blob: e649cdc08f142e6e14d4c6b23d1fdb0981a413aa [file] [log] [blame]
Chris Lattnerd6b65252001-10-24 01:15:12 +00001//===- Reader.cpp - Code to read bytecode files ---------------------------===//
Chris Lattner00950542001-06-06 20:29:01 +00002//
3// This library implements the functionality defined in llvm/Bytecode/Reader.h
4//
5// Note that this library should be as fast as possible, reentrant, and
6// threadsafe!!
7//
Chris Lattner74734132002-08-17 22:01:27 +00008// TODO: Return error messages to caller instead of printing them out directly.
Chris Lattner00950542001-06-06 20:29:01 +00009// TODO: Allow passing in an option to ignore the symbol table
10//
Chris Lattnerd6b65252001-10-24 01:15:12 +000011//===----------------------------------------------------------------------===//
Chris Lattner00950542001-06-06 20:29:01 +000012
Chris Lattner7061dc52001-12-03 18:02:31 +000013#include "ReaderInternals.h"
Chris Lattner00950542001-06-06 20:29:01 +000014#include "llvm/Bytecode/Reader.h"
15#include "llvm/Bytecode/Format.h"
Chris Lattner31bcdb82002-04-28 19:55:58 +000016#include "llvm/Constants.h"
Chris Lattner7061dc52001-12-03 18:02:31 +000017#include "llvm/iPHINode.h"
Chris Lattner00950542001-06-06 20:29:01 +000018#include "llvm/iOther.h"
Misha Brukman12c29d12003-09-22 23:38:23 +000019#include "llvm/Module.h"
20#include "Support/StringExtras.h"
John Criswell7a73b802003-06-30 21:59:07 +000021#include "Config/unistd.h"
Misha Brukman12c29d12003-09-22 23:38:23 +000022#include "Config/sys/mman.h"
23#include "Config/sys/stat.h"
24#include "Config/sys/types.h"
Chris Lattner00950542001-06-06 20:29:01 +000025#include <algorithm>
Misha Brukman12c29d12003-09-22 23:38:23 +000026#include <memory>
Chris Lattner00950542001-06-06 20:29:01 +000027
Misha Brukmane0dd0d42003-09-23 16:15:29 +000028static inline void ALIGN32(const unsigned char *&begin,
29 const unsigned char *end) {
30 if (align32(begin, end))
31 throw std::string("Alignment error in buffer: read past end of block.");
32}
Misha Brukman12c29d12003-09-22 23:38:23 +000033
Chris Lattner9e460f22003-10-04 20:00:03 +000034unsigned BytecodeParser::getTypeSlot(const Type *Ty) {
35 if (Ty->isPrimitiveType())
36 return Ty->getPrimitiveID();
37
38 // Check the function level types first...
39 TypeValuesListTy::iterator I = find(FunctionTypeValues.begin(),
40 FunctionTypeValues.end(), Ty);
41 if (I != FunctionTypeValues.end())
42 return FirstDerivedTyID + ModuleTypeValues.size() +
43 (&*I - &FunctionTypeValues[0]);
44
45 I = find(ModuleTypeValues.begin(), ModuleTypeValues.end(), Ty);
46 if (I == ModuleTypeValues.end())
47 throw std::string("Didn't find type in ModuleTypeValues.");
48 return FirstDerivedTyID + (&*I - &ModuleTypeValues[0]);
Chris Lattner00950542001-06-06 20:29:01 +000049}
50
51const Type *BytecodeParser::getType(unsigned ID) {
Chris Lattner927b1852003-10-09 20:22:47 +000052 if (ID < Type::NumPrimitiveIDs)
53 if (const Type *T = Type::getPrimitiveType((Type::PrimitiveID)ID))
54 return T;
Chris Lattner00950542001-06-06 20:29:01 +000055
Chris Lattner697954c2002-01-20 22:54:45 +000056 //cerr << "Looking up Type ID: " << ID << "\n";
Chris Lattner36392bc2003-10-08 21:18:57 +000057
Chris Lattner927b1852003-10-09 20:22:47 +000058 if (ID < Type::NumPrimitiveIDs)
59 if (const Type *T = Type::getPrimitiveType((Type::PrimitiveID)ID))
60 return T; // Asked for a primitive type...
Chris Lattner36392bc2003-10-08 21:18:57 +000061
62 // Otherwise, derived types need offset...
63 ID -= FirstDerivedTyID;
64
65 // Is it a module-level type?
66 if (ID < ModuleTypeValues.size())
67 return ModuleTypeValues[ID].get();
68
69 // Nope, is it a function-level type?
