blob: af775d282a4c00896763d8de1bb89e592da892a4 [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 Lattner8cdc6b72002-10-23 00:51:54 +000052 if (ID < Type::NumPrimitiveIDs) {
53 const Type *T = Type::getPrimitiveType((Type::PrimitiveID)ID);
54 if (T) return T;
55 }
Chris Lattner00950542001-06-06 20:29:01 +000056
Chris Lattner697954c2002-01-20 22:54:45 +000057 //cerr << "Looking up Type ID: " << ID << "\n";
Chris Lattner36392bc2003-10-08 21:18:57 +000058
59 if (ID < Type::NumPrimitiveIDs) {
60 const Type *T = Type::getPrimitiveType((Type::PrimitiveID)ID);
61 if (T) return T; // Asked for a primitive type...
62 }
63
64 // Otherwise, derived types need offset...
65 ID -= FirstDerivedTyID;
66
67 // Is it a module-level type?
68 if (ID < ModuleTypeValues.size())
69 return ModuleTypeValues[ID].get();
70
71 // Nope, is it a function-level type?
72 ID -= ModuleTypeValues.size();
73 if (ID < FunctionTypeValues.size())
74 return FunctionTypeValues[ID].get();
75
76 return 0;
Chris Lattner00950542001-06-06 20:29:01 +000077}
78
Chris Lattner52e20b02003-03-19 20:54:26 +000079int BytecodeParser::insertValue(Value *Val, ValueTable &ValueTab) {
80 assert((!HasImplicitZeroInitializer || !isa<Constant>(Val) ||
81 Val->getType()->isPrimitiveType() ||
82 !cast<Constant>(Val)->isNullValue()) &&
83 "Cannot read null values from bytecode!");
Chris Lattner9e460f22003-10-04 20:00:03 +000084 unsigned type = getTypeSlot(Val->getType());
Chris Lattner1d670cc2001-09-07 16:37:43 +000085 assert(type != Type::TypeTyID && "Types should never be insertValue'd!");
Chris Lattner00950542001-06-06 20:29:01 +000086
Chris Lattner52e20b02003-03-19 20:54:26 +000087 if (ValueTab.size() <= type) {
88 unsigned OldSize = ValueTab.size();
89 ValueTab.resize(type+1);
90 while (OldSize != type+1)
91 ValueTab[OldSize++] = new ValueList();
Chris Lattner036b8aa2003-03-06 17:55:45 +000092 }
Chris Lattner00950542001-06-06 20:29:01 +000093
94 //cerr << "insertValue Values[" << type << "][" << ValueTab[type].size()
Misha Brukman12c29d12003-09-22 23:38:23 +000095 // << "] = " << Val << "\n";
Chris Lattner52e20b02003-03-19 20:54:26 +000096 ValueTab[type]->push_back(Val);
Chris Lattner00950542001-06-06 20:29:01 +000097
Chris Lattner52e20b02003-03-19 20:54:26 +000098 bool HasOffset = HasImplicitZeroInitializer &&
Misha Brukman12c29d12003-09-22 23:38:23 +000099 !Val->getType()->isPrimitiveType();
Chris Lattner52e20b02003-03-19 20:54:26 +0000100
101 return ValueTab[type]->size()-1 + HasOffset;
102}
103
104
105void BytecodeParser::setValueTo(ValueTable &ValueTab, unsigned Slot,
106 Value *Val) {
107 assert(&ValueTab == &ModuleValues && "Can only setValueTo on Module values!");
Chris Lattner52e20b02003-03-19 20:54:26 +0000108 assert((!HasImplicitZeroInitializer || Slot != 0) &&
109 "Cannot change zero init");
Chris Lattner9e460f22003-10-04 20:00:03 +0000110 unsigned type = getTypeSlot(Val->getType());
Chris Lattner52e20b02003-03-19 20:54:26 +0000111 assert(type < ValueTab.size() && Slot <= ValueTab[type]->size());
112 ValueTab[type]->setOperand(Slot-HasImplicitZeroInitializer, Val);
Chris Lattner00950542001-06-06 20:29:01 +0000113}
114
Chris Lattner36392bc2003-10-08 21:18:57 +0000115
Chris Lattner00950542001-06-06 20:29:01 +0000116Value *BytecodeParser::getValue(const Type *Ty, unsigned oNum, bool Create) {
Chris Lattner36392bc2003-10-08 21:18:57 +0000117 return getValue(getTypeSlot(Ty), oNum, Create);
118}
119
120Value *BytecodeParser::getValue(unsigned type, unsigned oNum, bool Create) {
121 assert(type != Type::TypeTyID && "getValue() cannot get types!");
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000122 assert(type != Type::LabelTyID && "getValue() cannot get blocks!");
Chris Lattner00950542001-06-06 20:29:01 +0000123 unsigned Num = oNum;
Chris Lattner00950542001-06-06 20:29:01 +0000124
Chris Lattner52e20b02003-03-19 20:54:26 +0000125 if (HasImplicitZeroInitializer && type >= FirstDerivedTyID) {
126 if (Num == 0)
Chris Lattner36392bc2003-10-08 21:18:57 +0000127 return Constant::getNullValue(getType(type));
Chris Lattner52e20b02003-03-19 20:54:26 +0000128 --Num;
Chris Lattner00950542001-06-06 20:29:01 +0000129 }
130
Chris Lattner52e20b02003-03-19 20:54:26 +0000131 if (type < ModuleValues.size()) {
132 if (Num < ModuleValues[type]->size())
133 return ModuleValues[type]->getOperand(Num);
134 Num -= ModuleValues[type]->size();
135 }
136
137 if (Values.size() > type && Values[type]->size() > Num)
138 return Values[type]->getOperand(Num);
Chris Lattner00950542001-06-06 20:29:01 +0000139
Chris Lattner74734132002-08-17 22:01:27 +0000140 if (!Create) return 0; // Do not create a placeholder?
