tools/edis/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
  19 #include "llvm/ADT/OwningPtr.h"
  20 #include "llvm/ADT/SmallVector.h"
  21 #include "llvm/MC/MCAsmInfo.h"
  22 #include "llvm/MC/MCContext.h"
  23 #include "llvm/MC/MCDisassembler.h"
  24 #include "llvm/MC/MCExpr.h"
  25 #include "llvm/MC/MCInst.h"
  26 #include "llvm/MC/MCInstPrinter.h"
  27 #include "llvm/MC/MCStreamer.h"
  28 #include "llvm/MC/MCParser/AsmLexer.h"
  29 #include "llvm/MC/MCParser/AsmParser.h"
  30 #include "llvm/MC/MCParser/MCAsmParser.h"
  31 #include "llvm/MC/MCParser/MCParsedAsmOperand.h"
  32 #include "llvm/Support/MemoryBuffer.h"
  33 #include "llvm/Support/MemoryObject.h"
  34 #include "llvm/Support/SourceMgr.h"
  35 #include "llvm/Target/TargetAsmLexer.h"
  36 #include "llvm/Target/TargetAsmParser.h"
  37 #include "llvm/Target/TargetRegistry.h"
  38 #include "llvm/Target/TargetMachine.h"
  39 #include "llvm/Target/TargetRegisterInfo.h"
  40 #include "llvm/Target/TargetSelect.h"
  41
  42 #ifdef EDIS_X86
  43 #include "../../lib/Target/X86/X86GenEDInfo.inc"
  44 #endif
  45
  46 #ifdef EDIS_ARM
  47 #include "../../lib/Target/ARM/ARMGenEDInfo.inc"
  48 #endif
  49
  50 using namespace llvm;
  51
  52 bool EDDisassembler::sInitialized = false;
  53 EDDisassembler::DisassemblerMap_t EDDisassembler::sDisassemblers;
  54
  55 struct InfoMap {
  56   Triple::ArchType Arch;
  57   const char *String;
  58   const InstInfo *Info;
  59 };
  60
  61 static struct InfoMap infomap[] = {
  62 #ifdef EDIS_X86
  63   { Triple::x86,          "i386-unknown-unknown",   instInfoX86 },
  64   { Triple::x86_64,       "x86_64-unknown-unknown", instInfoX86 },
  65 #endif
  66 #ifdef EDIS_ARM
  67   { Triple::arm,          "arm-unknown-unknown",    instInfoARM },
  68   { Triple::thumb,        "thumb-unknown-unknown",  instInfoARM },
  69 #endif
  70   { Triple::InvalidArch,  NULL,                     NULL        }
  71 };
  72
  73 /// infoFromArch - Returns the InfoMap corresponding to a given architecture,
  74 ///   or NULL if there is an error
  75 ///
  76 /// @arg arch - The Triple::ArchType for the desired architecture
  77 static const InfoMap *infoFromArch(Triple::ArchType arch) {
  78   unsigned int infoIndex;
  79
  80   for (infoIndex = 0; infomap[infoIndex].String != NULL; ++infoIndex) {
  81     if (arch == infomap[infoIndex].Arch)
  82       return &infomap[infoIndex];
  83   }
  84
  85   return NULL;
  86 }
  87
  88 /// getLLVMSyntaxVariant - gets the constant to use to get an assembly printer
  89 ///   for the desired assembly syntax, suitable for passing to
  90 ///   Target::createMCInstPrinter()
  91 ///
  92 /// @arg arch   - The target architecture
  93 /// @arg syntax - The assembly syntax in sd form
  94 static int getLLVMSyntaxVariant(Triple::ArchType arch,
  95                                 EDAssemblySyntax_t syntax) {
  96   switch (syntax) {
  97   default:
  98     return -1;
  99   // Mappings below from X86AsmPrinter.cpp
 100   case kEDAssemblySyntaxX86ATT:
 101     if (arch == Triple::x86 || arch == Triple::x86_64)
 102       return 0;
 103     else
 104       return -1;
 105   case kEDAssemblySyntaxX86Intel:
 106     if (arch == Triple::x86 || arch == Triple::x86_64)
 107       return 1;
 108     else
 109       return -1;
 110   case kEDAssemblySyntaxARMUAL:
 111     if (arch == Triple::arm || arch == Triple::thumb)
 112       return 0;
 113     else
 114       return -1;
 115   }
 116 }
 117
 118 #define BRINGUP_TARGET(tgt)           \
