lib/Bitcode/Writer/BitcodeWriter.cpp

   1 //===--- Bitcode/Writer/BitcodeWriter.cpp - Bitcode Writer ----------------===//
   2 //
   3 //                     The LLVM Compiler Infrastructure
   4 //
   5 // This file is distributed under the University of Illinois Open Source
   6 // License. See LICENSE.TXT for details.
   7 //
   8 //===----------------------------------------------------------------------===//
   9 //
  10 // Bitcode writer implementation.
  11 //
  12 //===----------------------------------------------------------------------===//
  13
  14 #include "llvm/Bitcode/ReaderWriter.h"
  15 #include "llvm/Bitcode/BitstreamWriter.h"
  16 #include "llvm/Bitcode/LLVMBitCodes.h"
  17 #include "ValueEnumerator.h"
  18 #include "llvm/Constants.h"
  19 #include "llvm/DerivedTypes.h"
  20 #include "llvm/InlineAsm.h"
  21 #include "llvm/Instructions.h"
  22 #include "llvm/Module.h"
  23 #include "llvm/TypeSymbolTable.h"
  24 #include "llvm/ValueSymbolTable.h"
  25 #include "llvm/Support/MathExtras.h"
  26 #include "llvm/Support/Streams.h"
  27 #include "llvm/System/Program.h"
  28 using namespace llvm;
  29
  30 /// These are manifest constants used by the bitcode writer. They do not need to
  31 /// be kept in sync with the reader, but need to be consistent within this file.
  32 enum {
  33   CurVersion = 0,
  34
  35   // VALUE_SYMTAB_BLOCK abbrev id's.
  36   VST_ENTRY_8_ABBREV = bitc::FIRST_APPLICATION_ABBREV,
  37   VST_ENTRY_7_ABBREV,
  38   VST_ENTRY_6_ABBREV,
  39   VST_BBENTRY_6_ABBREV,
  40
  41   // CONSTANTS_BLOCK abbrev id's.
  42   CONSTANTS_SETTYPE_ABBREV = bitc::FIRST_APPLICATION_ABBREV,
  43   CONSTANTS_INTEGER_ABBREV,
  44   CONSTANTS_CE_CAST_Abbrev,
  45   CONSTANTS_NULL_Abbrev,
  46
  47   // FUNCTION_BLOCK abbrev id's.
  48   FUNCTION_INST_LOAD_ABBREV = bitc::FIRST_APPLICATION_ABBREV,
  49   FUNCTION_INST_BINOP_ABBREV,
  50   FUNCTION_INST_CAST_ABBREV,
  51   FUNCTION_INST_RET_VOID_ABBREV,
  52   FUNCTION_INST_RET_VAL_ABBREV,
  53   FUNCTION_INST_UNREACHABLE_ABBREV
  54 };
  55
  56
  57 static unsigned GetEncodedCastOpcode(unsigned Opcode) {
  58   switch (Opcode) {
  59   default: assert(0 && "Unknown cast instruction!");
  60   case Instruction::Trunc   : return bitc::CAST_TRUNC;
  61   case Instruction::ZExt    : return bitc::CAST_ZEXT;
  62   case Instruction::SExt    : return bitc::CAST_SEXT;
  63   case Instruction::FPToUI  : return bitc::CAST_FPTOUI;
  64   case Instruction::FPToSI  : return bitc::CAST_FPTOSI;
  65   case Instruction::UIToFP  : return bitc::CAST_UITOFP;
  66   case Instruction::SIToFP  : return bitc::CAST_SITOFP;
  67   case Instruction::FPTrunc : return bitc::CAST_FPTRUNC;
  68   case Instruction::FPExt   : return bitc::CAST_FPEXT;
  69   case Instruction::PtrToInt: return bitc::CAST_PTRTOINT;
  70   case Instruction::IntToPtr: return bitc::CAST_INTTOPTR;
  71   case Instruction::BitCast : return bitc::CAST_BITCAST;
  72   }
  73 }
  74
  75 static unsigned GetEncodedBinaryOpcode(unsigned Opcode) {
  76   switch (Opcode) {
  77   default: assert(0 && "Unknown binary instruction!");
  78   case Instruction::Add:  return bitc::BINOP_ADD;
  79   case Instruction::Sub:  return bitc::BINOP_SUB;
  80   case Instruction::Mul:  return bitc::BINOP_MUL;
  81   case Instruction::UDiv: return bitc::BINOP_UDIV;
  82   case Instruction::FDiv:
  83   case Instruction::SDiv: return bitc::BINOP_SDIV;
  84   case Instruction::URem: return bitc::BINOP_UREM;
  85   case Instruction::FRem:
  86   case Instruction::SRem: return bitc::BINOP_SREM;
  87   case Instruction::Shl:  return bitc::BINOP_SHL;
  88   case Instruction::LShr: return bitc::BINOP_LSHR;
  89   case Instruction::AShr: return bitc::BINOP_ASHR;
  90   case Instruction::And:  return bitc::BINOP_AND;
  91   case Instruction::Or:   return bitc::BINOP_OR;
  92   case Instruction::Xor:  return bitc::BINOP_XOR;
  93   }
  94 }
  95
  96
  97
  98 static void WriteStringRecord(unsigned Code, const std::string &Str,
  99                               unsigned AbbrevToUse, BitstreamWriter &Stream) {
 100   SmallVector<unsigned, 64> Vals;
 101
 102   // Code: [strchar x N]
 103   for (unsigned i = 0, e = Str.size(); i != e; ++i)
 104     Vals.push_back(Str[i]);
 105
 106   // Emit the finished record.
 107   Stream.EmitRecord(Code, Vals, AbbrevToUse);
 108 }
 109
 110 // Emit information about parameter attributes.
 111 static void WriteParamAttrTable(const ValueEnumerator &VE,
 112                                 BitstreamWriter &Stream) {
 113   const std::vector<PAListPtr> &Attrs = VE.getParamAttrs();
 114   if (Attrs.empty()) return;
 115
 116   Stream.EnterSubblock(bitc::PARAMATTR_BLOCK_ID, 3);
 117
 118   SmallVector<uint64_t, 64> Record;
 119   for (unsigned i = 0, e = Attrs.size(); i != e; ++i) {
 120     const PAListPtr &A = Attrs[i];
 121     for (unsigned i = 0, e = A.getNumSlots(); i != e; ++i) {
 122       const ParamAttrsWithIndex &PAWI = A.getSlot(i);
 123       Record.push_back(PAWI.Index);
 124       Record.push_back(PAWI.Attrs);
 125     }
 126
 127     Stream.EmitRecord(bitc::PARAMATTR_CODE_ENTRY, Record);
 128     Record.clear();
 129   }
 130
 131   Stream.ExitBlock();
 132 }
 133
 134 /// WriteTypeTable - Write out the type table for a module.
 135 static void WriteTypeTable(const ValueEnumerator &VE, BitstreamWriter &Stream) {
 136   const ValueEnumerator::TypeList &TypeList = VE.getTypes();
 137
 138   Stream.EnterSubblock(bitc::TYPE_BLOCK_ID, 4 /*count from # abbrevs */);
 139   SmallVector<uint64_t, 64> TypeVals;
 140
 141   // Abbrev for TYPE_CODE_POINTER.
 142   BitCodeAbbrev *Abbv = new BitCodeAbbrev();
 143   Abbv->Add(BitCodeAbbrevOp(bitc::TYPE_CODE_POINTER));
 144   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed,
 145                             Log2_32_Ceil(VE.getTypes().size()+1)));
 146   Abbv->Add(BitCodeAbbrevOp(0));  // Addrspace = 0
 147   unsigned PtrAbbrev = Stream.EmitAbbrev(Abbv);
 148
 149   // Abbrev for TYPE_CODE_FUNCTION.
 150   Abbv = new BitCodeAbbrev();
 151   Abbv->Add(BitCodeAbbrevOp(bitc::TYPE_CODE_FUNCTION));
 152   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 1));  // isvararg
 153   Abbv->Add(BitCodeAbbrevOp(0));  // FIXME: DEAD value, remove in LLVM 3.0
 154   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Array));
 155   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed,
 156                             Log2_32_Ceil(VE.getTypes().size()+1)));
 157   unsigned FunctionAbbrev = Stream.EmitAbbrev(Abbv);
 158
 159   // Abbrev for TYPE_CODE_STRUCT.
 160   Abbv = new BitCodeAbbrev();
 161   Abbv->Add(BitCodeAbbrevOp(bitc::TYPE_CODE_STRUCT));
 162   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 1));  // ispacked
 163   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Array));
 164   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed,
 165                             Log2_32_Ceil(VE.getTypes().size()+1)));
 166   unsigned StructAbbrev = Stream.EmitAbbrev(Abbv);
 167
 168   // Abbrev for TYPE_CODE_ARRAY.
