blob: 067b70b80625ef4341c44f1e336dc9887445cfef [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
34void
35BytecodeParser::getTypeSlot(const Type *Ty, unsigned &Slot) {
Chris Lattner00950542001-06-06 20:29:01 +000036 if (Ty->isPrimitiveType()) {
37 Slot = Ty->getPrimitiveID();
38 } else {
Chris Lattner2a7b6ba2003-03-06 17:15:19 +000039 // Check the function level types first...
Chris Lattner6e5a0e42003-03-06 17:18:14 +000040 TypeValuesListTy::iterator I = find(FunctionTypeValues.begin(),
Misha Brukman12c29d12003-09-22 23:38:23 +000041 FunctionTypeValues.end(), Ty);
Chris Lattner6e5a0e42003-03-06 17:18:14 +000042 if (I != FunctionTypeValues.end()) {
Misha Brukman12c29d12003-09-22 23:38:23 +000043 Slot = FirstDerivedTyID + ModuleTypeValues.size() +
44 (&*I - &FunctionTypeValues[0]);
Chris Lattner1d670cc2001-09-07 16:37:43 +000045 } else {
46 I = find(ModuleTypeValues.begin(), ModuleTypeValues.end(), Ty);
Misha Brukman12c29d12003-09-22 23:38:23 +000047 if (I == ModuleTypeValues.end())
48 throw std::string("Didn't find type in ModuleTypeValues.");
Chris Lattner1d670cc2001-09-07 16:37:43 +000049 Slot = FirstDerivedTyID + (&*I - &ModuleTypeValues[0]);
50 }
Chris Lattner00950542001-06-06 20:29:01 +000051 }
Chris Lattner697954c2002-01-20 22:54:45 +000052 //cerr << "getTypeSlot '" << Ty->getName() << "' = " << Slot << "\n";
Chris Lattner00950542001-06-06 20:29:01 +000053}
54
55const Type *BytecodeParser::getType(unsigned ID) {
Chris Lattner8cdc6b72002-10-23 00:51:54 +000056 if (ID < Type::NumPrimitiveIDs) {
57 const Type *T = Type::getPrimitiveType((Type::PrimitiveID)ID);
58 if (T) return T;
59 }
Chris Lattner00950542001-06-06 20:29:01 +000060
Chris Lattner697954c2002-01-20 22:54:45 +000061 //cerr << "Looking up Type ID: " << ID << "\n";
Chris Lattner8cdc6b72002-10-23 00:51:54 +000062 const Value *V = getValue(Type::TypeTy, ID, false);
63 return cast_or_null<Type>(V);
Chris Lattner00950542001-06-06 20:29:01 +000064}
65
Chris Lattner52e20b02003-03-19 20:54:26 +000066int BytecodeParser::insertValue(Value *Val, ValueTable &ValueTab) {
67 assert((!HasImplicitZeroInitializer || !isa<Constant>(Val) ||
68 Val->getType()->isPrimitiveType() ||
69 !cast<Constant>(Val)->isNullValue()) &&
70 "Cannot read null values from bytecode!");
Chris Lattner00950542001-06-06 20:29:01 +000071 unsigned type;
Misha Brukman12c29d12003-09-22 23:38:23 +000072 getTypeSlot(Val->getType(), type);
Chris Lattner1d670cc2001-09-07 16:37:43 +000073 assert(type != Type::TypeTyID && "Types should never be insertValue'd!");
Chris Lattner00950542001-06-06 20:29:01 +000074
Chris Lattner52e20b02003-03-19 20:54:26 +000075 if (ValueTab.size() <= type) {
76 unsigned OldSize = ValueTab.size();
77 ValueTab.resize(type+1);
78 while (OldSize != type+1)
79 ValueTab[OldSize++] = new ValueList();
Chris Lattner036b8aa2003-03-06 17:55:45 +000080 }
Chris Lattner00950542001-06-06 20:29:01 +000081
82 //cerr << "insertValue Values[" << type << "][" << ValueTab[type].size()
Misha Brukman12c29d12003-09-22 23:38:23 +000083 // << "] = " << Val << "\n";
Chris Lattner52e20b02003-03-19 20:54:26 +000084 ValueTab[type]->push_back(Val);
Chris Lattner00950542001-06-06 20:29:01 +000085
Chris Lattner52e20b02003-03-19 20:54:26 +000086 bool HasOffset = HasImplicitZeroInitializer &&
Misha Brukman12c29d12003-09-22 23:38:23 +000087 !Val->getType()->isPrimitiveType();
Chris Lattner52e20b02003-03-19 20:54:26 +000088
89 return ValueTab[type]->size()-1 + HasOffset;
90}
91
92
93void BytecodeParser::setValueTo(ValueTable &ValueTab, unsigned Slot,
94 Value *Val) {
95 assert(&ValueTab == &ModuleValues && "Can only setValueTo on Module values!");
96 unsigned type;
Misha Brukman12c29d12003-09-22 23:38:23 +000097 getTypeSlot(Val->getType(), type);
98
Chris Lattner52e20b02003-03-19 20:54:26 +000099 assert((!HasImplicitZeroInitializer || Slot != 0) &&
100 "Cannot change zero init");
101 assert(type < ValueTab.size() && Slot <= ValueTab[type]->size());
102 ValueTab[type]->setOperand(Slot-HasImplicitZeroInitializer, Val);
Chris Lattner00950542001-06-06 20:29:01 +0000103}
104
105Value *BytecodeParser::getValue(const Type *Ty, unsigned oNum, bool Create) {
106 unsigned Num = oNum;
107 unsigned type; // The type plane it lives in...
