blob: fb561aa16ce73e84b887b3744f45f22b69ca0c87 [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"
16#include "llvm/Module.h"
Chris Lattner31bcdb82002-04-28 19:55:58 +000017#include "llvm/Constants.h"
Chris Lattner7061dc52001-12-03 18:02:31 +000018#include "llvm/iPHINode.h"
Chris Lattner00950542001-06-06 20:29:01 +000019#include "llvm/iOther.h"
Chris Lattner00950542001-06-06 20:29:01 +000020#include <sys/types.h>
Chris Lattner00950542001-06-06 20:29:01 +000021#include <sys/stat.h>
Chris Lattner697954c2002-01-20 22:54:45 +000022#include <sys/mman.h>
Chris Lattner00950542001-06-06 20:29:01 +000023#include <fcntl.h>
24#include <unistd.h>
25#include <algorithm>
26
27bool BytecodeParser::getTypeSlot(const Type *Ty, unsigned &Slot) {
28 if (Ty->isPrimitiveType()) {
29 Slot = Ty->getPrimitiveID();
30 } else {
Chris Lattner2a7b6ba2003-03-06 17:15:19 +000031 // Check the function level types first...
Chris Lattner6e5a0e42003-03-06 17:18:14 +000032 TypeValuesListTy::iterator I = find(FunctionTypeValues.begin(),
33 FunctionTypeValues.end(), Ty);
34 if (I != FunctionTypeValues.end()) {
Chris Lattner1d670cc2001-09-07 16:37:43 +000035 Slot = FirstDerivedTyID+ModuleTypeValues.size()+
Chris Lattner6e5a0e42003-03-06 17:18:14 +000036 (&*I - &FunctionTypeValues[0]);
Chris Lattner1d670cc2001-09-07 16:37:43 +000037 } else {
38 I = find(ModuleTypeValues.begin(), ModuleTypeValues.end(), Ty);
39 if (I == ModuleTypeValues.end()) return true; // Didn't find type!
40 Slot = FirstDerivedTyID + (&*I - &ModuleTypeValues[0]);
41 }
Chris Lattner00950542001-06-06 20:29:01 +000042 }
Chris Lattner697954c2002-01-20 22:54:45 +000043 //cerr << "getTypeSlot '" << Ty->getName() << "' = " << Slot << "\n";
Chris Lattner00950542001-06-06 20:29:01 +000044 return false;
45}
46
47const Type *BytecodeParser::getType(unsigned ID) {
Chris Lattner8cdc6b72002-10-23 00:51:54 +000048 if (ID < Type::NumPrimitiveIDs) {
49 const Type *T = Type::getPrimitiveType((Type::PrimitiveID)ID);
50 if (T) return T;
51 }
Chris Lattner00950542001-06-06 20:29:01 +000052
Chris Lattner697954c2002-01-20 22:54:45 +000053 //cerr << "Looking up Type ID: " << ID << "\n";
Chris Lattner8cdc6b72002-10-23 00:51:54 +000054 const Value *V = getValue(Type::TypeTy, ID, false);
55 return cast_or_null<Type>(V);
Chris Lattner00950542001-06-06 20:29:01 +000056}
57
Chris Lattner52e20b02003-03-19 20:54:26 +000058int BytecodeParser::insertValue(Value *Val, ValueTable &ValueTab) {
59 assert((!HasImplicitZeroInitializer || !isa<Constant>(Val) ||
60 Val->getType()->isPrimitiveType() ||
61 !cast<Constant>(Val)->isNullValue()) &&
62 "Cannot read null values from bytecode!");
Chris Lattner00950542001-06-06 20:29:01 +000063 unsigned type;
Chris Lattner74734132002-08-17 22:01:27 +000064 if (getTypeSlot(Val->getType(), type)) return -1;
Chris Lattner1d670cc2001-09-07 16:37:43 +000065 assert(type != Type::TypeTyID && "Types should never be insertValue'd!");
Chris Lattner00950542001-06-06 20:29:01 +000066
Chris Lattner52e20b02003-03-19 20:54:26 +000067 if (ValueTab.size() <= type) {
68 unsigned OldSize = ValueTab.size();
69 ValueTab.resize(type+1);
70 while (OldSize != type+1)
71 ValueTab[OldSize++] = new ValueList();
Chris Lattner036b8aa2003-03-06 17:55:45 +000072 }
Chris Lattner00950542001-06-06 20:29:01 +000073
74 //cerr << "insertValue Values[" << type << "][" << ValueTab[type].size()
Chris Lattner697954c2002-01-20 22:54:45 +000075 // << "] = " << Val << "\n";
Chris Lattner52e20b02003-03-19 20:54:26 +000076 ValueTab[type]->push_back(Val);
Chris Lattner00950542001-06-06 20:29:01 +000077
Chris Lattner52e20b02003-03-19 20:54:26 +000078 bool HasOffset = HasImplicitZeroInitializer &&
79 !Val->getType()->isPrimitiveType();
80
81 return ValueTab[type]->size()-1 + HasOffset;
82}
83
84
85void BytecodeParser::setValueTo(ValueTable &ValueTab, unsigned Slot,
86 Value *Val) {
87 assert(&ValueTab == &ModuleValues && "Can only setValueTo on Module values!");
88 unsigned type;
89 if (getTypeSlot(Val->getType(), type))
90 assert(0 && "getTypeSlot failed!");
91
92 assert((!HasImplicitZeroInitializer || Slot != 0) &&
93 "Cannot change zero init");
94 assert(type < ValueTab.size() && Slot <= ValueTab[type]->size());
95 ValueTab[type]->setOperand(Slot-HasImplicitZeroInitializer, Val);
Chris Lattner00950542001-06-06 20:29:01 +000096}
97
98Value *BytecodeParser::getValue(const Type *Ty, unsigned oNum, bool Create) {
99 unsigned Num = oNum;
100 unsigned type; // The type plane it lives in...