 169   Abbv = new BitCodeAbbrev();
 170   Abbv->Add(BitCodeAbbrevOp(bitc::TYPE_CODE_ARRAY));
 171   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 8));   // size
 172   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed,
 173                             Log2_32_Ceil(VE.getTypes().size()+1)));
 174   unsigned ArrayAbbrev = Stream.EmitAbbrev(Abbv);
 175
 176   // Emit an entry count so the reader can reserve space.
 177   TypeVals.push_back(TypeList.size());
 178   Stream.EmitRecord(bitc::TYPE_CODE_NUMENTRY, TypeVals);
 179   TypeVals.clear();
 180
 181   // Loop over all of the types, emitting each in turn.
 182   for (unsigned i = 0, e = TypeList.size(); i != e; ++i) {
 183     const Type *T = TypeList[i].first;
 184     int AbbrevToUse = 0;
 185     unsigned Code = 0;
 186
 187     switch (T->getTypeID()) {
 188     default: assert(0 && "Unknown type!");
 189     case Type::VoidTyID:   Code = bitc::TYPE_CODE_VOID;   break;
 190     case Type::FloatTyID:  Code = bitc::TYPE_CODE_FLOAT;  break;
 191     case Type::DoubleTyID: Code = bitc::TYPE_CODE_DOUBLE; break;
 192     case Type::X86_FP80TyID: Code = bitc::TYPE_CODE_X86_FP80; break;
 193     case Type::FP128TyID: Code = bitc::TYPE_CODE_FP128; break;
 194     case Type::PPC_FP128TyID: Code = bitc::TYPE_CODE_PPC_FP128; break;
 195     case Type::LabelTyID:  Code = bitc::TYPE_CODE_LABEL;  break;
 196     case Type::OpaqueTyID: Code = bitc::TYPE_CODE_OPAQUE; break;
 197     case Type::IntegerTyID:
 198       // INTEGER: [width]
 199       Code = bitc::TYPE_CODE_INTEGER;
 200       TypeVals.push_back(cast<IntegerType>(T)->getBitWidth());
 201       break;
 202     case Type::PointerTyID: {
 203       const PointerType *PTy = cast<PointerType>(T);
 204       // POINTER: [pointee type, address space]
 205       Code = bitc::TYPE_CODE_POINTER;
 206       TypeVals.push_back(VE.getTypeID(PTy->getElementType()));
 207       unsigned AddressSpace = PTy->getAddressSpace();
 208       TypeVals.push_back(AddressSpace);
 209       if (AddressSpace == 0) AbbrevToUse = PtrAbbrev;
 210       break;
 211     }
 212     case Type::FunctionTyID: {
 213       const FunctionType *FT = cast<FunctionType>(T);
 214       // FUNCTION: [isvararg, attrid, retty, paramty x N]
 215       Code = bitc::TYPE_CODE_FUNCTION;
 216       TypeVals.push_back(FT->isVarArg());
 217       TypeVals.push_back(0);  // FIXME: DEAD: remove in llvm 3.0
 218       TypeVals.push_back(VE.getTypeID(FT->getReturnType()));
 219       for (unsigned i = 0, e = FT->getNumParams(); i != e; ++i)
 220         TypeVals.push_back(VE.getTypeID(FT->getParamType(i)));
 221       AbbrevToUse = FunctionAbbrev;
 222       break;
 223     }
 224     case Type::StructTyID: {
 225       const StructType *ST = cast<StructType>(T);
 226       // STRUCT: [ispacked, eltty x N]
 227       Code = bitc::TYPE_CODE_STRUCT;
 228       TypeVals.push_back(ST->isPacked());
 229       // Output all of the element types.
 230       for (StructType::element_iterator I = ST->element_begin(),
 231            E = ST->element_end(); I != E; ++I)
 232         TypeVals.push_back(VE.getTypeID(*I));
 233       AbbrevToUse = StructAbbrev;
 234       break;
 235     }
 236     case Type::ArrayTyID: {
 237       const ArrayType *AT = cast<ArrayType>(T);
 238       // ARRAY: [numelts, eltty]
 239       Code = bitc::TYPE_CODE_ARRAY;
 240       TypeVals.push_back(AT->getNumElements());
 241       TypeVals.push_back(VE.getTypeID(AT->getElementType()));
 242       AbbrevToUse = ArrayAbbrev;
 243       break;
 244     }
 245     case Type::VectorTyID: {
 246       const VectorType *VT = cast<VectorType>(T);
 247       // VECTOR [numelts, eltty]
 248       Code = bitc::TYPE_CODE_VECTOR;
 249       TypeVals.push_back(VT->getNumElements());
 250       TypeVals.push_back(VE.getTypeID(VT->getElementType()));
 251       break;
 252     }
 253     }
 254
 255     // Emit the finished record.
 256     Stream.EmitRecord(Code, TypeVals, AbbrevToUse);
 257     TypeVals.clear();
 258   }
 259
 260   Stream.ExitBlock();
 261 }
 262
 263 static unsigned getEncodedLinkage(const GlobalValue *GV) {
 264   switch (GV->getLinkage()) {
 265   default: assert(0 && "Invalid linkage!");
 266   case GlobalValue::GhostLinkage:  // Map ghost linkage onto external.
 267   case GlobalValue::ExternalLinkage:     return 0;
 268   case GlobalValue::WeakLinkage:         return 1;
 269   case GlobalValue::AppendingLinkage:    return 2;
 270   case GlobalValue::InternalLinkage:     return 3;
 271   case GlobalValue::LinkOnceLinkage:     return 4;
 272   case GlobalValue::DLLImportLinkage:    return 5;
 273   case GlobalValue::DLLExportLinkage:    return 6;
 274   case GlobalValue::ExternalWeakLinkage: return 7;
 275   case GlobalValue::CommonLinkage:       return 8;
 276   }
 277 }
 278
 279 static unsigned getEncodedVisibility(const GlobalValue *GV) {
 280   switch (GV->getVisibility()) {
 281   default: assert(0 && "Invalid visibility!");
 282   case GlobalValue::DefaultVisibility:   return 0;
 283   case GlobalValue::HiddenVisibility:    return 1;
 284   case GlobalValue::ProtectedVisibility: return 2;
 285   }
 286 }
 287
 288 // Emit top-level description of module, including target triple, inline asm,
 289 // descriptors for global variables, and function prototype info.
 290 static void WriteModuleInfo(const Module *M, const ValueEnumerator &VE,
 291                             BitstreamWriter &Stream) {
 292   // Emit the list of dependent libraries for the Module.
 293   for (Module::lib_iterator I = M->lib_begin(), E = M->lib_end(); I != E; ++I)
 294     WriteStringRecord(bitc::MODULE_CODE_DEPLIB, *I, 0/*TODO*/, Stream);
 295
 296   // Emit various pieces of data attached to a module.
 297   if (!M->getTargetTriple().empty())
 298     WriteStringRecord(bitc::MODULE_CODE_TRIPLE, M->getTargetTriple(),
 299                       0/*TODO*/, Stream);
 300   if (!M->getDataLayout().empty())
 301     WriteStringRecord(bitc::MODULE_CODE_DATALAYOUT, M->getDataLayout(),
 302                       0/*TODO*/, Stream);
 303   if (!M->getModuleInlineAsm().empty())
 304     WriteStringRecord(bitc::MODULE_CODE_ASM, M->getModuleInlineAsm(),
 305                       0/*TODO*/, Stream);
 306
 307   // Emit information about sections and GC, computing how many there are. Also
 308   // compute the maximum alignment value.
 309   std::map<std::string, unsigned> SectionMap;
 310   std::map<std::string, unsigned> GCMap;
 311   unsigned MaxAlignment = 0;
 312   unsigned MaxGlobalType = 0;
 313   for (Module::const_global_iterator GV = M->global_begin(),E = M->global_end();
 314        GV != E; ++GV) {
 315     MaxAlignment = std::max(MaxAlignment, GV->getAlignment());
 316     MaxGlobalType = std::max(MaxGlobalType, VE.getTypeID(GV->getType()));
 317
 318     if (!GV->hasSection()) continue;
 319     // Give section names unique ID's.
 320     unsigned &Entry = SectionMap[GV->getSection()];
 321     if (Entry != 0) continue;
 322     WriteStringRecord(bitc::MODULE_CODE_SECTIONNAME, GV->getSection(),
 323                       0/*TODO*/, Stream);
 324     Entry = SectionMap.size();
 325   }
 326   for (Module::const_iterator F = M->begin(), E = M->end(); F != E; ++F) {
 327     MaxAlignment = std::max(MaxAlignment, F->getAlignment());
 328     if (F->hasSection()) {
 329       // Give section names unique ID's.