Chris Lattner00950542001-06-06 20:29:01 +0000141
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000142 Value *Val = new ValPHolder(getType(type), oNum);
Chris Lattner36392bc2003-10-08 21:18:57 +0000143 if (insertValue(Val, LateResolveValues) == -1) return 0;
144 return Val;
Chris Lattner00950542001-06-06 20:29:01 +0000145}
146
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000147/// getBasicBlock - Get a particular numbered basic block, which might be a
148/// forward reference. This works together with ParseBasicBlock to handle these
149/// forward references in a clean manner.
150///
151BasicBlock *BytecodeParser::getBasicBlock(unsigned ID) {
152 // Make sure there is room in the table...
153 if (ParsedBasicBlocks.size() <= ID) ParsedBasicBlocks.resize(ID+1);
154
155 // First check to see if this is a backwards reference, i.e., ParseBasicBlock
156 // has already created this block, or if the forward reference has already
157 // been created.
158 if (ParsedBasicBlocks[ID])
159 return ParsedBasicBlocks[ID];
160
161 // Otherwise, the basic block has not yet been created. Do so and add it to
162 // the ParsedBasicBlocks list.
163 return ParsedBasicBlocks[ID] = new BasicBlock();
164}
165
Chris Lattnerbbd4b302002-10-14 03:33:02 +0000166/// getConstantValue - Just like getValue, except that it returns a null pointer
167/// only on error. It always returns a constant (meaning that if the value is
168/// defined, but is not a constant, that is an error). If the specified
169/// constant hasn't been parsed yet, a placeholder is defined and used. Later,
170/// after the real value is parsed, the placeholder is eliminated.
171///
172Constant *BytecodeParser::getConstantValue(const Type *Ty, unsigned Slot) {
173 if (Value *V = getValue(Ty, Slot, false))
174 return dyn_cast<Constant>(V); // If we already have the value parsed...
175
Chris Lattner52e20b02003-03-19 20:54:26 +0000176 std::pair<const Type*, unsigned> Key(Ty, Slot);
177 GlobalRefsType::iterator I = GlobalRefs.lower_bound(Key);
178
179 if (I != GlobalRefs.end() && I->first == Key) {
Chris Lattnerbbd4b302002-10-14 03:33:02 +0000180 BCR_TRACE(5, "Previous forward ref found!\n");
181 return cast<Constant>(I->second);
182 } else {
183 // Create a placeholder for the constant reference and
184 // keep track of the fact that we have a forward ref to recycle it
185 BCR_TRACE(5, "Creating new forward ref to a constant!\n");
186 Constant *C = new ConstPHolder(Ty, Slot);
187
188 // Keep track of the fact that we have a forward ref to recycle it
Chris Lattner52e20b02003-03-19 20:54:26 +0000189 GlobalRefs.insert(I, std::make_pair(Key, C));
Chris Lattnerbbd4b302002-10-14 03:33:02 +0000190 return C;
191 }
192}
193
194
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000195BasicBlock *BytecodeParser::ParseBasicBlock(const unsigned char *&Buf,
196 const unsigned char *EndBuf,
197 unsigned BlockNo) {
198 BasicBlock *BB;
199 if (ParsedBasicBlocks.size() == BlockNo)
200 ParsedBasicBlocks.push_back(BB = new BasicBlock());
201 else if (ParsedBasicBlocks[BlockNo] == 0)
202 BB = ParsedBasicBlocks[BlockNo] = new BasicBlock();
203 else
204 BB = ParsedBasicBlocks[BlockNo];
Chris Lattner00950542001-06-06 20:29:01 +0000205
206 while (Buf < EndBuf) {
Chris Lattner1d670cc2001-09-07 16:37:43 +0000207 Instruction *Inst;
Misha Brukman12c29d12003-09-22 23:38:23 +0000208 ParseInstruction(Buf, EndBuf, Inst);
209
210 if (Inst == 0) { throw std::string("Could not parse Instruction."); }
211 if (insertValue(Inst, Values) == -1) {
212 throw std::string("Could not insert value.");
Chris Lattner00950542001-06-06 20:29:01 +0000213 }
214
Chris Lattner1d670cc2001-09-07 16:37:43 +0000215 BB->getInstList().push_back(Inst);
Chris Lattner1d670cc2001-09-07 16:37:43 +0000216 BCR_TRACE(4, Inst);
Chris Lattner00950542001-06-06 20:29:01 +0000217 }
218
Misha Brukman12c29d12003-09-22 23:38:23 +0000219 return BB;
Chris Lattner00950542001-06-06 20:29:01 +0000220}
221
Misha Brukman12c29d12003-09-22 23:38:23 +0000222void BytecodeParser::ParseSymbolTable(const unsigned char *&Buf,
Chris Lattner12e64652003-05-22 18:08:30 +0000223 const unsigned char *EndBuf,
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000224 SymbolTable *ST,
225 Function *CurrentFunction) {
Chris Lattner00950542001-06-06 20:29:01 +0000226 while (Buf < EndBuf) {
227 // Symtab block header: [num entries][type id number]
228 unsigned NumEntries, Typ;
229 if (read_vbr(Buf, EndBuf, NumEntries) ||
Misha Brukman12c29d12003-09-22 23:38:23 +0000230 read_vbr(Buf, EndBuf, Typ)) throw Error_readvbr;