101
Chris Lattner74734132002-08-17 22:01:27 +0000102 if (getTypeSlot(Ty, type)) return 0;
Chris Lattner00950542001-06-06 20:29:01 +0000103
104 if (type == Type::TypeTyID) { // The 'type' plane has implicit values
Chris Lattner1d670cc2001-09-07 16:37:43 +0000105 assert(Create == false);
Chris Lattner8cdc6b72002-10-23 00:51:54 +0000106 if (Num < Type::NumPrimitiveIDs) {
107 const Type *T = Type::getPrimitiveType((Type::PrimitiveID)Num);
108 if (T) return (Value*)T; // Asked for a primitive type...
109 }
Chris Lattner00950542001-06-06 20:29:01 +0000110
111 // Otherwise, derived types need offset...
112 Num -= FirstDerivedTyID;
Chris Lattner1d670cc2001-09-07 16:37:43 +0000113
114 // Is it a module level type?
115 if (Num < ModuleTypeValues.size())
Chris Lattner05950c32001-10-13 06:47:01 +0000116 return (Value*)ModuleTypeValues[Num].get();
Chris Lattner1d670cc2001-09-07 16:37:43 +0000117
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000118 // Nope, is it a function level type?
Chris Lattner1d670cc2001-09-07 16:37:43 +0000119 Num -= ModuleTypeValues.size();
Chris Lattner6e5a0e42003-03-06 17:18:14 +0000120 if (Num < FunctionTypeValues.size())
121 return (Value*)FunctionTypeValues[Num].get();
Chris Lattner1d670cc2001-09-07 16:37:43 +0000122
123 return 0;
Chris Lattner00950542001-06-06 20:29:01 +0000124 }
125
Chris Lattner52e20b02003-03-19 20:54:26 +0000126 if (HasImplicitZeroInitializer && type >= FirstDerivedTyID) {
127 if (Num == 0)
128 return Constant::getNullValue(Ty);
129 --Num;
Chris Lattner00950542001-06-06 20:29:01 +0000130 }
131
Chris Lattner52e20b02003-03-19 20:54:26 +0000132 if (type < ModuleValues.size()) {
133 if (Num < ModuleValues[type]->size())
134 return ModuleValues[type]->getOperand(Num);
135 Num -= ModuleValues[type]->size();
136 }
137
138 if (Values.size() > type && Values[type]->size() > Num)
139 return Values[type]->getOperand(Num);
Chris Lattner00950542001-06-06 20:29:01 +0000140
Chris Lattner74734132002-08-17 22:01:27 +0000141 if (!Create) return 0; // Do not create a placeholder?
Chris Lattner00950542001-06-06 20:29:01 +0000142
143 Value *d = 0;
144 switch (Ty->getPrimitiveID()) {
Chris Lattner74734132002-08-17 22:01:27 +0000145 case Type::LabelTyID:
146 d = new BBPHolder(Ty, oNum);
147 break;
148 default:
149 d = new ValPHolder(Ty, oNum);
150 break;
Chris Lattner00950542001-06-06 20:29:01 +0000151 }
152
153 assert(d != 0 && "How did we not make something?");
Chris Lattner74734132002-08-17 22:01:27 +0000154 if (insertValue(d, LateResolveValues) == -1) return 0;
Chris Lattner00950542001-06-06 20:29:01 +0000155 return d;
156}
157
Chris Lattnerbbd4b302002-10-14 03:33:02 +0000158/// getConstantValue - Just like getValue, except that it returns a null pointer
159/// only on error. It always returns a constant (meaning that if the value is
160/// defined, but is not a constant, that is an error). If the specified
161/// constant hasn't been parsed yet, a placeholder is defined and used. Later,
162/// after the real value is parsed, the placeholder is eliminated.
163///
164Constant *BytecodeParser::getConstantValue(const Type *Ty, unsigned Slot) {
165 if (Value *V = getValue(Ty, Slot, false))
166 return dyn_cast<Constant>(V); // If we already have the value parsed...
167
Chris Lattner52e20b02003-03-19 20:54:26 +0000168 std::pair<const Type*, unsigned> Key(Ty, Slot);
169 GlobalRefsType::iterator I = GlobalRefs.lower_bound(Key);
170
171 if (I != GlobalRefs.end() && I->first == Key) {
Chris Lattnerbbd4b302002-10-14 03:33:02 +0000172 BCR_TRACE(5, "Previous forward ref found!\n");
173 return cast<Constant>(I->second);
174 } else {
175 // Create a placeholder for the constant reference and
176 // keep track of the fact that we have a forward ref to recycle it
177 BCR_TRACE(5, "Creating new forward ref to a constant!\n");
178 Constant *C = new ConstPHolder(Ty, Slot);
179
180 // Keep track of the fact that we have a forward ref to recycle it
Chris Lattner52e20b02003-03-19 20:54:26 +0000181 GlobalRefs.insert(I, std::make_pair(Key, C));
Chris Lattnerbbd4b302002-10-14 03:33:02 +0000182 return C;
183 }
184}
185
186
Chris Lattner00950542001-06-06 20:29:01 +0000187bool BytecodeParser::postResolveValues(ValueTable &ValTab) {
188 bool Error = false;
Chris Lattner52e20b02003-03-19 20:54:26 +0000189 while (!ValTab.empty()) {
190 ValueList &DL = *ValTab.back();
191 ValTab.pop_back();
Chris Lattner00950542001-06-06 20:29:01 +0000192
Chris Lattner52e20b02003-03-19 20:54:26 +0000193 while (!DL.empty()) {
194 Value *D = DL.back();
195 unsigned IDNumber = getValueIDNumberFromPlaceHolder(D);
Chris Lattner00950542001-06-06 20:29:01 +0000196 DL.pop_back();
197
198 Value *NewDef = getValue(D->getType(), IDNumber, false);
199 if (NewDef == 0) {
200 Error = true; // Unresolved thinger
Chris Lattner0d75d8d72003-03-06 16:32:25 +0000201 std::cerr << "Unresolvable reference found: <"
Chris Lattner52e20b02003-03-19 20:54:26 +0000202 << *D->getType() << ">:" << IDNumber <<"!\n";
Chris Lattner00950542001-06-06 20:29:01 +0000203 } else {
204 // Fixup all of the uses of this placeholder def...
