-//===- Reader.cpp - Code to read bytecode files -----------------------------===
+//===- Reader.cpp - Code to read bytecode files ---------------------------===//
//
// This library implements the functionality defined in llvm/Bytecode/Reader.h
//
// TODO: Make error message outputs be configurable depending on an option?
// TODO: Allow passing in an option to ignore the symbol table
//
-//===------------------------------------------------------------------------===
+//===----------------------------------------------------------------------===//
+#include "ReaderInternals.h"
#include "llvm/Bytecode/Reader.h"
#include "llvm/Bytecode/Format.h"
#include "llvm/GlobalVariable.h"
#include "llvm/Module.h"
#include "llvm/BasicBlock.h"
-#include "llvm/DerivedTypes.h"
-#include "llvm/ConstPoolVals.h"
+#include "llvm/ConstantVals.h"
+#include "llvm/iPHINode.h"
#include "llvm/iOther.h"
-#include "ReaderInternals.h"
#include <sys/types.h>
-#include <sys/mman.h>
#include <sys/stat.h>
+#include <sys/mman.h>
#include <fcntl.h>
#include <unistd.h>
#include <algorithm>
+#include <iostream>
+using std::cerr;
+using std::make_pair;
bool BytecodeParser::getTypeSlot(const Type *Ty, unsigned &Slot) {
if (Ty->isPrimitiveType()) {
Slot = FirstDerivedTyID + (&*I - &ModuleTypeValues[0]);
}
}
- //cerr << "getTypeSlot '" << Ty->getName() << "' = " << Slot << endl;
+ //cerr << "getTypeSlot '" << Ty->getName() << "' = " << Slot << "\n";
return false;
}
const Type *T = Type::getPrimitiveType((Type::PrimitiveID)ID);
if (T) return T;
- //cerr << "Looking up Type ID: " << ID << endl;
+ //cerr << "Looking up Type ID: " << ID << "\n";
const Value *D = getValue(Type::TypeTy, ID, false);
if (D == 0) return failure<const Type*>(0);
return cast<Type>(D);
}
-bool BytecodeParser::insertValue(Value *Val, vector<ValueList> &ValueTab) {
+int BytecodeParser::insertValue(Value *Val, std::vector<ValueList> &ValueTab) {
unsigned type;
- if (getTypeSlot(Val->getType(), type)) return failure(true);
+ if (getTypeSlot(Val->getType(), type)) return failure<int>(-1);
assert(type != Type::TypeTyID && "Types should never be insertValue'd!");
if (ValueTab.size() <= type)
ValueTab.resize(type+1, ValueList());
//cerr << "insertValue Values[" << type << "][" << ValueTab[type].size()
- // << "] = " << Val << endl;
+ // << "] = " << Val << "\n";
ValueTab[type].push_back(Val);
- return false;
+ return ValueTab[type].size()-1;
}
Value *BytecodeParser::getValue(const Type *Ty, unsigned oNum, bool Create) {
// Is it a module level type?
if (Num < ModuleTypeValues.size())
- return (Value*)(const Type*)ModuleTypeValues[Num];
+ return (Value*)ModuleTypeValues[Num].get();
// Nope, is it a method level type?
Num -= ModuleTypeValues.size();
if (Num < MethodTypeValues.size())
- return (Value*)(const Type*)MethodTypeValues[Num];
+ return (Value*)MethodTypeValues[Num].get();
return 0;
}
case Type::LabelTyID: d = new BBPHolder(Ty, oNum); break;
case Type::MethodTyID:
cerr << "Creating method pholder! : " << type << ":" << oNum << " "
- << Ty->getName() << endl;
+ << Ty->getName() << "\n";
d = new MethPHolder(Ty, oNum);
- insertValue(d, LateResolveModuleValues);
+ if (insertValue(d, LateResolveModuleValues) ==-1) return failure<Value*>(0);
return d;
default: d = new DefPHolder(Ty, oNum); break;
}
assert(d != 0 && "How did we not make something?");
- if (insertValue(d, LateResolveValues)) return failure<Value*>(0);
+ if (insertValue(d, LateResolveValues) == -1) return failure<Value*>(0);
return d;
}
Value *NewDef = getValue(D->getType(), IDNumber, false);
if (NewDef == 0) {
Error = true; // Unresolved thinger
- cerr << "Unresolvable reference found: <" << D->getType()->getName()
- << ">:" << IDNumber << "!\n";
+ cerr << "Unresolvable reference found: <"
+ << D->getType()->getDescription() << ">:" << IDNumber << "!\n";
} else {
// Fixup all of the uses of this placeholder def...