 330       unsigned &Entry = SectionMap[F->getSection()];
 331       if (!Entry) {
 332         WriteStringRecord(bitc::MODULE_CODE_SECTIONNAME, F->getSection(),
 333                           0/*TODO*/, Stream);
 334         Entry = SectionMap.size();
 335       }
 336     }
 337     if (F->hasGC()) {
 338       // Same for GC names.
 339       unsigned &Entry = GCMap[F->getGC()];
 340       if (!Entry) {
 341         WriteStringRecord(bitc::MODULE_CODE_GCNAME, F->getGC(),
 342                           0/*TODO*/, Stream);
 343         Entry = GCMap.size();
 344       }
 345     }
 346   }
 347
 348   // Emit abbrev for globals, now that we know # sections and max alignment.
 349   unsigned SimpleGVarAbbrev = 0;
 350   if (!M->global_empty()) {
 351     // Add an abbrev for common globals with no visibility or thread localness.
 352     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
 353     Abbv->Add(BitCodeAbbrevOp(bitc::MODULE_CODE_GLOBALVAR));
 354     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed,
 355                               Log2_32_Ceil(MaxGlobalType+1)));
 356     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 1));      // Constant.
 357     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 6));        // Initializer.
 358     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 4));      // Linkage.
 359     if (MaxAlignment == 0)                                      // Alignment.
 360       Abbv->Add(BitCodeAbbrevOp(0));
 361     else {
 362       unsigned MaxEncAlignment = Log2_32(MaxAlignment)+1;
 363       Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed,
 364                                Log2_32_Ceil(MaxEncAlignment+1)));
 365     }
 366     if (SectionMap.empty())                                    // Section.
 367       Abbv->Add(BitCodeAbbrevOp(0));
 368     else
 369       Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed,
 370                                Log2_32_Ceil(SectionMap.size()+1)));
 371     // Don't bother emitting vis + thread local.
 372     SimpleGVarAbbrev = Stream.EmitAbbrev(Abbv);
 373   }
 374
 375   // Emit the global variable information.
 376   SmallVector<unsigned, 64> Vals;
 377   for (Module::const_global_iterator GV = M->global_begin(),E = M->global_end();
 378        GV != E; ++GV) {
 379     unsigned AbbrevToUse = 0;
 380
 381     // GLOBALVAR: [type, isconst, initid,
 382     //             linkage, alignment, section, visibility, threadlocal]
 383     Vals.push_back(VE.getTypeID(GV->getType()));
 384     Vals.push_back(GV->isConstant());
 385     Vals.push_back(GV->isDeclaration() ? 0 :
 386                    (VE.getValueID(GV->getInitializer()) + 1));
 387     Vals.push_back(getEncodedLinkage(GV));
 388     Vals.push_back(Log2_32(GV->getAlignment())+1);
 389     Vals.push_back(GV->hasSection() ? SectionMap[GV->getSection()] : 0);
 390     if (GV->isThreadLocal() ||
 391         GV->getVisibility() != GlobalValue::DefaultVisibility) {
 392       Vals.push_back(getEncodedVisibility(GV));
 393       Vals.push_back(GV->isThreadLocal());
 394     } else {
 395       AbbrevToUse = SimpleGVarAbbrev;
 396     }
 397
 398     Stream.EmitRecord(bitc::MODULE_CODE_GLOBALVAR, Vals, AbbrevToUse);
 399     Vals.clear();
 400   }
 401
 402   // Emit the function proto information.
 403   for (Module::const_iterator F = M->begin(), E = M->end(); F != E; ++F) {
 404     // FUNCTION:  [type, callingconv, isproto, paramattr,
 405     //             linkage, alignment, section, visibility, gc]
 406     Vals.push_back(VE.getTypeID(F->getType()));
 407     Vals.push_back(F->getCallingConv());
 408     Vals.push_back(F->isDeclaration());
 409     Vals.push_back(getEncodedLinkage(F));
 410     Vals.push_back(VE.getParamAttrID(F->getParamAttrs()));
 411     Vals.push_back(Log2_32(F->getAlignment())+1);
 412     Vals.push_back(F->hasSection() ? SectionMap[F->getSection()] : 0);
 413     Vals.push_back(getEncodedVisibility(F));
 414     Vals.push_back(F->hasGC() ? GCMap[F->getGC()] : 0);
 415
 416     unsigned AbbrevToUse = 0;
 417     Stream.EmitRecord(bitc::MODULE_CODE_FUNCTION, Vals, AbbrevToUse);
 418     Vals.clear();
 419   }
 420
 421
 422   // Emit the alias information.
 423   for (Module::const_alias_iterator AI = M->alias_begin(), E = M->alias_end();
 424        AI != E; ++AI) {
 425     Vals.push_back(VE.getTypeID(AI->getType()));
 426     Vals.push_back(VE.getValueID(AI->getAliasee()));
 427     Vals.push_back(getEncodedLinkage(AI));
 428     Vals.push_back(getEncodedVisibility(AI));
 429     unsigned AbbrevToUse = 0;
 430     Stream.EmitRecord(bitc::MODULE_CODE_ALIAS, Vals, AbbrevToUse);
 431     Vals.clear();
 432   }
 433 }
 434
 435
 436 static void WriteConstants(unsigned FirstVal, unsigned LastVal,
 437                            const ValueEnumerator &VE,
 438                            BitstreamWriter &Stream, bool isGlobal) {
 439   if (FirstVal == LastVal) return;
 440
 441   Stream.EnterSubblock(bitc::CONSTANTS_BLOCK_ID, 4);
 442
 443   unsigned AggregateAbbrev = 0;
 444   unsigned String8Abbrev = 0;
 445   unsigned CString7Abbrev = 0;
 446   unsigned CString6Abbrev = 0;
 447   // If this is a constant pool for the module, emit module-specific abbrevs.
 448   if (isGlobal) {
 449     // Abbrev for CST_CODE_AGGREGATE.
 450     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
 451     Abbv->Add(BitCodeAbbrevOp(bitc::CST_CODE_AGGREGATE));
 452     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Array));
 453     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, Log2_32_Ceil(LastVal+1)));
 454     AggregateAbbrev = Stream.EmitAbbrev(Abbv);
 455
 456     // Abbrev for CST_CODE_STRING.
 457     Abbv = new BitCodeAbbrev();
 458     Abbv->Add(BitCodeAbbrevOp(bitc::CST_CODE_STRING));
 459     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Array));
 460     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 8));
 461     String8Abbrev = Stream.EmitAbbrev(Abbv);
 462     // Abbrev for CST_CODE_CSTRING.
 463     Abbv = new BitCodeAbbrev();
 464     Abbv->Add(BitCodeAbbrevOp(bitc::CST_CODE_CSTRING));
 465     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Array));
 466     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 7));
 467     CString7Abbrev = Stream.EmitAbbrev(Abbv);
 468     // Abbrev for CST_CODE_CSTRING.
 469     Abbv = new BitCodeAbbrev();
 470     Abbv->Add(BitCodeAbbrevOp(bitc::CST_CODE_CSTRING));
 471     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Array));
 472     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Char6));
 473     CString6Abbrev = Stream.EmitAbbrev(Abbv);
 474   }
 475
 476   SmallVector<uint64_t, 64> Record;
 477
 478   const ValueEnumerator::ValueList &Vals = VE.getValues();
 479   const Type *LastTy = 0;
 480   for (unsigned i = FirstVal; i != LastVal; ++i) {
 481     const Value *V = Vals[i].first;
 482     // If we need to switch types, do so now.
 483     if (V->getType() != LastTy) {
 484       LastTy = V->getType();
 485       Record.push_back(VE.getTypeID(LastTy));
 486       Stream.EmitRecord(bitc::CST_CODE_SETTYPE, Record,
 487                         CONSTANTS_SETTYPE_ABBREV);
 488       Record.clear();
 489     }
 490
 491     if (const InlineAsm *IA = dyn_cast<InlineAsm>(V)) {
 492       Record.push_back(unsigned(IA->hasSideEffects()));
 493
 494       // Add the asm string.
 495       const std::string &AsmStr = IA->getAsmString();
 496       Record.push_back(AsmStr.size());
 497       for (unsigned i = 0, e = AsmStr.size(); i != e; ++i)
 498         Record.push_back(AsmStr[i]);
 499
 500       // Add the constraint string.