205 D->replaceAllUsesWith(NewDef);
206
207 // Now that all the uses are gone, delete the placeholder...
208 // If we couldn't find a def (error case), then leak a little
209 delete D; // memory, 'cause otherwise we can't remove all uses!
210 }
211 }
Chris Lattner52e20b02003-03-19 20:54:26 +0000212 delete &DL;
Chris Lattner00950542001-06-06 20:29:01 +0000213 }
214
215 return Error;
216}
217
218bool BytecodeParser::ParseBasicBlock(const uchar *&Buf, const uchar *EndBuf,
219 BasicBlock *&BB) {
220 BB = new BasicBlock();
221
222 while (Buf < EndBuf) {
Chris Lattner1d670cc2001-09-07 16:37:43 +0000223 Instruction *Inst;
Chris Lattner036b8aa2003-03-06 17:55:45 +0000224 if (ParseInstruction(Buf, EndBuf, Inst, /*HACK*/BB)) {
Chris Lattner00950542001-06-06 20:29:01 +0000225 delete BB;
Chris Lattner74734132002-08-17 22:01:27 +0000226 return true;
Chris Lattner00950542001-06-06 20:29:01 +0000227 }
228
Chris Lattner74734132002-08-17 22:01:27 +0000229 if (Inst == 0) { delete BB; return true; }
230 if (insertValue(Inst, Values) == -1) { delete BB; return true; }
Chris Lattner00950542001-06-06 20:29:01 +0000231
Chris Lattner1d670cc2001-09-07 16:37:43 +0000232 BB->getInstList().push_back(Inst);
233
234 BCR_TRACE(4, Inst);
Chris Lattner00950542001-06-06 20:29:01 +0000235 }
236
237 return false;
238}
239
Chris Lattner1d670cc2001-09-07 16:37:43 +0000240bool BytecodeParser::ParseSymbolTable(const uchar *&Buf, const uchar *EndBuf,
241 SymbolTable *ST) {
Chris Lattner00950542001-06-06 20:29:01 +0000242 while (Buf < EndBuf) {
243 // Symtab block header: [num entries][type id number]
244 unsigned NumEntries, Typ;
245 if (read_vbr(Buf, EndBuf, NumEntries) ||
Chris Lattner74734132002-08-17 22:01:27 +0000246 read_vbr(Buf, EndBuf, Typ)) return true;
Chris Lattner00950542001-06-06 20:29:01 +0000247 const Type *Ty = getType(Typ);
Chris Lattner74734132002-08-17 22:01:27 +0000248 if (Ty == 0) return true;
Chris Lattner00950542001-06-06 20:29:01 +0000249
Chris Lattner1d670cc2001-09-07 16:37:43 +0000250 BCR_TRACE(3, "Plane Type: '" << Ty << "' with " << NumEntries <<
251 " entries\n");
252
Chris Lattner7fc9fe32001-06-27 23:41:11 +0000253 for (unsigned i = 0; i < NumEntries; ++i) {
Chris Lattner00950542001-06-06 20:29:01 +0000254 // Symtab entry: [def slot #][name]
255 unsigned slot;
Chris Lattner74734132002-08-17 22:01:27 +0000256 if (read_vbr(Buf, EndBuf, slot)) return true;
Chris Lattner697954c2002-01-20 22:54:45 +0000257 std::string Name;
Chris Lattner00950542001-06-06 20:29:01 +0000258 if (read(Buf, EndBuf, Name, false)) // Not aligned...
Chris Lattner74734132002-08-17 22:01:27 +0000259 return true;
Chris Lattner00950542001-06-06 20:29:01 +0000260
Chris Lattner52e20b02003-03-19 20:54:26 +0000261 Value *V = getValue(Ty, slot, false); // Find mapping...
262 if (V == 0) {
Chris Lattner697954c2002-01-20 22:54:45 +0000263 BCR_TRACE(3, "FAILED LOOKUP: Slot #" << slot << "\n");
Chris Lattner74734132002-08-17 22:01:27 +0000264 return true;
Chris Lattner1d670cc2001-09-07 16:37:43 +0000265 }
Chris Lattner52e20b02003-03-19 20:54:26 +0000266 BCR_TRACE(4, "Map: '" << Name << "' to #" << slot << ":" << *V;
267 if (!isa<Instruction>(V)) std::cerr << "\n");
Chris Lattner1d670cc2001-09-07 16:37:43 +0000268
Chris Lattner52e20b02003-03-19 20:54:26 +0000269 V->setName(Name, ST);
Chris Lattner00950542001-06-06 20:29:01 +0000270 }
271 }
272
Chris Lattner74734132002-08-17 22:01:27 +0000273 if (Buf > EndBuf) return true;
Chris Lattner3d3f2892001-07-28 17:50:18 +0000274 return false;
Chris Lattner00950542001-06-06 20:29:01 +0000275}
276
Chris Lattner74734132002-08-17 22:01:27 +0000277void BytecodeParser::ResolveReferencesToValue(Value *NewV, unsigned Slot) {
Chris Lattner0d75d8d72003-03-06 16:32:25 +0000278 GlobalRefsType::iterator I = GlobalRefs.find(std::make_pair(NewV->getType(),
279 Slot));
Chris Lattner74734132002-08-17 22:01:27 +0000280 if (I == GlobalRefs.end()) return; // Never forward referenced?