D->replaceAllUsesWith(NewDef);
}
if (Inst == 0) { delete BB; return failure(true); }
- if (insertValue(Inst, Values)) { delete BB; return failure(true); }
+ if (insertValue(Inst, Values) == -1) { delete BB; return failure(true); }
BB->getInstList().push_back(Inst);
// Symtab entry: [def slot #][name]
unsigned slot;
if (read_vbr(Buf, EndBuf, slot)) return failure(true);
- string Name;
+ std::string Name;
if (read(Buf, EndBuf, Name, false)) // Not aligned...
return failure(true);
Value *D = getValue(Ty, slot, false); // Find mapping...
if (D == 0) {
- BCR_TRACE(3, "FAILED LOOKUP: Slot #" << slot << endl);
+ BCR_TRACE(3, "FAILED LOOKUP: Slot #" << slot << "\n");
return failure(true);
}
BCR_TRACE(4, "Map: '" << Name << "' to #" << slot << ":" << D;
- if (!isa<Instruction>(D)) cerr << endl);
+ if (!isa<Instruction>(D)) cerr << "\n");
D->setName(Name, ST);
}
return false;
}
+// DeclareNewGlobalValue - Patch up forward references to global values in the
+// form of ConstantPointerRef.
+//
+void BytecodeParser::DeclareNewGlobalValue(GlobalValue *GV, unsigned Slot) {
+ // Check to see if there is a forward reference to this global variable...
+ // if there is, eliminate it and patch the reference to use the new def'n.
+ GlobalRefsType::iterator I = GlobalRefs.find(make_pair(GV->getType(), Slot));
+
+ if (I != GlobalRefs.end()) {
+ GlobalVariable *OldGV = I->second; // Get the placeholder...
+ BCR_TRACE(3, "Mutating CPPR Forward Ref!\n");
+
+ // Loop over all of the uses of the GlobalValue. The only thing they are
+ // allowed to be at this point is ConstantPointerRef's.
+ assert(OldGV->use_size() == 1 && "Only one reference should exist!");
+ while (!OldGV->use_empty()) {
+ User *U = OldGV->use_back(); // Must be a ConstantPointerRef...
+ ConstantPointerRef *CPPR = cast<ConstantPointerRef>(U);
+ assert(CPPR->getValue() == OldGV && "Something isn't happy");
+
+ BCR_TRACE(4, "Mutating Forward Ref!\n");
+
+ // Change the const pool reference to point to the real global variable
+ // now. This should drop a use from the OldGV.
+ CPPR->mutateReference(GV);
+ }
+
+ // Remove GV from the module...
+ GV->getParent()->getGlobalList().remove(OldGV);
+ delete OldGV; // Delete the old placeholder
+
+ // Remove the map entry for the global now that it has been created...
+ GlobalRefs.erase(I);
+ }
+}
bool BytecodeParser::ParseMethod(const uchar *&Buf, const uchar *EndBuf,
Module *C) {
// Clear out the local values table...
Values.clear();
- if (MethodSignatureList.empty()) return failure(true); // Unexpected method!
+ if (MethodSignatureList.empty()) {
+ Error = "Method found, but MethodSignatureList empty!";
+ return failure(true); // Unexpected method!
+ }
+
+ const PointerType *PMTy = MethodSignatureList.front().first; // PtrMeth
+ const MethodType *MTy = dyn_cast<const MethodType>(PMTy->getElementType());
+ if (MTy == 0) return failure(true); // Not ptr to method!