Chris Lattner00950542001-06-06 20:29:01 +0000231 const Type *Ty = getType(Typ);
Misha Brukman12c29d12003-09-22 23:38:23 +0000232 if (Ty == 0) throw std::string("Invalid type read in symbol table.");
Chris Lattner00950542001-06-06 20:29:01 +0000233
Chris Lattner1d670cc2001-09-07 16:37:43 +0000234 BCR_TRACE(3, "Plane Type: '" << Ty << "' with " << NumEntries <<
Misha Brukman12c29d12003-09-22 23:38:23 +0000235 " entries\n");
Chris Lattner1d670cc2001-09-07 16:37:43 +0000236
Chris Lattner7fc9fe32001-06-27 23:41:11 +0000237 for (unsigned i = 0; i < NumEntries; ++i) {
Chris Lattner00950542001-06-06 20:29:01 +0000238 // Symtab entry: [def slot #][name]
239 unsigned slot;
Misha Brukman12c29d12003-09-22 23:38:23 +0000240 if (read_vbr(Buf, EndBuf, slot)) throw Error_readvbr;
Chris Lattner697954c2002-01-20 22:54:45 +0000241 std::string Name;
Chris Lattner00950542001-06-06 20:29:01 +0000242 if (read(Buf, EndBuf, Name, false)) // Not aligned...
Misha Brukman12c29d12003-09-22 23:38:23 +0000243 throw std::string("Buffer not aligned.");
Chris Lattner00950542001-06-06 20:29:01 +0000244
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000245 Value *V = 0;
Chris Lattner36392bc2003-10-08 21:18:57 +0000246 if (Typ == Type::TypeTyID)
247 V = (Value*)getType(slot);
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000248 else if (Typ == Type::LabelTyID) {
249 if (CurrentFunction) {
250 // FIXME: THIS IS N^2!!!
251 Function::iterator BlockIterator = CurrentFunction->begin();
252 std::advance(BlockIterator, slot);
253 V = BlockIterator;
254 }
255 } else
Chris Lattner36392bc2003-10-08 21:18:57 +0000256 V = getValue(Typ, slot, false); // Find mapping...
257 if (V == 0) throw std::string("Failed value look-up.");
Chris Lattner52e20b02003-03-19 20:54:26 +0000258 BCR_TRACE(4, "Map: '" << Name << "' to #" << slot << ":" << *V;
Misha Brukman12c29d12003-09-22 23:38:23 +0000259 if (!isa<Instruction>(V)) std::cerr << "\n");
Chris Lattner1d670cc2001-09-07 16:37:43 +0000260
Chris Lattner52e20b02003-03-19 20:54:26 +0000261 V->setName(Name, ST);
Chris Lattner00950542001-06-06 20:29:01 +0000262 }
263 }
264
Misha Brukman12c29d12003-09-22 23:38:23 +0000265 if (Buf > EndBuf) throw std::string("Tried to read past end of buffer.");
Chris Lattner00950542001-06-06 20:29:01 +0000266}
267
Chris Lattner74734132002-08-17 22:01:27 +0000268void BytecodeParser::ResolveReferencesToValue(Value *NewV, unsigned Slot) {
Chris Lattner0d75d8d72003-03-06 16:32:25 +0000269 GlobalRefsType::iterator I = GlobalRefs.find(std::make_pair(NewV->getType(),
270 Slot));
Chris Lattner74734132002-08-17 22:01:27 +0000271 if (I == GlobalRefs.end()) return; // Never forward referenced?
Chris Lattner00950542001-06-06 20:29:01 +0000272
Chris Lattner74734132002-08-17 22:01:27 +0000273 BCR_TRACE(3, "Mutating forward refs!\n");
274 Value *VPH = I->second; // Get the placeholder...
Vikram S. Advec1e4a812002-07-14 23:04:18 +0000275
Chris Lattner52e20b02003-03-19 20:54:26 +0000276 VPH->replaceAllUsesWith(NewV);
277
278 // If this is a global variable being resolved, remove the placeholder from
279 // the module...
280 if (GlobalValue* GVal = dyn_cast<GlobalValue>(NewV))
281 GVal->getParent()->getGlobalList().remove(cast<GlobalVariable>(VPH));
Vikram S. Advec1e4a812002-07-14 23:04:18 +0000282
Chris Lattner74734132002-08-17 22:01:27 +0000283 delete VPH; // Delete the old placeholder
284 GlobalRefs.erase(I); // Remove the map entry for it
Vikram S. Advec1e4a812002-07-14 23:04:18 +0000285}
286
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000287void BytecodeParser::ParseFunction(const unsigned char *&Buf,
288 const unsigned char *EndBuf) {
Misha Brukman12c29d12003-09-22 23:38:23 +0000289 if (FunctionSignatureList.empty())
290 throw std::string("FunctionSignatureList empty!");
Chris Lattner52e20b02003-03-19 20:54:26 +0000291
Misha Brukman12c29d12003-09-22 23:38:23 +0000292 Function *F = FunctionSignatureList.back().first;
293 unsigned FunctionSlot = FunctionSignatureList.back().second;
294 FunctionSignatureList.pop_back();
295
296 // Save the information for future reading of the function
297 LazyFunctionInfo *LFI = new LazyFunctionInfo();
298 LFI->Buf = Buf; LFI->EndBuf = EndBuf; LFI->FunctionSlot = FunctionSlot;
299 LazyFunctionLoadMap[F] = LFI;
300 // Pretend we've `parsed' this function
301 Buf = EndBuf;
302}
303
304void BytecodeParser::materializeFunction(Function* F) {
305 // Find {start, end} pointers and slot in the map. If not there, we're done.