Chris Lattner00950542001-06-06 20:29:01 +0000281
Chris Lattner74734132002-08-17 22:01:27 +0000282 BCR_TRACE(3, "Mutating forward refs!\n");
283 Value *VPH = I->second; // Get the placeholder...
Vikram S. Advec1e4a812002-07-14 23:04:18 +0000284
Chris Lattner52e20b02003-03-19 20:54:26 +0000285 VPH->replaceAllUsesWith(NewV);
286
287 // If this is a global variable being resolved, remove the placeholder from
288 // the module...
289 if (GlobalValue* GVal = dyn_cast<GlobalValue>(NewV))
290 GVal->getParent()->getGlobalList().remove(cast<GlobalVariable>(VPH));
Vikram S. Advec1e4a812002-07-14 23:04:18 +0000291
Chris Lattner74734132002-08-17 22:01:27 +0000292 delete VPH; // Delete the old placeholder
293 GlobalRefs.erase(I); // Remove the map entry for it
Vikram S. Advec1e4a812002-07-14 23:04:18 +0000294}
295
Chris Lattner52e20b02003-03-19 20:54:26 +0000296
Chris Lattner6e5a0e42003-03-06 17:18:14 +0000297bool BytecodeParser::ParseFunction(const uchar *&Buf, const uchar *EndBuf) {
Chris Lattner00950542001-06-06 20:29:01 +0000298 // Clear out the local values table...
Chris Lattner74734132002-08-17 22:01:27 +0000299 if (FunctionSignatureList.empty()) {
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000300 Error = "Function found, but FunctionSignatureList empty!";
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000301 return true; // Unexpected function!
Chris Lattnerd6b65252001-10-24 01:15:12 +0000302 }
Chris Lattner00950542001-06-06 20:29:01 +0000303
Chris Lattnerd23b1d32001-11-26 18:56:10 +0000304 unsigned isInternal;
Chris Lattner74734132002-08-17 22:01:27 +0000305 if (read_vbr(Buf, EndBuf, isInternal)) return true;
Chris Lattnerd23b1d32001-11-26 18:56:10 +0000306
Chris Lattner52e20b02003-03-19 20:54:26 +0000307 Function *F = FunctionSignatureList.back().first;
308 unsigned FunctionSlot = FunctionSignatureList.back().second;
Chris Lattner74734132002-08-17 22:01:27 +0000309 FunctionSignatureList.pop_back();
Chris Lattner4ad02e72003-04-16 20:28:45 +0000310 F->setLinkage(isInternal ? GlobalValue::InternalLinkage :
311 GlobalValue::ExternalLinkage);
Chris Lattner00950542001-06-06 20:29:01 +0000312
Chris Lattner52e20b02003-03-19 20:54:26 +0000313 const FunctionType::ParamTypes &Params =F->getFunctionType()->getParamTypes();
314 Function::aiterator AI = F->abegin();
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000315 for (FunctionType::ParamTypes::const_iterator It = Params.begin();
Chris Lattner69da5cf2002-10-13 20:57:00 +0000316 It != Params.end(); ++It, ++AI) {
317 if (insertValue(AI, Values) == -1) {
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000318 Error = "Error reading function arguments!\n";
Chris Lattner52e20b02003-03-19 20:54:26 +0000319 return true;
Chris Lattnerd6b65252001-10-24 01:15:12 +0000320 }
Chris Lattner00950542001-06-06 20:29:01 +0000321 }
322
323 while (Buf < EndBuf) {
324 unsigned Type, Size;
Chris Lattnerb6c46952003-03-06 17:03:28 +0000325 const unsigned char *OldBuf = Buf;
Chris Lattnerd6b65252001-10-24 01:15:12 +0000326 if (readBlock(Buf, EndBuf, Type, Size)) {
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000327 Error = "Error reading Function level block!";
Chris Lattner52e20b02003-03-19 20:54:26 +0000328 return true;
Chris Lattnerd6b65252001-10-24 01:15:12 +0000329 }
Chris Lattner00950542001-06-06 20:29:01 +0000330
331 switch (Type) {
332 case BytecodeFormat::ConstantPool:
Chris Lattner1d670cc2001-09-07 16:37:43 +0000333 BCR_TRACE(2, "BLOCK BytecodeFormat::ConstantPool: {\n");
Chris Lattner52e20b02003-03-19 20:54:26 +0000334 if (ParseConstantPool(Buf, Buf+Size, Values, FunctionTypeValues))
335 return true;
Chris Lattner00950542001-06-06 20:29:01 +0000336 break;
337
338 case BytecodeFormat::BasicBlock: {
Chris Lattner1d670cc2001-09-07 16:37:43 +0000339 BCR_TRACE(2, "BLOCK BytecodeFormat::BasicBlock: {\n");
Chris Lattner00950542001-06-06 20:29:01 +0000340 BasicBlock *BB;
341 if (ParseBasicBlock(Buf, Buf+Size, BB) ||
Chris Lattner52e20b02003-03-19 20:54:26 +0000342 insertValue(BB, Values) == -1)
343 return true; // Parse error... :(
Chris Lattner00950542001-06-06 20:29:01 +0000344
Chris Lattner52e20b02003-03-19 20:54:26 +0000345 F->getBasicBlockList().push_back(BB);
Chris Lattner00950542001-06-06 20:29:01 +0000346 break;
347 }
348
349 case BytecodeFormat::SymbolTable:
Chris Lattner1d670cc2001-09-07 16:37:43 +0000350 BCR_TRACE(2, "BLOCK BytecodeFormat::SymbolTable: {\n");
Chris Lattner52e20b02003-03-19 20:54:26 +0000351 if (ParseSymbolTable(Buf, Buf+Size, &F->getSymbolTable()))
352 return true;
Chris Lattner00950542001-06-06 20:29:01 +0000353 break;
354
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;
Chris Lattner74734132002-08-17 22:01:27 +0000358 if (OldBuf > Buf) return true; // Wrap around!
