lib/MC/MCDisassembler/EDDisassembler.cpp

   1 //===-EDDisassembler.cpp - LLVM Enhanced Disassembler ---------------------===//
   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 // This file implements the Enhanced Disassembly library's  disassembler class.
  11 // The disassembler is responsible for vending individual instructions according
  12 // to a given architecture and disassembly syntax.
  13 //
  14 //===----------------------------------------------------------------------===//
  15
  16 #include "EDDisassembler.h"
  17 #include "EDInst.h"
  18 #include "llvm/MC/EDInstInfo.h"
  19 #include "llvm/MC/MCAsmInfo.h"
  20 #include "llvm/MC/MCContext.h"
  21 #include "llvm/MC/MCDisassembler.h"
  22 #include "llvm/MC/MCExpr.h"
  23 #include "llvm/MC/MCInst.h"
  24 #include "llvm/MC/MCInstPrinter.h"
  25 #include "llvm/MC/MCRegisterInfo.h"
  26 #include "llvm/MC/MCStreamer.h"
  27 #include "llvm/MC/MCSubtargetInfo.h"
  28 #include "llvm/MC/MCParser/AsmLexer.h"
  29 #include "llvm/MC/MCParser/MCAsmParser.h"
  30 #include "llvm/MC/MCParser/MCParsedAsmOperand.h"
  31 #include "llvm/Support/MemoryBuffer.h"
  32 #include "llvm/Support/MemoryObject.h"
  33 #include "llvm/Support/SourceMgr.h"
  34 #include "llvm/Target/TargetAsmLexer.h"
  35 #include "llvm/Target/TargetAsmParser.h"
  36 #include "llvm/Target/TargetRegistry.h"
  37 #include "llvm/Target/TargetMachine.h"
  38 #include "llvm/Target/TargetRegisterInfo.h"
  39 #include "llvm/Target/TargetSelect.h"
  40 using namespace llvm;
  41
  42 bool EDDisassembler::sInitialized = false;
  43 EDDisassembler::DisassemblerMap_t EDDisassembler::sDisassemblers;
  44
  45 struct TripleMap {
  46   Triple::ArchType Arch;
  47   const char *String;
  48 };
  49
  50 static struct TripleMap triplemap[] = {
  51   { Triple::x86,          "i386-unknown-unknown"    },
  52   { Triple::x86_64,       "x86_64-unknown-unknown"  },
  53   { Triple::arm,          "arm-unknown-unknown"     },
  54   { Triple::thumb,        "thumb-unknown-unknown"   },
  55   { Triple::InvalidArch,  NULL,                     }
  56 };
  57
  58 /// infoFromArch - Returns the TripleMap corresponding to a given architecture,
  59 ///   or NULL if there is an error
  60 ///
  61 /// @arg arch - The Triple::ArchType for the desired architecture
  62 static const char *tripleFromArch(Triple::ArchType arch) {
  63   unsigned int infoIndex;
  64
  65   for (infoIndex = 0; triplemap[infoIndex].String != NULL; ++infoIndex) {
  66     if (arch == triplemap[infoIndex].Arch)
  67       return triplemap[infoIndex].String;
  68   }
  69
  70   return NULL;
  71 }
  72
  73 /// getLLVMSyntaxVariant - gets the constant to use to get an assembly printer
  74 ///   for the desired assembly syntax, suitable for passing to
  75 ///   Target::createMCInstPrinter()
  76 ///
  77 /// @arg arch   - The target architecture
  78 /// @arg syntax - The assembly syntax in sd form
  79 static int getLLVMSyntaxVariant(Triple::ArchType arch,
  80                                 EDDisassembler::AssemblySyntax syntax) {
  81   switch (syntax) {
  82   default:
  83     return -1;
  84   // Mappings below from X86AsmPrinter.cpp
  85   case EDDisassembler::kEDAssemblySyntaxX86ATT:
  86     if (arch == Triple::x86 || arch == Triple::x86_64)
  87       return 0;
  88     else
  89       return -1;
  90   case EDDisassembler::kEDAssemblySyntaxX86Intel:
  91     if (arch == Triple::x86 || arch == Triple::x86_64)
  92       return 1;
  93     else
  94       return -1;
  95   case EDDisassembler::kEDAssemblySyntaxARMUAL:
  96     if (arch == Triple::arm || arch == Triple::thumb)
  97       return 0;
  98     else
  99       return -1;
 100   }
 101 }
 102
 103 void EDDisassembler::initialize() {
 104   if (sInitialized)
 105     return;
 106
 107   sInitialized = true;
 108
 109   InitializeAllTargetInfos();
 110   InitializeAllTargets();
 111   InitializeAllTargetMCs();
 112   InitializeAllAsmPrinters();
 113   InitializeAllAsmParsers();
 114   InitializeAllDisassemblers();
 115 }
 116
 117 #undef BRINGUP_TARGET
 118
 119 EDDisassembler *EDDisassembler::getDisassembler(Triple::ArchType arch,
 120                                                 AssemblySyntax syntax) {
 121   CPUKey key;
 122   key.Arch = arch;
 123   key.Syntax = syntax;
 124
 125   EDDisassembler::DisassemblerMap_t::iterator i = sDisassemblers.find(key);
 126
 127   if (i != sDisassemblers.end()) {
 128     return i->second;
 129   } else {
 130     EDDisassembler* sdd = new EDDisassembler(key);