 501       const std::string &ConstraintStr = IA->getConstraintString();
 502       Record.push_back(ConstraintStr.size());
 503       for (unsigned i = 0, e = ConstraintStr.size(); i != e; ++i)
 504         Record.push_back(ConstraintStr[i]);
 505       Stream.EmitRecord(bitc::CST_CODE_INLINEASM, Record);
 506       Record.clear();
 507       continue;
 508     }
 509     const Constant *C = cast<Constant>(V);
 510     unsigned Code = -1U;
 511     unsigned AbbrevToUse = 0;
 512     if (C->isNullValue()) {
 513       Code = bitc::CST_CODE_NULL;
 514     } else if (isa<UndefValue>(C)) {
 515       Code = bitc::CST_CODE_UNDEF;
 516     } else if (const ConstantInt *IV = dyn_cast<ConstantInt>(C)) {
 517       if (IV->getBitWidth() <= 64) {
 518         int64_t V = IV->getSExtValue();
 519         if (V >= 0)
 520           Record.push_back(V << 1);
 521         else
 522           Record.push_back((-V << 1) | 1);
 523         Code = bitc::CST_CODE_INTEGER;
 524         AbbrevToUse = CONSTANTS_INTEGER_ABBREV;
 525       } else {                             // Wide integers, > 64 bits in size.
 526         // We have an arbitrary precision integer value to write whose
 527         // bit width is > 64. However, in canonical unsigned integer
 528         // format it is likely that the high bits are going to be zero.
 529         // So, we only write the number of active words.
 530         unsigned NWords = IV->getValue().getActiveWords();
 531         const uint64_t *RawWords = IV->getValue().getRawData();
 532         for (unsigned i = 0; i != NWords; ++i) {
 533           int64_t V = RawWords[i];
 534           if (V >= 0)
 535             Record.push_back(V << 1);
 536           else
 537             Record.push_back((-V << 1) | 1);
 538         }
 539         Code = bitc::CST_CODE_WIDE_INTEGER;
 540       }
 541     } else if (const ConstantFP *CFP = dyn_cast<ConstantFP>(C)) {
 542       Code = bitc::CST_CODE_FLOAT;
 543       const Type *Ty = CFP->getType();
 544       if (Ty == Type::FloatTy || Ty == Type::DoubleTy) {
 545         Record.push_back(CFP->getValueAPF().convertToAPInt().getZExtValue());
 546       } else if (Ty == Type::X86_FP80Ty) {
 547         // api needed to prevent premature destruction
 548         APInt api = CFP->getValueAPF().convertToAPInt();
 549         const uint64_t *p = api.getRawData();
 550         Record.push_back(p[0]);
 551         Record.push_back((uint16_t)p[1]);
 552       } else if (Ty == Type::FP128Ty || Ty == Type::PPC_FP128Ty) {
 553         APInt api = CFP->getValueAPF().convertToAPInt();
 554         const uint64_t *p = api.getRawData();
 555         Record.push_back(p[0]);
 556         Record.push_back(p[1]);
 557       } else {
 558         assert (0 && "Unknown FP type!");
 559       }
 560     } else if (isa<ConstantArray>(C) && cast<ConstantArray>(C)->isString()) {
 561       // Emit constant strings specially.
 562       unsigned NumOps = C->getNumOperands();
 563       // If this is a null-terminated string, use the denser CSTRING encoding.
 564       if (C->getOperand(NumOps-1)->isNullValue()) {
 565         Code = bitc::CST_CODE_CSTRING;
 566         --NumOps;  // Don't encode the null, which isn't allowed by char6.
 567       } else {
 568         Code = bitc::CST_CODE_STRING;
 569         AbbrevToUse = String8Abbrev;
 570       }
 571       bool isCStr7 = Code == bitc::CST_CODE_CSTRING;
 572       bool isCStrChar6 = Code == bitc::CST_CODE_CSTRING;
 573       for (unsigned i = 0; i != NumOps; ++i) {
 574         unsigned char V = cast<ConstantInt>(C->getOperand(i))->getZExtValue();
 575         Record.push_back(V);
 576         isCStr7 &= (V & 128) == 0;
 577         if (isCStrChar6)
 578           isCStrChar6 = BitCodeAbbrevOp::isChar6(V);
 579       }
 580
 581       if (isCStrChar6)
 582         AbbrevToUse = CString6Abbrev;
 583       else if (isCStr7)
 584         AbbrevToUse = CString7Abbrev;
 585     } else if (isa<ConstantArray>(C) || isa<ConstantStruct>(V) ||
 586                isa<ConstantVector>(V)) {
 587       Code = bitc::CST_CODE_AGGREGATE;
 588       for (unsigned i = 0, e = C->getNumOperands(); i != e; ++i)
 589         Record.push_back(VE.getValueID(C->getOperand(i)));
 590       AbbrevToUse = AggregateAbbrev;
 591     } else if (const ConstantExpr *CE = dyn_cast<ConstantExpr>(C)) {
 592       switch (CE->getOpcode()) {
 593       default:
 594         if (Instruction::isCast(CE->getOpcode())) {
 595           Code = bitc::CST_CODE_CE_CAST;
 596           Record.push_back(GetEncodedCastOpcode(CE->getOpcode()));
 597           Record.push_back(VE.getTypeID(C->getOperand(0)->getType()));
 598           Record.push_back(VE.getValueID(C->getOperand(0)));
 599           AbbrevToUse = CONSTANTS_CE_CAST_Abbrev;
 600         } else {
 601           assert(CE->getNumOperands() == 2 && "Unknown constant expr!");
 602           Code = bitc::CST_CODE_CE_BINOP;
 603           Record.push_back(GetEncodedBinaryOpcode(CE->getOpcode()));
 604           Record.push_back(VE.getValueID(C->getOperand(0)));
 605           Record.push_back(VE.getValueID(C->getOperand(1)));
 606         }
 607         break;
 608       case Instruction::GetElementPtr:
 609         Code = bitc::CST_CODE_CE_GEP;
 610         for (unsigned i = 0, e = CE->getNumOperands(); i != e; ++i) {
 611           Record.push_back(VE.getTypeID(C->getOperand(i)->getType()));
 612           Record.push_back(VE.getValueID(C->getOperand(i)));
 613         }
 614         break;
 615       case Instruction::Select:
 616         Code = bitc::CST_CODE_CE_SELECT;
 617         Record.push_back(VE.getValueID(C->getOperand(0)));
 618         Record.push_back(VE.getValueID(C->getOperand(1)));
 619         Record.push_back(VE.getValueID(C->getOperand(2)));
 620         break;
 621       case Instruction::ExtractElement:
 622         Code = bitc::CST_CODE_CE_EXTRACTELT;
 623         Record.push_back(VE.getTypeID(C->getOperand(0)->getType()));
 624         Record.push_back(VE.getValueID(C->getOperand(0)));
 625         Record.push_back(VE.getValueID(C->getOperand(1)));
 626         break;
 627       case Instruction::InsertElement:
 628         Code = bitc::CST_CODE_CE_INSERTELT;
 629         Record.push_back(VE.getValueID(C->getOperand(0)));
 630         Record.push_back(VE.getValueID(C->getOperand(1)));
 631         Record.push_back(VE.getValueID(C->getOperand(2)));
 632         break;
 633       case Instruction::ShuffleVector:
 634         Code = bitc::CST_CODE_CE_SHUFFLEVEC;
 635         Record.push_back(VE.getValueID(C->getOperand(0)));
 636         Record.push_back(VE.getValueID(C->getOperand(1)));
 637         Record.push_back(VE.getValueID(C->getOperand(2)));
 638         break;
 639       case Instruction::ICmp:
 640       case Instruction::FCmp:
 641       case Instruction::VICmp:
 642       case Instruction::VFCmp:
 643         Code = bitc::CST_CODE_CE_CMP;
 644         Record.push_back(VE.getTypeID(C->getOperand(0)->getType()));
 645         Record.push_back(VE.getValueID(C->getOperand(0)));
 646         Record.push_back(VE.getValueID(C->getOperand(1)));
 647         Record.push_back(CE->getPredicate());
 648         break;
 649       }
 650     } else {
 651       assert(0 && "Unknown constant!");
 652     }
 653     Stream.EmitRecord(Code, Record, AbbrevToUse);
 654     Record.clear();
 655   }
 656
 657   Stream.ExitBlock();
 658 }
 659
 660 static void WriteModuleConstants(const ValueEnumerator &VE,
 661                                  BitstreamWriter &Stream) {
 662   const ValueEnumerator::ValueList &Vals = VE.getValues();
 663
 664   // Find the first constant to emit, which is the first non-globalvalue value.