Misha Brukman12c29d12003-09-22 23:38:23 +0000108 getTypeSlot(Ty, type);
Chris Lattner00950542001-06-06 20:29:01 +0000109
Misha Brukman12c29d12003-09-22 23:38:23 +0000110 if (type == Type::TypeTyID) { // The 'type' plane has implicit values
Chris Lattner1d670cc2001-09-07 16:37:43 +0000111 assert(Create == false);
Chris Lattner8cdc6b72002-10-23 00:51:54 +0000112 if (Num < Type::NumPrimitiveIDs) {
113 const Type *T = Type::getPrimitiveType((Type::PrimitiveID)Num);
114 if (T) return (Value*)T; // Asked for a primitive type...
115 }
Chris Lattner00950542001-06-06 20:29:01 +0000116
117 // Otherwise, derived types need offset...
118 Num -= FirstDerivedTyID;
Chris Lattner1d670cc2001-09-07 16:37:43 +0000119
Misha Brukman12c29d12003-09-22 23:38:23 +0000120 // Is it a module-level type?
Chris Lattner1d670cc2001-09-07 16:37:43 +0000121 if (Num < ModuleTypeValues.size())
Chris Lattner05950c32001-10-13 06:47:01 +0000122 return (Value*)ModuleTypeValues[Num].get();
Chris Lattner1d670cc2001-09-07 16:37:43 +0000123
Misha Brukman12c29d12003-09-22 23:38:23 +0000124 // Nope, is it a function-level type?
Chris Lattner1d670cc2001-09-07 16:37:43 +0000125 Num -= ModuleTypeValues.size();
Chris Lattner6e5a0e42003-03-06 17:18:14 +0000126 if (Num < FunctionTypeValues.size())
127 return (Value*)FunctionTypeValues[Num].get();
Chris Lattner1d670cc2001-09-07 16:37:43 +0000128
129 return 0;
Chris Lattner00950542001-06-06 20:29:01 +0000130 }
131
Chris Lattner52e20b02003-03-19 20:54:26 +0000132 if (HasImplicitZeroInitializer && type >= FirstDerivedTyID) {
133 if (Num == 0)
134 return Constant::getNullValue(Ty);
135 --Num;
Chris Lattner00950542001-06-06 20:29:01 +0000136 }
137
Chris Lattner52e20b02003-03-19 20:54:26 +0000138 if (type < ModuleValues.size()) {
139 if (Num < ModuleValues[type]->size())
140 return ModuleValues[type]->getOperand(Num);
141 Num -= ModuleValues[type]->size();
142 }
143
144 if (Values.size() > type && Values[type]->size() > Num)
145 return Values[type]->getOperand(Num);
Chris Lattner00950542001-06-06 20:29:01 +0000146
Chris Lattner74734132002-08-17 22:01:27 +0000147 if (!Create) return 0; // Do not create a placeholder?
Chris Lattner00950542001-06-06 20:29:01 +0000148
149 Value *d = 0;
150 switch (Ty->getPrimitiveID()) {
Chris Lattner74734132002-08-17 22:01:27 +0000151 case Type::LabelTyID:
152 d = new BBPHolder(Ty, oNum);
153 break;
154 default:
155 d = new ValPHolder(Ty, oNum);
156 break;
Chris Lattner00950542001-06-06 20:29:01 +0000157 }
158
159 assert(d != 0 && "How did we not make something?");
Chris Lattner74734132002-08-17 22:01:27 +0000160 if (insertValue(d, LateResolveValues) == -1) return 0;
Chris Lattner00950542001-06-06 20:29:01 +0000161 return d;
162}
163
Chris Lattnerbbd4b302002-10-14 03:33:02 +0000164/// getConstantValue - Just like getValue, except that it returns a null pointer
165/// only on error. It always returns a constant (meaning that if the value is
166/// defined, but is not a constant, that is an error). If the specified
167/// constant hasn't been parsed yet, a placeholder is defined and used. Later,
168/// after the real value is parsed, the placeholder is eliminated.
169///
170Constant *BytecodeParser::getConstantValue(const Type *Ty, unsigned Slot) {
171 if (Value *V = getValue(Ty, Slot, false))
172 return dyn_cast<Constant>(V); // If we already have the value parsed...
173
Chris Lattner52e20b02003-03-19 20:54:26 +0000174 std::pair<const Type*, unsigned> Key(Ty, Slot);
175 GlobalRefsType::iterator I = GlobalRefs.lower_bound(Key);
176
177 if (I != GlobalRefs.end() && I->first == Key) {
Chris Lattnerbbd4b302002-10-14 03:33:02 +0000178 BCR_TRACE(5, "Previous forward ref found!\n");
179 return cast<Constant>(I->second);
180 } else {
181 // Create a placeholder for the constant reference and
182 // keep track of the fact that we have a forward ref to recycle it
183 BCR_TRACE(5, "Creating new forward ref to a constant!\n");
184 Constant *C = new ConstPHolder(Ty, Slot);
185
186 // Keep track of the fact that we have a forward ref to recycle it
Chris Lattner52e20b02003-03-19 20:54:26 +0000187 GlobalRefs.insert(I, std::make_pair(Key, C));
Chris Lattnerbbd4b302002-10-14 03:33:02 +0000188 return C;
189 }
190}
191
192
Misha Brukman12c29d12003-09-22 23:38:23 +0000193void BytecodeParser::postResolveValues(ValueTable &ValTab) {
Chris Lattner52e20b02003-03-19 20:54:26 +0000194 while (!ValTab.empty()) {
195 ValueList &DL = *ValTab.back();
196 ValTab.pop_back();
Chris Lattner00950542001-06-06 20:29:01 +0000197
Chris Lattner52e20b02003-03-19 20:54:26 +0000198 while (!DL.empty()) {
199 Value *D = DL.back();
200 unsigned IDNumber = getValueIDNumberFromPlaceHolder(D);
Chris Lattner00950542001-06-06 20:29:01 +0000201 DL.pop_back();
202
203 Value *NewDef = getValue(D->getType(), IDNumber, false);
204 if (NewDef == 0) {
Misha Brukman12c29d12003-09-22 23:38:23 +0000205 throw std::string("Unresolvable reference found: <" +
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000206 D->getType()->getDescription() + ">:" +
207 utostr(IDNumber) + ".");
Chris Lattner00950542001-06-06 20:29:01 +0000208 } else {
Misha Brukman12c29d12003-09-22 23:38:23 +0000209 // Fixup all of the uses of this placeholder def...