+
+ unsigned isInternal;
+ if (read_vbr(Buf, EndBuf, isInternal)) return failure(true);
- const MethodType *MTy = MethodSignatureList.front().first;
unsigned MethSlot = MethodSignatureList.front().second;
MethodSignatureList.pop_front();
- Method *M = new Method(MTy);
+ Method *M = new Method(MTy, isInternal != 0);
- BCR_TRACE(2, "METHOD TYPE: " << MTy << endl);
+ BCR_TRACE(2, "METHOD TYPE: " << MTy << "\n");
const MethodType::ParamTypes &Params = MTy->getParamTypes();
for (MethodType::ParamTypes::const_iterator It = Params.begin();
It != Params.end(); ++It) {
- MethodArgument *MA = new MethodArgument(*It);
- if (insertValue(MA, Values)) { delete M; return failure(true); }
- M->getArgumentList().push_back(MA);
+ FunctionArgument *FA = new FunctionArgument(*It);
+ if (insertValue(FA, Values) == -1) {
+ Error = "Error reading method arguments!\n";
+ delete M; return failure(true);
+ }
+ M->getArgumentList().push_back(FA);
}
while (Buf < EndBuf) {
unsigned Type, Size;
const uchar *OldBuf = Buf;
- if (readBlock(Buf, EndBuf, Type, Size)) { delete M; return failure(true); }
+ if (readBlock(Buf, EndBuf, Type, Size)) {
+ Error = "Error reading Method level block!";
+ delete M; return failure(true);
+ }
switch (Type) {
case BytecodeFormat::ConstantPool:
BCR_TRACE(2, "BLOCK BytecodeFormat::BasicBlock: {\n");
BasicBlock *BB;
if (ParseBasicBlock(Buf, Buf+Size, BB) ||
- insertValue(BB, Values)) {
+ insertValue(BB, Values) == -1) {
delete M; return failure(true); // Parse error... :(
}
BCR_TRACE(2, "} end block\n");
if (align32(Buf, EndBuf)) {
+ Error = "Error aligning Method level block!";
delete M; // Malformed bc file, read past end of block.
return failure(true);
}
if (postResolveValues(LateResolveValues) ||
postResolveValues(LateResolveModuleValues)) {
+ Error = "Error resolving method values!";
delete M; return failure(true); // Unresolvable references!
}
- Value *MethPHolder = getValue(MTy, MethSlot, false);
+ Value *MethPHolder = getValue(PMTy, MethSlot, false);
assert(MethPHolder && "Something is broken no placeholder found!");
assert(isa<Method>(MethPHolder) && "Not a method?");
unsigned type; // Type slot
assert(!getTypeSlot(MTy, type) && "How can meth type not exist?");
- getTypeSlot(MTy, type);
+ getTypeSlot(PMTy, type);
- C->getMethodList().push_back(M);
+ C->getFunctionList().push_back(M);
// Replace placeholder with the real method pointer...
ModuleValues[type][MethSlot] = M;
// We don't need the placeholder anymore!
delete MethPHolder;
+ // If the method is empty, we don't need the method argument entries...
+ if (M->isExternal())
+ M->getArgumentList().delete_all();
+
+ DeclareNewGlobalValue(M, MethSlot);
+
return false;
}
bool BytecodeParser::ParseModuleGlobalInfo(const uchar *&Buf, const uchar *End,
- Module *C) {
- if (!MethodSignatureList.empty())
+ Module *Mod) {
+ if (!MethodSignatureList.empty()) {
+ Error = "Two ModuleGlobalInfo packets found!";
return failure(true); // Two ModuleGlobal blocks?
+ }
// Read global variables...
unsigned VarType;
if (read_vbr(Buf, End, VarType)) return failure(true);
while (VarType != Type::VoidTyID) { // List is terminated by Void
- // VarType Fields: bit0 = isConstant, bit1 = hasInitializer, bit2+ = slot#
- const Type *Ty = getType(VarType >> 2);
+ // VarType Fields: bit0 = isConstant, bit1 = hasInitializer,
+ // bit2 = isInternal, bit3+ = slot#
+ const Type *Ty = getType(VarType >> 3);
if (!Ty || !Ty->isPointerType()) {
- cerr << "Global not pointer type! Ty = " << Ty << endl;
+ Error = "Global not pointer type! Ty = " + Ty->getDescription();
return failure(true);
}
- ConstPoolVal *Initializer = 0;
+ const PointerType *PTy = cast<const PointerType>(Ty);
+ const Type *ElTy = PTy->getElementType();
+
+ Constant *Initializer = 0;
if (VarType & 2) { // Does it have an initalizer?