 119   LLVMInitialize##tgt##TargetInfo();  \
 120   LLVMInitialize##tgt##Target();      \
 121   LLVMInitialize##tgt##AsmPrinter();  \
 122   LLVMInitialize##tgt##AsmParser();   \
 123   LLVMInitialize##tgt##Disassembler();
 124
 125 void EDDisassembler::initialize() {
 126   if (sInitialized)
 127     return;
 128
 129   sInitialized = true;
 130
 131 #ifdef EDIS_X86
 132   BRINGUP_TARGET(X86)
 133 #endif
 134 #ifdef EDIS_ARM
 135   BRINGUP_TARGET(ARM)
 136 #endif
 137 }
 138
 139 #undef BRINGUP_TARGET
 140
 141 EDDisassembler *EDDisassembler::getDisassembler(Triple::ArchType arch,
 142                                                 EDAssemblySyntax_t syntax) {
 143   CPUKey key;
 144   key.Arch = arch;
 145   key.Syntax = syntax;
 146
 147   EDDisassembler::DisassemblerMap_t::iterator i = sDisassemblers.find(key);
 148
 149   if (i != sDisassemblers.end()) {
 150     return i->second;
 151   } else {
 152     EDDisassembler* sdd = new EDDisassembler(key);
 153     if (!sdd->valid()) {
 154       delete sdd;
 155       return NULL;
 156     }
 157
 158     sDisassemblers[key] = sdd;
 159
 160     return sdd;
 161   }
 162
 163   return NULL;
 164 }
 165
 166 EDDisassembler *EDDisassembler::getDisassembler(StringRef str,
 167                                                 EDAssemblySyntax_t syntax) {
 168   Triple triple(str);
 169
 170   return getDisassembler(triple.getArch(), syntax);
 171 }
 172
 173 EDDisassembler::EDDisassembler(CPUKey &key) :
 174   Valid(false),
 175   HasSemantics(false),
 176   ErrorStream(nulls()),
 177   Key(key) {
 178   const InfoMap *infoMap = infoFromArch(key.Arch);
 179
 180   if (!infoMap)
 181     return;
 182
 183   InstInfos = infoMap->Info;
 184
 185   const char *triple = infoMap->String;
 186
 187   LLVMSyntaxVariant = getLLVMSyntaxVariant(key.Arch, key.Syntax);
 188
 189   if (LLVMSyntaxVariant < 0)
 190     return;
 191
 192   std::string tripleString(triple);
 193   std::string errorString;
 194
 195   Tgt = TargetRegistry::lookupTarget(tripleString,
 196                                      errorString);
 197
 198   if (!Tgt)
 199     return;
 200
 201   std::string featureString;
 202
 203   OwningPtr<const TargetMachine>
 204     targetMachine(Tgt->createTargetMachine(tripleString,
 205                                            featureString));
 206
 207   const TargetRegisterInfo *registerInfo = targetMachine->getRegisterInfo();
 208
 209   if (!registerInfo)
 210     return;
 211
 212   initMaps(*registerInfo);
 213
 214   AsmInfo.reset(Tgt->createAsmInfo(tripleString));
 215
 216   if (!AsmInfo)
 217     return;
 218
 219   Disassembler.reset(Tgt->createMCDisassembler());
 220
 221   if (!Disassembler)
 222     return;
 223
 224   InstString.reset(new std::string);
 225   InstStream.reset(new raw_string_ostream(*InstString));
 226   InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo));
 227
 228   if (!InstPrinter)
 229     return;
 230
 231   GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
 232   SpecificAsmLexer.reset(Tgt->createAsmLexer(*AsmInfo));
 233   SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
 234
 235   InstInfos = infoMap->Info;
 236
 237   initMaps(*targetMachine->getRegisterInfo());
 238
 239   Valid = true;
 240 }
 241
 242 EDDisassembler::~EDDisassembler() {
 243   if (!valid())
 244     return;
 245 }
 246
 247 namespace {
 248   /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
 249   ///   as provided by the sd interface.  See MemoryObject.