70 ID -= ModuleTypeValues.size();
71 if (ID < FunctionTypeValues.size())
72 return FunctionTypeValues[ID].get();
73
Chris Lattner927b1852003-10-09 20:22:47 +000074 throw std::string("Illegal type reference!");
Chris Lattner00950542001-06-06 20:29:01 +000075}
76
Chris Lattner927b1852003-10-09 20:22:47 +000077unsigned BytecodeParser::insertValue(Value *Val, ValueTable &ValueTab) {
Chris Lattner52e20b02003-03-19 20:54:26 +000078 assert((!HasImplicitZeroInitializer || !isa<Constant>(Val) ||
79 Val->getType()->isPrimitiveType() ||
80 !cast<Constant>(Val)->isNullValue()) &&
81 "Cannot read null values from bytecode!");
Chris Lattner9e460f22003-10-04 20:00:03 +000082 unsigned type = getTypeSlot(Val->getType());
Chris Lattner1d670cc2001-09-07 16:37:43 +000083 assert(type != Type::TypeTyID && "Types should never be insertValue'd!");
Chris Lattner00950542001-06-06 20:29:01 +000084
Chris Lattner52e20b02003-03-19 20:54:26 +000085 if (ValueTab.size() <= type) {
86 unsigned OldSize = ValueTab.size();
87 ValueTab.resize(type+1);
88 while (OldSize != type+1)
89 ValueTab[OldSize++] = new ValueList();
Chris Lattner036b8aa2003-03-06 17:55:45 +000090 }
Chris Lattner00950542001-06-06 20:29:01 +000091
92 //cerr << "insertValue Values[" << type << "][" << ValueTab[type].size()
Misha Brukman12c29d12003-09-22 23:38:23 +000093 // << "] = " << Val << "\n";
Chris Lattner52e20b02003-03-19 20:54:26 +000094 ValueTab[type]->push_back(Val);
Chris Lattner00950542001-06-06 20:29:01 +000095
Chris Lattner52e20b02003-03-19 20:54:26 +000096 bool HasOffset = HasImplicitZeroInitializer &&
Misha Brukman12c29d12003-09-22 23:38:23 +000097 !Val->getType()->isPrimitiveType();
Chris Lattner52e20b02003-03-19 20:54:26 +000098
99 return ValueTab[type]->size()-1 + HasOffset;
100}
101
102
Chris Lattner00950542001-06-06 20:29:01 +0000103Value *BytecodeParser::getValue(const Type *Ty, unsigned oNum, bool Create) {
Chris Lattner36392bc2003-10-08 21:18:57 +0000104 return getValue(getTypeSlot(Ty), oNum, Create);
105}
106
107Value *BytecodeParser::getValue(unsigned type, unsigned oNum, bool Create) {
108 assert(type != Type::TypeTyID && "getValue() cannot get types!");
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000109 assert(type != Type::LabelTyID && "getValue() cannot get blocks!");
Chris Lattner00950542001-06-06 20:29:01 +0000110 unsigned Num = oNum;
Chris Lattner00950542001-06-06 20:29:01 +0000111
Chris Lattner52e20b02003-03-19 20:54:26 +0000112 if (HasImplicitZeroInitializer && type >= FirstDerivedTyID) {
113 if (Num == 0)
Chris Lattner36392bc2003-10-08 21:18:57 +0000114 return Constant::getNullValue(getType(type));
Chris Lattner52e20b02003-03-19 20:54:26 +0000115 --Num;
Chris Lattner00950542001-06-06 20:29:01 +0000116 }
117
Chris Lattner52e20b02003-03-19 20:54:26 +0000118 if (type < ModuleValues.size()) {
119 if (Num < ModuleValues[type]->size())
120 return ModuleValues[type]->getOperand(Num);
121 Num -= ModuleValues[type]->size();
122 }
123
124 if (Values.size() > type && Values[type]->size() > Num)
125 return Values[type]->getOperand(Num);
Chris Lattner00950542001-06-06 20:29:01 +0000126
Chris Lattner74734132002-08-17 22:01:27 +0000127 if (!Create) return 0; // Do not create a placeholder?
Chris Lattner00950542001-06-06 20:29:01 +0000128
Chris Lattner8eb10ce2003-10-09 06:05:40 +0000129 std::pair<unsigned,unsigned> KeyValue(type, oNum);
130 std::map<std::pair<unsigned,unsigned>, Value*>::iterator I =
131 ForwardReferences.lower_bound(KeyValue);
132 if (I != ForwardReferences.end() && I->first == KeyValue)
133 return I->second; // We have already created this placeholder
134
Chris Lattnerbf43ac62003-10-09 06:14:26 +0000135 Value *Val = new Argument(getType(type));
Chris Lattner8eb10ce2003-10-09 06:05:40 +0000136 ForwardReferences.insert(I, std::make_pair(KeyValue, Val));
Chris Lattner36392bc2003-10-08 21:18:57 +0000137 return Val;
Chris Lattner00950542001-06-06 20:29:01 +0000138}
139
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000140/// getBasicBlock - Get a particular numbered basic block, which might be a
141/// forward reference. This works together with ParseBasicBlock to handle these
142/// forward references in a clean manner.
143///
144BasicBlock *BytecodeParser::getBasicBlock(unsigned ID) {
145 // Make sure there is room in the table...
146 if (ParsedBasicBlocks.size() <= ID) ParsedBasicBlocks.resize(ID+1);
147
148 // First check to see if this is a backwards reference, i.e., ParseBasicBlock
149 // has already created this block, or if the forward reference has already
150 // been created.
151 if (ParsedBasicBlocks[ID])
152 return ParsedBasicBlocks[ID];
153
154 // Otherwise, the basic block has not yet been created. Do so and add it to
155 // the ParsedBasicBlocks list.
156 return ParsedBasicBlocks[ID] = new BasicBlock();
157}
158
Chris Lattnerbbd4b302002-10-14 03:33:02 +0000159/// getConstantValue - Just like getValue, except that it returns a null pointer
160/// only on error. It always returns a constant (meaning that if the value is
161/// defined, but is not a constant, that is an error). If the specified
162/// constant hasn't been parsed yet, a placeholder is defined and used. Later,
163/// after the real value is parsed, the placeholder is eliminated.
164///
165Constant *BytecodeParser::getConstantValue(const Type *Ty, unsigned Slot) {
166 if (Value *V = getValue(Ty, Slot, false))
167 return dyn_cast<Constant>(V); // If we already have the value parsed...