306 std::map<Function*, LazyFunctionInfo*>::iterator Fi =
307 LazyFunctionLoadMap.find(F);
308 if (Fi == LazyFunctionLoadMap.end()) return;
309
310 LazyFunctionInfo *LFI = Fi->second;
311 const unsigned char *Buf = LFI->Buf;
312 const unsigned char *EndBuf = LFI->EndBuf;
313 unsigned FunctionSlot = LFI->FunctionSlot;
314 LazyFunctionLoadMap.erase(Fi);
315 delete LFI;
Chris Lattner00950542001-06-06 20:29:01 +0000316
Chris Lattnere3869c82003-04-16 21:16:05 +0000317 GlobalValue::LinkageTypes Linkage = GlobalValue::ExternalLinkage;
318
319 if (!hasInternalMarkerOnly) {
320 unsigned LinkageType;
Misha Brukman12c29d12003-09-22 23:38:23 +0000321 if (read_vbr(Buf, EndBuf, LinkageType))
322 throw std::string("ParseFunction: Error reading from buffer.");
323 if (LinkageType & ~0x3)
324 throw std::string("Invalid linkage type for Function.");
Chris Lattnere3869c82003-04-16 21:16:05 +0000325 Linkage = (GlobalValue::LinkageTypes)LinkageType;
326 } else {
327 // We used to only support two linkage models: internal and external
328 unsigned isInternal;
Misha Brukman12c29d12003-09-22 23:38:23 +0000329 if (read_vbr(Buf, EndBuf, isInternal))
330 throw std::string("ParseFunction: Error reading from buffer.");
Chris Lattnere3869c82003-04-16 21:16:05 +0000331 if (isInternal) Linkage = GlobalValue::InternalLinkage;
332 }
Chris Lattnerd23b1d32001-11-26 18:56:10 +0000333
Chris Lattnere3869c82003-04-16 21:16:05 +0000334 F->setLinkage(Linkage);
Chris Lattner00950542001-06-06 20:29:01 +0000335
Chris Lattner52e20b02003-03-19 20:54:26 +0000336 const FunctionType::ParamTypes &Params =F->getFunctionType()->getParamTypes();
337 Function::aiterator AI = F->abegin();
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000338 for (FunctionType::ParamTypes::const_iterator It = Params.begin();
Chris Lattner69da5cf2002-10-13 20:57:00 +0000339 It != Params.end(); ++It, ++AI) {
Misha Brukman12c29d12003-09-22 23:38:23 +0000340 if (insertValue(AI, Values) == -1)
341 throw std::string("Error reading function arguments!");
Chris Lattner00950542001-06-06 20:29:01 +0000342 }
343
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000344 // Keep track of how many basic blocks we have read in...
345 unsigned BlockNum = 0;
346
Chris Lattner00950542001-06-06 20:29:01 +0000347 while (Buf < EndBuf) {
348 unsigned Type, Size;
Chris Lattnerb6c46952003-03-06 17:03:28 +0000349 const unsigned char *OldBuf = Buf;
Misha Brukman12c29d12003-09-22 23:38:23 +0000350 readBlock(Buf, EndBuf, Type, Size);
Chris Lattner00950542001-06-06 20:29:01 +0000351
352 switch (Type) {
Misha Brukman12c29d12003-09-22 23:38:23 +0000353 case BytecodeFormat::ConstantPool: {
Chris Lattner1d670cc2001-09-07 16:37:43 +0000354 BCR_TRACE(2, "BLOCK BytecodeFormat::ConstantPool: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000355 ParseConstantPool(Buf, Buf+Size, Values, FunctionTypeValues);
Chris Lattner00950542001-06-06 20:29:01 +0000356 break;
Misha Brukman12c29d12003-09-22 23:38:23 +0000357 }
Chris Lattner00950542001-06-06 20:29:01 +0000358
359 case BytecodeFormat::BasicBlock: {
Chris Lattner1d670cc2001-09-07 16:37:43 +0000360 BCR_TRACE(2, "BLOCK BytecodeFormat::BasicBlock: {\n");
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000361 BasicBlock *BB = ParseBasicBlock(Buf, Buf+Size, BlockNum++);
362 F->getBasicBlockList().push_back(BB);
Chris Lattner00950542001-06-06 20:29:01 +0000363 break;
364 }
365
Misha Brukman12c29d12003-09-22 23:38:23 +0000366 case BytecodeFormat::SymbolTable: {
Chris Lattner1d670cc2001-09-07 16:37:43 +0000367 BCR_TRACE(2, "BLOCK BytecodeFormat::SymbolTable: {\n");
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000368 ParseSymbolTable(Buf, Buf+Size, &F->getSymbolTable(), F);
Chris Lattner00950542001-06-06 20:29:01 +0000369 break;
Misha Brukman12c29d12003-09-22 23:38:23 +0000370 }
Chris Lattner00950542001-06-06 20:29:01 +0000371
372 default:
Chris Lattner1d670cc2001-09-07 16:37:43 +0000373 BCR_TRACE(2, "BLOCK <unknown>:ignored! {\n");
Chris Lattner00950542001-06-06 20:29:01 +0000374 Buf += Size;
Misha Brukman12c29d12003-09-22 23:38:23 +0000375 if (OldBuf > Buf)
376 throw std::string("Wrapped around reading bytecode.");
Chris Lattner00950542001-06-06 20:29:01 +0000377 break;
378 }
Chris Lattner1d670cc2001-09-07 16:37:43 +0000379 BCR_TRACE(2, "} end block\n");
380
Misha Brukman12c29d12003-09-22 23:38:23 +0000381 // Malformed bc file if read past end of block.