 665   // We know globalvalues have been emitted by WriteModuleInfo.
 666   for (unsigned i = 0, e = Vals.size(); i != e; ++i) {
 667     if (!isa<GlobalValue>(Vals[i].first)) {
 668       WriteConstants(i, Vals.size(), VE, Stream, true);
 669       return;
 670     }
 671   }
 672 }
 673
 674 /// PushValueAndType - The file has to encode both the value and type id for
 675 /// many values, because we need to know what type to create for forward
 676 /// references.  However, most operands are not forward references, so this type
 677 /// field is not needed.
 678 ///
 679 /// This function adds V's value ID to Vals.  If the value ID is higher than the
 680 /// instruction ID, then it is a forward reference, and it also includes the
 681 /// type ID.
 682 static bool PushValueAndType(Value *V, unsigned InstID,
 683                              SmallVector<unsigned, 64> &Vals,
 684                              ValueEnumerator &VE) {
 685   unsigned ValID = VE.getValueID(V);
 686   Vals.push_back(ValID);
 687   if (ValID >= InstID) {
 688     Vals.push_back(VE.getTypeID(V->getType()));
 689     return true;
 690   }
 691   return false;
 692 }
 693
 694 /// WriteInstruction - Emit an instruction to the specified stream.
 695 static void WriteInstruction(const Instruction &I, unsigned InstID,
 696                              ValueEnumerator &VE, BitstreamWriter &Stream,
 697                              SmallVector<unsigned, 64> &Vals) {
 698   unsigned Code = 0;
 699   unsigned AbbrevToUse = 0;
 700   switch (I.getOpcode()) {
 701   default:
 702     if (Instruction::isCast(I.getOpcode())) {
 703       Code = bitc::FUNC_CODE_INST_CAST;
 704       if (!PushValueAndType(I.getOperand(0), InstID, Vals, VE))
 705         AbbrevToUse = FUNCTION_INST_CAST_ABBREV;
 706       Vals.push_back(VE.getTypeID(I.getType()));
 707       Vals.push_back(GetEncodedCastOpcode(I.getOpcode()));
 708     } else {
 709       assert(isa<BinaryOperator>(I) && "Unknown instruction!");
 710       Code = bitc::FUNC_CODE_INST_BINOP;
 711       if (!PushValueAndType(I.getOperand(0), InstID, Vals, VE))
 712         AbbrevToUse = FUNCTION_INST_BINOP_ABBREV;
 713       Vals.push_back(VE.getValueID(I.getOperand(1)));
 714       Vals.push_back(GetEncodedBinaryOpcode(I.getOpcode()));
 715     }
 716     break;
 717
 718   case Instruction::GetElementPtr:
 719     Code = bitc::FUNC_CODE_INST_GEP;
 720     for (unsigned i = 0, e = I.getNumOperands(); i != e; ++i)
 721       PushValueAndType(I.getOperand(i), InstID, Vals, VE);
 722     break;
 723   case Instruction::ExtractValue: {
 724     Code = bitc::FUNC_CODE_INST_EXTRACTVAL;
 725     PushValueAndType(I.getOperand(0), InstID, Vals, VE);
 726     const ExtractValueInst *EVI = cast<ExtractValueInst>(&I);
 727     for (const unsigned *i = EVI->idx_begin(), *e = EVI->idx_end(); i != e; ++i)
 728       Vals.push_back(*i);
 729     break;
 730   }
 731   case Instruction::InsertValue: {
 732     Code = bitc::FUNC_CODE_INST_INSERTVAL;
 733     PushValueAndType(I.getOperand(0), InstID, Vals, VE);
 734     PushValueAndType(I.getOperand(1), InstID, Vals, VE);
 735     const InsertValueInst *IVI = cast<InsertValueInst>(&I);
 736     for (const unsigned *i = IVI->idx_begin(), *e = IVI->idx_end(); i != e; ++i)
 737       Vals.push_back(*i);
 738     break;
 739   }
 740   case Instruction::Select:
 741     Code = bitc::FUNC_CODE_INST_SELECT;
 742     PushValueAndType(I.getOperand(1), InstID, Vals, VE);
 743     Vals.push_back(VE.getValueID(I.getOperand(2)));
 744     Vals.push_back(VE.getValueID(I.getOperand(0)));
 745     break;
 746   case Instruction::ExtractElement:
 747     Code = bitc::FUNC_CODE_INST_EXTRACTELT;
 748     PushValueAndType(I.getOperand(0), InstID, Vals, VE);
 749     Vals.push_back(VE.getValueID(I.getOperand(1)));
 750     break;
 751   case Instruction::InsertElement:
 752     Code = bitc::FUNC_CODE_INST_INSERTELT;
 753     PushValueAndType(I.getOperand(0), InstID, Vals, VE);
 754     Vals.push_back(VE.getValueID(I.getOperand(1)));
 755     Vals.push_back(VE.getValueID(I.getOperand(2)));
 756     break;
 757   case Instruction::ShuffleVector:
 758     Code = bitc::FUNC_CODE_INST_SHUFFLEVEC;
 759     PushValueAndType(I.getOperand(0), InstID, Vals, VE);
 760     Vals.push_back(VE.getValueID(I.getOperand(1)));
 761     Vals.push_back(VE.getValueID(I.getOperand(2)));
 762     break;
 763   case Instruction::ICmp:
 764   case Instruction::FCmp:
 765   case Instruction::VICmp:
 766   case Instruction::VFCmp:
 767     Code = bitc::FUNC_CODE_INST_CMP;
 768     PushValueAndType(I.getOperand(0), InstID, Vals, VE);
 769     Vals.push_back(VE.getValueID(I.getOperand(1)));
 770     Vals.push_back(cast<CmpInst>(I).getPredicate());
 771     break;
 772
 773   case Instruction::Ret:
 774     {
 775       Code = bitc::FUNC_CODE_INST_RET;
 776       unsigned NumOperands = I.getNumOperands();
 777       if (NumOperands == 0)
 778         AbbrevToUse = FUNCTION_INST_RET_VOID_ABBREV;
 779       else if (NumOperands == 1) {
 780         if (!PushValueAndType(I.getOperand(0), InstID, Vals, VE))
 781           AbbrevToUse = FUNCTION_INST_RET_VAL_ABBREV;
 782       } else {
 783         for (unsigned i = 0, e = NumOperands; i != e; ++i)
 784           PushValueAndType(I.getOperand(i), InstID, Vals, VE);
 785       }
 786     }
 787     break;
 788   case Instruction::Br:
 789     Code = bitc::FUNC_CODE_INST_BR;
 790     Vals.push_back(VE.getValueID(I.getOperand(0)));
 791     if (cast<BranchInst>(I).isConditional()) {
 792       Vals.push_back(VE.getValueID(I.getOperand(1)));
 793       Vals.push_back(VE.getValueID(I.getOperand(2)));
 794     }
 795     break;
 796   case Instruction::Switch:
 797     Code = bitc::FUNC_CODE_INST_SWITCH;
 798     Vals.push_back(VE.getTypeID(I.getOperand(0)->getType()));
 799     for (unsigned i = 0, e = I.getNumOperands(); i != e; ++i)
 800       Vals.push_back(VE.getValueID(I.getOperand(i)));
 801     break;
 802   case Instruction::Invoke: {
 803     const PointerType *PTy = cast<PointerType>(I.getOperand(0)->getType());
 804     const FunctionType *FTy = cast<FunctionType>(PTy->getElementType());
 805     Code = bitc::FUNC_CODE_INST_INVOKE;
 806
 807     const InvokeInst *II = cast<InvokeInst>(&I);
 808     Vals.push_back(VE.getParamAttrID(II->getParamAttrs()));
 809     Vals.push_back(II->getCallingConv());
 810     Vals.push_back(VE.getValueID(I.getOperand(1)));      // normal dest
 811     Vals.push_back(VE.getValueID(I.getOperand(2)));      // unwind dest
 812     PushValueAndType(I.getOperand(0), InstID, Vals, VE); // callee
 813
 814     // Emit value #'s for the fixed parameters.
 815     for (unsigned i = 0, e = FTy->getNumParams(); i != e; ++i)
 816       Vals.push_back(VE.getValueID(I.getOperand(i+3)));  // fixed param.
 817
 818     // Emit type/value pairs for varargs params.