 131     if (!sdd->valid()) {
 132       delete sdd;
 133       return NULL;
 134     }
 135
 136     sDisassemblers[key] = sdd;
 137
 138     return sdd;
 139   }
 140
 141   return NULL;
 142 }
 143
 144 EDDisassembler *EDDisassembler::getDisassembler(StringRef str,
 145                                                 AssemblySyntax syntax) {
 146   return getDisassembler(Triple(str).getArch(), syntax);
 147 }
 148
 149 EDDisassembler::EDDisassembler(CPUKey &key) :
 150   Valid(false),
 151   HasSemantics(false),
 152   ErrorStream(nulls()),
 153   Key(key) {
 154   const char *triple = tripleFromArch(key.Arch);
 155
 156   if (!triple)
 157     return;
 158
 159   LLVMSyntaxVariant = getLLVMSyntaxVariant(key.Arch, key.Syntax);
 160
 161   if (LLVMSyntaxVariant < 0)
 162     return;
 163
 164   std::string tripleString(triple);
 165   std::string errorString;
 166
 167   Tgt = TargetRegistry::lookupTarget(tripleString,
 168                                      errorString);
 169
 170   if (!Tgt)
 171     return;
 172
 173   std::string CPU;
 174   std::string featureString;
 175   TargetMachine.reset(Tgt->createTargetMachine(tripleString, CPU,
 176                                                featureString));
 177
 178   // FIXME: It shouldn't be using TargetRegisterInfo!
 179   const TargetRegisterInfo *registerInfo = TargetMachine->getRegisterInfo();
 180
 181   if (!registerInfo)
 182     return;
 183
 184   initMaps(*registerInfo);
 185
 186   AsmInfo.reset(Tgt->createMCAsmInfo(tripleString));
 187
 188   if (!AsmInfo)
 189     return;
 190
 191   MRI.reset(Tgt->createMCRegInfo(tripleString));
 192
 193   if (!MRI)
 194     return;
 195
 196   Disassembler.reset(Tgt->createMCDisassembler());
 197
 198   if (!Disassembler)
 199     return;
 200
 201   InstInfos = Disassembler->getEDInfo();
 202
 203   InstString.reset(new std::string);
 204   InstStream.reset(new raw_string_ostream(*InstString));
 205   InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo));
 206
 207   if (!InstPrinter)
 208     return;
 209
 210   GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
 211   SpecificAsmLexer.reset(Tgt->createAsmLexer(*AsmInfo));
 212   SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
 213
 214   initMaps(*TargetMachine->getRegisterInfo());
 215
 216   Valid = true;
 217 }
 218
 219 EDDisassembler::~EDDisassembler() {
 220   if (!valid())
 221     return;
 222 }
 223
 224 namespace {
 225   /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
 226   ///   as provided by the sd interface.  See MemoryObject.
 227   class EDMemoryObject : public llvm::MemoryObject {
 228   private:
 229     EDByteReaderCallback Callback;
 230     void *Arg;
 231   public:
 232     EDMemoryObject(EDByteReaderCallback callback,
 233                    void *arg) : Callback(callback), Arg(arg) { }
 234     ~EDMemoryObject() { }
 235     uint64_t getBase() const { return 0x0; }
 236     uint64_t getExtent() const { return (uint64_t)-1; }
 237     int readByte(uint64_t address, uint8_t *ptr) const {
 238       if (!Callback)
 239         return -1;
 240
 241       if (Callback(ptr, address, Arg))
 242         return -1;
 243
 244       return 0;
 245     }
 246   };
 247 }
 248
 249 EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
 250                                    uint64_t address,
 251                                    void *arg) {
 252   EDMemoryObject memoryObject(byteReader, arg);
 253
 254   MCInst* inst = new MCInst;
 255   uint64_t byteSize;
 256
 257   if (!Disassembler->getInstruction(*inst,
 258                                     byteSize,
 259                                     memoryObject,
 260                                     address,
 261                                     ErrorStream)) {
 262     delete inst;
 263     return NULL;
 264   } else {
 265     const llvm::EDInstInfo *thisInstInfo = NULL;
 266
 267     if (InstInfos) {
 268       thisInstInfo = &InstInfos[inst->getOpcode()];
 269     }
 270
 271     EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
 272     return sdInst;
 273   }
 274 }
 275
 276 void EDDisassembler::initMaps(const TargetRegisterInfo &registerInfo) {
 277   unsigned numRegisters = registerInfo.getNumRegs();
 278   unsigned registerIndex;
 279
 280   for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
 281     const char* registerName = registerInfo.get(registerIndex).Name;
 282
 283     RegVec.push_back(registerName);
 284     RegRMap[registerName] = registerIndex;
 285   }
 286