168
Chris Lattner52e20b02003-03-19 20:54:26 +0000169 std::pair<const Type*, unsigned> Key(Ty, Slot);
170 GlobalRefsType::iterator I = GlobalRefs.lower_bound(Key);
171
172 if (I != GlobalRefs.end() && I->first == Key) {
Chris Lattnerbbd4b302002-10-14 03:33:02 +0000173 BCR_TRACE(5, "Previous forward ref found!\n");
174 return cast<Constant>(I->second);
175 } else {
176 // Create a placeholder for the constant reference and
177 // keep track of the fact that we have a forward ref to recycle it
178 BCR_TRACE(5, "Creating new forward ref to a constant!\n");
179 Constant *C = new ConstPHolder(Ty, Slot);
180
181 // Keep track of the fact that we have a forward ref to recycle it
Chris Lattner52e20b02003-03-19 20:54:26 +0000182 GlobalRefs.insert(I, std::make_pair(Key, C));
Chris Lattnerbbd4b302002-10-14 03:33:02 +0000183 return C;
184 }
185}
186
187
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000188BasicBlock *BytecodeParser::ParseBasicBlock(const unsigned char *&Buf,
189 const unsigned char *EndBuf,
190 unsigned BlockNo) {
191 BasicBlock *BB;
192 if (ParsedBasicBlocks.size() == BlockNo)
193 ParsedBasicBlocks.push_back(BB = new BasicBlock());
194 else if (ParsedBasicBlocks[BlockNo] == 0)
195 BB = ParsedBasicBlocks[BlockNo] = new BasicBlock();
196 else
197 BB = ParsedBasicBlocks[BlockNo];
Chris Lattner00950542001-06-06 20:29:01 +0000198
199 while (Buf < EndBuf) {
Chris Lattner3483f542003-10-09 18:25:19 +0000200 Instruction *Inst = ParseInstruction(Buf, EndBuf);
Chris Lattner927b1852003-10-09 20:22:47 +0000201 insertValue(Inst, Values);
Chris Lattner1d670cc2001-09-07 16:37:43 +0000202 BB->getInstList().push_back(Inst);
Chris Lattner1d670cc2001-09-07 16:37:43 +0000203 BCR_TRACE(4, Inst);
Chris Lattner00950542001-06-06 20:29:01 +0000204 }
205
Misha Brukman12c29d12003-09-22 23:38:23 +0000206 return BB;
Chris Lattner00950542001-06-06 20:29:01 +0000207}
208
Misha Brukman12c29d12003-09-22 23:38:23 +0000209void BytecodeParser::ParseSymbolTable(const unsigned char *&Buf,
Chris Lattner12e64652003-05-22 18:08:30 +0000210 const unsigned char *EndBuf,
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000211 SymbolTable *ST,
212 Function *CurrentFunction) {
Chris Lattner00950542001-06-06 20:29:01 +0000213 while (Buf < EndBuf) {
214 // Symtab block header: [num entries][type id number]
215 unsigned NumEntries, Typ;
216 if (read_vbr(Buf, EndBuf, NumEntries) ||
Misha Brukman12c29d12003-09-22 23:38:23 +0000217 read_vbr(Buf, EndBuf, Typ)) throw Error_readvbr;
Chris Lattner00950542001-06-06 20:29:01 +0000218 const Type *Ty = getType(Typ);
Chris Lattner927b1852003-10-09 20:22:47 +0000219 BCR_TRACE(3, "Plane Type: '" << *Ty << "' with " << NumEntries <<
Misha Brukman12c29d12003-09-22 23:38:23 +0000220 " entries\n");
Chris Lattner1d670cc2001-09-07 16:37:43 +0000221
Chris Lattner7fc9fe32001-06-27 23:41:11 +0000222 for (unsigned i = 0; i < NumEntries; ++i) {
Chris Lattner00950542001-06-06 20:29:01 +0000223 // Symtab entry: [def slot #][name]
224 unsigned slot;
Misha Brukman12c29d12003-09-22 23:38:23 +0000225 if (read_vbr(Buf, EndBuf, slot)) throw Error_readvbr;
Chris Lattner697954c2002-01-20 22:54:45 +0000226 std::string Name;
Chris Lattner00950542001-06-06 20:29:01 +0000227 if (read(Buf, EndBuf, Name, false)) // Not aligned...
Misha Brukman12c29d12003-09-22 23:38:23 +0000228 throw std::string("Buffer not aligned.");
Chris Lattner00950542001-06-06 20:29:01 +0000229
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000230 Value *V = 0;
Chris Lattner36392bc2003-10-08 21:18:57 +0000231 if (Typ == Type::TypeTyID)
232 V = (Value*)getType(slot);
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000233 else if (Typ == Type::LabelTyID) {
234 if (CurrentFunction) {
235 // FIXME: THIS IS N^2!!!
236 Function::iterator BlockIterator = CurrentFunction->begin();
237 std::advance(BlockIterator, slot);
238 V = BlockIterator;
239 }
240 } else
Chris Lattner36392bc2003-10-08 21:18:57 +0000241 V = getValue(Typ, slot, false); // Find mapping...
242 if (V == 0) throw std::string("Failed value look-up.");
Chris Lattner52e20b02003-03-19 20:54:26 +0000243 BCR_TRACE(4, "Map: '" << Name << "' to #" << slot << ":" << *V;
Misha Brukman12c29d12003-09-22 23:38:23 +0000244 if (!isa<Instruction>(V)) std::cerr << "\n");
Chris Lattner1d670cc2001-09-07 16:37:43 +0000245
Chris Lattner52e20b02003-03-19 20:54:26 +0000246 V->setName(Name, ST);
Chris Lattner00950542001-06-06 20:29:01 +0000247 }
248 }
249
Misha Brukman12c29d12003-09-22 23:38:23 +0000250 if (Buf > EndBuf) throw std::string("Tried to read past end of buffer.");
Chris Lattner00950542001-06-06 20:29:01 +0000251}
252
Chris Lattner74734132002-08-17 22:01:27 +0000253void BytecodeParser::ResolveReferencesToValue(Value *NewV, unsigned Slot) {
Chris Lattner0d75d8d72003-03-06 16:32:25 +0000254 GlobalRefsType::iterator I = GlobalRefs.find(std::make_pair(NewV->getType(),
255 Slot));
Chris Lattner74734132002-08-17 22:01:27 +0000256 if (I == GlobalRefs.end()) return; // Never forward referenced?