Chris Lattner00950542001-06-06 20:29:01 +0000210 D->replaceAllUsesWith(NewDef);
211
212 // Now that all the uses are gone, delete the placeholder...
213 // If we couldn't find a def (error case), then leak a little
Misha Brukman12c29d12003-09-22 23:38:23 +0000214 delete D; // memory, 'cause otherwise we can't remove all uses!
Chris Lattner00950542001-06-06 20:29:01 +0000215 }
216 }
Chris Lattner52e20b02003-03-19 20:54:26 +0000217 delete &DL;
Chris Lattner00950542001-06-06 20:29:01 +0000218 }
Chris Lattner00950542001-06-06 20:29:01 +0000219}
220
Misha Brukman12c29d12003-09-22 23:38:23 +0000221std::auto_ptr<BasicBlock>
222BytecodeParser::ParseBasicBlock(const unsigned char *&Buf,
223 const unsigned char *EndBuf) {
224 std::auto_ptr<BasicBlock> BB(new BasicBlock());
Chris Lattner00950542001-06-06 20:29:01 +0000225
226 while (Buf < EndBuf) {
Chris Lattner1d670cc2001-09-07 16:37:43 +0000227 Instruction *Inst;
Misha Brukman12c29d12003-09-22 23:38:23 +0000228 ParseInstruction(Buf, EndBuf, Inst);
229
230 if (Inst == 0) { throw std::string("Could not parse Instruction."); }
231 if (insertValue(Inst, Values) == -1) {
232 throw std::string("Could not insert value.");
Chris Lattner00950542001-06-06 20:29:01 +0000233 }
234
Chris Lattner1d670cc2001-09-07 16:37:43 +0000235 BB->getInstList().push_back(Inst);
Chris Lattner1d670cc2001-09-07 16:37:43 +0000236 BCR_TRACE(4, Inst);
Chris Lattner00950542001-06-06 20:29:01 +0000237 }
238
Misha Brukman12c29d12003-09-22 23:38:23 +0000239 return BB;
Chris Lattner00950542001-06-06 20:29:01 +0000240}
241
Misha Brukman12c29d12003-09-22 23:38:23 +0000242void BytecodeParser::ParseSymbolTable(const unsigned char *&Buf,
Chris Lattner12e64652003-05-22 18:08:30 +0000243 const unsigned char *EndBuf,
Misha Brukman12c29d12003-09-22 23:38:23 +0000244 SymbolTable *ST) {
Chris Lattner00950542001-06-06 20:29:01 +0000245 while (Buf < EndBuf) {
246 // Symtab block header: [num entries][type id number]
247 unsigned NumEntries, Typ;
248 if (read_vbr(Buf, EndBuf, NumEntries) ||
Misha Brukman12c29d12003-09-22 23:38:23 +0000249 read_vbr(Buf, EndBuf, Typ)) throw Error_readvbr;
Chris Lattner00950542001-06-06 20:29:01 +0000250 const Type *Ty = getType(Typ);
Misha Brukman12c29d12003-09-22 23:38:23 +0000251 if (Ty == 0) throw std::string("Invalid type read in symbol table.");
Chris Lattner00950542001-06-06 20:29:01 +0000252
Chris Lattner1d670cc2001-09-07 16:37:43 +0000253 BCR_TRACE(3, "Plane Type: '" << Ty << "' with " << NumEntries <<
Misha Brukman12c29d12003-09-22 23:38:23 +0000254 " entries\n");
Chris Lattner1d670cc2001-09-07 16:37:43 +0000255
Chris Lattner7fc9fe32001-06-27 23:41:11 +0000256 for (unsigned i = 0; i < NumEntries; ++i) {
Chris Lattner00950542001-06-06 20:29:01 +0000257 // Symtab entry: [def slot #][name]
258 unsigned slot;
Misha Brukman12c29d12003-09-22 23:38:23 +0000259 if (read_vbr(Buf, EndBuf, slot)) throw Error_readvbr;
Chris Lattner697954c2002-01-20 22:54:45 +0000260 std::string Name;
Chris Lattner00950542001-06-06 20:29:01 +0000261 if (read(Buf, EndBuf, Name, false)) // Not aligned...
Misha Brukman12c29d12003-09-22 23:38:23 +0000262 throw std::string("Buffer not aligned.");
Chris Lattner00950542001-06-06 20:29:01 +0000263
Chris Lattner52e20b02003-03-19 20:54:26 +0000264 Value *V = getValue(Ty, slot, false); // Find mapping...