// Do not improvise... values must have been stored in the constant pool,
// which should have been read before now.
unsigned InitSlot;
if (read_vbr(Buf, End, InitSlot)) return failure(true);
- Value *V = getValue(Ty->castPointerType()->getValueType(),
- InitSlot, false);
+ Value *V = getValue(ElTy, InitSlot, false);
if (V == 0) return failure(true);
- Initializer = cast<ConstPoolVal>(V);
+ Initializer = cast<Constant>(V);
}
// Create the global variable...
- GlobalVariable *GV = new GlobalVariable(Ty, VarType & 1, Initializer);
- insertValue(GV, ModuleValues);
- C->getGlobalList().push_back(GV);
+ GlobalVariable *GV = new GlobalVariable(ElTy, VarType & 1, VarType & 4,
+ Initializer);
+ int DestSlot = insertValue(GV, ModuleValues);
+ if (DestSlot == -1) return failure(true);
+
+ Mod->getGlobalList().push_back(GV);
+
+ DeclareNewGlobalValue(GV, unsigned(DestSlot));
+
+ BCR_TRACE(2, "Global Variable of type: " << PTy->getDescription()
+ << " into slot #" << DestSlot << "\n");
if (read_vbr(Buf, End, VarType)) return failure(true);
- BCR_TRACE(2, "Global Variable of type: " << Ty->getDescription() << endl);
}
// Read the method signatures for all of the methods that are coming, and
if (read_vbr(Buf, End, MethSignature)) return failure(true);
while (MethSignature != Type::VoidTyID) { // List is terminated by Void
const Type *Ty = getType(MethSignature);
- if (!Ty || !isa<MethodType>(Ty)) {
- cerr << "Method not meth type! Ty = " << Ty << endl;
+ if (!Ty || !isa<PointerType>(Ty) ||
+ !isa<MethodType>(cast<PointerType>(Ty)->getElementType())) {
+ Error = "Method not ptr to meth type! Ty = " + Ty->getDescription();
return failure(true);
}
+
+ // We create methods by passing the underlying MethodType to create...
+ Ty = cast<PointerType>(Ty)->getElementType();
// When the ModuleGlobalInfo section is read, we load the type of each
// method and the 'ModuleValues' slot that it lands in. We then load a
// placeholder is replaced.
// Insert the placeholder...
- Value *Def = new MethPHolder(Ty, 0);
- insertValue(Def, ModuleValues);
+ Value *Val = new MethPHolder(Ty, 0);
+ if (insertValue(Val, ModuleValues) == -1) return failure(true);
// Figure out which entry of its typeslot it went into...
unsigned TypeSlot;
- if (getTypeSlot(Def->getType(), TypeSlot)) return failure(true);
+ if (getTypeSlot(Val->getType(), TypeSlot)) return failure(true);
unsigned SlotNo = ModuleValues[TypeSlot].size()-1;
// Keep track of this information in a linked list that is emptied as
// methods are loaded...
//
- MethodSignatureList.push_back(make_pair((const MethodType*)Ty, SlotNo));
+ MethodSignatureList.push_back(
+ make_pair(cast<const PointerType>(Val->getType()), SlotNo));
if (read_vbr(Buf, End, MethSignature)) return failure(true);
- BCR_TRACE(2, "Method of type: " << Ty << endl);
+ BCR_TRACE(2, "Method of type: " << Ty << "\n");
}
if (align32(Buf, End)) return failure(true);
unsigned Type, Size;
if (readBlock(Buf, EndBuf, Type, Size)) return failure(true);
- if (Type != BytecodeFormat::Module || Buf+Size != EndBuf)
+ if (Type != BytecodeFormat::Module || Buf+Size != EndBuf) {
+ Error = "Expected Module packet!";
return failure(true); // Hrm, not a class?