Chris Lattner00950542001-06-06 20:29:01 +0000359 break;
360 }
Chris Lattner1d670cc2001-09-07 16:37:43 +0000361 BCR_TRACE(2, "} end block\n");
362
Chris Lattner00950542001-06-06 20:29:01 +0000363 if (align32(Buf, EndBuf)) {
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000364 Error = "Error aligning Function level block!";
Chris Lattner52e20b02003-03-19 20:54:26 +0000365 return true; // Malformed bc file, read past end of block.
Chris Lattner00950542001-06-06 20:29:01 +0000366 }
367 }
368
Chris Lattner52e20b02003-03-19 20:54:26 +0000369 if (postResolveValues(LateResolveValues)) {
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000370 Error = "Error resolving function values!";
Chris Lattner52e20b02003-03-19 20:54:26 +0000371 return true; // Unresolvable references!
Chris Lattner00950542001-06-06 20:29:01 +0000372 }
373
Chris Lattner52e20b02003-03-19 20:54:26 +0000374 ResolveReferencesToValue(F, FunctionSlot);
Chris Lattner00950542001-06-06 20:29:01 +0000375
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000376 // Clear out function level types...
Chris Lattner6e5a0e42003-03-06 17:18:14 +0000377 FunctionTypeValues.clear();
Chris Lattnere4d71a12001-09-14 22:03:42 +0000378
Chris Lattner52e20b02003-03-19 20:54:26 +0000379 freeTable(Values);
Chris Lattner00950542001-06-06 20:29:01 +0000380 return false;
381}
382
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000383bool BytecodeParser::ParseModuleGlobalInfo(const uchar *&Buf, const uchar *End){
Chris Lattner74734132002-08-17 22:01:27 +0000384 if (!FunctionSignatureList.empty()) {
Chris Lattnerd6b65252001-10-24 01:15:12 +0000385 Error = "Two ModuleGlobalInfo packets found!";
Chris Lattner74734132002-08-17 22:01:27 +0000386 return true; // Two ModuleGlobal blocks?
Chris Lattnerd6b65252001-10-24 01:15:12 +0000387 }
Chris Lattner00950542001-06-06 20:29:01 +0000388
Chris Lattner70cc3392001-09-10 07:58:01 +0000389 // Read global variables...
390 unsigned VarType;
Chris Lattner74734132002-08-17 22:01:27 +0000391 if (read_vbr(Buf, End, VarType)) return true;
Chris Lattner70cc3392001-09-10 07:58:01 +0000392 while (VarType != Type::VoidTyID) { // List is terminated by Void
Chris Lattnerd23b1d32001-11-26 18:56:10 +0000393 // VarType Fields: bit0 = isConstant, bit1 = hasInitializer,
394 // bit2 = isInternal, bit3+ = slot#
395 const Type *Ty = getType(VarType >> 3);
Chris Lattner9b625032002-05-06 16:15:30 +0000396 if (!Ty || !isa<PointerType>(Ty)) {
Chris Lattnerd6b65252001-10-24 01:15:12 +0000397 Error = "Global not pointer type! Ty = " + Ty->getDescription();
Chris Lattner74734132002-08-17 22:01:27 +0000398 return true;
Chris Lattner70cc3392001-09-10 07:58:01 +0000399 }
400
Chris Lattner52e20b02003-03-19 20:54:26 +0000401 const Type *ElTy = cast<PointerType>(Ty)->getElementType();
Chris Lattnerd70684f2001-09-18 04:01:05 +0000402
Chris Lattner4ad02e72003-04-16 20:28:45 +0000403
404 GlobalValue::LinkageTypes Linkage =
405 (VarType & 4) ? GlobalValue::InternalLinkage :
406 GlobalValue::ExternalLinkage;
407
Chris Lattner70cc3392001-09-10 07:58:01 +0000408 // Create the global variable...
Chris Lattner4ad02e72003-04-16 20:28:45 +0000409 GlobalVariable *GV = new GlobalVariable(ElTy, VarType & 1, Linkage,
Chris Lattner52e20b02003-03-19 20:54:26 +0000410 0, "", TheModule);
Chris Lattner05950c32001-10-13 06:47:01 +0000411 int DestSlot = insertValue(GV, ModuleValues);
Chris Lattner74734132002-08-17 22:01:27 +0000412 if (DestSlot == -1) return true;
Chris Lattner52e20b02003-03-19 20:54:26 +0000413 BCR_TRACE(2, "Global Variable of type: " << *Ty << "\n");
Chris Lattner74734132002-08-17 22:01:27 +0000414 ResolveReferencesToValue(GV, (unsigned)DestSlot);
Chris Lattner05950c32001-10-13 06:47:01 +0000415
Chris Lattner52e20b02003-03-19 20:54:26 +0000416 if (VarType & 2) { // Does it have an initalizer?
417 unsigned InitSlot;
418 if (read_vbr(Buf, End, InitSlot)) return true;
419 GlobalInits.push_back(std::make_pair(GV, InitSlot));
420 }
Chris Lattner74734132002-08-17 22:01:27 +0000421 if (read_vbr(Buf, End, VarType)) return true;
Chris Lattner70cc3392001-09-10 07:58:01 +0000422 }
423
Chris Lattner52e20b02003-03-19 20:54:26 +0000424 // Read the function objects for all of the functions that are coming
Chris Lattner74734132002-08-17 22:01:27 +0000425 unsigned FnSignature;
426 if (read_vbr(Buf, End, FnSignature)) return true;
427 while (FnSignature != Type::VoidTyID) { // List is terminated by Void
428 const Type *Ty = getType(FnSignature);
Chris Lattneref9c23f2001-10-03 14:53:21 +0000429 if (!Ty || !isa<PointerType>(Ty) ||
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000430 !isa<FunctionType>(cast<PointerType>(Ty)->getElementType())) {
431 Error = "Function not ptr to func type! Ty = " + Ty->getDescription();
Chris Lattner74734132002-08-17 22:01:27 +0000432 return true;
Chris Lattner00950542001-06-06 20:29:01 +0000433 }
Chris Lattner8cdc6b72002-10-23 00:51:54 +0000434
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000435 // We create functions by passing the underlying FunctionType to create...