265 if (V == 0) {
Misha Brukman12c29d12003-09-22 23:38:23 +0000266 BCR_TRACE(3, "FAILED LOOKUP: Slot #" << slot << "\n");
267 throw std::string("Failed value look-up.");
Chris Lattner1d670cc2001-09-07 16:37:43 +0000268 }
Chris Lattner52e20b02003-03-19 20:54:26 +0000269 BCR_TRACE(4, "Map: '" << Name << "' to #" << slot << ":" << *V;
Misha Brukman12c29d12003-09-22 23:38:23 +0000270 if (!isa<Instruction>(V)) std::cerr << "\n");
Chris Lattner1d670cc2001-09-07 16:37:43 +0000271
Chris Lattner52e20b02003-03-19 20:54:26 +0000272 V->setName(Name, ST);
Chris Lattner00950542001-06-06 20:29:01 +0000273 }
274 }
275
Misha Brukman12c29d12003-09-22 23:38:23 +0000276 if (Buf > EndBuf) throw std::string("Tried to read past end of buffer.");
Chris Lattner00950542001-06-06 20:29:01 +0000277}
278
Chris Lattner74734132002-08-17 22:01:27 +0000279void BytecodeParser::ResolveReferencesToValue(Value *NewV, unsigned Slot) {
Chris Lattner0d75d8d72003-03-06 16:32:25 +0000280 GlobalRefsType::iterator I = GlobalRefs.find(std::make_pair(NewV->getType(),
281 Slot));
Chris Lattner74734132002-08-17 22:01:27 +0000282 if (I == GlobalRefs.end()) return; // Never forward referenced?
Chris Lattner00950542001-06-06 20:29:01 +0000283
Chris Lattner74734132002-08-17 22:01:27 +0000284 BCR_TRACE(3, "Mutating forward refs!\n");
285 Value *VPH = I->second; // Get the placeholder...
Vikram S. Advec1e4a812002-07-14 23:04:18 +0000286
Chris Lattner52e20b02003-03-19 20:54:26 +0000287 VPH->replaceAllUsesWith(NewV);
288
289 // If this is a global variable being resolved, remove the placeholder from
290 // the module...
291 if (GlobalValue* GVal = dyn_cast<GlobalValue>(NewV))
292 GVal->getParent()->getGlobalList().remove(cast<GlobalVariable>(VPH));
Vikram S. Advec1e4a812002-07-14 23:04:18 +0000293
Chris Lattner74734132002-08-17 22:01:27 +0000294 delete VPH; // Delete the old placeholder
295 GlobalRefs.erase(I); // Remove the map entry for it
Vikram S. Advec1e4a812002-07-14 23:04:18 +0000296}
297
Misha Brukman12c29d12003-09-22 23:38:23 +0000298void
299BytecodeParser::ParseFunction(const unsigned char *&Buf,
300 const unsigned char *EndBuf) {
301 if (FunctionSignatureList.empty())
302 throw std::string("FunctionSignatureList empty!");
Chris Lattner52e20b02003-03-19 20:54:26 +0000303
Misha Brukman12c29d12003-09-22 23:38:23 +0000304 Function *F = FunctionSignatureList.back().first;
305 unsigned FunctionSlot = FunctionSignatureList.back().second;
306 FunctionSignatureList.pop_back();
307
308 // Save the information for future reading of the function
309 LazyFunctionInfo *LFI = new LazyFunctionInfo();
310 LFI->Buf = Buf; LFI->EndBuf = EndBuf; LFI->FunctionSlot = FunctionSlot;
311 LazyFunctionLoadMap[F] = LFI;
312 // Pretend we've `parsed' this function
313 Buf = EndBuf;
314}
315
316void BytecodeParser::materializeFunction(Function* F) {
317 // Find {start, end} pointers and slot in the map. If not there, we're done.
318 std::map<Function*, LazyFunctionInfo*>::iterator Fi =
319 LazyFunctionLoadMap.find(F);
320 if (Fi == LazyFunctionLoadMap.end()) return;
321
322 LazyFunctionInfo *LFI = Fi->second;
323 const unsigned char *Buf = LFI->Buf;
324 const unsigned char *EndBuf = LFI->EndBuf;
325 unsigned FunctionSlot = LFI->FunctionSlot;
326 LazyFunctionLoadMap.erase(Fi);
327 delete LFI;
Chris Lattner00950542001-06-06 20:29:01 +0000328
Chris Lattnere3869c82003-04-16 21:16:05 +0000329 GlobalValue::LinkageTypes Linkage = GlobalValue::ExternalLinkage;
330
331 if (!hasInternalMarkerOnly) {
332 unsigned LinkageType;
Misha Brukman12c29d12003-09-22 23:38:23 +0000333 if (read_vbr(Buf, EndBuf, LinkageType))
334 throw std::string("ParseFunction: Error reading from buffer.");
335 if (LinkageType & ~0x3)
336 throw std::string("Invalid linkage type for Function.");
Chris Lattnere3869c82003-04-16 21:16:05 +0000337 Linkage = (GlobalValue::LinkageTypes)LinkageType;
338 } else {
339 // We used to only support two linkage models: internal and external
340 unsigned isInternal;
Misha Brukman12c29d12003-09-22 23:38:23 +0000341 if (read_vbr(Buf, EndBuf, isInternal))
342 throw std::string("ParseFunction: Error reading from buffer.");
Chris Lattnere3869c82003-04-16 21:16:05 +0000343 if (isInternal) Linkage = GlobalValue::InternalLinkage;
344 }
Chris Lattnerd23b1d32001-11-26 18:56:10 +0000345
Chris Lattnere3869c82003-04-16 21:16:05 +0000346 F->setLinkage(Linkage);
Chris Lattner00950542001-06-06 20:29:01 +0000347
Chris Lattner52e20b02003-03-19 20:54:26 +0000348 const FunctionType::ParamTypes &Params =F->getFunctionType()->getParamTypes();
349 Function::aiterator AI = F->abegin();
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000350 for (FunctionType::ParamTypes::const_iterator It = Params.begin();
Chris Lattner69da5cf2002-10-13 20:57:00 +0000351 It != Params.end(); ++It, ++AI) {
Misha Brukman12c29d12003-09-22 23:38:23 +0000352 if (insertValue(AI, Values) == -1)
353 throw std::string("Error reading function arguments!");