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000382 ALIGN32(Buf, EndBuf);
Chris Lattner00950542001-06-06 20:29:01 +0000383 }
384
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000385 // Make sure there were no references to non-existant basic blocks.
386 if (BlockNum != ParsedBasicBlocks.size())
387 throw std::string("Illegal basic block operand reference");
388 ParsedBasicBlocks.clear();
389
390
Misha Brukman12c29d12003-09-22 23:38:23 +0000391 // Check for unresolvable references
Chris Lattner6e448022003-10-08 21:51:46 +0000392 while (!LateResolveValues.empty()) {
393 ValueList &VL = *LateResolveValues.back();
394 LateResolveValues.pop_back();
Chris Lattner00950542001-06-06 20:29:01 +0000395
Chris Lattner6e448022003-10-08 21:51:46 +0000396 while (!VL.empty()) {
397 Value *V = VL.back();
398 unsigned IDNumber = getValueIDNumberFromPlaceHolder(V);
399 VL.pop_back();
400
401 Value *NewVal = getValue(V->getType(), IDNumber, false);
402 if (NewVal == 0)
403 throw std::string("Unresolvable reference found: <" +
404 V->getType()->getDescription() + ">:" +
405 utostr(IDNumber) + ".");
406
407 // Fixup all of the uses of this placeholder def...
408 V->replaceAllUsesWith(NewVal);
409
410 // Now that all the uses are gone, delete the placeholder...
411 // If we couldn't find a def (error case), then leak a little
412 // memory, because otherwise we can't remove all uses!
413 delete V;
414 }
415 delete &VL;
416 }
Chris Lattner00950542001-06-06 20:29:01 +0000417
Misha Brukman12c29d12003-09-22 23:38:23 +0000418 // Clear out function-level types...
Chris Lattner6e5a0e42003-03-06 17:18:14 +0000419 FunctionTypeValues.clear();
Chris Lattnere4d71a12001-09-14 22:03:42 +0000420
Chris Lattner52e20b02003-03-19 20:54:26 +0000421 freeTable(Values);
Chris Lattner00950542001-06-06 20:29:01 +0000422}
423
Misha Brukman12c29d12003-09-22 23:38:23 +0000424void BytecodeParser::ParseModuleGlobalInfo(const unsigned char *&Buf,
425 const unsigned char *End) {
426 if (!FunctionSignatureList.empty())
427 throw std::string("Two ModuleGlobalInfo packets found!");
Chris Lattner00950542001-06-06 20:29:01 +0000428
Chris Lattner70cc3392001-09-10 07:58:01 +0000429 // Read global variables...
430 unsigned VarType;
Misha Brukman12c29d12003-09-22 23:38:23 +0000431 if (read_vbr(Buf, End, VarType)) throw Error_readvbr;
Chris Lattner70cc3392001-09-10 07:58:01 +0000432 while (VarType != Type::VoidTyID) { // List is terminated by Void
Chris Lattnere3869c82003-04-16 21:16:05 +0000433 unsigned SlotNo;
434 GlobalValue::LinkageTypes Linkage;
435
436 if (!hasInternalMarkerOnly) {
437 // VarType Fields: bit0 = isConstant, bit1 = hasInitializer,
438 // bit2,3 = Linkage, bit4+ = slot#
439 SlotNo = VarType >> 4;
440 Linkage = (GlobalValue::LinkageTypes)((VarType >> 2) & 3);
441 } else {
442 // VarType Fields: bit0 = isConstant, bit1 = hasInitializer,
443 // bit2 = isInternal, bit3+ = slot#
444 SlotNo = VarType >> 3;
445 Linkage = (VarType & 4) ? GlobalValue::InternalLinkage :
446 GlobalValue::ExternalLinkage;
447 }
448
449 const Type *Ty = getType(SlotNo);
Misha Brukman12c29d12003-09-22 23:38:23 +0000450 if (!Ty || !isa<PointerType>(Ty))
451 throw std::string("Global not pointer type! Ty = " +
452 Ty->getDescription());
Chris Lattner70cc3392001-09-10 07:58:01 +0000453
Chris Lattner52e20b02003-03-19 20:54:26 +0000454 const Type *ElTy = cast<PointerType>(Ty)->getElementType();
Chris Lattnerd70684f2001-09-18 04:01:05 +0000455
Chris Lattner70cc3392001-09-10 07:58:01 +0000456 // Create the global variable...