 819     if (FTy->isVarArg()) {
 820       for (unsigned i = 3+FTy->getNumParams(), e = I.getNumOperands();
 821            i != e; ++i)
 822         PushValueAndType(I.getOperand(i), InstID, Vals, VE); // vararg
 823     }
 824     break;
 825   }
 826   case Instruction::Unwind:
 827     Code = bitc::FUNC_CODE_INST_UNWIND;
 828     break;
 829   case Instruction::Unreachable:
 830     Code = bitc::FUNC_CODE_INST_UNREACHABLE;
 831     AbbrevToUse = FUNCTION_INST_UNREACHABLE_ABBREV;
 832     break;
 833
 834   case Instruction::PHI:
 835     Code = bitc::FUNC_CODE_INST_PHI;
 836     Vals.push_back(VE.getTypeID(I.getType()));
 837     for (unsigned i = 0, e = I.getNumOperands(); i != e; ++i)
 838       Vals.push_back(VE.getValueID(I.getOperand(i)));
 839     break;
 840
 841   case Instruction::Malloc:
 842     Code = bitc::FUNC_CODE_INST_MALLOC;
 843     Vals.push_back(VE.getTypeID(I.getType()));
 844     Vals.push_back(VE.getValueID(I.getOperand(0))); // size.
 845     Vals.push_back(Log2_32(cast<MallocInst>(I).getAlignment())+1);
 846     break;
 847
 848   case Instruction::Free:
 849     Code = bitc::FUNC_CODE_INST_FREE;
 850     PushValueAndType(I.getOperand(0), InstID, Vals, VE);
 851     break;
 852
 853   case Instruction::Alloca:
 854     Code = bitc::FUNC_CODE_INST_ALLOCA;
 855     Vals.push_back(VE.getTypeID(I.getType()));
 856     Vals.push_back(VE.getValueID(I.getOperand(0))); // size.
 857     Vals.push_back(Log2_32(cast<AllocaInst>(I).getAlignment())+1);
 858     break;
 859
 860   case Instruction::Load:
 861     Code = bitc::FUNC_CODE_INST_LOAD;
 862     if (!PushValueAndType(I.getOperand(0), InstID, Vals, VE))  // ptr
 863       AbbrevToUse = FUNCTION_INST_LOAD_ABBREV;
 864
 865     Vals.push_back(Log2_32(cast<LoadInst>(I).getAlignment())+1);
 866     Vals.push_back(cast<LoadInst>(I).isVolatile());
 867     break;
 868   case Instruction::Store:
 869     Code = bitc::FUNC_CODE_INST_STORE2;
 870     PushValueAndType(I.getOperand(1), InstID, Vals, VE);  // ptrty + ptr
 871     Vals.push_back(VE.getValueID(I.getOperand(0)));       // val.
 872     Vals.push_back(Log2_32(cast<StoreInst>(I).getAlignment())+1);
 873     Vals.push_back(cast<StoreInst>(I).isVolatile());
 874     break;
 875   case Instruction::Call: {
 876     const PointerType *PTy = cast<PointerType>(I.getOperand(0)->getType());
 877     const FunctionType *FTy = cast<FunctionType>(PTy->getElementType());
 878
 879     Code = bitc::FUNC_CODE_INST_CALL;
 880
 881     const CallInst *CI = cast<CallInst>(&I);
 882     Vals.push_back(VE.getParamAttrID(CI->getParamAttrs()));
 883     Vals.push_back((CI->getCallingConv() << 1) | unsigned(CI->isTailCall()));
 884     PushValueAndType(CI->getOperand(0), InstID, Vals, VE);  // Callee
 885
 886     // Emit value #'s for the fixed parameters.
 887     for (unsigned i = 0, e = FTy->getNumParams(); i != e; ++i)
 888       Vals.push_back(VE.getValueID(I.getOperand(i+1)));  // fixed param.
 889
 890     // Emit type/value pairs for varargs params.
 891     if (FTy->isVarArg()) {
 892       unsigned NumVarargs = I.getNumOperands()-1-FTy->getNumParams();
 893       for (unsigned i = I.getNumOperands()-NumVarargs, e = I.getNumOperands();
 894            i != e; ++i)
 895         PushValueAndType(I.getOperand(i), InstID, Vals, VE);  // varargs
 896     }
 897     break;
 898   }
 899   case Instruction::VAArg:
 900     Code = bitc::FUNC_CODE_INST_VAARG;
 901     Vals.push_back(VE.getTypeID(I.getOperand(0)->getType()));   // valistty
 902     Vals.push_back(VE.getValueID(I.getOperand(0))); // valist.
 903     Vals.push_back(VE.getTypeID(I.getType())); // restype.
 904     break;
 905   }
 906
 907   Stream.EmitRecord(Code, Vals, AbbrevToUse);
 908   Vals.clear();
 909 }
 910
 911 // Emit names for globals/functions etc.
 912 static void WriteValueSymbolTable(const ValueSymbolTable &VST,
 913                                   const ValueEnumerator &VE,
 914                                   BitstreamWriter &Stream) {
 915   if (VST.empty()) return;
 916   Stream.EnterSubblock(bitc::VALUE_SYMTAB_BLOCK_ID, 4);
 917
 918   // FIXME: Set up the abbrev, we know how many values there are!
 919   // FIXME: We know if the type names can use 7-bit ascii.
 920   SmallVector<unsigned, 64> NameVals;
 921
 922   for (ValueSymbolTable::const_iterator SI = VST.begin(), SE = VST.end();
 923        SI != SE; ++SI) {
 924
 925     const ValueName &Name = *SI;
 926
 927     // Figure out the encoding to use for the name.
 928     bool is7Bit = true;
 929     bool isChar6 = true;
 930     for (const char *C = Name.getKeyData(), *E = C+Name.getKeyLength();
 931          C != E; ++C) {
 932       if (isChar6)
 933         isChar6 = BitCodeAbbrevOp::isChar6(*C);
 934       if ((unsigned char)*C & 128) {
 935         is7Bit = false;
 936         break;  // don't bother scanning the rest.
 937       }
 938     }
 939
 940     unsigned AbbrevToUse = VST_ENTRY_8_ABBREV;
 941
 942     // VST_ENTRY:   [valueid, namechar x N]
 943     // VST_BBENTRY: [bbid, namechar x N]
 944     unsigned Code;
 945     if (isa<BasicBlock>(SI->getValue())) {
 946       Code = bitc::VST_CODE_BBENTRY;
 947       if (isChar6)
 948         AbbrevToUse = VST_BBENTRY_6_ABBREV;
 949     } else {
 950       Code = bitc::VST_CODE_ENTRY;
 951       if (isChar6)
 952         AbbrevToUse = VST_ENTRY_6_ABBREV;
 953       else if (is7Bit)
 954         AbbrevToUse = VST_ENTRY_7_ABBREV;
 955     }
 956
 957     NameVals.push_back(VE.getValueID(SI->getValue()));
 958     for (const char *P = Name.getKeyData(),
 959          *E = Name.getKeyData()+Name.getKeyLength(); P != E; ++P)
 960       NameVals.push_back((unsigned char)*P);
 961
 962     // Emit the finished record.
 963     Stream.EmitRecord(Code, NameVals, AbbrevToUse);
 964     NameVals.clear();
 965   }
 966   Stream.ExitBlock();
 967 }
 968
 969 /// WriteFunction - Emit a function body to the module stream.
 970 static void WriteFunction(const Function &F, ValueEnumerator &VE,
 971                           BitstreamWriter &Stream) {
 972   Stream.EnterSubblock(bitc::FUNCTION_BLOCK_ID, 4);
 973   VE.incorporateFunction(F);
 974
 975   SmallVector<unsigned, 64> Vals;
 976
 977   // Emit the number of basic blocks, so the reader can create them ahead of
 978   // time.
 979   Vals.push_back(VE.getBasicBlocks().size());
 980   Stream.EmitRecord(bitc::FUNC_CODE_DECLAREBLOCKS, Vals);
 981   Vals.clear();
 982
 983   // If there are function-local constants, emit them now.
 984   unsigned CstStart, CstEnd;
 985   VE.getFunctionConstantRange(CstStart, CstEnd);
 986   WriteConstants(CstStart, CstEnd, VE, Stream, false);
 987
 988   // Keep a running idea of what the instruction ID is.
 989   unsigned InstID = CstEnd;
 990
 991   // Finally, emit all the instructions, in order.
 992   for (Function::const_iterator BB = F.begin(), E = F.end(); BB != E; ++BB)
 993     for (BasicBlock::const_iterator I = BB->begin(), E = BB->end();
 994          I != E; ++I) {
 995       WriteInstruction(*I, InstID, VE, Stream, Vals);
 996       if (I->getType() != Type::VoidTy)
 997         ++InstID;
 998     }
 999
1000   // Emit names for all the instructions etc.
