From 12c29d10bb5dc73b9c67dc63051c714b29de547d Mon Sep 17 00:00:00 2001 From: Misha Brukman Date: Mon, 22 Sep 2003 23:38:23 +0000 Subject: [PATCH] * Cleaned up code: - no more passing around a string pointer to set errors - no more returning booleans and checking for errors, we use C++ exceptions * Broke functionality into 2 new classes, one reads from file, one from a stream * Implemented lazy function streaming - the parser can read in a function at-a-time git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@8671 91177308-0d34-0410-b5e6-96231b3b80d8 --- lib/Bytecode/Reader/ConstantReader.cpp | 95 +++--- lib/Bytecode/Reader/Reader.cpp | 423 ++++++++++--------------- lib/Bytecode/Reader/ReaderInternals.h | 95 ++++-- lib/Bytecode/Reader/ReaderWrappers.cpp | 170 ++++++++++ 4 files changed, 443 insertions(+), 340 deletions(-) create mode 100644 lib/Bytecode/Reader/ReaderWrappers.cpp diff --git a/lib/Bytecode/Reader/ConstantReader.cpp b/lib/Bytecode/Reader/ConstantReader.cpp index 25abb79b461..91baa566d7f 100644 --- a/lib/Bytecode/Reader/ConstantReader.cpp +++ b/lib/Bytecode/Reader/ConstantReader.cpp @@ -16,7 +16,7 @@ const Type *BytecodeParser::parseTypeConstant(const unsigned char *&Buf, const unsigned char *EndBuf) { unsigned PrimType; - if (read_vbr(Buf, EndBuf, PrimType)) return 0; + if (read_vbr(Buf, EndBuf, PrimType)) throw Error_readvbr; const Type *Val = 0; if ((Val = Type::getPrimitiveType((Type::PrimitiveID)PrimType))) @@ -124,7 +124,7 @@ void BytecodeParser::refineAbstractType(const DerivedType *OldType, // with a new resolved concrete type. // void debug_type_tables(); -bool BytecodeParser::parseTypeConstants(const unsigned char *&Buf, +void BytecodeParser::parseTypeConstants(const unsigned char *&Buf, const unsigned char *EndBuf, TypeValuesListTy &Tab, unsigned NumEntries) { @@ -139,7 +139,7 @@ bool BytecodeParser::parseTypeConstants(const unsigned char *&Buf, // for (unsigned i = 0; i < NumEntries; ++i) { const Type *NewTy = parseTypeConstant(Buf, EndBuf), *OldTy = Tab[i].get(); - if (NewTy == 0) return true; + if (NewTy == 0) throw std::string("Parsed invalid type."); BCR_TRACE(4, "#" << i << ": Read Type Constant: '" << NewTy << "' Replacing: " << OldTy << "\n"); @@ -163,11 +163,10 @@ bool BytecodeParser::parseTypeConstants(const unsigned char *&Buf, BCR_TRACE(5, (void*)Tab[i].get() << " - " << Tab[i].get() << "\n"); } debug_type_tables(); - return false; } -bool BytecodeParser::parseConstantValue(const unsigned char *&Buf, +void BytecodeParser::parseConstantValue(const unsigned char *&Buf, const unsigned char *EndBuf, const Type *Ty, Constant *&V) { @@ -175,28 +174,28 @@ bool BytecodeParser::parseConstantValue(const unsigned char *&Buf, // a ConstantExpr can be of any type, and has no explicit value. // unsigned isExprNumArgs; // 0 if not expr; numArgs if is expr - if (read_vbr(Buf, EndBuf, isExprNumArgs)) return true; + if (read_vbr(Buf, EndBuf, isExprNumArgs)) throw Error_readvbr; if (isExprNumArgs) { // FIXME: Encoding of constant exprs could be much more compact! unsigned Opcode; std::vector ArgVec; ArgVec.reserve(isExprNumArgs); - if (read_vbr(Buf, EndBuf, Opcode)) return true; + if (read_vbr(Buf, EndBuf, Opcode)) throw Error_readvbr; // Read the slot number and types of each of the arguments for (unsigned i = 0; i != isExprNumArgs; ++i) { unsigned ArgValSlot, ArgTypeSlot; - if (read_vbr(Buf, EndBuf, ArgValSlot)) return true; - if (read_vbr(Buf, EndBuf, ArgTypeSlot)) return true; + if (read_vbr(Buf, EndBuf, ArgValSlot)) throw Error_readvbr; + if (read_vbr(Buf, EndBuf, ArgTypeSlot)) throw Error_readvbr; const Type *ArgTy = getType(ArgTypeSlot); - if (ArgTy == 0) return true; + if (ArgTy == 0) throw std::string("Argument type slot not found."); BCR_TRACE(4, "CE Arg " << i << ": Type: '" << ArgTy << "' slot: " << ArgValSlot << "\n"); // Get the arg value from its slot if it exists, otherwise a placeholder Constant *C = getConstantValue(ArgTy, ArgValSlot); - if (C == 0) return true; + if (C == 0) throw std::string("No arg value or placeholder found."); ArgVec.push_back(C); } @@ -210,15 +209,15 @@ bool BytecodeParser::parseConstantValue(const unsigned char *&Buf, } else { // All other 2-operand expressions V = ConstantExpr::get(Opcode, ArgVec[0], ArgVec[1]); } - return false; + return; } // Ok, not an ConstantExpr. We now know how to read the given type... switch (Ty->getPrimitiveID()) { case Type::BoolTyID: { unsigned Val; - if (read_vbr(Buf, EndBuf, Val)) return true; - if (Val != 0 && Val != 1) return true; + if (read_vbr(Buf, EndBuf, Val)) throw Error_readvbr; + if (Val != 0 && Val != 1) throw std::string("Invalid boolean value read."); V = ConstantBool::get(Val == 1); break; } @@ -227,15 +226,16 @@ bool BytecodeParser::parseConstantValue(const unsigned char *&Buf, case Type::UShortTyID: case Type::UIntTyID: { unsigned Val; - if (read_vbr(Buf, EndBuf, Val)) return true; - if (!ConstantUInt::isValueValidForType(Ty, Val)) return true; + if (read_vbr(Buf, EndBuf, Val)) throw Error_readvbr; + if (!ConstantUInt::isValueValidForType(Ty, Val)) + throw std::string("Invalid unsigned byte/short/int read."); V = ConstantUInt::get(Ty, Val); break; } case Type::ULongTyID: { uint64_t Val; - if (read_vbr(Buf, EndBuf, Val)) return true; + if (read_vbr(Buf, EndBuf, Val)) throw Error_readvbr; V = ConstantUInt::get(Ty, Val); break; } @@ -245,22 +245,23 @@ bool BytecodeParser::parseConstantValue(const unsigned char *&Buf, case Type::IntTyID: { case Type::LongTyID: int64_t Val; - if (read_vbr(Buf, EndBuf, Val)) return true; - if (!ConstantSInt::isValueValidForType(Ty, Val)) return true; + if (read_vbr(Buf, EndBuf, Val)) throw Error_readvbr; + if (!ConstantSInt::isValueValidForType(Ty, Val)) + throw std::string("Invalid signed byte/short/int/long read."); V = ConstantSInt::get(Ty, Val); break; } case Type::FloatTyID: { float F; - if (input_data(Buf, EndBuf, &F, &F+1)) return true; + if (input_data(Buf, EndBuf, &F, &F+1)) throw Error_inputdata; V = ConstantFP::get(Ty, F); break; } case