Chris Lattner4ad02e72003-04-16 20:28:45 +0000457 GlobalVariable *GV = new GlobalVariable(ElTy, VarType & 1, Linkage,
Chris Lattner52e20b02003-03-19 20:54:26 +0000458 0, "", TheModule);
Chris Lattner05950c32001-10-13 06:47:01 +0000459 int DestSlot = insertValue(GV, ModuleValues);
Misha Brukman12c29d12003-09-22 23:38:23 +0000460 if (DestSlot == -1) throw Error_DestSlot;
Chris Lattner52e20b02003-03-19 20:54:26 +0000461 BCR_TRACE(2, "Global Variable of type: " << *Ty << "\n");
Chris Lattner74734132002-08-17 22:01:27 +0000462 ResolveReferencesToValue(GV, (unsigned)DestSlot);
Chris Lattner05950c32001-10-13 06:47:01 +0000463
Misha Brukman37f92e22003-09-11 22:34:13 +0000464 if (VarType & 2) { // Does it have an initializer?
Chris Lattner52e20b02003-03-19 20:54:26 +0000465 unsigned InitSlot;
Misha Brukman12c29d12003-09-22 23:38:23 +0000466 if (read_vbr(Buf, End, InitSlot)) throw Error_readvbr;
Chris Lattner52e20b02003-03-19 20:54:26 +0000467 GlobalInits.push_back(std::make_pair(GV, InitSlot));
468 }
Misha Brukman12c29d12003-09-22 23:38:23 +0000469 if (read_vbr(Buf, End, VarType)) throw Error_readvbr;
Chris Lattner70cc3392001-09-10 07:58:01 +0000470 }
471
Chris Lattner52e20b02003-03-19 20:54:26 +0000472 // Read the function objects for all of the functions that are coming
Chris Lattner74734132002-08-17 22:01:27 +0000473 unsigned FnSignature;
Misha Brukman12c29d12003-09-22 23:38:23 +0000474 if (read_vbr(Buf, End, FnSignature)) throw Error_readvbr;
Chris Lattner74734132002-08-17 22:01:27 +0000475 while (FnSignature != Type::VoidTyID) { // List is terminated by Void
476 const Type *Ty = getType(FnSignature);
Chris Lattneref9c23f2001-10-03 14:53:21 +0000477 if (!Ty || !isa<PointerType>(Ty) ||
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000478 !isa<FunctionType>(cast<PointerType>(Ty)->getElementType())) {
Misha Brukman12c29d12003-09-22 23:38:23 +0000479 throw std::string("Function not ptr to func type! Ty = " +
480 Ty->getDescription());
Chris Lattner00950542001-06-06 20:29:01 +0000481 }
Chris Lattner8cdc6b72002-10-23 00:51:54 +0000482
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000483 // We create functions by passing the underlying FunctionType to create...
Chris Lattner7a176752001-12-04 00:03:30 +0000484 Ty = cast<PointerType>(Ty)->getElementType();
Chris Lattner00950542001-06-06 20:29:01 +0000485
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000486 // When the ModuleGlobalInfo section is read, we load the type of each
487 // function and the 'ModuleValues' slot that it lands in. We then load a
488 // placeholder into its slot to reserve it. When the function is loaded,
489 // this placeholder is replaced.
Chris Lattner00950542001-06-06 20:29:01 +0000490
491 // Insert the placeholder...
Chris Lattner4ad02e72003-04-16 20:28:45 +0000492 Function *Func = new Function(cast<FunctionType>(Ty),
493 GlobalValue::InternalLinkage, "", TheModule);
Chris Lattner52e20b02003-03-19 20:54:26 +0000494 int DestSlot = insertValue(Func, ModuleValues);
Misha Brukman12c29d12003-09-22 23:38:23 +0000495 if (DestSlot == -1) throw Error_DestSlot;
Chris Lattner52e20b02003-03-19 20:54:26 +0000496 ResolveReferencesToValue(Func, (unsigned)DestSlot);
Chris Lattner00950542001-06-06 20:29:01 +0000497
Chris Lattner52e20b02003-03-19 20:54:26 +0000498 // Keep track of this information in a list that is emptied as functions are
499 // loaded...
Chris Lattner00950542001-06-06 20:29:01 +0000500 //
Chris Lattner52e20b02003-03-19 20:54:26 +0000501 FunctionSignatureList.push_back(std::make_pair(Func, DestSlot));
502
Misha Brukman12c29d12003-09-22 23:38:23 +0000503 if (read_vbr(Buf, End, FnSignature)) throw Error_readvbr;
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000504 BCR_TRACE(2, "Function of type: " << Ty << "\n");
Chris Lattner00950542001-06-06 20:29:01 +0000505 }
506
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000507 ALIGN32(Buf, End);
Chris Lattner74734132002-08-17 22:01:27 +0000508
509 // Now that the function signature list is set up, reverse it so that we can
510 // remove elements efficiently from the back of the vector.
511 std::reverse(FunctionSignatureList.begin(), FunctionSignatureList.end());
Chris Lattner00950542001-06-06 20:29:01 +0000512
513 // This is for future proofing... in the future extra fields may be added that
514 // we don't understand, so we transparently ignore them.