Chris Lattner00950542001-06-06 20:29:01 +0000354 }
355
356 while (Buf < EndBuf) {
357 unsigned Type, Size;
Chris Lattnerb6c46952003-03-06 17:03:28 +0000358 const unsigned char *OldBuf = Buf;
Misha Brukman12c29d12003-09-22 23:38:23 +0000359 readBlock(Buf, EndBuf, Type, Size);
Chris Lattner00950542001-06-06 20:29:01 +0000360
361 switch (Type) {
Misha Brukman12c29d12003-09-22 23:38:23 +0000362 case BytecodeFormat::ConstantPool: {
Chris Lattner1d670cc2001-09-07 16:37:43 +0000363 BCR_TRACE(2, "BLOCK BytecodeFormat::ConstantPool: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000364 ParseConstantPool(Buf, Buf+Size, Values, FunctionTypeValues);
Chris Lattner00950542001-06-06 20:29:01 +0000365 break;
Misha Brukman12c29d12003-09-22 23:38:23 +0000366 }
Chris Lattner00950542001-06-06 20:29:01 +0000367
368 case BytecodeFormat::BasicBlock: {
Chris Lattner1d670cc2001-09-07 16:37:43 +0000369 BCR_TRACE(2, "BLOCK BytecodeFormat::BasicBlock: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000370 std::auto_ptr<BasicBlock> BB = ParseBasicBlock(Buf, Buf+Size);
371 if (!BB.get() || insertValue(BB.get(), Values) == -1)
372 throw std::string("Parse error: BasicBlock");
Chris Lattner00950542001-06-06 20:29:01 +0000373
Misha Brukman12c29d12003-09-22 23:38:23 +0000374 F->getBasicBlockList().push_back(BB.release());
Chris Lattner00950542001-06-06 20:29:01 +0000375 break;
376 }
377
Misha Brukman12c29d12003-09-22 23:38:23 +0000378 case BytecodeFormat::SymbolTable: {
Chris Lattner1d670cc2001-09-07 16:37:43 +0000379 BCR_TRACE(2, "BLOCK BytecodeFormat::SymbolTable: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000380 ParseSymbolTable(Buf, Buf+Size, &F->getSymbolTable());
Chris Lattner00950542001-06-06 20:29:01 +0000381 break;
Misha Brukman12c29d12003-09-22 23:38:23 +0000382 }
Chris Lattner00950542001-06-06 20:29:01 +0000383
384 default:
Chris Lattner1d670cc2001-09-07 16:37:43 +0000385 BCR_TRACE(2, "BLOCK <unknown>:ignored! {\n");
Chris Lattner00950542001-06-06 20:29:01 +0000386 Buf += Size;
Misha Brukman12c29d12003-09-22 23:38:23 +0000387 if (OldBuf > Buf)
388 throw std::string("Wrapped around reading bytecode.");
Chris Lattner00950542001-06-06 20:29:01 +0000389 break;
390 }
Chris Lattner1d670cc2001-09-07 16:37:43 +0000391 BCR_TRACE(2, "} end block\n");
392
Misha Brukman12c29d12003-09-22 23:38:23 +0000393 // Malformed bc file if read past end of block.
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000394 ALIGN32(Buf, EndBuf);
Chris Lattner00950542001-06-06 20:29:01 +0000395 }
396
Misha Brukman12c29d12003-09-22 23:38:23 +0000397 // Check for unresolvable references
398 postResolveValues(LateResolveValues);
Chris Lattner00950542001-06-06 20:29:01 +0000399
Misha Brukman12c29d12003-09-22 23:38:23 +0000400 //ResolveReferencesToValue(F, FunctionSlot);
Chris Lattner00950542001-06-06 20:29:01 +0000401
Misha Brukman12c29d12003-09-22 23:38:23 +0000402 // Clear out function-level types...
Chris Lattner6e5a0e42003-03-06 17:18:14 +0000403 FunctionTypeValues.clear();
Chris Lattnere4d71a12001-09-14 22:03:42 +0000404
Chris Lattner52e20b02003-03-19 20:54:26 +0000405 freeTable(Values);
Chris Lattner00950542001-06-06 20:29:01 +0000406}
407
Misha Brukman12c29d12003-09-22 23:38:23 +0000408void BytecodeParser::ParseModuleGlobalInfo(const unsigned char *&Buf,
409 const unsigned char *End) {
410 if (!FunctionSignatureList.empty())
411 throw std::string("Two ModuleGlobalInfo packets found!");
Chris Lattner00950542001-06-06 20:29:01 +0000412
Chris Lattner70cc3392001-09-10 07:58:01 +0000413 // Read global variables...
414 unsigned VarType;
Misha Brukman12c29d12003-09-22 23:38:23 +0000415 if (read_vbr(Buf, End, VarType)) throw Error_readvbr;
Chris Lattner70cc3392001-09-10 07:58:01 +0000416 while (VarType != Type::VoidTyID) { // List is terminated by Void
Chris Lattnere3869c82003-04-16 21:16:05 +0000417 unsigned SlotNo;
418 GlobalValue::LinkageTypes Linkage;
419
420 if (!hasInternalMarkerOnly) {
421 // VarType Fields: bit0 = isConstant, bit1 = hasInitializer,
422 // bit2,3 = Linkage, bit4+ = slot#
423 SlotNo = VarType >> 4;
424 Linkage = (GlobalValue::LinkageTypes)((VarType >> 2) & 3);
425 } else {
426 // VarType Fields: bit0 = isConstant, bit1 = hasInitializer,
427 // bit2 = isInternal, bit3+ = slot#
428 SlotNo = VarType >> 3;
429 Linkage = (VarType & 4) ? GlobalValue::InternalLinkage :
430 GlobalValue::ExternalLinkage;
431 }
432
433 const Type *Ty = getType(SlotNo);
Misha Brukman12c29d12003-09-22 23:38:23 +0000434 if (!Ty || !isa<PointerType>(Ty))
435 throw std::string("Global not pointer type! Ty = " +
436 Ty->getDescription());
Chris Lattner70cc3392001-09-10 07:58:01 +0000437
Chris Lattner52e20b02003-03-19 20:54:26 +0000438 const Type *ElTy = cast<PointerType>(Ty)->getElementType();
Chris Lattnerd70684f2001-09-18 04:01:05 +0000439
Chris Lattner70cc3392001-09-10 07:58:01 +0000440 // Create the global variable...