Chris Lattner00950542001-06-06 20:29:01 +0000257
Chris Lattner74734132002-08-17 22:01:27 +0000258 BCR_TRACE(3, "Mutating forward refs!\n");
259 Value *VPH = I->second; // Get the placeholder...
Vikram S. Advec1e4a812002-07-14 23:04:18 +0000260
Chris Lattner52e20b02003-03-19 20:54:26 +0000261 VPH->replaceAllUsesWith(NewV);
262
263 // If this is a global variable being resolved, remove the placeholder from
264 // the module...
265 if (GlobalValue* GVal = dyn_cast<GlobalValue>(NewV))
266 GVal->getParent()->getGlobalList().remove(cast<GlobalVariable>(VPH));
Vikram S. Advec1e4a812002-07-14 23:04:18 +0000267
Chris Lattner74734132002-08-17 22:01:27 +0000268 delete VPH; // Delete the old placeholder
269 GlobalRefs.erase(I); // Remove the map entry for it
Vikram S. Advec1e4a812002-07-14 23:04:18 +0000270}
271
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000272void BytecodeParser::ParseFunction(const unsigned char *&Buf,
273 const unsigned char *EndBuf) {
Misha Brukman12c29d12003-09-22 23:38:23 +0000274 if (FunctionSignatureList.empty())
275 throw std::string("FunctionSignatureList empty!");
Chris Lattner52e20b02003-03-19 20:54:26 +0000276
Misha Brukman12c29d12003-09-22 23:38:23 +0000277 Function *F = FunctionSignatureList.back().first;
278 unsigned FunctionSlot = FunctionSignatureList.back().second;
279 FunctionSignatureList.pop_back();
280
281 // Save the information for future reading of the function
282 LazyFunctionInfo *LFI = new LazyFunctionInfo();
283 LFI->Buf = Buf; LFI->EndBuf = EndBuf; LFI->FunctionSlot = FunctionSlot;
284 LazyFunctionLoadMap[F] = LFI;
285 // Pretend we've `parsed' this function
286 Buf = EndBuf;
287}
288
289void BytecodeParser::materializeFunction(Function* F) {
290 // Find {start, end} pointers and slot in the map. If not there, we're done.
291 std::map<Function*, LazyFunctionInfo*>::iterator Fi =
292 LazyFunctionLoadMap.find(F);
293 if (Fi == LazyFunctionLoadMap.end()) return;
294
295 LazyFunctionInfo *LFI = Fi->second;
296 const unsigned char *Buf = LFI->Buf;
297 const unsigned char *EndBuf = LFI->EndBuf;
298 unsigned FunctionSlot = LFI->FunctionSlot;
299 LazyFunctionLoadMap.erase(Fi);
300 delete LFI;
Chris Lattner00950542001-06-06 20:29:01 +0000301
Chris Lattnere3869c82003-04-16 21:16:05 +0000302 GlobalValue::LinkageTypes Linkage = GlobalValue::ExternalLinkage;
303
304 if (!hasInternalMarkerOnly) {
305 unsigned LinkageType;
Misha Brukman12c29d12003-09-22 23:38:23 +0000306 if (read_vbr(Buf, EndBuf, LinkageType))
307 throw std::string("ParseFunction: Error reading from buffer.");
308 if (LinkageType & ~0x3)
309 throw std::string("Invalid linkage type for Function.");
Chris Lattnere3869c82003-04-16 21:16:05 +0000310 Linkage = (GlobalValue::LinkageTypes)LinkageType;
311 } else {
312 // We used to only support two linkage models: internal and external
313 unsigned isInternal;
Misha Brukman12c29d12003-09-22 23:38:23 +0000314 if (read_vbr(Buf, EndBuf, isInternal))
315 throw std::string("ParseFunction: Error reading from buffer.");
Chris Lattnere3869c82003-04-16 21:16:05 +0000316 if (isInternal) Linkage = GlobalValue::InternalLinkage;
317 }
Chris Lattnerd23b1d32001-11-26 18:56:10 +0000318
Chris Lattnere3869c82003-04-16 21:16:05 +0000319 F->setLinkage(Linkage);
Chris Lattner00950542001-06-06 20:29:01 +0000320
Chris Lattner52e20b02003-03-19 20:54:26 +0000321 const FunctionType::ParamTypes &Params =F->getFunctionType()->getParamTypes();
322 Function::aiterator AI = F->abegin();
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000323 for (FunctionType::ParamTypes::const_iterator It = Params.begin();
Chris Lattner927b1852003-10-09 20:22:47 +0000324 It != Params.end(); ++It, ++AI)
325 insertValue(AI, Values);
Chris Lattner00950542001-06-06 20:29:01 +0000326
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000327 // Keep track of how many basic blocks we have read in...
