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