Chris Lattner7a176752001-12-04 00:03:30 +0000436 Ty = cast<PointerType>(Ty)->getElementType();
Chris Lattner00950542001-06-06 20:29:01 +0000437
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000438 // When the ModuleGlobalInfo section is read, we load the type of each
439 // function and the 'ModuleValues' slot that it lands in. We then load a
440 // placeholder into its slot to reserve it. When the function is loaded,
441 // this placeholder is replaced.
Chris Lattner00950542001-06-06 20:29:01 +0000442
443 // Insert the placeholder...
Chris Lattner4ad02e72003-04-16 20:28:45 +0000444 Function *Func = new Function(cast<FunctionType>(Ty),
445 GlobalValue::InternalLinkage, "", TheModule);
Chris Lattner52e20b02003-03-19 20:54:26 +0000446 int DestSlot = insertValue(Func, ModuleValues);
447 if (DestSlot == -1) return true;
448 ResolveReferencesToValue(Func, (unsigned)DestSlot);
Chris Lattner00950542001-06-06 20:29:01 +0000449
Chris Lattner52e20b02003-03-19 20:54:26 +0000450 // Keep track of this information in a list that is emptied as functions are
451 // loaded...
Chris Lattner00950542001-06-06 20:29:01 +0000452 //
Chris Lattner52e20b02003-03-19 20:54:26 +0000453 FunctionSignatureList.push_back(std::make_pair(Func, DestSlot));
454
Chris Lattner74734132002-08-17 22:01:27 +0000455 if (read_vbr(Buf, End, FnSignature)) return true;
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000456 BCR_TRACE(2, "Function of type: " << Ty << "\n");
Chris Lattner00950542001-06-06 20:29:01 +0000457 }
458
Chris Lattner74734132002-08-17 22:01:27 +0000459 if (align32(Buf, End)) return true;
460
461 // Now that the function signature list is set up, reverse it so that we can
462 // remove elements efficiently from the back of the vector.
463 std::reverse(FunctionSignatureList.begin(), FunctionSignatureList.end());
Chris Lattner00950542001-06-06 20:29:01 +0000464
465 // This is for future proofing... in the future extra fields may be added that
466 // we don't understand, so we transparently ignore them.
467 //
468 Buf = End;
469 return false;
470}
471
Chris Lattner036b8aa2003-03-06 17:55:45 +0000472bool BytecodeParser::ParseVersionInfo(const uchar *&Buf, const uchar *EndBuf) {
473 unsigned Version;
474 if (read_vbr(Buf, EndBuf, Version)) return true;
475
476 // Unpack version number: low four bits are for flags, top bits = version
477 isBigEndian = Version & 1;
478 hasLongPointers = Version & 2;
479 RevisionNum = Version >> 4;
480 HasImplicitZeroInitializer = true;
481
482 switch (RevisionNum) {
483 case 0: // Initial revision
Chris Lattner52e20b02003-03-19 20:54:26 +0000484 // Version #0 didn't have any of the flags stored correctly, and in fact as
485 // only valid with a 14 in the flags values. Also, it does not support
486 // encoding zero initializers for arrays compactly.
487 //
Chris Lattner036b8aa2003-03-06 17:55:45 +0000488 if (Version != 14) return true; // Unknown revision 0 flags?
489 FirstDerivedTyID = 14;
490 HasImplicitZeroInitializer = false;
491 isBigEndian = hasLongPointers = true;
492 break;
493 case 1:
Chris Lattner52e20b02003-03-19 20:54:26 +0000494 // Version #1 has two bit fields: isBigEndian and hasLongPointers
Chris Lattner036b8aa2003-03-06 17:55:45 +0000495 FirstDerivedTyID = 14;
496 break;
497 default:
498 Error = "Unknown bytecode version number!";
499 return true;
500 }
501
502 BCR_TRACE(1, "Bytecode Rev = " << (unsigned)RevisionNum << "\n");
503 BCR_TRACE(1, "BigEndian/LongPointers = " << isBigEndian << ","
504 << hasLongPointers << "\n");
505 BCR_TRACE(1, "HasImplicitZeroInit = " << HasImplicitZeroInitializer << "\n");
506 return false;
507}
508
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000509bool BytecodeParser::ParseModule(const uchar *Buf, const uchar *EndBuf) {
Chris Lattner00950542001-06-06 20:29:01 +0000510 unsigned Type, Size;
Chris Lattner74734132002-08-17 22:01:27 +0000511 if (readBlock(Buf, EndBuf, Type, Size)) return true;
Chris Lattnerd6b65252001-10-24 01:15:12 +0000512 if (Type != BytecodeFormat::Module || Buf+Size != EndBuf) {
513 Error = "Expected Module packet!";
Chris Lattner74734132002-08-17 22:01:27 +0000514 return true; // Hrm, not a class?
Chris Lattnerd6b65252001-10-24 01:15:12 +0000515 }
Chris Lattner00950542001-06-06 20:29:01 +0000516
Chris Lattner1d670cc2001-09-07 16:37:43 +0000517 BCR_TRACE(0, "BLOCK BytecodeFormat::Module: {\n");
Chris Lattner74734132002-08-17 22:01:27 +0000518 FunctionSignatureList.clear(); // Just in case...
Chris Lattner00950542001-06-06 20:29:01 +0000519
520 // Read into instance variables...