328 unsigned BlockNum = 0;
329
Chris Lattner00950542001-06-06 20:29:01 +0000330 while (Buf < EndBuf) {
331 unsigned Type, Size;
Chris Lattnerb6c46952003-03-06 17:03:28 +0000332 const unsigned char *OldBuf = Buf;
Misha Brukman12c29d12003-09-22 23:38:23 +0000333 readBlock(Buf, EndBuf, Type, Size);
Chris Lattner00950542001-06-06 20:29:01 +0000334
335 switch (Type) {
Misha Brukman12c29d12003-09-22 23:38:23 +0000336 case BytecodeFormat::ConstantPool: {
Chris Lattner1d670cc2001-09-07 16:37:43 +0000337 BCR_TRACE(2, "BLOCK BytecodeFormat::ConstantPool: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000338 ParseConstantPool(Buf, Buf+Size, Values, FunctionTypeValues);
Chris Lattner00950542001-06-06 20:29:01 +0000339 break;
Misha Brukman12c29d12003-09-22 23:38:23 +0000340 }
Chris Lattner00950542001-06-06 20:29:01 +0000341
342 case BytecodeFormat::BasicBlock: {
Chris Lattner1d670cc2001-09-07 16:37:43 +0000343 BCR_TRACE(2, "BLOCK BytecodeFormat::BasicBlock: {\n");
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000344 BasicBlock *BB = ParseBasicBlock(Buf, Buf+Size, BlockNum++);
345 F->getBasicBlockList().push_back(BB);
Chris Lattner00950542001-06-06 20:29:01 +0000346 break;
347 }
348
Misha Brukman12c29d12003-09-22 23:38:23 +0000349 case BytecodeFormat::SymbolTable: {
Chris Lattner1d670cc2001-09-07 16:37:43 +0000350 BCR_TRACE(2, "BLOCK BytecodeFormat::SymbolTable: {\n");
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000351 ParseSymbolTable(Buf, Buf+Size, &F->getSymbolTable(), F);
Chris Lattner00950542001-06-06 20:29:01 +0000352 break;
Misha Brukman12c29d12003-09-22 23:38:23 +0000353 }
Chris Lattner00950542001-06-06 20:29:01 +0000354
355 default:
Chris Lattner1d670cc2001-09-07 16:37:43 +0000356 BCR_TRACE(2, "BLOCK <unknown>:ignored! {\n");
Chris Lattner00950542001-06-06 20:29:01 +0000357 Buf += Size;
Misha Brukman12c29d12003-09-22 23:38:23 +0000358 if (OldBuf > Buf)
359 throw std::string("Wrapped around reading bytecode.");
Chris Lattner00950542001-06-06 20:29:01 +0000360 break;
361 }
Chris Lattner1d670cc2001-09-07 16:37:43 +0000362 BCR_TRACE(2, "} end block\n");
363
Misha Brukman12c29d12003-09-22 23:38:23 +0000364 // Malformed bc file if read past end of block.
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000365 ALIGN32(Buf, EndBuf);
Chris Lattner00950542001-06-06 20:29:01 +0000366 }
367
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000368 // Make sure there were no references to non-existant basic blocks.
369 if (BlockNum != ParsedBasicBlocks.size())
370 throw std::string("Illegal basic block operand reference");
371 ParsedBasicBlocks.clear();
372
373
Chris Lattner8eb10ce2003-10-09 06:05:40 +0000374 // Resolve forward references
375 while (!ForwardReferences.empty()) {
376 std::map<std::pair<unsigned,unsigned>, Value*>::iterator I = ForwardReferences.begin();
377 unsigned type = I->first.first;
378 unsigned Slot = I->first.second;
379 Value *PlaceHolder = I->second;
380 ForwardReferences.erase(I);
Chris Lattner00950542001-06-06 20:29:01 +0000381
Chris Lattner8eb10ce2003-10-09 06:05:40 +0000382 Value *NewVal = getValue(type, Slot, false);
383 if (NewVal == 0)
384 throw std::string("Unresolvable reference found: <" +
385 PlaceHolder->getType()->getDescription() + ">:" +
386 utostr(Slot) + ".");
Chris Lattner6e448022003-10-08 21:51:46 +0000387
Chris Lattner8eb10ce2003-10-09 06:05:40 +0000388 // Fixup all of the uses of this placeholder def...
389 PlaceHolder->replaceAllUsesWith(NewVal);
Chris Lattner6e448022003-10-08 21:51:46 +0000390
Chris Lattner8eb10ce2003-10-09 06:05:40 +0000391 // Now that all the uses are gone, delete the placeholder...
392 // If we couldn't find a def (error case), then leak a little
393 // memory, because otherwise we can't remove all uses!
394 delete PlaceHolder;
Chris Lattner6e448022003-10-08 21:51:46 +0000395 }
Chris Lattner00950542001-06-06 20:29:01 +0000396
Misha Brukman12c29d12003-09-22 23:38:23 +0000397 // Clear out function-level types...
Chris Lattner6e5a0e42003-03-06 17:18:14 +0000398 FunctionTypeValues.clear();
Chris Lattnere4d71a12001-09-14 22:03:42 +0000399
Chris Lattner52e20b02003-03-19 20:54:26 +0000400 freeTable(Values);
Chris Lattner00950542001-06-06 20:29:01 +0000401}
402
Misha Brukman12c29d12003-09-22 23:38:23 +0000403void BytecodeParser::ParseModuleGlobalInfo(const unsigned char *&Buf,
404 const unsigned char *End) {
405 if (!FunctionSignatureList.empty())
406 throw std::string("Two ModuleGlobalInfo packets found!");
Chris Lattner00950542001-06-06 20:29:01 +0000407
Chris Lattner70cc3392001-09-10 07:58:01 +0000408 // Read global variables...