Type::DoubleTyID: { double Val; - if (input_data(Buf, EndBuf, &Val, &Val+1)) return true; + if (input_data(Buf, EndBuf, &Val, &Val+1)) throw Error_inputdata; V = ConstantFP::get(Ty, Val); break; } @@ -276,9 +277,9 @@ bool BytecodeParser::parseConstantValue(const unsigned char *&Buf, std::vector Elements; while (NumElements--) { // Read all of the elements of the constant. unsigned Slot; - if (read_vbr(Buf, EndBuf, Slot)) return true; + if (read_vbr(Buf, EndBuf, Slot)) throw Error_readvbr; Constant *C = getConstantValue(AT->getElementType(), Slot); - if (!C) return true; + if (!C) throw std::string("Unable to get const value of array slot."); Elements.push_back(C); } V = ConstantArray::get(AT, Elements); @@ -292,9 +293,9 @@ bool BytecodeParser::parseConstantValue(const unsigned char *&Buf, std::vector Elements; for (unsigned i = 0; i < ET.size(); ++i) { unsigned Slot; - if (read_vbr(Buf, EndBuf, Slot)) return true; + if (read_vbr(Buf, EndBuf, Slot)) throw Error_readvbr; Constant *C = getConstantValue(ET[i], Slot); - if (!C) return true; + if (!C) throw std::string("Could not read const value in struct slot."); Elements.push_back(C); } @@ -308,7 +309,7 @@ bool BytecodeParser::parseConstantValue(const unsigned char *&Buf, if (HasImplicitZeroInitializer) SubClass = 1; else - if (read_vbr(Buf, EndBuf, SubClass)) return true; + if (read_vbr(Buf, EndBuf, SubClass)) throw Error_readvbr; switch (SubClass) { case 0: // ConstantPointerNull value... @@ -317,19 +318,20 @@ bool BytecodeParser::parseConstantValue(const unsigned char *&Buf, case 1: { // ConstantPointerRef value... unsigned Slot; - if (read_vbr(Buf, EndBuf, Slot)) return true; + if (read_vbr(Buf, EndBuf, Slot)) throw Error_readvbr; BCR_TRACE(4, "CPR: Type: '" << Ty << "' slot: " << Slot << "\n"); // Check to see if we have already read this global variable... Value *Val = getValue(PT, Slot, false); GlobalValue *GV; if (Val) { - if (!(GV = dyn_cast(Val))) return true; + if (!(GV = dyn_cast(Val))) + throw std::string("Value of ConstantPointerRef not in ValueTable!"); BCR_TRACE(5, "Value Found in ValueTable!\n"); } else if (RevisionNum > 0) { // Revision #0 could have forward references to globals that were weird. // We got rid of this in subsequent revs. - return true; + throw std::string("Forward references to globals not allowed."); } else { // Nope... find or create a forward ref. for it GlobalRefsType::iterator I = GlobalRefs.find(std::make_pair(PT, Slot)); @@ -359,7 +361,7 @@ bool BytecodeParser::parseConstantValue(const unsigned char *&Buf, default: BCR_TRACE(5, "UNKNOWN Pointer Constant Type!\n"); - return true; + throw std::string("Unknown pointer constant type."); } break; } @@ -368,41 +370,41 @@ bool BytecodeParser::parseConstantValue(const unsigned char *&Buf, std::cerr << __FILE__ << ":" << __LINE__ << ": Don't know how to deserialize constant value of type '" << Ty->getName() << "'\n"; - return true; + throw std::string("Don't know how to deserialize constant value of type '"+ + Ty->getName()); } - - return false; } -bool BytecodeParser::ParseGlobalTypes(const unsigned char *&Buf, +void BytecodeParser::ParseGlobalTypes(const unsigned char *&Buf, const unsigned char *EndBuf) { ValueTable T; - return ParseConstantPool(Buf, EndBuf, T, ModuleTypeValues); + ParseConstantPool(Buf, EndBuf, T, ModuleTypeValues); } -bool BytecodeParser::ParseConstantPool(const unsigned char *&Buf, +void BytecodeParser::ParseConstantPool(const unsigned char *&Buf, const unsigned char *EndBuf, - ValueTable &Tab, - TypeValuesListTy &TypeTab) { + ValueTable &Tab, + TypeValuesListTy &TypeTab) { while (Buf < EndBuf) { unsigned NumEntries, Typ; if (read_vbr(Buf, EndBuf, NumEntries) || - read_vbr(Buf, EndBuf, Typ)) return true; + read_vbr(Buf, EndBuf, Typ)) throw Error_readvbr; const Type *Ty = getType(Typ); - if (Ty == 0) return true; + if (Ty == 0) throw std::string("Invalid type read."); BCR_TRACE(3, "Type: '" << Ty << "' NumEntries: " << NumEntries << "\n"); if (Typ == Type::TypeTyID) { - if (parseTypeConstants(Buf, EndBuf, TypeTab, NumEntries)) return true; + parseTypeConstants(Buf, EndBuf, TypeTab, NumEntries); } else { for (unsigned i = 0; i < NumEntries; ++i) { - Constant *C; + Constant *C; int Slot; - if (parseConstantValue(Buf, EndBuf, Ty, C)) return true; + parseConstantValue(Buf, EndBuf, Ty, C); assert(C && "parseConstantValue returned NULL!"); - BCR_TRACE(4, "Read Constant: '" << *C << "'\n"); - if ((Slot = insertValue(C, Tab)) == -1) return true; + BCR_TRACE(4, "Read Constant: '" << *C << "'\n"); + if ((Slot = insertValue(C, Tab)) == -1) + throw std::string("Could not insert value into ValueTable."); // If we are reading a function constant table, make sure that we adjust // the slot number to be the real global constant number. @@ -414,6 +416,5 @@ bool BytecodeParser::ParseConstantPool(const unsigned char *&Buf, } } - if (Buf > EndBuf) return true; - return false; + if (Buf > EndBuf) throw std::string("Read past end of buffer."); } diff --git a/lib/Bytecode/Reader/Reader.cpp b/lib/Bytecode/Reader/Reader.cpp index 5277806cff5..071b1da40f5 100644 --- a/lib/Bytecode/Reader/Reader.cpp +++ b/lib/Bytecode/Reader/Reader.cpp @@ -11,37 +11,44 @@ //===----------------------------------------------------------------------===// #include "ReaderInternals.h" -#include "Config/sys/mman.h" #include "llvm/Bytecode/Reader.h" #include "llvm/Bytecode/Format.h" -#include "llvm/Module.h" #include "llvm/Constants.h" #include "llvm/iPHINode.h" #include "llvm/iOther.h" -#include "Config/sys/types.h" -#include "Config/sys/stat.h" -#include "Config/fcntl.h" +#include "llvm/Module.h" +#include "Support/StringExtras.h" #include "Config/unistd.h" +#include "Config/sys/mman.h" +#include "Config/sys/stat.h" +#include "Config/sys/types.h" #include +#include -bool BytecodeParser::getTypeSlot(const Type *Ty, unsigned &Slot) { +#define CHECK_ALIGN32(begin,end) \ + if (align32(begin,end)) \ + throw std::string("Alignment error: Reader.cpp:" + \ + utostr((unsigned)__LINE__)); + +void +BytecodeParser::getTypeSlot(const Type *Ty, unsigned &Slot) { if (Ty->isPrimitiveType()) { Slot = Ty->getPrimitiveID(); } else { // Check the function level types first... TypeValuesListTy::iterator I = find(FunctionTypeValues.begin(), - FunctionTypeValues.end(), Ty); + FunctionTypeValues.end(), Ty); if (I != FunctionTypeValues.end()) { - Slot = FirstDerivedTyID+ModuleTypeValues.size()+ - (&*I - &FunctionTypeValues[0]); + Slot = FirstDerivedTyID + ModuleTypeValues.size() + + (&*I - &FunctionTypeValues[0]); } else { I = find(ModuleTypeValues.begin(), ModuleTypeValues.end(), Ty); - if (I == ModuleTypeValues.end()) return true; // Didn't find type! + if (I == ModuleTypeValues.end()) + throw std::string("Didn't find type in ModuleTypeValues."); Slot = FirstDerivedTyID + (&*I - &ModuleTypeValues[0]); } } //cerr << "getTypeSlot '" << Ty->getName() << "' = " << Slot << "\n"; - return false; } const Type *BytecodeParser::getType(unsigned ID) { @@ -61,7 +68,7 @@ int BytecodeParser::insertValue(Value *Val, ValueTable &ValueTab) { !cast(Val)->isNullValue()) && "Cannot read null values from bytecode!"); unsigned type; - if (getTypeSlot(Val->getType(), type)) return -1; + getTypeSlot(Val->getType(), type); assert(type != Type::TypeTyID && "Types should never be insertValue'd!"); if (ValueTab.size() <= type) { @@ -72,11 +79,11 @@ int BytecodeParser::insertValue(Value *Val, ValueTable &ValueTab) { } //cerr << "insertValue Values[" << type << "][" << ValueTab[type].size() - // << "] = " << Val << "\n"; + // << "] = " << Val << "\n"; ValueTab[type]->push_back(Val); bool HasOffset = HasImplicitZeroInitializer && - !Val->getType()->isPrimitiveType(); + !Val->getType()->isPrimitiveType(); return ValueTab[type]->size()-1 + HasOffset; } @@ -86,9 +93,8 @@ void BytecodeParser::setValueTo(ValueTable &ValueTab, unsigned Slot, Value *Val) { assert(&ValueTab == &ModuleValues && "Can only setValueTo on Module values!"); unsigned type; - if (getTypeSlot(Val->getType(), type)) - assert(0 && "getTypeSlot failed!"); - + getTypeSlot(Val->getType(), type); + assert((!HasImplicitZeroInitializer || Slot != 0) && "Cannot change zero init"); assert(type < ValueTab.size() && Slot <= ValueTab[type]->size()); @@ -98,10 +104,9 @@ void BytecodeParser::setValueTo(ValueTable &ValueTab, unsigned Slot, Value *BytecodeParser::getValue(const Type *Ty, unsigned oNum, bool Create) { unsigned Num = oNum; unsigned type; // The type plane it lives in... + getTypeSlot(Ty, type); - if (getTypeSlot(Ty, type)) return 0; - - if (type == Type::TypeTyID) { // The 'type' plane has implicit values + if (type == Type::TypeTyID) { // The 'type' plane has implicit values assert(Create == false); if (Num < Type::NumPrimitiveIDs) { const Type *T = Type::getPrimitiveType((Type::PrimitiveID)Num); @@ -111,11 +116,11 @@ Value *BytecodeParser::getValue(const Type *Ty, unsigned oNum, bool Create) { // Otherwise, derived types need offset... Num -= FirstDerivedTyID; - // Is it a module level type? + // Is it a module-level type? if (Num < ModuleTypeValues.size()) return (Value*)ModuleTypeValues[Num].get(); - // Nope, is it a function level type? + // Nope, is it a function-level type? Num -= ModuleTypeValues.size(); if (Num < FunctionTypeValues.size()) return (Value*)FunctionTypeValues[Num].get(); @@ -184,8 +189,7 @@ Constant *BytecodeParser::getConstantValue(const Type *Ty, unsigned Slot) { } -bool BytecodeParser::postResolveValues(ValueTable &ValTab) { - bool Error = false; +void BytecodeParser::postResolveValues(ValueTable &ValTab) { while (!ValTab.empty()) { ValueList &DL = *ValTab.back(); ValTab.pop_back(); @@ -197,83 +201,77 @@ bool BytecodeParser::postResolveValues(ValueTable &ValTab) { Value *NewDef = getValue(D->getType(), IDNumber, false); if (NewDef == 0) { - Error = true; // Unresolved thinger - std::cerr << "Unresolvable reference found: <" - << *D->getType() << ">:" << IDNumber <<"!\n"; + throw std::string("Unresolvable reference found: <" + + D->getType()->getName() + ">:" +utostr(IDNumber)+"."); } else { - // Fixup all of the uses of this placeholder def... + // Fixup all of the uses of this placeholder def... D->replaceAllUsesWith(NewDef); // Now that all the uses are gone, delete the placeholder... // If we couldn't find a def (error case), then leak a little - delete D; // memory, 'cause otherwise we can't remove all uses! + delete D; // memory, 'cause otherwise we can't remove all uses! } } delete &DL; } - - return Error; } -bool BytecodeParser::ParseBasicBlock(const unsigned char *&Buf, - const unsigned char *EndBuf, - BasicBlock *&BB) { - BB = new BasicBlock(); +std::auto_ptr +BytecodeParser::ParseBasicBlock(const unsigned char *&Buf, + const unsigned char *EndBuf) { + std::auto_ptr BB(new BasicBlock()); while (Buf < EndBuf) { Instruction *Inst; - if (ParseInstruction(Buf, EndBuf, Inst)) { - delete BB; - return true; + ParseInstruction(Buf, EndBuf, Inst); + + if (Inst == 0) { throw std::string("Could not parse Instruction."); } + if (insertValue(Inst, Values) == -1) { + throw std::string("Could not insert value."); } - if (Inst == 0) { delete BB; return true; } - if (insertValue(Inst, Values) == -1) { delete BB; return true; } - BB->getInstList().push_back(Inst); - BCR_TRACE(4, Inst); } - return false; + return BB; } -bool BytecodeParser::ParseSymbolTable(const unsigned char *&Buf, +void BytecodeParser::ParseSymbolTable(const unsigned char *&Buf, const unsigned char *EndBuf, - SymbolTable *ST) { + SymbolTable *ST) { while (Buf < EndBuf) { // Symtab block header: [num entries][type id number] unsigned NumEntries, Typ; if (read_vbr(Buf, EndBuf, NumEntries) || - read_vbr(Buf, EndBuf, Typ)) return true; + read_vbr(Buf, EndBuf, Typ)) throw Error_readvbr; const Type *Ty = getType(Typ); - if (Ty == 0) return true; + if (Ty == 0) throw std::string("Invalid type read in symbol table."); BCR_TRACE(3, "Plane Type: '" << Ty << "' with " << NumEntries << - " entries\n"); + " entries\n"); for (unsigned i = 0; i < NumEntries; ++i) { // Symtab entry: [def slot #][name] unsigned slot; - if (read_vbr(Buf, EndBuf, slot)) return true; + if (read_vbr(Buf, EndBuf, slot)) throw Error_readvbr; std::string Name; if (read(Buf, EndBuf, Name, false)) // Not aligned... - return true; + throw std::string("Buffer not aligned."); Value *V = getValue(Ty, slot, false); // Find mapping... if (V == 0) { - BCR_TRACE(3, "FAILED LOOKUP: Slot #" << slot << "\n"); - return true; + BCR_TRACE(3, "FAILED LOOKUP: Slot #" << slot << "\n"); + throw std::string("Failed value look-up."); } BCR_TRACE(4, "Map: '" << Name << "' to #" << slot << ":" << *V; - if (!isa(V)) std::cerr << "\n"); + if (!isa(V)) std::cerr << "\n"); V->setName(Name, ST); } } - if (Buf > EndBuf) return true; - return false; + if (Buf > EndBuf) throw std::string("Tried to read past end of buffer."); } void BytecodeParser::ResolveReferencesToValue(Value *NewV, unsigned Slot) { @@ -295,114 +293,124 @@ void BytecodeParser::ResolveReferencesToValue(Value *NewV, unsigned Slot) { GlobalRefs.erase(I); // Remove the map entry for it } +void +BytecodeParser::ParseFunction(const unsigned char *&Buf, + const unsigned char *EndBuf) { + if (FunctionSignatureList.empty()) + throw std::string("FunctionSignatureList empty!"); -bool BytecodeParser::ParseFunction(const unsigned char *&Buf, - const unsigned char *EndBuf) { - // Clear out the local values table... - if (FunctionSignatureList.empty()) { - Error = "Function found, but FunctionSignatureList empty!"; - return true; // Unexpected function! - } + Function *F = FunctionSignatureList.back().first; + unsigned FunctionSlot = FunctionSignatureList.back().second; + FunctionSignatureList.pop_back(); + + // Save the information for future reading of the function + LazyFunctionInfo *LFI = new LazyFunctionInfo(); + LFI->Buf = Buf; LFI->EndBuf = EndBuf; LFI->FunctionSlot = FunctionSlot; + LazyFunctionLoadMap[F] = LFI; + // Pretend we've `parsed' this function + Buf = EndBuf; +} + +void BytecodeParser::materializeFunction(Function* F) { + // Find {start, end} pointers and slot in the map. If not there, we're done. + std::map::iterator Fi = + LazyFunctionLoadMap.find(F); + if (Fi == LazyFunctionLoadMap.end()) return; + + LazyFunctionInfo *LFI = Fi->second; + const unsigned char *Buf = LFI->Buf; + const unsigned char *EndBuf = LFI->EndBuf; + unsigned FunctionSlot = LFI->FunctionSlot; + LazyFunctionLoadMap.erase(Fi); + delete LFI; GlobalValue::LinkageTypes Linkage = GlobalValue::ExternalLinkage; if (!hasInternalMarkerOnly) { unsigned LinkageType; - if (read_vbr(Buf, EndBuf, LinkageType)) return true; - if (LinkageType & ~0x3) return true; + if (read_vbr(Buf, EndBuf, LinkageType)) + throw std::string("ParseFunction: Error reading from buffer."); + if (LinkageType & ~0x3) + throw std::string("Invalid linkage type for Function."); Linkage = (GlobalValue::LinkageTypes)LinkageType; } else { // We used to only support two linkage models: internal and external unsigned isInternal; - if (read_vbr(Buf, EndBuf, isInternal)) return true; + if (read_vbr(Buf, EndBuf, isInternal)) + throw std::string("ParseFunction: Error reading from buffer."); if (isInternal) Linkage = GlobalValue::InternalLinkage; } - Function *F = FunctionSignatureList.back().first; - unsigned FunctionSlot = FunctionSignatureList.back().second; - FunctionSignatureList.pop_back(); F->setLinkage(Linkage); const FunctionType::ParamTypes &Params =F->getFunctionType()->getParamTypes(); Function::aiterator AI = F->abegin(); for (FunctionType::ParamTypes::const_iterator It = Params.begin(); It != Params.end(); ++It, ++AI) { - if (insertValue(AI, Values) == -1) { - Error = "Error reading function arguments!\n"; - return true; - } + if (insertValue(AI, Values) == -1) + throw std::string("Error reading function arguments!"); } while (Buf < EndBuf) { unsigned Type, Size; const unsigned char *OldBuf = Buf; - if (readBlock(Buf, EndBuf, Type, Size)) { - Error = "Error reading Function level block!"; - return true; - } + readBlock(Buf, EndBuf, Type, Size); switch (Type) { - case BytecodeFormat::ConstantPool: + case BytecodeFormat::ConstantPool: { BCR_TRACE(2, "BLOCK BytecodeFormat::ConstantPool: {\n"); - if (ParseConstantPool(Buf, Buf+Size, Values, FunctionTypeValues)) - return true; + ParseConstantPool(Buf, Buf+Size, Values, FunctionTypeValues); break; + } case BytecodeFormat::BasicBlock: { BCR_TRACE(2, "BLOCK BytecodeFormat::BasicBlock: {\n"); - BasicBlock *BB; - if (ParseBasicBlock(Buf, Buf+Size, BB) || - insertValue(BB, Values) == -1) - return true; // Parse error... :( + std::auto_ptr BB = ParseBasicBlock(Buf, Buf+Size); + if (!BB.get() || insertValue(BB.get(), Values) == -1) + throw std::string("Parse error: BasicBlock"); - F->getBasicBlockList().push_back(BB); + F->getBasicBlockList().push_back(BB.release()); break; } - case BytecodeFormat::SymbolTable: + case BytecodeFormat::SymbolTable: { BCR_TRACE(2, "BLOCK BytecodeFormat::SymbolTable: {\n"); - if (ParseSymbolTable(Buf, Buf+Size, &F->getSymbolTable())) - return true; + ParseSymbolTable(Buf, Buf+Size, &F->getSymbolTable()); break; + } default: BCR_TRACE(2, "BLOCK :ignored! {\n"); Buf += Size; - if (OldBuf > Buf) return true; // Wrap around! + if (OldBuf > Buf) + throw std::string("Wrapped around reading bytecode."); break; } BCR_TRACE(2, "} end block\n"); - if (align32(Buf, EndBuf)) { - Error = "Error aligning Function level block!"; - return true; // Malformed bc file, read past end of block. - } + // Malformed bc file if read past end of block. + CHECK_ALIGN32(Buf, EndBuf); } - if (postResolveValues(LateResolveValues)) { - Error = "Error resolving function values!"; - return true; // Unresolvable references! - } + // Check for unresolvable references + postResolveValues(LateResolveValues); - ResolveReferencesToValue(F, FunctionSlot); + //ResolveReferencesToValue(F, FunctionSlot); - // Clear out function level types... + // Clear out function-level types... FunctionTypeValues.clear(); freeTable(Values); - return