515 //
516 Buf = End;
Chris Lattner00950542001-06-06 20:29:01 +0000517}
518
Misha Brukman12c29d12003-09-22 23:38:23 +0000519void BytecodeParser::ParseVersionInfo(const unsigned char *&Buf,
Chris Lattner12e64652003-05-22 18:08:30 +0000520 const unsigned char *EndBuf) {
Chris Lattner036b8aa2003-03-06 17:55:45 +0000521 unsigned Version;
Misha Brukman12c29d12003-09-22 23:38:23 +0000522 if (read_vbr(Buf, EndBuf, Version)) throw Error_readvbr;
Chris Lattner036b8aa2003-03-06 17:55:45 +0000523
524 // Unpack version number: low four bits are for flags, top bits = version
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000525 Module::Endianness Endianness;
526 Module::PointerSize PointerSize;
527 Endianness = (Version & 1) ? Module::BigEndian : Module::LittleEndian;
528 PointerSize = (Version & 2) ? Module::Pointer64 : Module::Pointer32;
529
530 bool hasNoEndianness = Version & 4;
531 bool hasNoPointerSize = Version & 8;
532
533 RevisionNum = Version >> 4;
Chris Lattnere3869c82003-04-16 21:16:05 +0000534
535 // Default values for the current bytecode version
Chris Lattner036b8aa2003-03-06 17:55:45 +0000536 HasImplicitZeroInitializer = true;
Chris Lattnere3869c82003-04-16 21:16:05 +0000537 hasInternalMarkerOnly = false;
538 FirstDerivedTyID = 14;
Chris Lattner036b8aa2003-03-06 17:55:45 +0000539
540 switch (RevisionNum) {
541 case 0: // Initial revision
Chris Lattner52e20b02003-03-19 20:54:26 +0000542 // Version #0 didn't have any of the flags stored correctly, and in fact as
543 // only valid with a 14 in the flags values. Also, it does not support
544 // encoding zero initializers for arrays compactly.
545 //
Misha Brukman12c29d12003-09-22 23:38:23 +0000546 if (Version != 14) throw std::string("Unknown revision 0 flags?");
Chris Lattner036b8aa2003-03-06 17:55:45 +0000547 HasImplicitZeroInitializer = false;
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000548 Endianness = Module::BigEndian;
549 PointerSize = Module::Pointer64;
Chris Lattnere3869c82003-04-16 21:16:05 +0000550 hasInternalMarkerOnly = true;
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000551 hasNoEndianness = hasNoPointerSize = false;
Chris Lattner036b8aa2003-03-06 17:55:45 +0000552 break;
553 case 1:
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000554 // Version #1 has four bit fields: isBigEndian, hasLongPointers,
555 // hasNoEndianness, and hasNoPointerSize.
Chris Lattnere3869c82003-04-16 21:16:05 +0000556 hasInternalMarkerOnly = true;
557 break;
558 case 2:
559 // Version #2 added information about all 4 linkage types instead of just
560 // having internal and external.
Chris Lattner036b8aa2003-03-06 17:55:45 +0000561 break;
562 default:
Misha Brukman12c29d12003-09-22 23:38:23 +0000563 throw std::string("Unknown bytecode version number!");
Chris Lattner036b8aa2003-03-06 17:55:45 +0000564 }
565
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000566 if (hasNoEndianness) Endianness = Module::AnyEndianness;
567 if (hasNoPointerSize) PointerSize = Module::AnyPointerSize;
Chris Lattner76e38962003-04-22 18:15:10 +0000568
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000569 TheModule->setEndianness(Endianness);
570 TheModule->setPointerSize(PointerSize);
Chris Lattner036b8aa2003-03-06 17:55:45 +0000571 BCR_TRACE(1, "Bytecode Rev = " << (unsigned)RevisionNum << "\n");
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000572 BCR_TRACE(1, "Endianness/PointerSize = " << Endianness << ","
573 << PointerSize << "\n");
Chris Lattner036b8aa2003-03-06 17:55:45 +0000574 BCR_TRACE(1, "HasImplicitZeroInit = " << HasImplicitZeroInitializer << "\n");
Chris Lattner036b8aa2003-03-06 17:55:45 +0000575}
576
Misha Brukman12c29d12003-09-22 23:38:23 +0000577void BytecodeParser::ParseModule(const unsigned char *Buf,
Chris Lattner12e64652003-05-22 18:08:30 +0000578 const unsigned char *EndBuf) {
Chris Lattner00950542001-06-06 20:29:01 +0000579 unsigned Type, Size;
Misha Brukman12c29d12003-09-22 23:38:23 +0000580 readBlock(Buf, EndBuf, Type, Size);
581 if (Type != BytecodeFormat::Module || Buf+Size != EndBuf)
582 throw std::string("Expected Module packet! B: "+
583 utostr((unsigned)(intptr_t)Buf) + ", S: "+utostr(Size)+
584 " E: "+utostr((unsigned)(intptr_t)EndBuf)); // Hrm, not a class?
Chris Lattner00950542001-06-06 20:29:01 +0000585
Chris Lattner1d670cc2001-09-07 16:37:43 +0000586 BCR_TRACE(0, "BLOCK BytecodeFormat::Module: {\n");
Chris Lattner74734132002-08-17 22:01:27 +0000587 FunctionSignatureList.clear(); // Just in case...
Chris Lattner00950542001-06-06 20:29:01 +0000588
589 // Read into instance variables...