+ }
BCR_TRACE(0, "BLOCK BytecodeFormat::Module: {\n");
MethodSignatureList.clear(); // Just in case...
if (align32(Buf, EndBuf)) return failure(true);
BCR_TRACE(1, "FirstDerivedTyID = " << FirstDerivedTyID << "\n");
- C = new Module();
+ TheModule = C = new Module();
while (Buf < EndBuf) {
const uchar *OldBuf = Buf;
if (readBlock(Buf, EndBuf, Type, Size)) { delete C; return failure(true); }
break;
default:
- cerr << " Unknown class block: " << Type << endl;
+ Error = "Expected Module Block!";
Buf += Size;
if (OldBuf > Buf) return failure(true); // Wrap around!
break;
if (align32(Buf, EndBuf)) { delete C; return failure(true); }
}
- if (!MethodSignatureList.empty()) // Expected more methods!
+ if (!MethodSignatureList.empty()) { // Expected more methods!
+ Error = "Method expected, but bytecode stream at end!";
return failure(true);
+ }
BCR_TRACE(0, "} end block\n\n");
return false;
unsigned Sig;
// Read and check signature...
if (read(Buf, EndBuf, Sig) ||
- Sig != ('l' | ('l' << 8) | ('v' << 16) | 'm' << 24))
+ Sig != ('l' | ('l' << 8) | ('v' << 16) | 'm' << 24)) {
+ Error = "Invalid bytecode signature!";
return failure<Module*>(0); // Invalid signature!
+ }
Module *Result;
if (ParseModule(Buf, EndBuf, Result)) return 0;
// Parse and return a class file...
//
-Module *ParseBytecodeFile(const string &Filename) {
+Module *ParseBytecodeFile(const std::string &Filename, std::string *ErrorStr) {
struct stat StatBuf;
Module *Result = 0;
- if (Filename != string("-")) { // Read from a file...
+ if (Filename != std::string("-")) { // Read from a file...
int FD = open(Filename.c_str(), O_RDONLY);
- if (FD == -1) return failure<Module*>(0);
+ if (FD == -1) {
+ if (ErrorStr) *ErrorStr = "Error opening file!";
+ return failure<Module*>(0);
+ }
if (fstat(FD, &StatBuf) == -1) { close(FD); return failure<Module*>(0); }
int Length = StatBuf.st_size;
- if (Length == 0) { close(FD); return failure<Module*>(0); }
+ if (Length == 0) {
+ if (ErrorStr) *ErrorStr = "Error stat'ing file!";
+ close(FD); return failure<Module*>(0);
+ }
uchar *Buffer = (uchar*)mmap(0, Length, PROT_READ,
MAP_PRIVATE, FD, 0);
- if (Buffer == (uchar*)-1) { close(FD); return failure<Module*>(0); }
+ if (Buffer == (uchar*)-1) {
+ if (ErrorStr) *ErrorStr = "Error mmapping file!";
+ close(FD); return failure<Module*>(0);
+ }
BytecodeParser Parser;
Result = Parser.ParseBytecode(Buffer, Buffer+Length);
munmap((char*)Buffer, Length);
close(FD);
+ if (ErrorStr) *ErrorStr = Parser.getError();
} else { // Read from stdin
size_t FileSize = 0;
int BlockSize;
FileSize += BlockSize;
}
- if (FileSize == 0) { free(FileData); return failure<Module*>(0); }
+ if (FileSize == 0) {
+ if (ErrorStr) *ErrorStr = "Standard Input empty!";
+ free(FileData); return failure<Module*>(0);
+ }
#define ALIGN_PTRS 1
#if ALIGN_PTRS
uchar *Buf = (uchar*)mmap(0, FileSize, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
assert((Buf != (uchar*)-1) && "mmap returned error!");
- free(FileData);
memcpy(Buf, FileData, FileSize);
+ free(FileData);
#else
uchar *Buf = FileData;
#endif
#else
free(FileData); // Free realloc'd block of memory
#endif
+
+ if (ErrorStr) *ErrorStr = Parser.getError();
}
return Result;