false; } -bool BytecodeParser::ParseModuleGlobalInfo(const unsigned char *&Buf, - const unsigned char *End){ - if (!FunctionSignatureList.empty()) { - Error = "Two ModuleGlobalInfo packets found!"; - return true; // Two ModuleGlobal blocks? - } +void BytecodeParser::ParseModuleGlobalInfo(const unsigned char *&Buf, + const unsigned char *End) { + if (!FunctionSignatureList.empty()) + throw std::string("Two ModuleGlobalInfo packets found!"); // Read global variables... unsigned VarType; - if (read_vbr(Buf, End, VarType)) return true; + if (read_vbr(Buf, End, VarType)) throw Error_readvbr; while (VarType != Type::VoidTyID) { // List is terminated by Void unsigned SlotNo; GlobalValue::LinkageTypes Linkage; @@ -421,10 +429,9 @@ bool BytecodeParser::ParseModuleGlobalInfo(const unsigned char *&Buf, } const Type *Ty = getType(SlotNo); - if (!Ty || !isa(Ty)) { - Error = "Global not pointer type! Ty = " + Ty->getDescription(); - return true; - } + if (!Ty || !isa(Ty)) + throw std::string("Global not pointer type! Ty = " + + Ty->getDescription()); const Type *ElTy = cast(Ty)->getElementType(); @@ -432,27 +439,27 @@ bool BytecodeParser::ParseModuleGlobalInfo(const unsigned char *&Buf, GlobalVariable *GV = new GlobalVariable(ElTy, VarType & 1, Linkage, 0, "", TheModule); int DestSlot = insertValue(GV, ModuleValues); - if (DestSlot == -1) return true; + if (DestSlot == -1) throw Error_DestSlot; BCR_TRACE(2, "Global Variable of type: " << *Ty << "\n"); ResolveReferencesToValue(GV, (unsigned)DestSlot); if (VarType & 2) { // Does it have an initializer? unsigned InitSlot; - if (read_vbr(Buf, End, InitSlot)) return true; + if (read_vbr(Buf, End, InitSlot)) throw Error_readvbr; GlobalInits.push_back(std::make_pair(GV, InitSlot)); } - if (read_vbr(Buf, End, VarType)) return true; + if (read_vbr(Buf, End, VarType)) throw Error_readvbr; } // Read the function objects for all of the functions that are coming unsigned FnSignature; - if (read_vbr(Buf, End, FnSignature)) return true; + if (read_vbr(Buf, End, FnSignature)) throw Error_readvbr; while (FnSignature != Type::VoidTyID) { // List is terminated by Void const Type *Ty = getType(FnSignature); if (!Ty || !isa(Ty) || !isa(cast(Ty)->getElementType())) { - Error = "Function not ptr to func type! Ty = " + Ty->getDescription(); - return true; + throw std::string("Function not ptr to func type! Ty = " + + Ty->getDescription()); } // We create functions by passing the underlying FunctionType to create... @@ -467,7 +474,7 @@ bool BytecodeParser::ParseModuleGlobalInfo(const unsigned char *&Buf, Function *Func = new Function(cast(Ty), GlobalValue::InternalLinkage, "", TheModule); int DestSlot = insertValue(Func, ModuleValues); - if (DestSlot == -1) return true; + if (DestSlot == -1) throw Error_DestSlot; ResolveReferencesToValue(Func, (unsigned)DestSlot); // Keep track of this information in a list that is emptied as functions are @@ -475,11 +482,11 @@ bool BytecodeParser::ParseModuleGlobalInfo(const unsigned char *&Buf, // FunctionSignatureList.push_back(std::make_pair(Func, DestSlot)); - if (read_vbr(Buf, End, FnSignature)) return true; + if (read_vbr(Buf, End, FnSignature)) throw Error_readvbr; BCR_TRACE(2, "Function of type: " << Ty << "\n"); } - if (align32(Buf, End)) return true; + CHECK_ALIGN32(Buf, End); // Now that the function signature list is set up, reverse it so that we can // remove elements efficiently from the back of the vector. @@ -489,13 +496,12 @@ bool BytecodeParser::ParseModuleGlobalInfo(const unsigned char *&Buf, // we don't understand, so we transparently ignore them. // Buf = End; - return false; } -bool BytecodeParser::ParseVersionInfo(const unsigned char *&Buf, +void BytecodeParser::ParseVersionInfo(const unsigned char *&Buf, const unsigned char *EndBuf) { unsigned Version; - if (read_vbr(Buf, EndBuf, Version)) return true; + if (read_vbr(Buf, EndBuf, Version)) throw Error_readvbr; // Unpack version number: low four bits are for flags, top bits = version Module::Endianness Endianness; @@ -519,7 +525,7 @@ bool BytecodeParser::ParseVersionInfo(const unsigned char *&Buf, // only valid with a 14 in the flags values. Also, it does not support // encoding zero initializers for arrays compactly. // - if (Version != 14) return true; // Unknown revision 0 flags? + if (Version != 14) throw std::string("Unknown revision 0 flags?"); HasImplicitZeroInitializer = false; Endianness = Module::BigEndian; PointerSize = Module::Pointer64; @@ -536,8 +542,7 @@ bool BytecodeParser::ParseVersionInfo(const unsigned char *&Buf, // having internal and external. break; default: - Error = "Unknown bytecode version number!"; - return true; + throw std::string("Unknown bytecode version number!"); } if (hasNoEndianness) Endianness = Module::AnyEndianness; @@ -549,66 +554,61 @@ bool BytecodeParser::ParseVersionInfo(const unsigned char *&Buf, BCR_TRACE(1, "Endianness/PointerSize = " << Endianness << "," << PointerSize << "\n"); BCR_TRACE(1, "HasImplicitZeroInit = " << HasImplicitZeroInitializer << "\n"); - return false; } -bool BytecodeParser::ParseModule(const unsigned char *Buf, +void BytecodeParser::ParseModule(const unsigned char *Buf, const unsigned char *EndBuf) { unsigned Type, Size; - if (readBlock(Buf, EndBuf, Type, Size)) return true; - if (Type != BytecodeFormat::Module || Buf+Size != EndBuf) { - Error = "Expected Module packet!"; - return true; // Hrm, not a class? - } + readBlock(Buf, EndBuf, Type, Size); + if (Type != BytecodeFormat::Module || Buf+Size != EndBuf) + throw std::string("Expected Module packet! B: "+ + utostr((unsigned)(intptr_t)Buf) + ", S: "+utostr(Size)+ + " E: "+utostr((unsigned)(intptr_t)EndBuf)); // Hrm, not a class? BCR_TRACE(0, "BLOCK BytecodeFormat::Module: {\n"); FunctionSignatureList.clear(); // Just in case... // Read into instance variables... - if (ParseVersionInfo(Buf, EndBuf)) return true; - if (align32(Buf, EndBuf)) return true; + ParseVersionInfo(Buf, EndBuf); + CHECK_ALIGN32(Buf, EndBuf); while (Buf < EndBuf) { const