Chris Lattner4ad02e72003-04-16 20:28:45 +0000441 GlobalVariable *GV = new GlobalVariable(ElTy, VarType & 1, Linkage,
Chris Lattner52e20b02003-03-19 20:54:26 +0000442 0, "", TheModule);
Chris Lattner05950c32001-10-13 06:47:01 +0000443 int DestSlot = insertValue(GV, ModuleValues);
Misha Brukman12c29d12003-09-22 23:38:23 +0000444 if (DestSlot == -1) throw Error_DestSlot;
Chris Lattner52e20b02003-03-19 20:54:26 +0000445 BCR_TRACE(2, "Global Variable of type: " << *Ty << "\n");
Chris Lattner74734132002-08-17 22:01:27 +0000446 ResolveReferencesToValue(GV, (unsigned)DestSlot);
Chris Lattner05950c32001-10-13 06:47:01 +0000447
Misha Brukman37f92e22003-09-11 22:34:13 +0000448 if (VarType & 2) { // Does it have an initializer?
Chris Lattner52e20b02003-03-19 20:54:26 +0000449 unsigned InitSlot;
Misha Brukman12c29d12003-09-22 23:38:23 +0000450 if (read_vbr(Buf, End, InitSlot)) throw Error_readvbr;
Chris Lattner52e20b02003-03-19 20:54:26 +0000451 GlobalInits.push_back(std::make_pair(GV, InitSlot));
452 }
Misha Brukman12c29d12003-09-22 23:38:23 +0000453 if (read_vbr(Buf, End, VarType)) throw Error_readvbr;
Chris Lattner70cc3392001-09-10 07:58:01 +0000454 }
455
Chris Lattner52e20b02003-03-19 20:54:26 +0000456 // Read the function objects for all of the functions that are coming
Chris Lattner74734132002-08-17 22:01:27 +0000457 unsigned FnSignature;
Misha Brukman12c29d12003-09-22 23:38:23 +0000458 if (read_vbr(Buf, End, FnSignature)) throw Error_readvbr;
Chris Lattner74734132002-08-17 22:01:27 +0000459 while (FnSignature != Type::VoidTyID) { // List is terminated by Void
460 const Type *Ty = getType(FnSignature);
Chris Lattneref9c23f2001-10-03 14:53:21 +0000461 if (!Ty || !isa<PointerType>(Ty) ||
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000462 !isa<FunctionType>(cast<PointerType>(Ty)->getElementType())) {
Misha Brukman12c29d12003-09-22 23:38:23 +0000463 throw std::string("Function not ptr to func type! Ty = " +
464 Ty->getDescription());
Chris Lattner00950542001-06-06 20:29:01 +0000465 }
Chris Lattner8cdc6b72002-10-23 00:51:54 +0000466
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000467 // We create functions by passing the underlying FunctionType to create...
Chris Lattner7a176752001-12-04 00:03:30 +0000468 Ty = cast<PointerType>(Ty)->getElementType();
Chris Lattner00950542001-06-06 20:29:01 +0000469
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000470 // When the ModuleGlobalInfo section is read, we load the type of each
471 // function and the 'ModuleValues' slot that it lands in. We then load a
472 // placeholder into its slot to reserve it. When the function is loaded,
473 // this placeholder is replaced.
Chris Lattner00950542001-06-06 20:29:01 +0000474
475 // Insert the placeholder...
Chris Lattner4ad02e72003-04-16 20:28:45 +0000476 Function *Func = new Function(cast<FunctionType>(Ty),
477 GlobalValue::InternalLinkage, "", TheModule);
Chris Lattner52e20b02003-03-19 20:54:26 +0000478 int DestSlot = insertValue(Func, ModuleValues);
Misha Brukman12c29d12003-09-22 23:38:23 +0000479 if (DestSlot == -1) throw Error_DestSlot;
Chris Lattner52e20b02003-03-19 20:54:26 +0000480 ResolveReferencesToValue(Func, (unsigned)DestSlot);
Chris Lattner00950542001-06-06 20:29:01 +0000481
Chris Lattner52e20b02003-03-19 20:54:26 +0000482 // Keep track of this information in a list that is emptied as functions are
483 // loaded...
Chris Lattner00950542001-06-06 20:29:01 +0000484 //
Chris Lattner52e20b02003-03-19 20:54:26 +0000485 FunctionSignatureList.push_back(std::make_pair(Func, DestSlot));
486
Misha Brukman12c29d12003-09-22 23:38:23 +0000487 if (read_vbr(Buf, End, FnSignature)) throw Error_readvbr;
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000488 BCR_TRACE(2, "Function of type: " << Ty << "\n");
Chris Lattner00950542001-06-06 20:29:01 +0000489 }
490
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000491 ALIGN32(Buf, End);
Chris Lattner74734132002-08-17 22:01:27 +0000492
493 // Now that the function signature list is set up, reverse it so that we can
494 // remove elements efficiently from the back of the vector.
495 std::reverse(FunctionSignatureList.begin(), FunctionSignatureList.end());
Chris Lattner00950542001-06-06 20:29:01 +0000496
497 // This is for future proofing... in the future extra fields may be added that
498 // we don't understand, so we transparently ignore them.
