attrid is dead, remove it in LLVM 3.0 - // FUNCTION: [vararg, attrid, retty, paramty x N] - if (Record.size() < 3) - return Error("Invalid FUNCTION type record"); - std::vector<Type*> ArgTys; - for (unsigned i = 3, e = Record.size(); i != e; ++i) { - if (Type *Elt = getTypeByIDOrNull(Record[i])) - ArgTys.push_back(Elt); - else - break; - } - if (ArgTys.size()+3 != Record.size()) - break; // Something was null. - if ((ResultTy = getTypeByIDOrNull(Record[2]))) - ResultTy = FunctionType::get(ResultTy, ArgTys, Record[0]); - break; - } - case bitc::TYPE_CODE_ARRAY: // ARRAY: [numelts, eltty] - if (Record.size() < 2) - return Error("Invalid ARRAY type record"); - if ((ResultTy = getTypeByIDOrNull(Record[1]))) - ResultTy = ArrayType::get(ResultTy, Record[0]); - break; - case bitc::TYPE_CODE_VECTOR: // VECTOR: [numelts, eltty] - if (Record.size() < 2) - return Error("Invalid VECTOR type record"); - if ((ResultTy = getTypeByIDOrNull(Record[1]))) - ResultTy = VectorType::get(ResultTy, Record[0]); - break; - } - - if (NextTypeID >= TypeList.size()) - return Error("invalid TYPE table"); - - if (ResultTy && TypeList[NextTypeID] == 0) { - ++NumTypesRead; - ReadAnyTypes = true; - - TypeList[NextTypeID] = ResultTy; - } - - ++NextTypeID; - } -} - - -bool BitcodeReader::ParseOldTypeSymbolTable() { - if (Stream.EnterSubBlock(bitc::TYPE_SYMTAB_BLOCK_ID_OLD)) - return Error("Malformed block record"); - - SmallVector<uint64_t, 64> Record; - - // Read all the records for this type table. - std::string TypeName; - while (1) { - unsigned Code = Stream.ReadCode(); - if (Code == bitc::END_BLOCK) { - if (Stream.ReadBlockEnd()) - return Error("Error at end of type symbol table block"); - return false; - } - - if (Code == bitc::ENTER_SUBBLOCK) { - // No known subblocks, always skip them. - Stream.ReadSubBlockID(); - if (Stream.SkipBlock()) - return Error("Malformed block record"); - continue; - } - - if (Code == bitc::DEFINE_ABBREV) { - Stream.ReadAbbrevRecord(); - continue; - } - - // Read a record. - Record.clear(); - switch (Stream.ReadRecord(Code, Record)) { - default: // Default behavior: unknown type. - break; - case bitc::TST_CODE_ENTRY: // TST_ENTRY: [typeid, namechar x N] - if (ConvertToString(Record, 1, TypeName)) - return Error("Invalid TST_ENTRY record"); - unsigned TypeID = Record[0]; - if (TypeID >= TypeList.size()) - return Error("Invalid Type ID in TST_ENTRY record"); - - // Only apply the type name to a struct type with no name. - if (StructType *STy = dyn_cast<StructType>(TypeList[TypeID])) - if (!STy->isLiteral() && !STy->hasName()) - STy->setName(TypeName); - TypeName.clear(); - break; - } - } -} - bool BitcodeReader::ParseValueSymbolTable() { if (Stream.EnterSubBlock(bitc::VALUE_SYMTAB_BLOCK_ID)) return Error("Malformed block record"); @@ -1262,7 +1045,9 @@ bool BitcodeReader::ParseConstants() { case bitc::CST_CODE_FLOAT: { // FLOAT: [fpval] if (Record.empty()) return Error("Invalid FLOAT record"); - if (CurTy->isFloatTy()) + if (CurTy->isHalfTy()) + V = ConstantFP::get(Context, APFloat(APInt(16, (uint16_t)Record[0]))); + else if (CurTy->isFloatTy()) V = ConstantFP::get(Context, APFloat(APInt(32, (uint32_t)Record[0]))); else if (CurTy->isDoubleTy()) V = ConstantFP::get(Context, APFloat(APInt(64, Record[0]))); @@ -1286,7 +1071,7 @@ bool BitcodeReader::ParseConstants() { return Error("Invalid CST_AGGREGATE record"); unsigned Size = Record.size(); - std::vector<Constant*> Elts; + SmallVector<Constant*, 16> Elts; if (StructType *STy = dyn_cast<StructType>(CurTy)) { for (unsigned i = 0; i != Size; ++i) @@ -1308,35 +1093,78 @@ bool BitcodeReader::ParseConstants() { } break; } - case bitc::CST_CODE_STRING: { // STRING: [values] + case bitc::CST_CODE_STRING: // STRING: [values] + case bitc::CST_CODE_CSTRING: { // CSTRING: [values] if (Record.empty()) - return Error("Invalid CST_AGGREGATE record"); - - ArrayType *ATy = cast<ArrayType>(CurTy); - Type *EltTy = ATy->getElementType(); + return Error("Invalid CST_STRING record"); unsigned Size = Record.size(); - std::vector<Constant*> Elts; + SmallString<16> Elts; for (unsigned i = 0; i != Size; ++i) - Elts.push_back(ConstantInt::get(EltTy, Record[i])); - V = ConstantArray::get(ATy, Elts); + Elts.push_back(Record[i]); + V = ConstantDataArray::getString(Context, Elts, + BitCode == bitc::CST_CODE_CSTRING); break; } - case bitc::CST_CODE_CSTRING: { // CSTRING: [values] + case bitc::CST_CODE_DATA: {// DATA: [n x value] if (Record.empty()) - return Error("Invalid CST_AGGREGATE record"); - - ArrayType *ATy = cast<ArrayType>(CurTy); - Type *EltTy = ATy->getElementType(); - + return Error("Invalid CST_DATA record"); + + Type *EltTy = cast<SequentialType>(CurTy)->getElementType(); unsigned Size = Record.size(); - std::vector<Constant*> Elts; - for (unsigned i = 0; i != Size; ++i) - Elts.push_back(ConstantInt::get(EltTy, Record[i])); - Elts.push_back(Constant::getNullValue(EltTy)); - V = ConstantArray::get(ATy, Elts); + + if (EltTy->isIntegerTy(8)) { + SmallVector<uint8_t, 16> Elts(Record.begin(), Record.end()); + if (isa<VectorType>(CurTy)) + V = ConstantDataVector::get(Context, Elts); + else + V = ConstantDataArray::get(Context, Elts); + } else if (EltTy->isIntegerTy(16)) { + SmallVector<uint16_t, 16> Elts(Record.begin(), Record.end()); + if (isa<VectorType>(CurTy)) + V = ConstantDataVector::get(Context, Elts); + else + V = ConstantDataArray::get(Context, Elts); + } else if (EltTy->isIntegerTy(32)) { + SmallVector<uint32_t, 16> Elts(Record.begin(), Record.end()); + if (isa<VectorType>(CurTy)) + V = ConstantDataVector::get(Context, Elts); + else + V = ConstantDataArray::get(Context, Elts); + } else if (EltTy->isIntegerTy(64)) { + SmallVector<uint64_t, 16> Elts(Record.begin(), Record.end()); + if (isa<VectorType>(CurTy)) + V = ConstantDataVector::get(Context, Elts); + else + V = ConstantDataArray::get(Context, Elts); + } else if (EltTy->isFloatTy()) { + SmallVector<float, 16> Elts; + for (unsigned i = 0; i != Size; ++i) { + union { uint32_t I; float F; }; + I = Record[i]; + Elts.push_back(F); + } + if (isa<VectorType>(CurTy)) + V = ConstantDataVector::get(Context, Elts); + else + V = ConstantDataArray::get(Context, Elts); + } else if (EltTy->isDoubleTy()) { + SmallVector<double, 16> Elts; + for (unsigned i = 0; i != Size; ++i) { + union { uint64_t I; double F; }; + I = Record[i]; + Elts.push_back(F); + } + if (isa<VectorType>(CurTy)) + V = ConstantDataVector::get(Context, Elts); + else + V = ConstantDataArray::get(Context, Elts); + } else { + return Error("Unknown element type in CE_DATA"); + } break; } + case bitc::CST_CODE_CE_BINOP: { // CE_BINOP: [opcode, opval, opval] if (Record.size() < 3) return Error("Invalid CE_BINOP record"); int Opc = GetDecodedBinaryOpcode(Record[0], CurTy); @@ -1517,6 +1345,50 @@ bool BitcodeReader::ParseConstants() { return false; } +bool BitcodeReader::ParseUseLists() { + if (Stream.EnterSubBlock(bitc::USELIST_BLOCK_ID)) + return Error("Malformed block record"); + + SmallVector<uint64_t, 64> Record; + + // Read all the records. + while (1) { + unsigned Code = Stream.ReadCode(); + if (Code == bitc::END_BLOCK) { + if (Stream.ReadBlockEnd()) + return Error("Error at end of use-list table block"); + return false; + } + + if (Code == bitc::ENTER_SUBBLOCK) { + // No known subblocks, always skip them. + Stream.ReadSubBlockID(); + if (Stream.SkipBlock()) + return Error("Malformed block record"); + continue; + } + + if (Code == bitc::DEFINE_ABBREV) { + Stream.ReadAbbrevRecord(); + continue; + } + + // Read a use list record. + Record.clear(); + switch (Stream.ReadRecord(Code, Record)) { + default: // Default behavior: unknown type. + break; + case bitc::USELIST_CODE_ENTRY: { // USELIST_CODE_ENTRY: TBD. + unsigned RecordLength = Record.size(); + if (RecordLength < 1) + return Error ("Invalid UseList reader!"); + UseListRecords.push_back(Record); + break; + } + } + } +} + /// RememberAndSkipFunctionBody - When we see the block for a function body, /// remember where it is and then skip it. This lets us lazily deserialize the /// functions. @@ -1538,8 +1410,36 @@ bool BitcodeReader::RememberAndSkipFunctionBody() { return false; } -bool BitcodeReader::ParseModule() { - if (Stream.EnterSubBlock(bitc::MODULE_BLOCK_ID)) +bool BitcodeReader::GlobalCleanup() { + // Patch the initializers for globals and aliases up. + ResolveGlobalAndAliasInits(); + if (!GlobalInits.empty() || !AliasInits.empty()) + return Error("Malformed global initializer set"); + + // Look for intrinsic functions which need to be upgraded at some point + for (Module::iterator FI = TheModule->begin(), FE = TheModule->end(); + FI != FE; ++FI) { + Function *NewFn; + if (UpgradeIntrinsicFunction(FI, NewFn)) + UpgradedIntrinsics.push_back(std::make_pair(FI, NewFn)); + } + + // Look for global variables which need to be renamed. + for (Module::global_iterator + GI = TheModule->global_begin(), GE = TheModule->global_end(); + GI != GE; ++GI) + UpgradeGlobalVariable(GI); + // Force deallocation of memory for these vectors to favor the client that + // want lazy deserialization. + std::vector<std::pair<GlobalVariable*, unsigned> >().swap(GlobalInits); + std::vector<std::pair<GlobalAlias*, unsigned> >().swap(AliasInits); + return false; +} + +bool BitcodeReader::ParseModule(bool Resume) { + if (Resume) + Stream.JumpToBit(NextUnreadBit); + else if (Stream.EnterSubBlock(bitc::MODULE_BLOCK_ID)) return Error("Malformed block record"); SmallVector<uint64_t, 64> Record; @@ -1553,33 +1453,7 @@ bool BitcodeReader::ParseModule() { if (Stream.ReadBlockEnd()) return Error("Error at end of module block"); - // Patch the initializers for globals and aliases up. - ResolveGlobalAndAliasInits(); - if (!GlobalInits.empty() || !AliasInits.empty()) - return Error("Malformed global initializer set"); - if (!FunctionsWithBodies.empty()) - return Error("Too few function bodies found"); - - // Look for intrinsic functions which need to be upgraded at some point - for (Module::iterator FI = TheModule->begin(), FE = TheModule->end(); - FI != FE; ++FI) { - Function* NewFn; - if (UpgradeIntrinsicFunction(FI, NewFn)) - UpgradedIntrinsics.push_back(std::make_pair(FI, NewFn)); - } - - // Look for global variables which need to be renamed. - for (Module::global_iterator - GI = TheModule->global_begin(), GE = TheModule->global_end(); - GI != GE; ++GI) - UpgradeGlobalVariable(GI); - - // Force deallocation of memory for these vectors to favor the client that - // want lazy deserialization. - std::vector<std::pair<GlobalVariable*, unsigned> >().swap(GlobalInits); - std::vector<std::pair<GlobalAlias*, unsigned> >().swap(AliasInits); - std::vector<Function*>().swap(FunctionsWithBodies); - return false; + return GlobalCleanup(); } if (Code == bitc::ENTER_SUBBLOCK) { @@ -1600,17 +1474,10 @@ bool BitcodeReader::ParseModule() { if (ParseTypeTable()) return true; break; - case bitc::TYPE_BLOCK_ID_OLD: - if (ParseOldTypeTable()) - return true; - break; - case bitc::TYPE_SYMTAB_BLOCK_ID_OLD: - if (ParseOldTypeSymbolTable()) - return true; - break; case bitc::VALUE_SYMTAB_BLOCK_ID: if (ParseValueSymbolTable()) return true; + SeenValueSymbolTable = true; break; case bitc::CONSTANTS_BLOCK_ID: if (ParseConstants() || ResolveGlobalAndAliasInits()) @@ -1623,13 +1490,29 @@ bool BitcodeReader::ParseModule() { case bitc::FUNCTION_BLOCK_ID: // If this is the first function body we've seen, reverse the // FunctionsWithBodies list. - if (!HasReversedFunctionsWithBodies) { + if (!SeenFirstFunctionBody) { std::reverse(FunctionsWithBodies.begin(), FunctionsWithBodies.end()); - HasReversedFunctionsWithBodies = true; + if (GlobalCleanup()) + return true; + SeenFirstFunctionBody = true; } if (RememberAndSkipFunctionBody()) return true; + // For streaming bitcode, suspend parsing when we reach the function + // bodies. Subsequent materialization calls will resume it when + // necessary. For streaming, the function bodies