unsigned char *OldBuf = Buf; - if (readBlock(Buf, EndBuf, Type, Size)) return true; + readBlock(Buf, EndBuf, Type, Size); switch (Type) { case BytecodeFormat::GlobalTypePlane: BCR_TRACE(1, "BLOCK BytecodeFormat::GlobalTypePlane: {\n"); - if (ParseGlobalTypes(Buf, Buf+Size)) return true; + ParseGlobalTypes(Buf, Buf+Size); break; case BytecodeFormat::ModuleGlobalInfo: BCR_TRACE(1, "BLOCK BytecodeFormat::ModuleGlobalInfo: {\n"); - if (ParseModuleGlobalInfo(Buf, Buf+Size)) return true; + ParseModuleGlobalInfo(Buf, Buf+Size); break; case BytecodeFormat::ConstantPool: BCR_TRACE(1, "BLOCK BytecodeFormat::ConstantPool: {\n"); - if (ParseConstantPool(Buf, Buf+Size, ModuleValues, ModuleTypeValues)) - return true; + ParseConstantPool(Buf, Buf+Size, ModuleValues, ModuleTypeValues); break; case BytecodeFormat::Function: { BCR_TRACE(1, "BLOCK BytecodeFormat::Function: {\n"); - if (ParseFunction(Buf, Buf+Size)) - return true; // Error parsing function + ParseFunction(Buf, Buf+Size); break; } case BytecodeFormat::SymbolTable: BCR_TRACE(1, "BLOCK BytecodeFormat::SymbolTable: {\n"); - if (ParseSymbolTable(Buf, Buf+Size, &TheModule->getSymbolTable())) - return true; + ParseSymbolTable(Buf, Buf+Size, &TheModule->getSymbolTable()); break; default: - Error = "Expected Module Block!"; Buf += Size; - if (OldBuf > Buf) return true; // Wrap around! + if (OldBuf > Buf) throw std::string("Expected Module Block!"); break; } BCR_TRACE(1, "} end block\n"); - if (align32(Buf, EndBuf)) return true; + CHECK_ALIGN32(Buf, EndBuf); } // After the module constant pool has been read, we can safely initialize @@ -620,137 +620,36 @@ bool BytecodeParser::ParseModule(const unsigned char *Buf, // Look up the initializer value... if (Value *V = getValue(GV->getType()->getElementType(), Slot, false)) { - if (GV->hasInitializer()) return true; + if (GV->hasInitializer()) + throw std::string("Global *already* has an initializer?!"); GV->setInitializer(cast(V)); } else - return true; + throw std::string("Cannot find initializer value."); } - if (!FunctionSignatureList.empty()) { // Expected more functions! - Error = "Function expected, but bytecode stream at end!"; - return true; - } + if (!FunctionSignatureList.empty()) + throw std::string("Function expected, but bytecode stream ended!"); BCR_TRACE(0, "} end block\n\n"); - return false; } -static inline Module *Error(std::string *ErrorStr, const char *Message) { - if (ErrorStr) *ErrorStr = Message; - return 0; -} - -Module *BytecodeParser::ParseBytecode(const unsigned char *Buf, - const unsigned char *EndBuf, - const std::string &ModuleID) { +void +BytecodeParser::ParseBytecode(const unsigned char *Buf, unsigned Length, + const std::string &ModuleID) { unsigned Sig; + unsigned char *EndBuf = (unsigned char*)(Buf + Length); // Read and check signature... if (read(Buf, EndBuf, Sig) || - Sig != ('l' | ('l' << 8) | ('v' << 16) | 'm' << 24)) - return ::Error(&Error, "Invalid bytecode signature!"); + Sig != ('l' | ('l' << 8) | ('v' << 16) | ('m' << 24))) + throw std::string("Invalid bytecode signature!"); TheModule = new Module(ModuleID); - if (ParseModule(Buf, EndBuf)) { + try { + ParseModule(Buf, EndBuf); + } catch (std::string &Error) { freeState(); // Must destroy handles before deleting module! delete TheModule; TheModule = 0; + throw Error; } - return TheModule; -} - - -Module *ParseBytecodeBuffer(const unsigned char *Buffer, unsigned Length, - const std::string &ModuleID, std::string *ErrorStr){ - BytecodeParser Parser; - unsigned char *PtrToDelete = 0; - if ((intptr_t)Buffer & 3) { // If the buffer is not 4 byte aligned... - // Allocate a new buffer to hold the bytecode... - PtrToDelete = new unsigned char[Length+4]; - unsigned Offset = 4-((intptr_t)PtrToDelete & 3); // Make sure it's aligned - memcpy(PtrToDelete+Offset, Buffer, Length); // Copy it over - Buffer = PtrToDelete+Offset; - } - - Module *R = Parser.ParseBytecode(Buffer, Buffer+Length, ModuleID); - if (ErrorStr) *ErrorStr = Parser.getError(); - - delete [] PtrToDelete; // Delete alignment buffer if necessary - return R; -} - - -/// FDHandle - Simple handle class to make sure a file descriptor gets closed -/// when the object is destroyed. -class FDHandle { - int FD; -public: - FDHandle(int fd) : FD(fd) {} - operator int() const { return FD; } - ~FDHandle() { - if (FD != -1) close(FD); - } -}; - -// Parse and return a class file... -// -Module *ParseBytecodeFile(const std::string &Filename, std::string *ErrorStr) { - Module *Result = 0; - - if (Filename != std::string("-")) { // Read from a file... - FDHandle FD = open(Filename.c_str(), O_RDONLY); - if (FD == -1) - return Error(ErrorStr, "Error opening file!"); - - // Stat the file to get its length... - struct stat StatBuf; - if (fstat(FD, &StatBuf) == -1 || StatBuf.st_size == 0) - return Error(ErrorStr, "Error stat'ing file!"); - - // mmap in the file all at once... - int Length = StatBuf.st_size; - unsigned char *Buffer = (unsigned char*)mmap(0, Length, PROT_READ, - MAP_PRIVATE, FD, 0); - if (Buffer == (unsigned char*)MAP_FAILED) - return Error(ErrorStr, "Error mmapping file!"); - - // Parse the bytecode we mmapped in - Result = ParseBytecodeBuffer(Buffer, Length, Filename, ErrorStr); - - // Unmmap the bytecode... - munmap((char*)Buffer, Length); - } else { // Read from stdin - int BlockSize; - unsigned char Buffer[4096*4]; - std::vector FileData; - - // Read in all of the data from stdin, we cannot mmap stdin... - while ((BlockSize = read(0 /*stdin*/, Buffer, 4096*4))) { - if (BlockSize == -1) - return Error(ErrorStr, "Error reading from stdin!"); - - FileData.insert(FileData.end(), Buffer, Buffer+BlockSize); - } - - if (FileData.empty()) - return Error(ErrorStr, "Standard Input empty!"); - -#define ALIGN_PTRS 0 -#if ALIGN_PTRS - unsigned char *Buf = - (unsigned char*)mmap(0, FileData.size(), PROT_READ|PROT_WRITE, - MAP_PRIVATE|MAP_ANONYMOUS, -1, 0); - assert((Buf != (unsigned char*)-1) && "mmap returned error!"); - memcpy(Buf, &FileData[0], FileData.size()); -#else - unsigned char *Buf = &FileData[0]; -#endif - - Result = ParseBytecodeBuffer(Buf, FileData.size(), "", ErrorStr); - -#if ALIGN_PTRS - munmap((char*)Buf, FileData.size()); // Free mmap'd data area -#endif - } - - return Result; } diff --git a/lib/Bytecode/Reader/ReaderInternals.h b/lib/Bytecode/Reader/ReaderInternals.h index a64794a7018..bf3524d7c24 100644 --- a/lib/Bytecode/Reader/ReaderInternals.h +++ b/lib/Bytecode/Reader/ReaderInternals.h @@ -7,17 +7,20 @@ #ifndef READER_INTERNALS_H #define READER_INTERNALS_H -#include "llvm/Bytecode/Primitives.h" +#include "llvm/Constant.h" #include "llvm/DerivedTypes.h" #include "llvm/Function.h" -#include "llvm/Constant.h" +#include "llvm/ModuleProvider.h" +#include "llvm/Bytecode/Primitives.h" #include #include +#include +class Module; // Enable to trace to figure out what the heck is going on when parsing fails //#define TRACE_LEVEL 10 -#if TRACE_LEVEL // ByteCodeReading_TRACEer +#if TRACE_LEVEL // ByteCodeReading_TRACEr #define BCR_TRACE(n, X) \ if (n < TRACE_LEVEL) std::cerr << std::string(n*2, ' ') << X #else @@ -35,15 +38,21 @@ struct RawInst { // The raw fields out of the bytecode stream... }; }; -class BytecodeParser : public AbstractTypeUser { - std::string Error; // Error message string goes here... +struct LazyFunctionInfo { + const unsigned char *Buf, *EndBuf; + unsigned FunctionSlot; +}; + +class BytecodeParser : public AbstractTypeUser, public AbstractModuleProvider { + unsigned char *Buffer; BytecodeParser(const BytecodeParser &); // DO NOT IMPLEMENT void operator=(const BytecodeParser &); // DO NOT IMPLEMENT public: - BytecodeParser() { + BytecodeParser() : Buffer(0) { // Define this in case we don't see a ModuleGlobalInfo block. FirstDerivedTyID = Type::FirstDerivedTyID; } + ~BytecodeParser() { freeState(); } @@ -51,12 +60,21 @@ public: freeTable(Values); freeTable(LateResolveValues); freeTable(ModuleValues); + delete [] Buffer; + Buffer = 0; } - Module *ParseBytecode(const unsigned char *Buf, const unsigned char *EndBuf, - const std::string &ModuleID); + Module* releaseModule() { + // Since we're losing control of this Module, we must hand it back complete + materializeModule(); + freeState(); + Module *tempM = TheModule; + TheModule = 0; + return tempM; + } - std::string getError() const { return Error; } + void ParseBytecode(const unsigned char *Buf, unsigned Length, + const std::string &ModuleID); void dump() const { std::cerr << "BytecodeParser instance!\n"; @@ -80,8 +98,6 @@ private: // All of this data is transient across calls to ParseBytecode } }; - Module *TheModule; // Current Module being read into... - // Information about the module, extracted from the bytecode revision number. unsigned char RevisionNum; // The rev # itself unsigned char FirstDerivedTyID; // First variable index to use for type @@ -120,6 +136,12 @@ private: // All of this data is transient across calls to ParseBytecode // std::vector > GlobalInits; + // For lazy reading-in of functions, we need to save away several pieces of + // information about each function: its begin and end pointer in the buffer + // and its FunctionSlot. + // + std::map LazyFunctionLoadMap; + private: void freeTable(ValueTable &Tab) { while (!Tab.empty()) { @@ -128,27 +150,33 @@ private: } } - bool ParseModule (const unsigned char * Buf, const unsigned char *End); - bool ParseVersionInfo (const unsigned char *&Buf, const unsigned char *End); - bool ParseModuleGlobalInfo(const unsigned char *&Buf, const unsigned char *E); - bool ParseSymbolTable (const unsigned char *&Buf, const unsigned char *End, - SymbolTable *); - bool ParseFunction (const unsigned char *&Buf, const unsigned char *End); - bool ParseBasicBlock (const unsigned char *&Buf, const unsigned char *End, - BasicBlock *&); +public: + void ParseModule(const unsigned char * Buf, const unsigned char *End); + void materializeFunction(Function *F); + +private: + void ParseVersionInfo (const unsigned char *&Buf, const unsigned char *End); + void ParseModuleGlobalInfo(const unsigned char *&Buf, const unsigned char *E); + void ParseSymbolTable(const unsigned char *&Buf, const unsigned char *End, + SymbolTable *); + void ParseFunction(const unsigned char *&Buf, const unsigned char *End); + void ParseGlobalTypes(const unsigned char *&Buf, const unsigned char *EndBuf); + + std::auto_ptr + ParseBasicBlock(const unsigned char *&Buf, const unsigned char *End); + bool ParseInstruction (const unsigned char *&Buf, const unsigned char *End, Instruction *&); bool ParseRawInst (const unsigned char *&Buf, const unsigned char *End, RawInst &); - bool ParseGlobalTypes(const unsigned char *&Buf, const unsigned char *EndBuf); - bool ParseConstantPool(const unsigned char *&Buf, const unsigned char *EndBuf, - ValueTable &Tab, TypeValuesListTy &TypeTab); - bool parseConstantValue(const unsigned char *&Buf, const unsigned char *End, + void ParseConstantPool(const unsigned char *&Buf, const unsigned char *EndBuf, + ValueTable &Tab, TypeValuesListTy &TypeTab); + void parseConstantValue(const unsigned char *&Buf, const unsigned char *End, const Type *Ty, Constant *&V); - bool parseTypeConstants(const unsigned char *&Buf, + void parseTypeConstants(const unsigned char *&Buf, const unsigned char *EndBuf, - TypeValuesListTy &Tab, unsigned NumEntries); + TypeValuesListTy &Tab, unsigned NumEntries); const Type *parseTypeConstant(const unsigned char *&Buf, const unsigned char *EndBuf); @@ -158,9 +186,9 @@ private: int insertValue(Value *V, ValueTable &Table); // -1 = Failure void setValueTo(ValueTable &D, unsigned Slot, Value *V); - bool postResolveValues(ValueTable &ValTab); + void postResolveValues(ValueTable &ValTab); - bool getTypeSlot(const Type *Ty, unsigned &Slot); + void getTypeSlot(const Type *Ty, unsigned &Slot); // resolve all references to the placeholder (if any) for the given value void ResolveReferencesToValue(Value *Val, unsigned Slot); @@ -205,6 +233,11 @@ typedef PlaceholderDef ValPHolder; typedef PlaceholderDef