409 unsigned VarType;
Misha Brukman12c29d12003-09-22 23:38:23 +0000410 if (read_vbr(Buf, End, VarType)) throw Error_readvbr;
Chris Lattner70cc3392001-09-10 07:58:01 +0000411 while (VarType != Type::VoidTyID) { // List is terminated by Void
Chris Lattnere3869c82003-04-16 21:16:05 +0000412 unsigned SlotNo;
413 GlobalValue::LinkageTypes Linkage;
414
415 if (!hasInternalMarkerOnly) {
416 // VarType Fields: bit0 = isConstant, bit1 = hasInitializer,
417 // bit2,3 = Linkage, bit4+ = slot#
418 SlotNo = VarType >> 4;
419 Linkage = (GlobalValue::LinkageTypes)((VarType >> 2) & 3);
420 } else {
421 // VarType Fields: bit0 = isConstant, bit1 = hasInitializer,
422 // bit2 = isInternal, bit3+ = slot#
423 SlotNo = VarType >> 3;
424 Linkage = (VarType & 4) ? GlobalValue::InternalLinkage :
425 GlobalValue::ExternalLinkage;
426 }
427
428 const Type *Ty = getType(SlotNo);
Chris Lattner927b1852003-10-09 20:22:47 +0000429 if (!isa<PointerType>(Ty))
Misha Brukman12c29d12003-09-22 23:38:23 +0000430 throw std::string("Global not pointer type! Ty = " +
431 Ty->getDescription());
Chris Lattner70cc3392001-09-10 07:58:01 +0000432
Chris Lattner52e20b02003-03-19 20:54:26 +0000433 const Type *ElTy = cast<PointerType>(Ty)->getElementType();
Chris Lattnerd70684f2001-09-18 04:01:05 +0000434
Chris Lattner70cc3392001-09-10 07:58:01 +0000435 // Create the global variable...
Chris Lattner4ad02e72003-04-16 20:28:45 +0000436 GlobalVariable *GV = new GlobalVariable(ElTy, VarType & 1, Linkage,
Chris Lattner52e20b02003-03-19 20:54:26 +0000437 0, "", TheModule);
Chris Lattner52e20b02003-03-19 20:54:26 +0000438 BCR_TRACE(2, "Global Variable of type: " << *Ty << "\n");
Chris Lattner927b1852003-10-09 20:22:47 +0000439 ResolveReferencesToValue(GV, insertValue(GV, ModuleValues));
Chris Lattner05950c32001-10-13 06:47:01 +0000440
Misha Brukman37f92e22003-09-11 22:34:13 +0000441 if (VarType & 2) { // Does it have an initializer?
Chris Lattner52e20b02003-03-19 20:54:26 +0000442 unsigned InitSlot;
Misha Brukman12c29d12003-09-22 23:38:23 +0000443 if (read_vbr(Buf, End, InitSlot)) throw Error_readvbr;
Chris Lattner52e20b02003-03-19 20:54:26 +0000444 GlobalInits.push_back(std::make_pair(GV, InitSlot));
445 }
Misha Brukman12c29d12003-09-22 23:38:23 +0000446 if (read_vbr(Buf, End, VarType)) throw Error_readvbr;
Chris Lattner70cc3392001-09-10 07:58:01 +0000447 }
448
Chris Lattner52e20b02003-03-19 20:54:26 +0000449 // Read the function objects for all of the functions that are coming
Chris Lattner74734132002-08-17 22:01:27 +0000450 unsigned FnSignature;
Misha Brukman12c29d12003-09-22 23:38:23 +0000451 if (read_vbr(Buf, End, FnSignature)) throw Error_readvbr;
Chris Lattner74734132002-08-17 22:01:27 +0000452 while (FnSignature != Type::VoidTyID) { // List is terminated by Void
453 const Type *Ty = getType(FnSignature);
Chris Lattner927b1852003-10-09 20:22:47 +0000454 if (!isa<PointerType>(Ty) ||
455 !isa<FunctionType>(cast<PointerType>(Ty)->getElementType()))
Misha Brukman12c29d12003-09-22 23:38:23 +0000456 throw std::string("Function not ptr to func type! Ty = " +
457 Ty->getDescription());
Chris Lattner8cdc6b72002-10-23 00:51:54 +0000458
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000459 // We create functions by passing the underlying FunctionType to create...
Chris Lattner7a176752001-12-04 00:03:30 +0000460 Ty = cast<PointerType>(Ty)->getElementType();
Chris Lattner00950542001-06-06 20:29:01 +0000461
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000462 // When the ModuleGlobalInfo section is read, we load the type of each
463 // function and the 'ModuleValues' slot that it lands in. We then load a
464 // placeholder into its slot to reserve it. When the function is loaded,
465 // this placeholder is replaced.
Chris Lattner00950542001-06-06 20:29:01 +0000466
467 // Insert the placeholder...
Chris Lattner4ad02e72003-04-16 20:28:45 +0000468 Function *Func = new Function(cast<FunctionType>(Ty),
469 GlobalValue::InternalLinkage, "", TheModule);
Chris Lattner927b1852003-10-09 20:22:47 +0000470 unsigned DestSlot = insertValue(Func, ModuleValues);
471 ResolveReferencesToValue(Func, DestSlot);
Chris Lattner00950542001-06-06 20:29:01 +0000472
Chris Lattner52e20b02003-03-19 20:54:26 +0000473 // Keep track of this information in a list that is emptied as functions are
474 // loaded...
Chris Lattner00950542001-06-06 20:29:01 +0000475 //
Chris Lattner52e20b02003-03-19 20:54:26 +0000476 FunctionSignatureList.push_back(std::make_pair(Func, DestSlot));
477
Misha Brukman12c29d12003-09-22 23:38:23 +0000478 if (read_vbr(Buf, End, FnSignature)) throw Error_readvbr;
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000479 BCR_TRACE(2, "Function of type: " << Ty << "\n");
Chris Lattner00950542001-06-06 20:29:01 +0000480 }
481
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000482 ALIGN32(Buf, End);
Chris Lattner74734132002-08-17 22:01:27 +0000483
484 // Now that the function signature list is set up, reverse it so that we can
485 // remove elements efficiently from the back of the vector.
486 std::reverse(FunctionSignatureList.begin(), FunctionSignatureList.end());
Chris Lattner00950542001-06-06 20:29:01 +0000487
488 // This is for future proofing... in the future extra fields may be added that
489 // we don't understand, so we transparently ignore them.