 287   switch (Key.Arch) {
 288   default:
 289     break;
 290   case Triple::x86:
 291   case Triple::x86_64:
 292     stackPointers.insert(registerIDWithName("SP"));
 293     stackPointers.insert(registerIDWithName("ESP"));
 294     stackPointers.insert(registerIDWithName("RSP"));
 295
 296     programCounters.insert(registerIDWithName("IP"));
 297     programCounters.insert(registerIDWithName("EIP"));
 298     programCounters.insert(registerIDWithName("RIP"));
 299     break;
 300   case Triple::arm:
 301   case Triple::thumb:
 302     stackPointers.insert(registerIDWithName("SP"));
 303
 304     programCounters.insert(registerIDWithName("PC"));
 305     break;
 306   }
 307 }
 308
 309 const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
 310   if (registerID >= RegVec.size())
 311     return NULL;
 312   else
 313     return RegVec[registerID].c_str();
 314 }
 315
 316 unsigned EDDisassembler::registerIDWithName(const char *name) const {
 317   regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
 318   if (iter == RegRMap.end())
 319     return 0;
 320   else
 321     return (*iter).second;
 322 }
 323
 324 bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
 325   return (stackPointers.find(registerID) != stackPointers.end());
 326 }
 327
 328 bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
 329   return (programCounters.find(registerID) != programCounters.end());
 330 }
 331
 332 int EDDisassembler::printInst(std::string &str, MCInst &inst) {
 333   PrinterMutex.acquire();
 334
 335   InstPrinter->printInst(&inst, *InstStream);
 336   InstStream->flush();
 337   str = *InstString;
 338   InstString->clear();
 339
 340   PrinterMutex.release();
 341
 342   return 0;
 343 }
 344
 345 static void diag_handler(const SMDiagnostic &diag,
 346                          void *context)
 347 {
 348   if (context) {
 349     EDDisassembler *disassembler = static_cast<EDDisassembler*>(context);
 350     diag.Print("", disassembler->ErrorStream);
 351   }
 352 }
 353
 354 int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
 355                               SmallVectorImpl<AsmToken> &tokens,
 356                               const std::string &str) {
 357   int ret = 0;
 358
 359   switch (Key.Arch) {
 360   default:
 361     return -1;
 362   case Triple::x86:
 363   case Triple::x86_64:
 364   case Triple::arm:
 365   case Triple::thumb:
 366     break;
 367   }
 368
 369   const char *cStr = str.c_str();
 370   MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
 371
 372   StringRef instName;
 373   SMLoc instLoc;
 374
 375   SourceMgr sourceMgr;
 376   sourceMgr.setDiagHandler(diag_handler, static_cast<void*>(this));
 377   sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
 378   MCContext context(*AsmInfo, *MRI, NULL);
 379   OwningPtr<MCStreamer> streamer(createNullStreamer(context));
 380   OwningPtr<MCAsmParser> genericParser(createMCAsmParser(*Tgt, sourceMgr,
 381                                                          context, *streamer,
 382                                                          *AsmInfo));
 383
 384   StringRef triple = tripleFromArch(Key.Arch);
 385   OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(triple, "", ""));
 386   OwningPtr<TargetAsmParser> TargetParser(Tgt->createAsmParser(*STI,
 387                                                                *genericParser));
 388
 389   AsmToken OpcodeToken = genericParser->Lex();
 390   AsmToken NextToken = genericParser->Lex();  // consume next token, because specificParser expects us to
 391
 392   if (OpcodeToken.is(AsmToken::Identifier)) {
 393     instName = OpcodeToken.getString();
 394     instLoc = OpcodeToken.getLoc();
 395
 396     if (NextToken.isNot(AsmToken::Eof) &&
 397         TargetParser->ParseInstruction(instName, instLoc, operands))
 398       ret = -1;
 399   } else {
 400     ret = -1;
 401   }
 402
 403   ParserMutex.acquire();
 404
 405   if (!ret) {
 406     GenericAsmLexer->setBuffer(buf);
 407
 408     while (SpecificAsmLexer->Lex(),
 409            SpecificAsmLexer->isNot(AsmToken::Eof) &&
 410            SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
 411       if (SpecificAsmLexer->is(AsmToken::Error)) {
 412         ret = -1;
 413         break;
 414       }
 415       tokens.push_back(SpecificAsmLexer->getTok());
 416     }
 417   }
 418
 419   ParserMutex.release();
 420
 421   return ret;
 422 }
 423
 424 int EDDisassembler::llvmSyntaxVariant() const {
 425   return LLVMSyntaxVariant;
 426 }