must be at the end of + // the bitcode. If the bitcode file is old, the symbol table will be + // at the end instead and will not have been seen yet. In this case, + // just finish the parse now. + if (LazyStreamer && SeenValueSymbolTable) { + NextUnreadBit = Stream.GetCurrentBitNo(); + return false; + } + break; + case bitc::USELIST_BLOCK_ID: + if (ParseUseLists()) + return true; break; } continue; @@ -1784,8 +1667,10 @@ bool BitcodeReader::ParseModule() { // If this is a function with a body, remember the prototype we are // creating now, so that we can match up the body with them later. - if (!isProto) + if (!isProto) { FunctionsWithBodies.push_back(Func); + if (LazyStreamer) DeferredFunctionInfo[Func] = 0; + } break; } // ALIAS: [alias type, aliasee val#, linkage] @@ -1824,24 +1709,7 @@ bool BitcodeReader::ParseModule() { bool BitcodeReader::ParseBitcodeInto(Module *M) { TheModule = 0; - unsigned char *BufPtr = (unsigned char *)Buffer->getBufferStart(); - unsigned char *BufEnd = BufPtr+Buffer->getBufferSize(); - - if (Buffer->getBufferSize() & 3) { - if (!isRawBitcode(BufPtr, BufEnd) && !isBitcodeWrapper(BufPtr, BufEnd)) - return Error("Invalid bitcode signature"); - else - return Error("Bitcode stream should be a multiple of 4 bytes in length"); - } - - // If we have a wrapper header, parse it and ignore the non-bc file contents. - // The magic number is 0x0B17C0DE stored in little endian. - if (isBitcodeWrapper(BufPtr, BufEnd)) - if (SkipBitcodeWrapperHeader(BufPtr, BufEnd)) - return Error("Invalid bitcode wrapper header"); - - StreamFile.init(BufPtr, BufEnd); - Stream.init(StreamFile); + if (InitStream()) return true; // Sniff for the signature. if (Stream.Read(8) != 'B' || @@ -1883,8 +1751,9 @@ bool BitcodeReader::ParseBitcodeInto(Module *M) { if (TheModule) return Error("Multiple MODULE_BLOCKs in same stream"); TheModule = M; - if (ParseModule()) + if (ParseModule(false)) return true; + if (LazyStreamer) return false; break; default: if (Stream.SkipBlock()) @@ -1952,20 +1821,7 @@ bool BitcodeReader::ParseModuleTriple(std::string &Triple) { } bool BitcodeReader::ParseTriple(std::string &Triple) { - if (Buffer->getBufferSize() & 3) - return Error("Bitcode stream should be a multiple of 4 bytes in length"); - - unsigned char *BufPtr = (unsigned char *)Buffer->getBufferStart(); - unsigned char *BufEnd = BufPtr+Buffer->getBufferSize(); - - // If we have a wrapper header, parse it and ignore the non-bc file contents. - // The magic number is 0x0B17C0DE stored in little endian. - if (isBitcodeWrapper(BufPtr, BufEnd)) - if (SkipBitcodeWrapperHeader(BufPtr, BufEnd)) - return Error("Invalid bitcode wrapper header"); - - StreamFile.init(BufPtr, BufEnd); - Stream.init(StreamFile); + if (InitStream()) return true; // Sniff for the signature. if (Stream.Read(8) != 'B' || @@ -2517,10 +2373,6 @@ bool BitcodeReader::ParseFunctionBody(Function *F) { InstructionList.push_back(I); break; } - case bitc::FUNC_CODE_INST_UNWIND: // UNWIND - I = new UnwindInst(Context); - InstructionList.push_back(I); - break; case bitc::FUNC_CODE_INST_UNREACHABLE: // UNREACHABLE I = new UnreachableInst(Context); InstructionList.push_back(I); @@ -2845,6 +2697,19 @@ bool BitcodeReader::ParseFunctionBody(Function *F) { return false; } +/// FindFunctionInStream - Find the function body in the bitcode stream +bool BitcodeReader::FindFunctionInStream(Function *F, + DenseMap<Function*, uint64_t>::iterator DeferredFunctionInfoIterator) { + while (DeferredFunctionInfoIterator->second == 0) { + if (Stream.AtEndOfStream()) + return Error("Could not find Function in stream"); + // ParseModule will parse the next body in the stream and set its + // position in the DeferredFunctionInfo map. + if (ParseModule(true)) return true; + } + return false; +} + //===----------------------------------------------------------------------===// // GVMaterializer implementation //===----------------------------------------------------------------------===// @@ -2865,6 +2730,10 @@ bool BitcodeReader::Materialize(GlobalValue *GV, std::string *ErrInfo) { DenseMap<Function*, uint64_t>::iterator DFII = DeferredFunctionInfo.find(F); assert(DFII != DeferredFunctionInfo.end() && "Deferred function not found!"); + // If its position is recorded as 0, its body is somewhere in the stream + // but we haven't seen it yet. + if (DFII->second == 0) + if (LazyStreamer && FindFunctionInStream(F, DFII)) return true; // Move the bit stream to the saved position of the deferred function body. Stream.JumpToBit(DFII->second); @@ -2920,6 +2789,12 @@ bool BitcodeReader::MaterializeModule(Module *M, std::string *ErrInfo) { Materialize(F, ErrInfo)) return true; + // At this point, if there are any function bodies, the current bit is + // pointing to the END_BLOCK record after them. Now make sure the rest + // of the bits in the module have been read. + if (NextUnreadBit) + ParseModule(true); + // Upgrade any intrinsic calls that slipped through (should not happen!) and // delete the old functions to clean up. We can't do this unless the entire // module is materialized because there could always be another function body @@ -2939,15 +2814,60 @@ bool BitcodeReader::MaterializeModule(Module *M, std::string *ErrInfo) { } std::vector<std::pair<Function*, Function*> >().swap(UpgradedIntrinsics); - // Upgrade to new EH scheme. N.B. This will go away in 3.1. - UpgradeExceptionHandling(M); + return false; +} + +bool BitcodeReader::InitStream() { + if (LazyStreamer) return InitLazyStream(); + return InitStreamFromBuffer(); +} + +bool BitcodeReader::InitStreamFromBuffer() { + const unsigned char *BufPtr = (unsigned char *)Buffer->getBufferStart(); + const unsigned char *BufEnd = BufPtr+Buffer->getBufferSize(); + + if (Buffer->getBufferSize() & 3) { + if (!isRawBitcode(BufPtr, BufEnd) && !isBitcodeWrapper(BufPtr, BufEnd)) + return Error("Invalid bitcode signature"); + else + return Error("Bitcode stream should be a multiple of 