 250   class EDMemoryObject : public llvm::MemoryObject {
 251   private:
 252     EDByteReaderCallback Callback;
 253     void *Arg;
 254   public:
 255     EDMemoryObject(EDByteReaderCallback callback,
 256                    void *arg) : Callback(callback), Arg(arg) { }
 257     ~EDMemoryObject() { }
 258     uint64_t getBase() const { return 0x0; }
 259     uint64_t getExtent() const { return (uint64_t)-1; }
 260     int readByte(uint64_t address, uint8_t *ptr) const {
 261       if (!Callback)
 262         return -1;
 263
 264       if (Callback(ptr, address, Arg))
 265         return -1;
 266
 267       return 0;
 268     }
 269   };
 270 }
 271
 272 EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
 273                                    uint64_t address,
 274                                    void *arg) {
 275   EDMemoryObject memoryObject(byteReader, arg);
 276
 277   MCInst* inst = new MCInst;
 278   uint64_t byteSize;
 279
 280   if (!Disassembler->getInstruction(*inst,
 281                                     byteSize,
 282                                     memoryObject,
 283                                     address,
 284                                     ErrorStream)) {
 285     delete inst;
 286     return NULL;
 287   } else {
 288     const InstInfo *thisInstInfo;
 289
 290     thisInstInfo = &InstInfos[inst->getOpcode()];
 291
 292     EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
 293     return sdInst;
 294   }
 295 }
 296
 297 void EDDisassembler::initMaps(const TargetRegisterInfo &registerInfo) {
 298   unsigned numRegisters = registerInfo.getNumRegs();
 299   unsigned registerIndex;
 300
 301   for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
 302     const char* registerName = registerInfo.get(registerIndex).Name;
 303
 304     RegVec.push_back(registerName);
 305     RegRMap[registerName] = registerIndex;
 306   }
 307
 308   switch (Key.Arch) {
 309   default:
 310     break;
 311 #ifdef EDIS_X86
 312   case Triple::x86:
 313   case Triple::x86_64:
 314     stackPointers.insert(registerIDWithName("SP"));
 315     stackPointers.insert(registerIDWithName("ESP"));
 316     stackPointers.insert(registerIDWithName("RSP"));
 317
 318     programCounters.insert(registerIDWithName("IP"));
 319     programCounters.insert(registerIDWithName("EIP"));
 320     programCounters.insert(registerIDWithName("RIP"));
 321     break;
 322 #endif
 323 #ifdef EDIS_ARM
 324   case Triple::arm:
 325   case Triple::thumb:
 326     stackPointers.insert(registerIDWithName("SP"));
 327
 328     programCounters.insert(registerIDWithName("PC"));
 329     break;
 330 #endif
 331   }
 332 }
 333
 334 const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
 335   if (registerID >= RegVec.size())
 336     return NULL;
 337   else
 338     return RegVec[registerID].c_str();
 339 }
 340
 341 unsigned EDDisassembler::registerIDWithName(const char *name) const {
 342   regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
 343   if (iter == RegRMap.end())
 344     return 0;
 345   else
 346     return (*iter).second;
 347 }
 348
 349 bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
 350   return (stackPointers.find(registerID) != stackPointers.end());
 351 }
 352
 353 bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
 354   return (programCounters.find(registerID) != programCounters.end());
 355 }
 356
 357 int EDDisassembler::printInst(std::string &str, MCInst &inst) {
 358   PrinterMutex.acquire();
 359
 360   InstPrinter->printInst(&inst, *InstStream);
 361   InstStream->flush();
 362   str = *InstString;
 363   InstString->clear();
 364
 365   PrinterMutex.release();
 366
 367   return 0;
 368 }
 369
 370 int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
 371                               SmallVectorImpl<AsmToken> &tokens,
 372                               const std::string &str) {
 373   int ret = 0;
 374
 375   switch (Key.Arch) {
 376   default:
 377     return -1;
 378   case Triple::x86:
 379   case Triple::x86_64:
 380   case Triple::arm:
 381   case Triple::thumb:
 382     break;
 383   }
 384
 385   const char *cStr = str.c_str();
 386   MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
 387
 388   StringRef instName;
 389   SMLoc instLoc;
 390
 391   SourceMgr sourceMgr;
 392   sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
 393   MCContext context(*AsmInfo);
 394   OwningPtr<MCStreamer> streamer(createNullStreamer(context));
 395   AsmParser genericParser(sourceMgr, context, *streamer, *AsmInfo);
 396   OwningPtr<TargetAsmParser> TargetParser(Tgt->createAsmParser(genericParser));
 397
 398   AsmToken OpcodeToken = genericParser.Lex();
 399   genericParser.Lex();  // consume next token, because specificParser expects us to
 400
 401   if (OpcodeToken.is(AsmToken::Identifier)) {
 402     instName = OpcodeToken.getString();
 403     instLoc = OpcodeToken.getLoc();
 404     if (TargetParser->ParseInstruction(instName, instLoc, operands))
 405       ret = -1;
 406   } else {
 407     ret = -1;
 408   }
 409
 410   ParserMutex.acquire();
 411
 412   if (!ret) {
 413     GenericAsmLexer->setBuffer(buf);
 414
 415     while (SpecificAsmLexer->Lex(),
 416            SpecificAsmLexer->isNot(AsmToken::Eof) &&
 417            SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
 418       if (SpecificAsmLexer->is(AsmToken::Error)) {
 419         ret = -1;
 420         break;
 421       }
 422       tokens.push_back(SpecificAsmLexer->getTok());
 423     }
 424   }
 425
 426   ParserMutex.release();
 427
 428   return ret;
 429 }
 430
 431 int EDDisassembler::llvmSyntaxVariant() const {
 432   return LLVMSyntaxVariant;
 433 }