BBPHolder; typedef PlaceholderDef ConstPHolder; +// Some common errors we find +static const std::string Error_readvbr = "read_vbr(): error reading."; +static const std::string Error_read = "read(): error reading."; +static const std::string Error_inputdata = "input_data(): error reading."; +static const std::string Error_DestSlot = "No destination slot found."; static inline unsigned getValueIDNumberFromPlaceHolder(Value *Val) { if (isa(Val)) @@ -217,16 +250,16 @@ static inline unsigned getValueIDNumberFromPlaceHolder(Value *Val) { } } -static inline bool readBlock(const unsigned char *&Buf, +static inline void readBlock(const unsigned char *&Buf, const unsigned char *EndBuf, - unsigned &Type, unsigned &Size) { + unsigned &Type, unsigned &Size) { #if DEBUG_OUTPUT bool Result = read(Buf, EndBuf, Type) || read(Buf, EndBuf, Size); std::cerr << "StartLoc = " << ((unsigned)Buf & 4095) << " Type = " << Type << " Size = " << Size << endl; - return Result; + if (Result) throw Error_read; #else - return read(Buf, EndBuf, Type) || read(Buf, EndBuf, Size); + if (read(Buf, EndBuf, Type) || read(Buf, EndBuf, Size)) throw Error_read; #endif } diff --git a/lib/Bytecode/Reader/ReaderWrappers.cpp b/lib/Bytecode/Reader/ReaderWrappers.cpp new file mode 100644 index 00000000000..218c694f2e4 --- /dev/null +++ b/lib/Bytecode/Reader/ReaderWrappers.cpp @@ -0,0 +1,170 @@ +#include "ReaderInternals.h" +#include "Support/StringExtras.h" +#include "Config/fcntl.h" +#include "Config/unistd.h" +#include "Config/sys/mman.h" + +#define CHECK_ALIGN32(begin,end) \ + if (align32(begin,end)) \ + throw std::string("Alignment error: ReaderWrappers.cpp:" + \ + utostr((unsigned)__LINE__)); + +namespace { + + /// BytecodeFileReader - parses a bytecode file from a file + /// + class BytecodeFileReader : public BytecodeParser { + private: + unsigned char *Buffer; + int Length; + + BytecodeFileReader(const BytecodeFileReader&); // Do not implement + void operator=(BytecodeFileReader &BFR); // Do not implement + + public: + BytecodeFileReader(const std::string &Filename); + ~BytecodeFileReader(); + + }; + + /// BytecodeStdinReader - parses a bytecode file from stdin + /// + class BytecodeStdinReader : public BytecodeParser { + private: + std::vector FileData; + unsigned char *FileBuf; + + BytecodeStdinReader(const BytecodeStdinReader&); // Do not implement + void operator=(BytecodeStdinReader &BFR); // Do not implement + + public: + BytecodeStdinReader(); + ~BytecodeStdinReader(); + }; + + /// FDHandle - Simple handle class to make sure a file descriptor gets closed + /// when the object is destroyed. + /// + class FDHandle { + int FD; + public: + FDHandle(int fd) : FD(fd) {} + operator int() const { return FD; } + ~FDHandle() { + if (FD != -1) close(FD); + } + }; +} + +BytecodeFileReader::BytecodeFileReader(const std::string &Filename) { + FDHandle FD = open(Filename.c_str(), O_RDONLY); + if (FD == -1) + throw std::string("Error opening file!"); + + // Stat the file to get its length... + struct stat StatBuf; + if (fstat(FD, &StatBuf) == -1 || StatBuf.st_size == 0) + throw std::string("Error stat'ing file!"); + + // mmap in the file all at once... + Length = StatBuf.st_size; + unsigned char *Buffer = (unsigned char*)mmap(0, Length, PROT_READ, + MAP_PRIVATE, FD, 0); + if (Buffer == (unsigned char*)MAP_FAILED) + throw std::string("Error mmapping file!"); + + // Parse the bytecode we mmapped in + ParseBytecode(Buffer, Length, Filename); +} + +BytecodeFileReader::~BytecodeFileReader() { + // Unmmap the bytecode... + munmap((char*)Buffer, Length); +} + + +#define ALIGN_PTRS 0 + +BytecodeStdinReader::BytecodeStdinReader() { + int BlockSize; + unsigned char Buffer[4096*4]; + + // Read in all of the data from stdin, we cannot mmap stdin... + while ((BlockSize = read(0 /*stdin*/, Buffer, 4096*4))) { + if (BlockSize == -1) + throw std::string("Error reading from stdin!"); + + FileData.insert(FileData.end(), Buffer, Buffer+BlockSize); + } + + if (FileData.empty()) + throw std::string("Standard Input empty!"); + +#if ALIGN_PTRS + FileBuf = (unsigned char*)mmap(0, FileData.size(), PROT_READ|PROT_WRITE, + MAP_PRIVATE|MAP_ANONYMOUS, -1, 0); + assert((Buf != (unsigned char*)-1) && "mmap returned error!"); + memcpy(Buf, &FileData[0], FileData.size()); +#else + FileBuf = &FileData[0]; +#endif + +#if 0 + // Allocate a new buffer to hold the bytecode... + unsigned char *ParseBegin=0; + unsigned Offset=0; + if ((intptr_t)Buffer & 3) { + delete [] Buffer; + Buffer = new unsigned char[Length+4]; + Offset = 4-((intptr_t)Buffer & 3); // Make sure it's aligned + } + memcpy(Buffer+Offset, Buf, Length); // Copy it over + ParseBegin = Buffer+Offset; +#endif + + ParseBytecode(FileBuf, FileData.size(), ""); +} + +BytecodeStdinReader::~BytecodeStdinReader() { +#if ALIGN_PTRS + munmap((char*)FileBuf, FileData.size()); // Free mmap'd data area +#endif +} + +/// +/// +AbstractModuleProvider* +getBytecodeBufferModuleProvider(const unsigned char *Buffer, unsigned Length, + const std::string &ModuleID) { + CHECK_ALIGN32(Buffer, Buffer+Length); + BytecodeParser *Parser = new BytecodeParser(); + Parser->ParseBytecode(Buffer, Length, ModuleID); + return Parser; +} + +Module *ParseBytecodeBuffer(const unsigned char *Buffer, unsigned Length, + const std::string &ModuleID, std::string *ErrorStr){ + AbstractModuleProvider *AMP = + getBytecodeBufferModuleProvider(Buffer, Length, ModuleID); + Module *M = AMP->releaseModule(); + delete AMP; + return M; +} + + +/// Parse and return a class file... +/// +AbstractModuleProvider* +getBytecodeModuleProvider(const std::string &Filename) { + if (Filename != std::string("-")) // Read from a file... + return new BytecodeFileReader(Filename); + else // Read from stdin + return new BytecodeStdinReader(); +} + +Module *ParseBytecodeFile(const std::string &Filename, std::string *ErrorStr) { + AbstractModuleProvider *AMP = getBytecodeModuleProvider(Filename); + Module *M = AMP->releaseModule(); + delete AMP; + return M; +}