499 //
500 Buf = End;
Chris Lattner00950542001-06-06 20:29:01 +0000501}
502
Misha Brukman12c29d12003-09-22 23:38:23 +0000503void BytecodeParser::ParseVersionInfo(const unsigned char *&Buf,
Chris Lattner12e64652003-05-22 18:08:30 +0000504 const unsigned char *EndBuf) {
Chris Lattner036b8aa2003-03-06 17:55:45 +0000505 unsigned Version;
Misha Brukman12c29d12003-09-22 23:38:23 +0000506 if (read_vbr(Buf, EndBuf, Version)) throw Error_readvbr;
Chris Lattner036b8aa2003-03-06 17:55:45 +0000507
508 // Unpack version number: low four bits are for flags, top bits = version
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000509 Module::Endianness Endianness;
510 Module::PointerSize PointerSize;
511 Endianness = (Version & 1) ? Module::BigEndian : Module::LittleEndian;
512 PointerSize = (Version & 2) ? Module::Pointer64 : Module::Pointer32;
513
514 bool hasNoEndianness = Version & 4;
515 bool hasNoPointerSize = Version & 8;
516
517 RevisionNum = Version >> 4;
Chris Lattnere3869c82003-04-16 21:16:05 +0000518
519 // Default values for the current bytecode version
Chris Lattner036b8aa2003-03-06 17:55:45 +0000520 HasImplicitZeroInitializer = true;
Chris Lattnere3869c82003-04-16 21:16:05 +0000521 hasInternalMarkerOnly = false;
522 FirstDerivedTyID = 14;
Chris Lattner036b8aa2003-03-06 17:55:45 +0000523
524 switch (RevisionNum) {
525 case 0: // Initial revision
Chris Lattner52e20b02003-03-19 20:54:26 +0000526 // Version #0 didn't have any of the flags stored correctly, and in fact as
527 // only valid with a 14 in the flags values. Also, it does not support
528 // encoding zero initializers for arrays compactly.
529 //
Misha Brukman12c29d12003-09-22 23:38:23 +0000530 if (Version != 14) throw std::string("Unknown revision 0 flags?");
Chris Lattner036b8aa2003-03-06 17:55:45 +0000531 HasImplicitZeroInitializer = false;
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000532 Endianness = Module::BigEndian;
533 PointerSize = Module::Pointer64;
Chris Lattnere3869c82003-04-16 21:16:05 +0000534 hasInternalMarkerOnly = true;
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000535 hasNoEndianness = hasNoPointerSize = false;
Chris Lattner036b8aa2003-03-06 17:55:45 +0000536 break;
537 case 1:
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000538 // Version #1 has four bit fields: isBigEndian, hasLongPointers,
539 // hasNoEndianness, and hasNoPointerSize.
Chris Lattnere3869c82003-04-16 21:16:05 +0000540 hasInternalMarkerOnly = true;
541 break;
542 case 2:
543 // Version #2 added information about all 4 linkage types instead of just
544 // having internal and external.
Chris Lattner036b8aa2003-03-06 17:55:45 +0000545 break;
546 default:
Misha Brukman12c29d12003-09-22 23:38:23 +0000547 throw std::string("Unknown bytecode version number!");
Chris Lattner036b8aa2003-03-06 17:55:45 +0000548 }
549
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000550 if (hasNoEndianness) Endianness = Module::AnyEndianness;
551 if (hasNoPointerSize) PointerSize = Module::AnyPointerSize;
Chris Lattner76e38962003-04-22 18:15:10 +0000552
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000553 TheModule->setEndianness(Endianness);
554 TheModule->setPointerSize(PointerSize);
Chris Lattner036b8aa2003-03-06 17:55:45 +0000555 BCR_TRACE(1, "Bytecode Rev = " << (unsigned)RevisionNum << "\n");
Chris Lattnerd445c6b2003-08-24 13:47:36 +0000556 BCR_TRACE(1, "Endianness/PointerSize = " << Endianness << ","
557 << PointerSize << "\n");
Chris Lattner036b8aa2003-03-06 17:55:45 +0000558 BCR_TRACE(1, "HasImplicitZeroInit = " << HasImplicitZeroInitializer << "\n");
Chris Lattner036b8aa2003-03-06 17:55:45 +0000559}
560
Misha Brukman12c29d12003-09-22 23:38:23 +0000561void BytecodeParser::ParseModule(const unsigned char *Buf,
Chris Lattner12e64652003-05-22 18:08:30 +0000562 const unsigned char *EndBuf) {
Chris Lattner00950542001-06-06 20:29:01 +0000563 unsigned Type, Size;
Misha Brukman12c29d12003-09-22 23:38:23 +0000564 readBlock(Buf, EndBuf, Type, Size);
565 if (Type != BytecodeFormat::Module || Buf+Size != EndBuf)
566 throw std::string("Expected Module packet! B: "+
567 utostr((unsigned)(intptr_t)Buf) + ", S: "+utostr(Size)+
568 " E: "+utostr((unsigned)(intptr_t)EndBuf)); // Hrm, not a class?
Chris Lattner00950542001-06-06 20:29:01 +0000569
Chris Lattner1d670cc2001-09-07 16:37:43 +0000570 BCR_TRACE(0, "BLOCK BytecodeFormat::Module: {\n");
Chris Lattner74734132002-08-17 22:01:27 +0000571 FunctionSignatureList.clear(); // Just in case...
Chris Lattner00950542001-06-06 20:29:01 +0000572
573 // Read into instance variables...
