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