490 //
491 Buf = End;
Chris Lattner00950542001-06-06 20:29:01 +0000492}
493
Misha Brukman12c29d12003-09-22 23:38:23 +0000494void BytecodeParser::ParseVersionInfo(const unsigned char *&Buf,
Chris Lattner12e64652003-05-22 18:08:30 +0000495 const unsigned char *EndBuf) {
Chris Lattner036b8aa2003-03-06 17:55:45 +0000496 unsigned Version;
Misha Brukman12c29d12003-09-22 23:38:23 +0000497 if (read_vbr(Buf, EndBuf, Version)) throw Error_readvbr;
Chris Lattner036b8aa2003-03-06 17:55:45 +0000498
499 // Unpack version number: low four bits are for flags, top bits = version
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000500 Module::Endianness Endianness;
501 Module::PointerSize PointerSize;
502 Endianness = (Version & 1) ? Module::BigEndian : Module::LittleEndian;
503 PointerSize = (Version & 2) ? Module::Pointer64 : Module::Pointer32;
504
505 bool hasNoEndianness = Version & 4;
506 bool hasNoPointerSize = Version & 8;
507
508 RevisionNum = Version >> 4;
Chris Lattnere3869c82003-04-16 21:16:05 +0000509
510 // Default values for the current bytecode version
Chris Lattner036b8aa2003-03-06 17:55:45 +0000511 HasImplicitZeroInitializer = true;
Chris Lattnere3869c82003-04-16 21:16:05 +0000512 hasInternalMarkerOnly = false;
513 FirstDerivedTyID = 14;
Chris Lattner036b8aa2003-03-06 17:55:45 +0000514
515 switch (RevisionNum) {
516 case 0: // Initial revision
Chris Lattner52e20b02003-03-19 20:54:26 +0000517 // Version #0 didn't have any of the flags stored correctly, and in fact as
518 // only valid with a 14 in the flags values. Also, it does not support
519 // encoding zero initializers for arrays compactly.
520 //
Misha Brukman12c29d12003-09-22 23:38:23 +0000521 if (Version != 14) throw std::string("Unknown revision 0 flags?");
Chris Lattner036b8aa2003-03-06 17:55:45 +0000522 HasImplicitZeroInitializer = false;
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000523 Endianness = Module::BigEndian;
524 PointerSize = Module::Pointer64;
Chris Lattnere3869c82003-04-16 21:16:05 +0000525 hasInternalMarkerOnly = true;
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000526 hasNoEndianness = hasNoPointerSize = false;
Chris Lattner036b8aa2003-03-06 17:55:45 +0000527 break;
528 case 1:
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000529 // Version #1 has four bit fields: isBigEndian, hasLongPointers,
530 // hasNoEndianness, and hasNoPointerSize.
Chris Lattnere3869c82003-04-16 21:16:05 +0000531 hasInternalMarkerOnly = true;
532 break;
533 case 2:
534 // Version #2 added information about all 4 linkage types instead of just
535 // having internal and external.
Chris Lattner036b8aa2003-03-06 17:55:45 +0000536 break;
537 default:
Misha Brukman12c29d12003-09-22 23:38:23 +0000538 throw std::string("Unknown bytecode version number!");
Chris Lattner036b8aa2003-03-06 17:55:45 +0000539 }
540
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000541 if (hasNoEndianness) Endianness = Module::AnyEndianness;
542 if (hasNoPointerSize) PointerSize = Module::AnyPointerSize;
Chris Lattner76e38962003-04-22 18:15:10 +0000543
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000544 TheModule->setEndianness(Endianness);
545 TheModule->setPointerSize(PointerSize);
Chris Lattner036b8aa2003-03-06 17:55:45 +0000546 BCR_TRACE(1, "Bytecode Rev = " << (unsigned)RevisionNum << "\n");
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000547 BCR_TRACE(1, "Endianness/PointerSize = " << Endianness << ","
548 << PointerSize << "\n");
Chris Lattner036b8aa2003-03-06 17:55:45 +0000549 BCR_TRACE(1, "HasImplicitZeroInit = " << HasImplicitZeroInitializer << "\n");
Chris Lattner036b8aa2003-03-06 17:55:45 +0000550}
551
Misha Brukman12c29d12003-09-22 23:38:23 +0000552void BytecodeParser::ParseModule(const unsigned char *Buf,
Chris Lattner12e64652003-05-22 18:08:30 +0000553 const unsigned char *EndBuf) {
Chris Lattner00950542001-06-06 20:29:01 +0000554 unsigned Type, Size;
Misha Brukman12c29d12003-09-22 23:38:23 +0000555 readBlock(Buf, EndBuf, Type, Size);
556 if (Type != BytecodeFormat::Module || Buf+Size != EndBuf)
557 throw std::string("Expected Module packet! B: "+
558 utostr((unsigned)(intptr_t)Buf) + ", S: "+utostr(Size)+
559 " E: "+utostr((unsigned)(intptr_t)EndBuf)); // Hrm, not a class?
Chris Lattner00950542001-06-06 20:29:01 +0000560
Chris Lattner1d670cc2001-09-07 16:37:43 +0000561 BCR_TRACE(0, "BLOCK BytecodeFormat::Module: {\n");
Chris Lattner74734132002-08-17 22:01:27 +0000562 FunctionSignatureList.clear(); // Just in case...
Chris Lattner00950542001-06-06 20:29:01 +0000563
564 // Read into instance variables...