Misha Brukman12c29d12003-09-22 23:38:23 +0000590 ParseVersionInfo(Buf, EndBuf);
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000591 ALIGN32(Buf, EndBuf);
Chris Lattner00950542001-06-06 20:29:01 +0000592
Chris Lattner00950542001-06-06 20:29:01 +0000593 while (Buf < EndBuf) {
Chris Lattnerb6c46952003-03-06 17:03:28 +0000594 const unsigned char *OldBuf = Buf;
Misha Brukman12c29d12003-09-22 23:38:23 +0000595 readBlock(Buf, EndBuf, Type, Size);
Chris Lattner00950542001-06-06 20:29:01 +0000596 switch (Type) {
Chris Lattner52e20b02003-03-19 20:54:26 +0000597 case BytecodeFormat::GlobalTypePlane:
598 BCR_TRACE(1, "BLOCK BytecodeFormat::GlobalTypePlane: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000599 ParseGlobalTypes(Buf, Buf+Size);
Chris Lattner52e20b02003-03-19 20:54:26 +0000600 break;
601
602 case BytecodeFormat::ModuleGlobalInfo:
603 BCR_TRACE(1, "BLOCK BytecodeFormat::ModuleGlobalInfo: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000604 ParseModuleGlobalInfo(Buf, Buf+Size);
Chris Lattner52e20b02003-03-19 20:54:26 +0000605 break;
606
Chris Lattner1d670cc2001-09-07 16:37:43 +0000607 case BytecodeFormat::ConstantPool:
608 BCR_TRACE(1, "BLOCK BytecodeFormat::ConstantPool: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000609 ParseConstantPool(Buf, Buf+Size, ModuleValues, ModuleTypeValues);
Chris Lattner00950542001-06-06 20:29:01 +0000610 break;
611
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000612 case BytecodeFormat::Function: {
613 BCR_TRACE(1, "BLOCK BytecodeFormat::Function: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000614 ParseFunction(Buf, Buf+Size);
Chris Lattner00950542001-06-06 20:29:01 +0000615 break;
616 }
617
618 case BytecodeFormat::SymbolTable:
Chris Lattner1d670cc2001-09-07 16:37:43 +0000619 BCR_TRACE(1, "BLOCK BytecodeFormat::SymbolTable: {\n");
Chris Lattner4ee8ef22003-10-08 22:52:54 +0000620 ParseSymbolTable(Buf, Buf+Size, &TheModule->getSymbolTable(), 0);
Chris Lattner00950542001-06-06 20:29:01 +0000621 break;
622
623 default:
Chris Lattner00950542001-06-06 20:29:01 +0000624 Buf += Size;
Misha Brukman12c29d12003-09-22 23:38:23 +0000625 if (OldBuf > Buf) throw std::string("Expected Module Block!");
Chris Lattner00950542001-06-06 20:29:01 +0000626 break;
627 }
Chris Lattner1d670cc2001-09-07 16:37:43 +0000628 BCR_TRACE(1, "} end block\n");
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000629 ALIGN32(Buf, EndBuf);
Chris Lattner00950542001-06-06 20:29:01 +0000630 }
631
Chris Lattner52e20b02003-03-19 20:54:26 +0000632 // After the module constant pool has been read, we can safely initialize
633 // global variables...
634 while (!GlobalInits.empty()) {
635 GlobalVariable *GV = GlobalInits.back().first;
636 unsigned Slot = GlobalInits.back().second;
637 GlobalInits.pop_back();
638
639 // Look up the initializer value...
640 if (Value *V = getValue(GV->getType()->getElementType(), Slot, false)) {
Misha Brukman12c29d12003-09-22 23:38:23 +0000641 if (GV->hasInitializer())
642 throw std::string("Global *already* has an initializer?!");
Chris Lattner52e20b02003-03-19 20:54:26 +0000643 GV->setInitializer(cast<Constant>(V));
644 } else
Misha Brukman12c29d12003-09-22 23:38:23 +0000645 throw std::string("Cannot find initializer value.");
Chris Lattner52e20b02003-03-19 20:54:26 +0000646 }
647
Misha Brukman12c29d12003-09-22 23:38:23 +0000648 if (!FunctionSignatureList.empty())
649 throw std::string("Function expected, but bytecode stream ended!");
Chris Lattner1d670cc2001-09-07 16:37:43 +0000650
651 BCR_TRACE(0, "} end block\n\n");
Chris Lattner00950542001-06-06 20:29:01 +0000652}
653
Misha Brukman12c29d12003-09-22 23:38:23 +0000654void
655BytecodeParser::ParseBytecode(const unsigned char *Buf, unsigned Length,
656 const std::string &ModuleID) {
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000657
Misha Brukman12c29d12003-09-22 23:38:23 +0000658 unsigned char *EndBuf = (unsigned char*)(Buf + Length);
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000659
Chris Lattner00950542001-06-06 20:29:01 +0000660 // Read and check signature...
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000661 unsigned Sig;
Chris Lattner00950542001-06-06 20:29:01 +0000662 if (read(Buf, EndBuf, Sig) ||
Misha Brukman12c29d12003-09-22 23:38:23 +0000663 Sig != ('l' | ('l' << 8) | ('v' << 16) | ('m' << 24)))
664 throw std::string("Invalid bytecode signature!");
Chris Lattner00950542001-06-06 20:29:01 +0000665
Chris Lattner75f20532003-04-22 18:02:52 +0000666 TheModule = new Module(ModuleID);
Misha Brukman12c29d12003-09-22 23:38:23 +0000667 try {
668 ParseModule(Buf, EndBuf);
669 } catch (std::string &Error) {
Chris Lattnera2602f32003-05-22 18:26:48 +0000670 freeState(); // Must destroy handles before deleting module!
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000671 delete TheModule;
672 TheModule = 0;
Chris Lattnerb0b7c0d2003-09-26 14:44:52 +0000673 throw;
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000674 }
Chris Lattner00950542001-06-06 20:29:01 +0000675}