1001   WriteValueSymbolTable(F.getValueSymbolTable(), VE, Stream);
1002
1003   VE.purgeFunction();
1004   Stream.ExitBlock();
1005 }
1006
1007 /// WriteTypeSymbolTable - Emit a block for the specified type symtab.
1008 static void WriteTypeSymbolTable(const TypeSymbolTable &TST,
1009                                  const ValueEnumerator &VE,
1010                                  BitstreamWriter &Stream) {
1011   if (TST.empty()) return;
1012
1013   Stream.EnterSubblock(bitc::TYPE_SYMTAB_BLOCK_ID, 3);
1014
1015   // 7-bit fixed width VST_CODE_ENTRY strings.
1016   BitCodeAbbrev *Abbv = new BitCodeAbbrev();
1017   Abbv->Add(BitCodeAbbrevOp(bitc::VST_CODE_ENTRY));
1018   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed,
1019                             Log2_32_Ceil(VE.getTypes().size()+1)));
1020   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Array));
1021   Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 7));
1022   unsigned V7Abbrev = Stream.EmitAbbrev(Abbv);
1023
1024   SmallVector<unsigned, 64> NameVals;
1025
1026   for (TypeSymbolTable::const_iterator TI = TST.begin(), TE = TST.end();
1027        TI != TE; ++TI) {
1028     // TST_ENTRY: [typeid, namechar x N]
1029     NameVals.push_back(VE.getTypeID(TI->second));
1030
1031     const std::string &Str = TI->first;
1032     bool is7Bit = true;
1033     for (unsigned i = 0, e = Str.size(); i != e; ++i) {
1034       NameVals.push_back((unsigned char)Str[i]);
1035       if (Str[i] & 128)
1036         is7Bit = false;
1037     }
1038
1039     // Emit the finished record.
1040     Stream.EmitRecord(bitc::VST_CODE_ENTRY, NameVals, is7Bit ? V7Abbrev : 0);
1041     NameVals.clear();
1042   }
1043
1044   Stream.ExitBlock();
1045 }
1046
1047 // Emit blockinfo, which defines the standard abbreviations etc.
1048 static void WriteBlockInfo(const ValueEnumerator &VE, BitstreamWriter &Stream) {
1049   // We only want to emit block info records for blocks that have multiple
1050   // instances: CONSTANTS_BLOCK, FUNCTION_BLOCK and VALUE_SYMTAB_BLOCK.  Other
1051   // blocks can defined their abbrevs inline.
1052   Stream.EnterBlockInfoBlock(2);
1053
1054   { // 8-bit fixed-width VST_ENTRY/VST_BBENTRY strings.
1055     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
1056     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 3));
1057     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 8));
1058     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Array));
1059     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 8));
1060     if (Stream.EmitBlockInfoAbbrev(bitc::VALUE_SYMTAB_BLOCK_ID,
1061                                    Abbv) != VST_ENTRY_8_ABBREV)
1062       assert(0 && "Unexpected abbrev ordering!");
1063   }
1064
1065   { // 7-bit fixed width VST_ENTRY strings.
1066     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
1067     Abbv->Add(BitCodeAbbrevOp(bitc::VST_CODE_ENTRY));
1068     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 8));
1069     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Array));
1070     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 7));
1071     if (Stream.EmitBlockInfoAbbrev(bitc::VALUE_SYMTAB_BLOCK_ID,
1072                                    Abbv) != VST_ENTRY_7_ABBREV)
1073       assert(0 && "Unexpected abbrev ordering!");
1074   }
1075   { // 6-bit char6 VST_ENTRY strings.
1076     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
1077     Abbv->Add(BitCodeAbbrevOp(bitc::VST_CODE_ENTRY));
1078     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 8));
1079     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Array));
1080     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Char6));
1081     if (Stream.EmitBlockInfoAbbrev(bitc::VALUE_SYMTAB_BLOCK_ID,
1082                                    Abbv) != VST_ENTRY_6_ABBREV)
1083       assert(0 && "Unexpected abbrev ordering!");
1084   }
1085   { // 6-bit char6 VST_BBENTRY strings.
1086     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
1087     Abbv->Add(BitCodeAbbrevOp(bitc::VST_CODE_BBENTRY));
1088     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 8));
1089     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Array));
1090     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Char6));
1091     if (Stream.EmitBlockInfoAbbrev(bitc::VALUE_SYMTAB_BLOCK_ID,
1092                                    Abbv) != VST_BBENTRY_6_ABBREV)
1093       assert(0 && "Unexpected abbrev ordering!");
1094   }
1095
1096
1097
1098   { // SETTYPE abbrev for CONSTANTS_BLOCK.
1099     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
1100     Abbv->Add(BitCodeAbbrevOp(bitc::CST_CODE_SETTYPE));
1101     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed,
1102                               Log2_32_Ceil(VE.getTypes().size()+1)));
1103     if (Stream.EmitBlockInfoAbbrev(bitc::CONSTANTS_BLOCK_ID,
1104                                    Abbv) != CONSTANTS_SETTYPE_ABBREV)
1105       assert(0 && "Unexpected abbrev ordering!");
1106   }
1107
1108   { // INTEGER abbrev for CONSTANTS_BLOCK.
1109     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
1110     Abbv->Add(BitCodeAbbrevOp(bitc::CST_CODE_INTEGER));
1111     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 8));
1112     if (Stream.EmitBlockInfoAbbrev(bitc::CONSTANTS_BLOCK_ID,
1113                                    Abbv) != CONSTANTS_INTEGER_ABBREV)
1114       assert(0 && "Unexpected abbrev ordering!");
1115   }
1116
1117   { // CE_CAST abbrev for CONSTANTS_BLOCK.
1118     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
1119     Abbv->Add(BitCodeAbbrevOp(bitc::CST_CODE_CE_CAST));
1120     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 4));  // cast opc
1121     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed,       // typeid
1122                               Log2_32_Ceil(VE.getTypes().size()+1)));
1123     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 8));    // value id
1124
1125     if (Stream.EmitBlockInfoAbbrev(bitc::CONSTANTS_BLOCK_ID,
1126                                    Abbv) != CONSTANTS_CE_CAST_Abbrev)
1127       assert(0 && "Unexpected abbrev ordering!");
1128   }
1129   { // NULL abbrev for CONSTANTS_BLOCK.
1130     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
1131     Abbv->Add(BitCodeAbbrevOp(bitc::CST_CODE_NULL));
1132     if (Stream.EmitBlockInfoAbbrev(bitc::CONSTANTS_BLOCK_ID,
1133                                    Abbv) != CONSTANTS_NULL_Abbrev)
1134       assert(0 && "Unexpected abbrev ordering!");
1135   }
1136
1137   // FIXME: This should only use space for first class types!
1138
1139   { // INST_LOAD abbrev for FUNCTION_BLOCK.
1140     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
1141     Abbv->Add(BitCodeAbbrevOp(bitc::FUNC_CODE_INST_LOAD));
1142     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 6)); // Ptr
1143     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 4)); // Align
1144     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 1)); // volatile
1145     if (Stream.EmitBlockInfoAbbrev(bitc::FUNCTION_BLOCK_ID,
1146                                    Abbv) != FUNCTION_INST_LOAD_ABBREV)
1147       assert(0 && "Unexpected abbrev ordering!");
1148   }
1149   { // INST_BINOP abbrev for FUNCTION_BLOCK.
1150     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
1151     Abbv->Add(BitCodeAbbrevOp(bitc::FUNC_CODE_INST_BINOP));
1152     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 6)); // LHS
1153     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 6)); // RHS
1154     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 4)); // opc
1155     if (Stream.EmitBlockInfoAbbrev(bitc::FUNCTION_BLOCK_ID,
1156                                    Abbv) != FUNCTION_INST_BINOP_ABBREV)
1157       assert(0 && "Unexpected abbrev ordering!");
1158   }
1159   { // INST_CAST abbrev for FUNCTION_BLOCK.
1160     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
1161     Abbv->Add(BitCodeAbbrevOp(bitc::FUNC_CODE_INST_CAST));
1162     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 6));    // OpVal
1163     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed,       // dest ty
1164                               Log2_32_Ceil(VE.getTypes().size()+1)));
1165     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 4));  // opc
1166     if (Stream.EmitBlockInfoAbbrev(bitc::FUNCTION_BLOCK_ID,
1167                                    Abbv) != FUNCTION_INST_CAST_ABBREV)
1168       assert(0 && "Unexpected abbrev ordering!");
1169   }
1170
1171   { // INST_RET abbrev for FUNCTION_BLOCK.