Chris Lattner036b8aa2003-03-06 17:55:45 +0000521 if (ParseVersionInfo(Buf, EndBuf)) return true;
Chris Lattner74734132002-08-17 22:01:27 +0000522 if (align32(Buf, EndBuf)) return true;
Chris Lattner00950542001-06-06 20:29:01 +0000523
Chris Lattner00950542001-06-06 20:29:01 +0000524 while (Buf < EndBuf) {
Chris Lattnerb6c46952003-03-06 17:03:28 +0000525 const unsigned char *OldBuf = Buf;
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000526 if (readBlock(Buf, EndBuf, Type, Size)) return true;
Chris Lattner00950542001-06-06 20:29:01 +0000527 switch (Type) {
Chris Lattner52e20b02003-03-19 20:54:26 +0000528 case BytecodeFormat::GlobalTypePlane:
529 BCR_TRACE(1, "BLOCK BytecodeFormat::GlobalTypePlane: {\n");
530 if (ParseGlobalTypes(Buf, Buf+Size)) return true;
531 break;
532
533 case BytecodeFormat::ModuleGlobalInfo:
534 BCR_TRACE(1, "BLOCK BytecodeFormat::ModuleGlobalInfo: {\n");
535 if (ParseModuleGlobalInfo(Buf, Buf+Size)) return true;
536 break;
537
Chris Lattner1d670cc2001-09-07 16:37:43 +0000538 case BytecodeFormat::ConstantPool:
539 BCR_TRACE(1, "BLOCK BytecodeFormat::ConstantPool: {\n");
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000540 if (ParseConstantPool(Buf, Buf+Size, ModuleValues, ModuleTypeValues))
541 return true;
Chris Lattner00950542001-06-06 20:29:01 +0000542 break;
543
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000544 case BytecodeFormat::Function: {
545 BCR_TRACE(1, "BLOCK BytecodeFormat::Function: {\n");
Chris Lattner52e20b02003-03-19 20:54:26 +0000546 if (ParseFunction(Buf, Buf+Size))
547 return true; // Error parsing function
Chris Lattner00950542001-06-06 20:29:01 +0000548 break;
549 }
550
551 case BytecodeFormat::SymbolTable:
Chris Lattner1d670cc2001-09-07 16:37:43 +0000552 BCR_TRACE(1, "BLOCK BytecodeFormat::SymbolTable: {\n");
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000553 if (ParseSymbolTable(Buf, Buf+Size, &TheModule->getSymbolTable()))
554 return true;
Chris Lattner00950542001-06-06 20:29:01 +0000555 break;
556
557 default:
Chris Lattnerd6b65252001-10-24 01:15:12 +0000558 Error = "Expected Module Block!";
Chris Lattner00950542001-06-06 20:29:01 +0000559 Buf += Size;
Chris Lattner74734132002-08-17 22:01:27 +0000560 if (OldBuf > Buf) return true; // Wrap around!
Chris Lattner00950542001-06-06 20:29:01 +0000561 break;
562 }
Chris Lattner1d670cc2001-09-07 16:37:43 +0000563 BCR_TRACE(1, "} end block\n");
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000564 if (align32(Buf, EndBuf)) return true;
Chris Lattner00950542001-06-06 20:29:01 +0000565 }
566
Chris Lattner52e20b02003-03-19 20:54:26 +0000567 // After the module constant pool has been read, we can safely initialize
568 // global variables...
569 while (!GlobalInits.empty()) {
570 GlobalVariable *GV = GlobalInits.back().first;
571 unsigned Slot = GlobalInits.back().second;
572 GlobalInits.pop_back();
573
574 // Look up the initializer value...
575 if (Value *V = getValue(GV->getType()->getElementType(), Slot, false)) {
576 if (GV->hasInitializer()) return true;
577 GV->setInitializer(cast<Constant>(V));
578 } else
579 return true;
580 }
581
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000582 if (!FunctionSignatureList.empty()) { // Expected more functions!
Chris Lattnerc9aa7df2002-03-29 03:51:11 +0000583 Error = "Function expected, but bytecode stream at end!";
Chris Lattner74734132002-08-17 22:01:27 +0000584 return true;
Chris Lattnerd6b65252001-10-24 01:15:12 +0000585 }
Chris Lattner1d670cc2001-09-07 16:37:43 +0000586
587 BCR_TRACE(0, "} end block\n\n");
Chris Lattner00950542001-06-06 20:29:01 +0000588 return false;
589}
590
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000591static inline Module *Error(std::string *ErrorStr, const char *Message) {
592 if (ErrorStr) *ErrorStr = Message;
593 return 0;
594}
595
Chris Lattner00950542001-06-06 20:29:01 +0000596Module *BytecodeParser::ParseBytecode(const uchar *Buf, const uchar *EndBuf) {
Chris Lattner00950542001-06-06 20:29:01 +0000597 unsigned Sig;
598 // Read and check signature...
599 if (read(Buf, EndBuf, Sig) ||
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000600 Sig != ('l' | ('l' << 8) | ('v' << 16) | 'm' << 24))
601 return ::Error(&Error, "Invalid bytecode signature!");
Chris Lattner00950542001-06-06 20:29:01 +0000602
Chris Lattner2a7b6ba2003-03-06 17:15:19 +0000603 TheModule = new Module();
604 if (ParseModule(Buf, EndBuf)) {
605 delete TheModule;
606 TheModule = 0;
607 }
608 return TheModule;
Chris Lattner00950542001-06-06 20:29:01 +0000609}
610
611
Chris Lattner09abe6a2003-03-06 16:50:32 +0000612Module *ParseBytecodeBuffer(const unsigned char *Buffer, unsigned Length,
613 std::string *ErrorStr) {
Chris Lattner00950542001-06-06 20:29:01 +0000614 BytecodeParser Parser;
Chris Lattner09abe6a2003-03-06 16:50:32 +0000615 Module *R = Parser.ParseBytecode(Buffer, Buffer+Length);
616 if (ErrorStr) *ErrorStr = Parser.getError();
617 return R;
Chris Lattner00950542001-06-06 20:29:01 +0000618}
619
Chris Lattnerb6c46952003-03-06 17:03:28 +0000620
621/// FDHandle - Simple handle class to make sure a file descriptor gets closed
622/// when the object is destroyed.