Misha Brukman12c29d12003-09-22 23:38:23 +0000574 ParseVersionInfo(Buf, EndBuf);
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000575 ALIGN32(Buf, EndBuf);
Chris Lattner00950542001-06-06 20:29:01 +0000576
Chris Lattner00950542001-06-06 20:29:01 +0000577 while (Buf < EndBuf) {
Chris Lattnerb6c46952003-03-06 17:03:28 +0000578 const unsigned char *OldBuf = Buf;
Misha Brukman12c29d12003-09-22 23:38:23 +0000579 readBlock(Buf, EndBuf, Type, Size);
Chris Lattner00950542001-06-06 20:29:01 +0000580 switch (Type) {
Chris Lattner52e20b02003-03-19 20:54:26 +0000581 case BytecodeFormat::GlobalTypePlane:
582 BCR_TRACE(1, "BLOCK BytecodeFormat::GlobalTypePlane: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000583 ParseGlobalTypes(Buf, Buf+Size);
Chris Lattner52e20b02003-03-19 20:54:26 +0000584 break;
585
586 case BytecodeFormat::ModuleGlobalInfo:
587 BCR_TRACE(1, "BLOCK BytecodeFormat::ModuleGlobalInfo: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000588 ParseModuleGlobalInfo(Buf, Buf+Size);
Chris Lattner52e20b02003-03-19 20:54:26 +0000589 break;
590
Chris Lattner1d670cc2001-09-07 16:37:43 +0000591 case BytecodeFormat::ConstantPool:
592 BCR_TRACE(1, "BLOCK BytecodeFormat::ConstantPool: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000593 ParseConstantPool(Buf, Buf+Size, ModuleValues, ModuleTypeValues);
Chris Lattner00950542001-06-06 20:29:01 +0000594 break;
595
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000596 case BytecodeFormat::Function: {
597 BCR_TRACE(1, "BLOCK BytecodeFormat::Function: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000598 ParseFunction(Buf, Buf+Size);
Chris Lattner00950542001-06-06 20:29:01 +0000599 break;
600 }
601
602 case BytecodeFormat::SymbolTable:
Chris Lattner1d670cc2001-09-07 16:37:43 +0000603 BCR_TRACE(1, "BLOCK BytecodeFormat::SymbolTable: {\n");
Misha Brukman12c29d12003-09-22 23:38:23 +0000604 ParseSymbolTable(Buf, Buf+Size, &TheModule->getSymbolTable());
Chris Lattner00950542001-06-06 20:29:01 +0000605 break;
606
607 default:
Chris Lattner00950542001-06-06 20:29:01 +0000608 Buf += Size;
Misha Brukman12c29d12003-09-22 23:38:23 +0000609 if (OldBuf > Buf) throw std::string("Expected Module Block!");
Chris Lattner00950542001-06-06 20:29:01 +0000610 break;
611 }
Chris Lattner1d670cc2001-09-07 16:37:43 +0000612 BCR_TRACE(1, "} end block\n");
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000613 ALIGN32(Buf, EndBuf);
Chris Lattner00950542001-06-06 20:29:01 +0000614 }
615
Chris Lattner52e20b02003-03-19 20:54:26 +0000616 // After the module constant pool has been read, we can safely initialize
617 // global variables...
618 while (!GlobalInits.empty()) {
619 GlobalVariable *GV = GlobalInits.back().first;
620 unsigned Slot = GlobalInits.back().second;
621 GlobalInits.pop_back();
622
623 // Look up the initializer value...
624 if (Value *V = getValue(GV->getType()->getElementType(), Slot, false)) {
Misha Brukman12c29d12003-09-22 23:38:23 +0000625 if (GV->hasInitializer())
626 throw std::string("Global *already* has an initializer?!");
Chris Lattner52e20b02003-03-19 20:54:26 +0000627 GV->setInitializer(cast<Constant>(V));
628 } else
Misha Brukman12c29d12003-09-22 23:38:23 +0000629 throw std::string("Cannot find initializer value.");
Chris Lattner52e20b02003-03-19 20:54:26 +0000630 }
631
Misha Brukman12c29d12003-09-22 23:38:23 +0000632 if (!FunctionSignatureList.empty())
633 throw std::string("Function expected, but bytecode stream ended!");
Chris Lattner1d670cc2001-09-07 16:37:43 +0000634
635 BCR_TRACE(0, "} end block\n\n");
Chris Lattner00950542001-06-06 20:29:01 +0000636}
637
Misha Brukman12c29d12003-09-22 23:38:23 +0000638void
639BytecodeParser::ParseBytecode(const unsigned char *Buf, unsigned Length,
640 const std::string &ModuleID) {
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000641
Misha Brukman12c29d12003-09-22 23:38:23 +0000642 unsigned char *EndBuf = (unsigned char*)(Buf + Length);
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000643
Chris Lattner00950542001-06-06 20:29:01 +0000644 // Read and check signature...
Misha Brukmane0dd0d42003-09-23 16:15:29 +0000645 unsigned Sig;
Chris Lattner00950542001-06-06 20:29:01 +0000646 if (read(Buf, EndBuf, Sig) ||
Misha Brukman12c29d12003-09-22 23:38:23 +0000647 Sig != ('l' | ('l' << 8) | ('v' << 16) | ('m' << 24)))
648 throw std::string("Invalid bytecode signature!");
Chris Lattner00950542001-06-06 20:29:01 +0000649
Chris Lattner75f20532003-04-22 18:02:52 +0000650 TheModule = new Module(ModuleID);
Misha Brukman12c29d12003-09-22 23:38:23 +0000651 try {
652 ParseModule(Buf, EndBuf);
653 } catch (std::string &Error) {
Chris Lattnera2602f32003-05-22 18:26:48 +0000654 freeState(); // Must destroy handles before deleting module!
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000655 delete TheModule;
656 TheModule = 0;
Misha Brukman12c29d12003-09-22 23:38:23 +0000657 throw Error;
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000658 }
Chris Lattner00950542001-06-06 20:29:01 +0000659}