1172     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
1173     Abbv->Add(BitCodeAbbrevOp(bitc::FUNC_CODE_INST_RET));
1174     if (Stream.EmitBlockInfoAbbrev(bitc::FUNCTION_BLOCK_ID,
1175                                    Abbv) != FUNCTION_INST_RET_VOID_ABBREV)
1176       assert(0 && "Unexpected abbrev ordering!");
1177   }
1178   { // INST_RET abbrev for FUNCTION_BLOCK.
1179     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
1180     Abbv->Add(BitCodeAbbrevOp(bitc::FUNC_CODE_INST_RET));
1181     Abbv->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::VBR, 6)); // ValID
1182     if (Stream.EmitBlockInfoAbbrev(bitc::FUNCTION_BLOCK_ID,
1183                                    Abbv) != FUNCTION_INST_RET_VAL_ABBREV)
1184       assert(0 && "Unexpected abbrev ordering!");
1185   }
1186   { // INST_UNREACHABLE abbrev for FUNCTION_BLOCK.
1187     BitCodeAbbrev *Abbv = new BitCodeAbbrev();
1188     Abbv->Add(BitCodeAbbrevOp(bitc::FUNC_CODE_INST_UNREACHABLE));
1189     if (Stream.EmitBlockInfoAbbrev(bitc::FUNCTION_BLOCK_ID,
1190                                    Abbv) != FUNCTION_INST_UNREACHABLE_ABBREV)
1191       assert(0 && "Unexpected abbrev ordering!");
1192   }
1193
1194   Stream.ExitBlock();
1195 }
1196
1197
1198 /// WriteModule - Emit the specified module to the bitstream.
1199 static void WriteModule(const Module *M, BitstreamWriter &Stream) {
1200   Stream.EnterSubblock(bitc::MODULE_BLOCK_ID, 3);
1201
1202   // Emit the version number if it is non-zero.
1203   if (CurVersion) {
1204     SmallVector<unsigned, 1> Vals;
1205     Vals.push_back(CurVersion);
1206     Stream.EmitRecord(bitc::MODULE_CODE_VERSION, Vals);
1207   }
1208
1209   // Analyze the module, enumerating globals, functions, etc.
1210   ValueEnumerator VE(M);
1211
1212   // Emit blockinfo, which defines the standard abbreviations etc.
1213   WriteBlockInfo(VE, Stream);
1214
1215   // Emit information about parameter attributes.
1216   WriteParamAttrTable(VE, Stream);
1217
1218   // Emit information describing all of the types in the module.
1219   WriteTypeTable(VE, Stream);
1220
1221   // Emit top-level description of module, including target triple, inline asm,
1222   // descriptors for global variables, and function prototype info.
1223   WriteModuleInfo(M, VE, Stream);
1224
1225   // Emit constants.
1226   WriteModuleConstants(VE, Stream);
1227
1228   // If we have any aggregate values in the value table, purge them - these can
1229   // only be used to initialize global variables.  Doing so makes the value
1230   // namespace smaller for code in functions.
1231   int NumNonAggregates = VE.PurgeAggregateValues();
1232   if (NumNonAggregates != -1) {
1233     SmallVector<unsigned, 1> Vals;
1234     Vals.push_back(NumNonAggregates);
1235     Stream.EmitRecord(bitc::MODULE_CODE_PURGEVALS, Vals);
1236   }
1237
1238   // Emit function bodies.
1239   for (Module::const_iterator I = M->begin(), E = M->end(); I != E; ++I)
1240     if (!I->isDeclaration())
1241       WriteFunction(*I, VE, Stream);
1242
1243   // Emit the type symbol table information.
1244   WriteTypeSymbolTable(M->getTypeSymbolTable(), VE, Stream);
1245
1246   // Emit names for globals/functions etc.
1247   WriteValueSymbolTable(M->getValueSymbolTable(), VE, Stream);
1248
1249   Stream.ExitBlock();
1250 }
1251
1252 /// EmitDarwinBCHeader - If generating a bc file on darwin, we have to emit a
1253 /// header and trailer to make it compatible with the system archiver.  To do
1254 /// this we emit the following header, and then emit a trailer that pads the
1255 /// file out to be a multiple of 16 bytes.
1256 ///
1257 /// struct bc_header {
1258 ///   uint32_t Magic;         // 0x0B17C0DE
1259 ///   uint32_t Version;       // Version, currently always 0.
1260 ///   uint32_t BitcodeOffset; // Offset to traditional bitcode file.
1261 ///   uint32_t BitcodeSize;   // Size of traditional bitcode file.
1262 ///   uint32_t CPUType;       // CPU specifier.
1263 ///   ... potentially more later ...
1264 /// };
1265 enum {
1266   DarwinBCSizeFieldOffset = 3*4, // Offset to bitcode_size.
1267   DarwinBCHeaderSize = 5*4
1268 };
1269
1270 static void EmitDarwinBCHeader(BitstreamWriter &Stream,
1271                                const std::string &TT) {
1272   unsigned CPUType = ~0U;
1273
1274   // Match x86_64-*, i[3-9]86-*, powerpc-*, powerpc64-*.  The CPUType is a
1275   // magic number from /usr/include/mach/machine.h.  It is ok to reproduce the
1276   // specific constants here because they are implicitly part of the Darwin ABI.
1277   enum {
1278     DARWIN_CPU_ARCH_ABI64      = 0x01000000,
1279     DARWIN_CPU_TYPE_X86        = 7,
1280     DARWIN_CPU_TYPE_POWERPC    = 18
1281   };
1282
1283   if (TT.find("x86_64-") == 0)
1284     CPUType = DARWIN_CPU_TYPE_X86 | DARWIN_CPU_ARCH_ABI64;
1285   else if (TT.size() >= 5 && TT[0] == 'i' && TT[2] == '8' && TT[3] == '6' &&
1286            TT[4] == '-' && TT[1] - '3' < 6)
1287     CPUType = DARWIN_CPU_TYPE_X86;
1288   else if (TT.find("powerpc-") == 0)
1289     CPUType = DARWIN_CPU_TYPE_POWERPC;
1290   else if (TT.find("powerpc64-") == 0)
1291     CPUType = DARWIN_CPU_TYPE_POWERPC | DARWIN_CPU_ARCH_ABI64;
1292
1293   // Traditional Bitcode starts after header.
1294   unsigned BCOffset = DarwinBCHeaderSize;
1295
1296   Stream.Emit(0x0B17C0DE, 32);
1297   Stream.Emit(0         , 32);  // Version.
1298   Stream.Emit(BCOffset  , 32);
1299   Stream.Emit(0         , 32);  // Filled in later.
1300   Stream.Emit(CPUType   , 32);
1301 }
1302
1303 /// EmitDarwinBCTrailer - Emit the darwin epilog after the bitcode file and
1304 /// finalize the header.
1305 static void EmitDarwinBCTrailer(BitstreamWriter &Stream, unsigned BufferSize) {
1306   // Update the size field in the header.
1307   Stream.BackpatchWord(DarwinBCSizeFieldOffset, BufferSize-DarwinBCHeaderSize);
1308
1309   // If the file is not a multiple of 16 bytes, insert dummy padding.
1310   while (BufferSize & 15) {
1311     Stream.Emit(0, 8);
1312     ++BufferSize;
1313   }
1314 }
1315
1316
1317 /// WriteBitcodeToFile - Write the specified module to the specified output
1318 /// stream.
1319 void llvm::WriteBitcodeToFile(const Module *M, std::ostream &Out) {
1320   std::vector<unsigned char> Buffer;
1321   BitstreamWriter Stream(Buffer);
1322
1323   Buffer.reserve(256*1024);
1324
1325   // If this is darwin, emit a file header and trailer if needed.
1326   bool isDarwin = M->getTargetTriple().find("-darwin") != std::string::npos;
1327   if (isDarwin)
1328     EmitDarwinBCHeader(Stream, M->getTargetTriple());
1329
1330   // Emit the file header.
1331   Stream.Emit((unsigned)'B', 8);
1332   Stream.Emit((unsigned)'C', 8);
1333   Stream.Emit(0x0, 4);
1334   Stream.Emit(0xC, 4);
1335   Stream.Emit(0xE, 4);
1336   Stream.Emit(0xD, 4);
1337
1338   // Emit the module.
1339   WriteModule(M, Stream);
1340
1341   if (isDarwin)
1342     EmitDarwinBCTrailer(Stream, Buffer.size());
1343
1344
1345   // If writing to stdout, set binary mode.
1346   if (llvm::cout == Out)
1347     sys::Program::ChangeStdoutToBinary();
1348
1349   // Write the generated bitstream to "Out".
1350   Out.write((char*)&Buffer.front(), Buffer.size());
1351
1352   // Make sure it hits disk now.
1353   Out.flush();
1354 }