Misha Brukman12c29d12003-09-22 23:38:23 +0000565 ParseVersionInfo(Buf, EndBuf);
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000566 ALIGN32(Buf, EndBuf);
Chris Lattner00950542001-06-06 20:29:01 +0000567
Chris Lattner00950542001-06-06 20:29:01 +0000568 while (Buf < EndBuf) {
Chris Lattnerb6c46952003-03-06 17:03:28 +0000569 const unsigned char *OldBuf = Buf;
Misha Brukman12c29d12003-09-22 23:38:23 +0000570 readBlock(Buf, EndBuf, Type, Size);
Chris Lattner00950542001-06-06 20:29:01 +0000571 switch (Type) {
Chris Lattner52e20b02003-03-19 20:54:26 +0000572 case BytecodeFormat::GlobalTypePlane:
573 BCR_TRACE(1, "BLOCK BytecodeFormat::GlobalTypePlane: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000574 ParseGlobalTypes(Buf, Buf+Size);
Chris Lattner52e20b02003-03-19 20:54:26 +0000575 break;
576
577 case BytecodeFormat::ModuleGlobalInfo:
578 BCR_TRACE(1, "BLOCK BytecodeFormat::ModuleGlobalInfo: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000579 ParseModuleGlobalInfo(Buf, Buf+Size);
Chris Lattner52e20b02003-03-19 20:54:26 +0000580 break;
581
Chris Lattner1d670cc2001-09-07 16:37:43 +0000582 case BytecodeFormat::ConstantPool:
583 BCR_TRACE(1, "BLOCK BytecodeFormat::ConstantPool: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000584 ParseConstantPool(Buf, Buf+Size, ModuleValues, ModuleTypeValues);
Chris Lattner00950542001-06-06 20:29:01 +0000585 break;
586
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000587 case BytecodeFormat::Function: {
588 BCR_TRACE(1, "BLOCK BytecodeFormat::Function: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000589 ParseFunction(Buf, Buf+Size);
Chris Lattner00950542001-06-06 20:29:01 +0000590 break;
591 }
592
593 case BytecodeFormat::SymbolTable:
Chris Lattner1d670cc2001-09-07 16:37:43 +0000594 BCR_TRACE(1, "BLOCK BytecodeFormat::SymbolTable: {\n");
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000595 ParseSymbolTable(Buf, Buf+Size, &TheModule->getSymbolTable(), 0);
Chris Lattner00950542001-06-06 20:29:01 +0000596 break;
597
598 default:
Chris Lattner00950542001-06-06 20:29:01 +0000599 Buf += Size;
Misha Brukman12c29d12003-09-22 23:38:23 +0000600 if (OldBuf > Buf) throw std::string("Expected Module Block!");
Chris Lattner00950542001-06-06 20:29:01 +0000601 break;
602 }
Chris Lattner1d670cc2001-09-07 16:37:43 +0000603 BCR_TRACE(1, "} end block\n");
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000604 ALIGN32(Buf, EndBuf);
Chris Lattner00950542001-06-06 20:29:01 +0000605 }
606
Chris Lattner52e20b02003-03-19 20:54:26 +0000607 // After the module constant pool has been read, we can safely initialize
608 // global variables...
609 while (!GlobalInits.empty()) {
610 GlobalVariable *GV = GlobalInits.back().first;
611 unsigned Slot = GlobalInits.back().second;
612 GlobalInits.pop_back();
613
614 // Look up the initializer value...
615 if (Value *V = getValue(GV->getType()->getElementType(), Slot, false)) {
Misha Brukman12c29d12003-09-22 23:38:23 +0000616 if (GV->hasInitializer())
617 throw std::string("Global *already* has an initializer?!");
Chris Lattner52e20b02003-03-19 20:54:26 +0000618 GV->setInitializer(cast<Constant>(V));
619 } else
Misha Brukman12c29d12003-09-22 23:38:23 +0000620 throw std::string("Cannot find initializer value.");
Chris Lattner52e20b02003-03-19 20:54:26 +0000621 }
622
Misha Brukman12c29d12003-09-22 23:38:23 +0000623 if (!FunctionSignatureList.empty())
624 throw std::string("Function expected, but bytecode stream ended!");
Chris Lattner1d670cc2001-09-07 16:37:43 +0000625
626 BCR_TRACE(0, "} end block\n\n");
Chris Lattner00950542001-06-06 20:29:01 +0000627}
628
Misha Brukman12c29d12003-09-22 23:38:23 +0000629void
630BytecodeParser::ParseBytecode(const unsigned char *Buf, unsigned Length,
631 const std::string &ModuleID) {
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000632
Misha Brukman12c29d12003-09-22 23:38:23 +0000633 unsigned char *EndBuf = (unsigned char*)(Buf + Length);
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000634
Chris Lattner00950542001-06-06 20:29:01 +0000635 // Read and check signature...
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000636 unsigned Sig;
Chris Lattner00950542001-06-06 20:29:01 +0000637 if (read(Buf, EndBuf, Sig) ||
Misha Brukman12c29d12003-09-22 23:38:23 +0000638 Sig != ('l' | ('l' << 8) | ('v' << 16) | ('m' << 24)))
639 throw std::string("Invalid bytecode signature!");
Chris Lattner00950542001-06-06 20:29:01 +0000640
Chris Lattner75f20532003-04-22 18:02:52 +0000641 TheModule = new Module(ModuleID);
Misha Brukman12c29d12003-09-22 23:38:23 +0000642 try {
643 ParseModule(Buf, EndBuf);
644 } catch (std::string &Error) {
Chris Lattnera2602f32003-05-22 18:26:48 +0000645 freeState(); // Must destroy handles before deleting module!
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000646 delete TheModule;
647 TheModule = 0;
Chris Lattnerb0b7c0d2003-09-26 14:44:52 +0000648 throw;
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000649 }
Chris Lattner00950542001-06-06 20:29:01 +0000650}