623class FDHandle {
624 int FD;
625public:
626 FDHandle(int fd) : FD(fd) {}
627 operator int() const { return FD; }
628 ~FDHandle() {
629 if (FD != -1) close(FD);
630 }
631};
632
Chris Lattner00950542001-06-06 20:29:01 +0000633// Parse and return a class file...
634//
Chris Lattner697954c2002-01-20 22:54:45 +0000635Module *ParseBytecodeFile(const std::string &Filename, std::string *ErrorStr) {
Chris Lattner00950542001-06-06 20:29:01 +0000636 Module *Result = 0;
637
Chris Lattner697954c2002-01-20 22:54:45 +0000638 if (Filename != std::string("-")) { // Read from a file...
Chris Lattnerb6c46952003-03-06 17:03:28 +0000639 FDHandle FD = open(Filename.c_str(), O_RDONLY);
640 if (FD == -1)
641 return Error(ErrorStr, "Error opening file!");
Chris Lattner00950542001-06-06 20:29:01 +0000642
Chris Lattnerb6c46952003-03-06 17:03:28 +0000643 // Stat the file to get its length...
Chris Lattner09abe6a2003-03-06 16:50:32 +0000644 struct stat StatBuf;
Chris Lattnerb6c46952003-03-06 17:03:28 +0000645 if (fstat(FD, &StatBuf) == -1 || StatBuf.st_size == 0)
646 return Error(ErrorStr, "Error stat'ing file!");
Chris Lattner09abe6a2003-03-06 16:50:32 +0000647
Chris Lattnerb6c46952003-03-06 17:03:28 +0000648 // mmap in the file all at once...
Chris Lattner09abe6a2003-03-06 16:50:32 +0000649 int Length = StatBuf.st_size;
650 unsigned char *Buffer = (unsigned char*)mmap(0, Length, PROT_READ,
651 MAP_PRIVATE, FD, 0);
Chris Lattnerb6c46952003-03-06 17:03:28 +0000652 if (Buffer == (unsigned char*)MAP_FAILED)
653 return Error(ErrorStr, "Error mmapping file!");
Chris Lattner00950542001-06-06 20:29:01 +0000654
Chris Lattnerb6c46952003-03-06 17:03:28 +0000655 // Parse the bytecode we mmapped in
Chris Lattner09abe6a2003-03-06 16:50:32 +0000656 Result = ParseBytecodeBuffer(Buffer, Length, ErrorStr);
Chris Lattner00950542001-06-06 20:29:01 +0000657
Chris Lattnerb6c46952003-03-06 17:03:28 +0000658 // Unmmap the bytecode...
Chris Lattner00950542001-06-06 20:29:01 +0000659 munmap((char*)Buffer, Length);
Chris Lattner00950542001-06-06 20:29:01 +0000660 } else { // Read from stdin
Chris Lattner00950542001-06-06 20:29:01 +0000661 int BlockSize;
Chris Lattnerb6c46952003-03-06 17:03:28 +0000662 uchar Buffer[4096*4];
663 std::vector<unsigned char> FileData;
Chris Lattner00950542001-06-06 20:29:01 +0000664
Chris Lattnerb6c46952003-03-06 17:03:28 +0000665 // Read in all of the data from stdin, we cannot mmap stdin...
666 while ((BlockSize = read(0 /*stdin*/, Buffer, 4096*4))) {
667 if (BlockSize == -1)
668 return Error(ErrorStr, "Error reading from stdin!");
669
670 FileData.insert(FileData.end(), Buffer, Buffer+BlockSize);
Chris Lattner00950542001-06-06 20:29:01 +0000671 }
672
Chris Lattnerb6c46952003-03-06 17:03:28 +0000673 if (FileData.empty())
674 return Error(ErrorStr, "Standard Input empty!");
Chris Lattner00950542001-06-06 20:29:01 +0000675
Chris Lattner915ce8a2002-08-18 00:38:32 +0000676#define ALIGN_PTRS 0
Chris Lattner00950542001-06-06 20:29:01 +0000677#if ALIGN_PTRS
Chris Lattnerb6c46952003-03-06 17:03:28 +0000678 uchar *Buf = (uchar*)mmap(0, FileData.size(), PROT_READ|PROT_WRITE,
Chris Lattner00950542001-06-06 20:29:01 +0000679 MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
680 assert((Buf != (uchar*)-1) && "mmap returned error!");
Chris Lattnerb6c46952003-03-06 17:03:28 +0000681 memcpy(Buf, &FileData[0], FileData.size());
Chris Lattner00950542001-06-06 20:29:01 +0000682#else
Chris Lattnerb6c46952003-03-06 17:03:28 +0000683 unsigned char *Buf = &FileData[0];
Chris Lattner00950542001-06-06 20:29:01 +0000684#endif
685
Chris Lattnerb6c46952003-03-06 17:03:28 +0000686 Result = ParseBytecodeBuffer(Buf, FileData.size(), ErrorStr);
Chris Lattner00950542001-06-06 20:29:01 +0000687
688#if ALIGN_PTRS
Chris Lattnerb6c46952003-03-06 17:03:28 +0000689 munmap((char*)Buf, FileData.size()); // Free mmap'd data area
Chris Lattner00950542001-06-06 20:29:01 +0000690#endif
Chris Lattner00950542001-06-06 20:29:01 +0000691 }
692
693 return Result;
694}