4 bytes in length"); + } + + // If we have a wrapper header, parse it and ignore the non-bc file contents. + // The magic number is 0x0B17C0DE stored in little endian. + if (isBitcodeWrapper(BufPtr, BufEnd)) + if (SkipBitcodeWrapperHeader(BufPtr, BufEnd, true)) + return Error("Invalid bitcode wrapper header"); - // Check debug info intrinsics. - CheckDebugInfoIntrinsics(TheModule); + StreamFile.reset(new BitstreamReader(BufPtr, BufEnd)); + Stream.init(*StreamFile); return false; } +bool BitcodeReader::InitLazyStream() { + // Check and strip off the bitcode wrapper; BitstreamReader expects never to + // see it. + StreamingMemoryObject *Bytes = new StreamingMemoryObject(LazyStreamer); + StreamFile.reset(new BitstreamReader(Bytes)); + Stream.init(*StreamFile); + + unsigned char buf[16]; + if (Bytes->readBytes(0, 16, buf, NULL) == -1) + return Error("Bitcode stream must be at least 16 bytes in length"); + + if (!isBitcode(buf, buf + 16)) + return Error("Invalid bitcode signature"); + + if (isBitcodeWrapper(buf, buf + 4)) { + const unsigned char *bitcodeStart = buf; + const unsigned char *bitcodeEnd = buf + 16; + SkipBitcodeWrapperHeader(bitcodeStart, bitcodeEnd, false); + Bytes->dropLeadingBytes(bitcodeStart - buf); + Bytes->setKnownObjectSize(bitcodeEnd - bitcodeStart); + } + return false; +} //===----------------------------------------------------------------------===// // External interface @@ -2970,6 +2890,27 @@ Module *llvm::getLazyBitcodeModule(MemoryBuffer *Buffer, } // Have the BitcodeReader dtor delete 'Buffer'. R->setBufferOwned(true); + + R->materializeForwardReferencedFunctions(); + + return M; +} + + +Module *llvm::getStreamedBitcodeModule(const std::string &name, + DataStreamer *streamer, + LLVMContext &Context, + std::string *ErrMsg) { + Module *M = new Module(name, Context); + BitcodeReader *R = new BitcodeReader(streamer, Context); + M->setMaterializer(R); + if (R->ParseBitcodeInto(M)) { + if (ErrMsg) + *ErrMsg = R->getErrorString(); + delete M; // Also deletes R. + return 0; + } + R->setBufferOwned(false); // no buffer to delete return M; } @@ -2990,6 +2931,9 @@ Module *llvm::ParseBitcodeFile(MemoryBuffer *Buffer, LLVMContext& Context, return 0; } + // TODO: Restore the use-lists to the in-memory state when the bitcode was + // written. We must defer until the Module has been fully materialized. + return M; } diff --git a/contrib/llvm/lib/Bitcode/Reader/BitcodeReader.h b/contrib/llvm/lib/Bitcode/Reader/BitcodeReader.h index 6e6118c..e7c4e94 100644 --- a/contrib/llvm/lib/Bitcode/Reader/BitcodeReader.h +++ b/contrib/llvm/lib/Bitcode/Reader/BitcodeReader.h @@ -126,8 +126,11 @@ class BitcodeReader : public GVMaterializer { Module *TheModule; MemoryBuffer *Buffer; bool BufferOwned; - BitstreamReader StreamFile; + OwningPtr<BitstreamReader> StreamFile; BitstreamCursor Stream; + DataStreamer *LazyStreamer; + uint64_t NextUnreadBit; + bool SeenValueSymbolTable; const char *ErrorString; @@ -135,6 +138,7 @@ class BitcodeReader : public GVMaterializer { BitcodeReaderValueList ValueList; BitcodeReaderMDValueList MDValueList; SmallVector<Instruction *, 64> InstructionList; + SmallVector<SmallVector<uint64_t, 64>, 64> UseListRecords; std::vector<std::pair<GlobalVariable*, unsigned> > GlobalInits; std::vector<std::pair<GlobalAlias*, unsigned> > AliasInits; @@ -160,9 +164,10 @@ class BitcodeReader : public GVMaterializer { // Map the bitcode's custom MDKind ID to the Module's MDKind ID. DenseMap<unsigned, unsigned> MDKindMap; - // After the module header has been read, the FunctionsWithBodies list is - // reversed. This keeps track of whether we've done this yet. - bool HasReversedFunctionsWithBodies; + // Several operations happen after the module header has been read, but + // before function bodies are processed. This keeps track of whether + // we've done this yet. + bool SeenFirstFunctionBody; /// DeferredFunctionInfo - When function bodies are initially scanned, this /// map contains info about where to find deferred function body in the @@ -177,13 +182,22 @@ class BitcodeReader : public GVMaterializer { public: explicit BitcodeReader(MemoryBuffer *buffer, LLVMContext &C) : Context(C), TheModule(0), Buffer(buffer), BufferOwned(false), - ErrorString(0), ValueList(C), MDValueList(C) { - HasReversedFunctionsWithBodies = false; + LazyStreamer(0), NextUnreadBit(0), SeenValueSymbolTable(false), + ErrorString(0), ValueList(C), MDValueList(C), + SeenFirstFunctionBody(false) { + } + explicit BitcodeReader(DataStreamer *streamer, LLVMContext &C) + : Context(C), TheModule(0), Buffer(0), BufferOwned(false), + LazyStreamer(streamer), NextUnreadBit(0), SeenValueSymbolTable(false), + ErrorString(0), ValueList(C), MDValueList(C), + SeenFirstFunctionBody(false) { } ~BitcodeReader() { FreeState(); } - + + void materializeForwardReferencedFunctions(); + void FreeState(); /// setBufferOwned - If this is true, the reader will destroy the MemoryBuffer @@ -211,7 +225,6 @@ public: bool ParseTriple(std::string &Triple); private: Type *getTypeByID(unsigned ID); - Type *getTypeByIDOrNull(unsigned ID); Value *getFnValueByID(unsigned ID, Type *Ty) { if (Ty && Ty->isMetadataTy()) return MDValueList.getValueFwdRef(ID); @@ -256,21 +269,26 @@ private: } - bool ParseModule(); + bool ParseModule(bool Resume); bool ParseAttributeBlock(); bool ParseTypeTable(); - bool ParseOldTypeTable(); // FIXME: Remove in LLVM 3.1 bool ParseTypeTableBody(); - bool ParseOldTypeSymbolTable(); // FIXME: Remove in LLVM 3.1 bool ParseValueSymbolTable(); bool ParseConstants(); bool RememberAndSkipFunctionBody(); bool ParseFunctionBody(Function *F); + bool GlobalCleanup(); bool ResolveGlobalAndAliasInits(); bool ParseMetadata(); bool ParseMetadataAttachment(); bool ParseModuleTriple(std::string &Triple); + bool ParseUseLists(); + bool InitStream(); + bool InitStreamFromBuffer(); + bool InitLazyStream(); + bool FindFunctionInStream(Function *F, + DenseMap<Function*, uint64_t>::iterator DeferredFunctionInfoIterator); }; } // End llvm namespace diff --git a/contrib/llvm/lib/Bitcode/Writer/BitcodeWriter.cpp b/contrib/llvm/lib/Bitcode/Writer/BitcodeWriter.cpp index 5b3d969..b25d2e9 100644 --- a/contrib/llvm/lib/Bitcode/Writer/BitcodeWriter.cpp +++ b/contrib/llvm/lib/Bitcode/Writer/BitcodeWriter.cpp @@ -23,6 +23,7 @@ #include "llvm/Operator.h" #include "llvm/ValueSymbolTable.h" #include "llvm/ADT/Triple.h" +#include "llvm/Support/CommandLine.h" #include "llvm/Support/ErrorHandling.h" #include "llvm/Support/MathExtras.h" #include "llvm/Support/raw_ostream.h" @@ -31,6 +32,12 @@ #include <map> using namespace llvm; +static cl::opt<bool> +EnablePreserveUseListOrdering("enable-bc-uselist-preserve", + cl::desc("Turn on experimental support for " + "use-list order preservation."), + cl::init(false), cl::Hidden); + /// These are manifest constants used by the bitcode writer. They do not need to /// be kept in sync with the reader, but need to be consistent within this file. enum { @@ -119,7 +126,6 @@ static unsigned GetEncodedRMWOperation(AtomicRMWInst::BinOp Op) { static unsigned GetEncodedOrdering(AtomicOrdering Ordering) { switch (Ordering) { - default: llvm_unreachable("Unknown atomic ordering"); case NotAtomic: return bitc::ORDERING_NOTATOMIC; case Unordered: return bitc::ORDERING_UNORDERED; case Monotonic: return bitc::ORDERING_MONOTONIC; @@ -128,14 +134,15 @@ static unsigned GetEncodedOrdering(AtomicOrdering Ordering) { case AcquireRelease: return bitc::ORDERING_ACQREL; case SequentiallyConsistent: return bitc::ORDERING_SEQCST; } + llvm_unreachable("Invalid ordering"); } static unsigned GetEncodedSynchScope(SynchronizationScope SynchScope) { switch (SynchScope) { - default: llvm_unreachable("Unknown synchronization scope"); case SingleThread: return bitc::SYNCHSCOPE_SINGLETHREAD; case CrossThread: return bitc::SYNCHSCOPE_CROSSTHREAD; } + llvm_unreachable("Invalid synch scope"); } static void WriteStringRecord(unsigned Code, StringRef Str, @@ -172,10 +179,11 @@ static void WriteAttributeTable(const ValueEnumerator &VE, // Store the alignment in the bitcode as a 16-bit raw value instead of a // 5-bit log2 encoded value. Shift the bits above the alignment up by // 11 bits. - uint64_t FauxAttr = PAWI.Attrs & 0xffff; + uint64_t FauxAttr = PAWI.Attrs.Raw() & 0xffff; if (PAWI.Attrs & Attribute::Alignment) - FauxAttr |= (1ull<<16)<<(((PAWI.Attrs & Attribute::Alignment)-1) >> 16); - FauxAttr |= (PAWI.Attrs & (0x3FFull << 21)) << 11; + FauxAttr |= (1ull<<16)<< + (((PAWI.Attrs & Attribute::Alignment).Raw()-1) >> 16); + FauxAttr |= (PAWI.Attrs.Raw() & (0x3FFull << 21)) << 11; Record.push_back(FauxAttr); } @@ -194,11 +202,12 @@ static void WriteTypeTable(const ValueEnumerator &VE, BitstreamWriter &Stream) { Stream.EnterSubblock(bitc::TYPE_BLOCK_ID_NEW, 4 /*count from # abbrevs */); SmallVector<uint64_t, 64> TypeVals; + uint64_t NumBits = Log2_32_Ceil(VE.getTypes().size()+1); + // Abbrev for TYPE_CODE_POINTER. BitCodeAbbrev *Abbv = new BitCodeAbbrev(); Abbv->Add(BitCodeAbbrevOp(bitc::TYPE_CODE_POINTER)); - Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, - Log2_32_Ceil(VE.getTypes().size()+1))); + Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, NumBits)); Abbv->Add(BitCodeAbbrevOp(0)); // Addrspace = 0 unsigned PtrAbbrev = Stream.EmitAbbrev(Abbv); @@ -206,10 +215,9 @@ static void WriteTypeTable(const ValueEnumerator &VE, BitstreamWriter &Stream) { Abbv = new BitCodeAbbrev(); Abbv->Add(BitCodeAbbrevOp(bitc::TYPE_CODE_FUNCTION)); Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 1)); // isvararg - Abbv->Add(BitCodeAbbrevOp(0)); // FIXME: DEAD value, remove in LLVM 3.0 Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Array)); - Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, - Log2_32_Ceil(VE.getTypes().size()+1))); + Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, NumBits)); + unsigned FunctionAbbrev = Stream.EmitAbbrev(Abbv); // Abbrev for TYPE_CODE_STRUCT_ANON. @@ -217,8 +225,8 @@ static void WriteTypeTable(const ValueEnumerator &VE, BitstreamWriter &Stream) { Abbv->Add(BitCodeAbbrevOp(bitc::TYPE_CODE_STRUCT_ANON)); Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 1)); // ispacked Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Array)); - Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, - Log2_32_Ceil(VE.getTypes().size()+1))); + Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, NumBits)); + unsigned StructAnonAbbrev = Stream.EmitAbbrev(Abbv); // Abbrev for TYPE_CODE_STRUCT_NAME. @@ -233,16 +241,16 @@ static void WriteTypeTable(const ValueEnumerator &VE, BitstreamWriter &Stream) { Abbv->Add(BitCodeAbbrevOp(bitc::TYPE_CODE_STRUCT_NAMED)); Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 1)); // ispacked Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Array)); - Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, - Log2_32_Ceil(VE.getTypes().size()+1))); + Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, NumBits)); + unsigned StructNamedAbbrev = Stream.EmitAbbrev(Abbv); // Abbrev for TYPE_CODE_ARRAY. Abbv = new BitCodeAbbrev(); Abbv->Add(BitCodeAbbrevOp(bitc::TYPE_CODE_ARRAY)); Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 8)); // size - Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, - Log2_32_Ceil(VE.getTypes().size()+1))); + Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, NumBits)); + unsigned ArrayAbbrev = Stream.EmitAbbrev(Abbv); // Emit an entry count so the reader can reserve space. @@ -259,6 +267,7 @@ static void WriteTypeTable(const ValueEnumerator &VE, BitstreamWriter &Stream) { switch (T->getTypeID()) { default: llvm_unreachable("Unknown type!"); case Type::VoidTyID: Code = bitc::TYPE_CODE_VOID; break; + case Type::HalfTyID: Code = bitc::TYPE_CODE_HALF; break; case Type::FloatTyID: Code = bitc::TYPE_CODE_FLOAT; break; case Type::DoubleTyID: Code = bitc::TYPE_CODE_DOUBLE; break; case Type::X86_FP80TyID: Code = bitc::TYPE_CODE_X86_FP80; break; @@ -284,10 +293,9 @@ static void WriteTypeTable(const ValueEnumerator &VE, BitstreamWriter &Stream) { } case Type::FunctionTyID: { FunctionType *FT = cast<FunctionType>(T); - // FUNCTION: [isvararg, attrid, retty, paramty x N] + // FUNCTION: [isvararg, retty, paramty x N] Code = bitc::TYPE_CODE_FUNCTION; TypeVals.push_back(FT->isVarArg()); - TypeVals.push_back(0); // FIXME: DEAD: remove in llvm 3.0 TypeVals.push_back(VE.getTypeID(FT->getReturnType())); for (unsigned i = 0, e = FT->getNumParams(); i != e; ++i) TypeVals.push_back(VE.getTypeID(FT->getParamType(i))); @@ -350,7 +358,6 @@ static void WriteTypeTable(const ValueEnumerator &VE, BitstreamWriter &Stream) { static unsigned getEncodedLinkage(const GlobalValue *GV) { switch (GV->getLinkage()) { - default: llvm_unreachable("Invalid linkage!"); case GlobalValue::ExternalLinkage: return 0; case GlobalValue::WeakAnyLinkage: return 1; case GlobalValue::AppendingLinkage: return 2; @@ -368,15 +375,16 @@ static unsigned getEncodedLinkage(const GlobalValue *GV) { case GlobalValue::LinkerPrivateWeakLinkage: return 14; case GlobalValue::LinkerPrivateWeakDefAutoLinkage: return 15; } + llvm_unreachable("Invalid linkage"); } static unsigned getEncodedVisibility(const GlobalValue *GV) { switch (GV->getVisibility()) { - default: llvm_unreachable("Invalid visibility!"); case GlobalValue::DefaultVisibility: return 0; case GlobalValue::HiddenVisibility: return 1; case GlobalValue::ProtectedVisibility: return 2; } + llvm_unreachable("Invalid visibility"); } // Emit top-level description of module, including target triple, inline asm, @@ -499,8 +507,8 @@ static void WriteModuleInfo(const Module *M, const ValueEnumerator &VE, // Emit the function proto information. for (Module::const_iterator F = M->begin(), E = M->end(); F != E; ++F) { - // FUNCTION: [type, callingconv, isproto, paramattr, - // linkage, alignment, section, visibility, gc, unnamed_addr] + // FUNCTION: [type, callingconv, isproto, linkage, paramattrs, alignment, + // section, visibility, gc, unnamed_addr] Vals.push_back(VE.getTypeID(F->getType())); Vals.push_back(F->getCallingConv()); Vals.push_back(F->isDeclaration()); @@ -520,6 +528,7 @@ static void WriteModuleInfo(const Module *M, const ValueEnumerator &VE, // Emit the alias information. for (Module::const_alias_iterator AI = M->alias_begin(), E = M->alias_end(); AI != E; ++AI) { + // ALIAS: [alias type, aliasee val#, linkage, visibility] Vals.push_back(VE.getTypeID(AI->getType())); Vals.push_back(VE.getValueID(AI->getAliasee())); Vals.push_back(getEncodedLinkage(AI)); @@ -819,7 +828,7 @@ static void WriteConstants(unsigned FirstVal, unsigned LastVal, } else if (const ConstantFP *CFP = dyn_cast<ConstantFP>(C)) { Code = bitc::CST_CODE_FLOAT; Type *Ty = CFP->getType(); - if (Ty->isFloatTy() || Ty->isDoubleTy()) { + if (Ty->isHalfTy() || Ty->isFloatTy() || Ty->isDoubleTy()) { Record.push_back(CFP->getValueAPF().bitcastToAPInt().getZExtValue()); } else if (Ty->isX86_FP80Ty()) { // api needed to prevent premature destruction @@ -836,34 +845,56 @@ static void WriteConstants(unsigned FirstVal, unsigned LastVal, } else { assert (0 && "Unknown FP type!"); } - } else if (isa<ConstantArray>(C) && cast<ConstantArray>(C)->isString()) { - const ConstantArray *CA = cast<ConstantArray>(C); + } else if (isa<ConstantDataSequential>(C) && + cast<ConstantDataSequential>(C)->isString()) { + const ConstantDataSequential *Str = cast<ConstantDataSequential>(C); // Emit constant strings specially. - unsigned NumOps = CA->getNumOperands(); + unsigned NumElts = Str->getNumElements(); // If this is a null-terminated string, use the denser CSTRING encoding. - if (CA->getOperand(NumOps-1)->isNullValue()) { + if (Str->isCString()) { Code = bitc::CST_CODE_CSTRING; - --NumOps; // Don't encode the null, which isn't allowed by char6. + --NumElts; // Don't encode the null, which isn't allowed by char6. } else { Code = bitc::CST_CODE_STRING; AbbrevToUse = String8Abbrev; } bool isCStr7 = Code == bitc::CST_CODE_CSTRING; bool isCStrChar6 = Code == bitc::CST_CODE_CSTRING; - for (unsigned i = 0; i != NumOps; ++i) { - unsigned char V = cast<ConstantInt>(CA->getOperand(i))->getZExtValue(); + for (unsigned i = 0; i != NumElts; ++i) { + unsigned char V = Str->getElementAsInteger(i); Record.push_back(V); isCStr7 &= (V & 128) == 0; if (isCStrChar6) isCStrChar6 = BitCodeAbbrevOp::isChar6(V); } - + if (isCStrChar6) AbbrevToUse = CString6Abbrev; else if (isCStr7) AbbrevToUse = CString7Abbrev; - } else if (isa<ConstantArray>(C) || isa<ConstantStruct>(V) || - isa<ConstantVector>(V)) { + } else if (const ConstantDataSequential *CDS = + dyn_cast<ConstantDataSequential>(C)) { + Code = bitc::CST_CODE_DATA; + Type *EltTy = CDS->getType()->getElementType(); + if (isa<IntegerType>(EltTy)) { + for (unsigned i = 0, e = CDS->getNumElements(); i != e; ++i) + Record.push_back(CDS->getElementAsInteger(i)); + } else if (EltTy->isFloatTy()) { + for (unsigned i = 0, e = CDS->getNumElements(); i != e; ++i) { + union { float F; uint32_t I; }; + F = CDS->getElementAsFloat(i); + Record.push_back(I); + } + } else { + assert(EltTy->isDoubleTy() && "Unknown ConstantData element type"); + for (unsigned i = 0, e = CDS->getNumElements(); i != e; ++i) { + union { double F; uint64_t I; }; + F = CDS->getElementAsDouble(i); + Record.push_back(I); + } + } + } else if (isa<ConstantArray>(C) || isa<ConstantStruct>(C) || + isa<ConstantVector>(C)) { Code = bitc::CST_CODE_AGGREGATE; for (unsigned i = 0, e = C->getNumOperands(); i != e; ++i) Record.push_back(VE.getValueID(C->getOperand(i))); @@ -1105,10 +1136,18 @@ static void WriteInstruction(const Instruction &I, unsigned InstID, } break; case Instruction::Switch: - Code = bitc::FUNC_CODE_INST_SWITCH; - Vals.push_back(VE.getTypeID(I.getOperand(0)->getType())); - for (unsigned i = 0, e = I.getNumOperands(); i != e; ++i) - Vals.push_back(VE.getValueID(I.getOperand(i))); + { + Code = bitc::FUNC_CODE_INST_SWITCH; + SwitchInst &SI = cast<SwitchInst>(I); + Vals.push_back(VE.getTypeID(SI.getCondition()->getType())); + Vals.push_back(VE.getValueID(SI.getCondition())); + Vals.push_back(VE.getValueID(SI.getDefaultDest())); + for (SwitchInst::CaseIt i = SI.case_begin(), e = SI.case_end(); + i != e; ++i) { + Vals.push_back(VE.getValueID(i.getCaseValue())); + Vals.push_back(VE.getValueID(i.getCaseSuccessor())); + } + } break; case Instruction::IndirectBr: Code = bitc::FUNC_CODE_INST_INDIRECTBR; @@ -1146,9 +1185,6 @@ static void WriteInstruction(const Instruction &I, unsigned InstID, Code = bitc::FUNC_CODE_INST_RESUME; PushValueAndType(I.getOperand(0), InstID, Vals, VE); break; - case Instruction::Unwind: - Code = bitc::FUNC_CODE_INST_UNWIND; - break; case Instruction::Unreachable: Code = bitc::FUNC_CODE_INST_UNREACHABLE; AbbrevToUse = FUNCTION_INST_UNREACHABLE_ABBREV; @@ -1573,6 +1609,102 @@ static void WriteBlockInfo(const ValueEnumerator &VE, BitstreamWriter &Stream) { Stream.ExitBlock(); } +// Sort the Users based on the order in which the reader parses the bitcode +// file. +static bool bitcodereader_order(const User *lhs, const User *rhs) { + // TODO: Implement. + return true; +} + +static void WriteUseList(const Value *V, const ValueEnumerator &VE, + BitstreamWriter &Stream) { + + // One or zero uses can't get out of order. + if (V->use_empty() || V->hasNUses(1)) + return; + + // Make a copy of the in-memory use-list for sorting. + unsigned UseListSize = std::distance(V->use_begin(), V->use_end()); + SmallVector<const User*, 8> UseList; + UseList.reserve(UseListSize); + for (Value::const_use_iterator I = V->use_begin(), E = V->use_end(); + I != E; ++I) { + const User *U = *I; + UseList.push_back(U); + } + + // Sort the copy based on the order read by the BitcodeReader. + std::sort(UseList.begin(), UseList.end(), bitcodereader_order); + + // TODO: Generate a diff between the BitcodeWriter in-memory use-list and the + // sorted list (i.e., the expected BitcodeReader in-memory use-list). + + // TODO: Emit the USELIST_CODE_ENTRYs. +} + +static void WriteFunctionUseList(const Function *F, ValueEnumerator &VE, + BitstreamWriter &Stream) { + VE.incorporateFunction(*F); + + for (Function::const_arg_iterator AI = F->arg_begin(), AE = F->arg_end(); + AI != AE; ++AI) + WriteUseList(AI, VE, Stream); + for (Function::const_iterator BB = F->begin(), FE = F->end(); BB != FE; + ++BB) { + WriteUseList(BB, VE, Stream); + for (BasicBlock::const_iterator II = BB->begin(), IE = BB->end(); II != IE; + ++II) { + WriteUseList(II, VE, Stream); + for (User::const_op_iterator OI = II->op_begin(), E = II->op_end(); + OI != E; ++OI) { + if ((isa<Constant>(*OI) && !isa<GlobalValue>(*OI)) || + isa<InlineAsm>(*OI)) + WriteUseList(*OI, VE, Stream); + } + } + } + VE.purgeFunction(); +} + +// Emit use-lists. +static void WriteModuleUseLists(const Module *M, ValueEnumerator &VE, + BitstreamWriter &Stream) { + Stream.EnterSubblock(bitc::USELIST_BLOCK_ID, 3); + + // XXX: this modifies the module, but in a way that should never change the + // behavior of any pass or codegen in LLVM. The problem is that GVs may + // contain entries in the use_list that do not exist in the Module and are + // not stored in the .bc file. + for (Module::const_global_iterator I = M->global_begin(), E = M->global_end(); + I != E; ++I) + I->removeDeadConstantUsers(); + + // Write the global variables. + for (Module::const_global_iterator GI = M->global_begin(), + GE = M->global_end(); GI != GE; ++GI) { + WriteUseList(GI, VE, Stream); + + // Write the global variable initializers. + if (GI->hasInitializer()) + WriteUseList(GI->getInitializer(), VE, Stream); + } + + // Write the functions. + for (Module::const_iterator FI = M->begin(), FE = M->end(); FI != FE; ++FI) { + WriteUseList(FI, VE, Stream); + if (!FI->isDeclaration()) + WriteFunctionUseList(FI, VE, Stream); + } + + // Write the aliases. + for (Module::const_alias_iterator AI = M->alias_begin(), AE = M->alias_end(); + AI != AE; ++AI) { + WriteUseList(AI, VE, Stream); + WriteUseList(AI->getAliasee(), VE, Stream); + } + + Stream.ExitBlock(); +} /// WriteModule - Emit the specified module to the bitstream. static void WriteModule(const Module *M, BitstreamWriter &Stream) { @@ -1607,17 +1739,21 @@ static void WriteModule(const Module *M, BitstreamWriter &Stream) { // Emit metadata. WriteModuleMetadata(M, VE, Stream); - // Emit function bodies. - for (Module::const_iterator F = M->begin(), E = M->end(); F != E; ++F) - if (!F->isDeclaration()) - WriteFunction(*F, VE, Stream); - // Emit metadata. WriteModuleMetadataStore(M, Stream); // Emit names for globals/functions etc. WriteValueSymbolTable(M->getValueSymbolTable(), VE, Stream); + // Emit use-lists. + if (EnablePreserveUseListOrdering) + WriteModuleUseLists(M, VE, Stream); + + // Emit function bodies. + for (Module::const_iterator F = M->begin(), E = M->end(); F != E; ++F) + if (!F->isDeclaration()) + WriteFunction(*F, VE, Stream); + Stream.ExitBlock(); } @@ -1639,7 +1775,17 @@ enum { DarwinBCHeaderSize = 5*4 }; -static void EmitDarwinBCHeader(BitstreamWriter &Stream, const Triple &TT) { +static void WriteInt32ToBuffer(uint32_t Value, SmallVectorImpl<char> &Buffer, + uint32_t &Position) { + Buffer[Position + 0] = (unsigned char) (Value >> 0); + Buffer[Position + 1] = (unsigned char) (Value >> 8); + Buffer[Position + 2] = (unsigned char) (Value >> 16); + Buffer[Position + 3] = (unsigned char) (Value >> 24); + Position += 4; +} + +static void EmitDarwinBCHeaderAndTrailer(SmallVectorImpl<char> &Buffer, + const Triple &TT) { unsigned CPUType = ~0U; // Match x86_64-*, i[3-9]86-*, powerpc-*, powerpc64-*, arm-*, thumb-*, @@ -1666,63 +1812,55 @@ static void EmitDarwinBCHeader(BitstreamWriter &Stream, const Triple &TT) { CPUType = DARWIN_CPU_TYPE_ARM; // Traditional Bitcode starts after header. + assert(Buffer.size() >= DarwinBCHeaderSize && + "Expected header size to be reserved"); unsigned BCOffset = DarwinBCHeaderSize; + unsigned BCSize = Buffer.size()-DarwinBCHeaderSize; - Stream.Emit(0x0B17C0DE, 32); - Stream.Emit(0 , 32); // Version. - Stream.Emit(BCOffset , 32); - Stream.Emit(0 , 32); // Filled in later. - Stream.Emit(CPUType , 32); -} - -/// EmitDarwinBCTrailer - Emit the darwin epilog after the bitcode file and -/// finalize the header. -static void EmitDarwinBCTrailer(BitstreamWriter &Stream, unsigned BufferSize) { - // Update the size field in the header. - Stream.BackpatchWord(DarwinBCSizeFieldOffset, BufferSize-DarwinBCHeaderSize); + // Write the magic and version. + unsigned Position = 0; + WriteInt32ToBuffer(0x0B17C0DE , Buffer, Position); + WriteInt32ToBuffer(0 , Buffer, Position); // Version. + WriteInt32ToBuffer(BCOffset , Buffer, Position); + WriteInt32ToBuffer(BCSize , Buffer, Position); + WriteInt32ToBuffer(CPUType , Buffer, Position); // If the file is not a multiple of 16 bytes, insert dummy padding. - while (BufferSize & 15) { - Stream.Emit(0, 8); - ++BufferSize; - } + while (Buffer.size() & 15) + Buffer.push_back(0); } - /// WriteBitcodeToFile - Write the specified module to the specified output /// stream. void llvm::WriteBitcodeToFile(const Module *M, raw_ostream &Out) { - std::vector<unsigned char> Buffer; - BitstreamWriter Stream(Buffer); - + SmallVector<char, 1024> Buffer; Buffer.reserve(256*1024); - WriteBitcodeToStream( M, Stream ); - - // Write the generated bitstream to "Out". - Out.write((char*)&Buffer.front(), Buffer.size()); -} - -/// WriteBitcodeToStream - Write the specified module to the specified output -/// stream. -void llvm::WriteBitcodeToStream(const Module *M, BitstreamWriter &Stream) { - // If this is darwin or another generic macho target, emit a file header and - // trailer if needed. + // If this is darwin or another generic macho target, reserve space for the + // header. Triple TT(M->getTargetTriple()); if (TT.isOSDarwin()) - EmitDarwinBCHeader(Stream, TT); - - // Emit the file header. - Stream.Emit((unsigned)'B', 8); - Stream.Emit((unsigned)'C', 8); - Stream.Emit(0x0, 4); - Stream.Emit(0xC, 4); - Stream.Emit(0xE, 4); - Stream.Emit(0xD, 4); - - // Emit the module. - WriteModule(M, Stream); + Buffer.insert(Buffer.begin(), DarwinBCHeaderSize, 0); + + // Emit the module into the buffer. + { + BitstreamWriter Stream(Buffer); + + // Emit the file header. + Stream.Emit((unsigned)'B', 8); + Stream.Emit((unsigned)'C', 8); + Stream.Emit(0x0, 4); + Stream.Emit(0xC, 4); + Stream.Emit(0xE, 4); + Stream.Emit(0xD, 4); + + // Emit the module. + WriteModule(M, Stream); + } if (TT.isOSDarwin()) - EmitDarwinBCTrailer(Stream, Stream.getBuffer().size()); + EmitDarwinBCHeaderAndTrailer(Buffer, TT); + + // Write the generated bitstream to "Out". + Out.write((char*)&Buffer.front(), Buffer.size()); } diff --git a/contrib/llvm/lib/Bitcode/Writer/ValueEnumerator.cpp b/contrib/llvm/lib/Bitcode/Writer/ValueEnumerator.cpp index 9ae9905..1ed9004 100644 --- a/contrib/llvm/lib/Bitcode/Writer/ValueEnumerator.cpp +++ b/contrib/llvm/lib/Bitcode/Writer/ValueEnumerator.cpp @@ -19,6 +19,8 @@ #include "llvm/Module.h" #include "llvm/ValueSymbolTable.h" #include "llvm/Instructions.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/raw_ostream.h" #include <algorithm> using namespace llvm; @@ -107,7 +109,6 @@ ValueEnumerator::ValueEnumerator(const Module *M) { OptimizeConstants(FirstConstant, Values.size()); } - unsigned ValueEnumerator::getInstructionID(const Instruction *Inst) const { InstructionMapType::const_iterator I = InstructionMap.find(Inst); assert(I != InstructionMap.end() && "Instruction is not mapped!"); @@ -130,6 +131,43 @@ unsigned ValueEnumerator::getValueID(const Value *V) const { return I->second-1; } +void ValueEnumerator::dump() const { + print(dbgs(), ValueMap, "Default"); + dbgs() << '\n'; + print(dbgs(), MDValueMap, "MetaData"); + dbgs() << '\n'; +} + +void ValueEnumerator::print(raw_ostream &OS, const ValueMapType &Map, + const char *Name) const { + + OS << "Map Name: " << Name << "\n"; + OS << "Size: " << Map.size() << "\n"; + for (ValueMapType::const_iterator I = Map.begin(), + E = Map.end(); I != E; ++I) { + + const Value *V = I->first; + if (V->hasName()) + OS << "Value: " << V->getName(); + else + OS << "Value: [null]\n"; + V->dump(); + + OS << " Uses(" << std::distance(V->use_begin(),V->use_end()) << "):"; + for (Value::const_use_iterator UI = V->use_begin(), UE = V->use_end(); + UI != UE; ++UI) { + if (UI != V->use_begin()) + OS << ","; + if((*UI)->hasName()) + OS << " " << (*UI)->getName(); + else + OS << " [null]"; + + } + OS << "\n\n"; + } +} + // Optimize constant ordering. namespace { struct CstSortPredicate { @@ -283,10 +321,6 @@ void ValueEnumerator::EnumerateValue(const Value *V) { if (const Constant *C = dyn_cast<Constant>(V)) { if (isa<GlobalValue>(C)) { // Initializers for globals are handled explicitly elsewhere. - } else if (isa<ConstantArray>(C) && cast<ConstantArray>(C)->isString()) { - // Do not enumerate the initializers for an array of simple characters. - // The initializers just pollute the value table, and we emit the strings - // specially. } else if (C->getNumOperands()) { // If a constant has operands, enumerate them. This makes sure that if a // constant has uses (for example an array of const ints), that they are diff --git a/contrib/llvm/lib/Bitcode/Writer/ValueEnumerator.h b/contrib/llvm/lib/Bitcode/Writer/ValueEnumerator.h index b6fc920..a6ca536 100644 --- a/contrib/llvm/lib/Bitcode/Writer/ValueEnumerator.h +++ b/contrib/llvm/lib/Bitcode/Writer/ValueEnumerator.h @@ -32,6 +32,7 @@ class NamedMDNode; class AttrListPtr; class ValueSymbolTable; class MDSymbolTable; +class raw_ostream; class ValueEnumerator { public: @@ -83,6 +84,9 @@ private: public: ValueEnumerator(const Module *M); + void dump() const; + void print(raw_ostream &OS, const ValueMapType &Map, const char *Name) const; + unsigned getValueID(const Value *V) const; unsigned getTypeID(Type *T) const { |