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/MC/MCTargetAsmLexer.h"
  32 #include "llvm/MC/MCTargetAsmParser.h"
  33 #include "llvm/Support/MemoryBuffer.h"
  34 #include "llvm/Support/MemoryObject.h"
  35 #include "llvm/Support/SourceMgr.h"
  36 #include "llvm/Target/TargetRegistry.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   InitializeAllTargetMCs();
 109   InitializeAllAsmParsers();
 110   InitializeAllDisassemblers();
 111 }
 112
 113 #undef BRINGUP_TARGET
 114
 115 EDDisassembler *EDDisassembler::getDisassembler(Triple::ArchType arch,
 116                                                 AssemblySyntax syntax) {
 117   CPUKey key;
 118   key.Arch = arch;
 119   key.Syntax = syntax;
 120
 121   EDDisassembler::DisassemblerMap_t::iterator i = sDisassemblers.find(key);
 122
 123   if (i != sDisassemblers.end()) {
 124     return i->second;
 125   } else {
 126     EDDisassembler* sdd = new EDDisassembler(key);
 127     if (!sdd->valid()) {
 128       delete sdd;
 129       return NULL;
 130     }
 131
 132     sDisassemblers[key] = sdd;
 133
 134     return sdd;
 135   }
 136
 137   return NULL;
 138 }
 139
 140 EDDisassembler *EDDisassembler::getDisassembler(StringRef str,
 141                                                 AssemblySyntax syntax) {
 142   return getDisassembler(Triple(str).getArch(), syntax);
 143 }
 144
 145 EDDisassembler::EDDisassembler(CPUKey &key) :
 146   Valid(false),
 147   HasSemantics(false),
 148   ErrorStream(nulls()),
 149   Key(key) {
 150   const char *triple = tripleFromArch(key.Arch);
 151
 152   if (!triple)
 153     return;
 154
 155   LLVMSyntaxVariant = getLLVMSyntaxVariant(key.Arch, key.Syntax);
 156
 157   if (LLVMSyntaxVariant < 0)
 158     return;
 159
 160   std::string tripleString(triple);
 161   std::string errorString;
 162
 163   Tgt = TargetRegistry::lookupTarget(tripleString,
 164                                      errorString);
 165
 166   if (!Tgt)
 167     return;
 168
 169   MRI.reset(Tgt->createMCRegInfo(tripleString));
 170
 171   if (!MRI)
 172     return;
 173
 174   initMaps(*MRI);
 175
 176   AsmInfo.reset(Tgt->createMCAsmInfo(tripleString));
 177
 178   if (!AsmInfo)
 179     return;
 180
 181   Disassembler.reset(Tgt->createMCDisassembler());
 182
 183   if (!Disassembler)
 184     return;
 185
 186   InstInfos = Disassembler->getEDInfo();
 187
 188   InstString.reset(new std::string);
 189   InstStream.reset(new raw_string_ostream(*InstString));
 190   InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo));
 191
 192   if (!InstPrinter)
 193     return;
 194
 195   GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
 196   SpecificAsmLexer.reset(Tgt->createMCAsmLexer(*MRI, *AsmInfo));
 197   SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
 198
 199   initMaps(*MRI);
 200
 201   Valid = true;
 202 }
 203
 204 EDDisassembler::~EDDisassembler() {
 205   if (!valid())
 206     return;
 207 }
 208
 209 namespace {
 210   /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
 211   ///   as provided by the sd interface.  See MemoryObject.
 212   class EDMemoryObject : public llvm::MemoryObject {
 213   private:
 214     EDByteReaderCallback Callback;
 215     void *Arg;
 216   public:
 217     EDMemoryObject(EDByteReaderCallback callback,
 218                    void *arg) : Callback(callback), Arg(arg) { }
 219     ~EDMemoryObject() { }
 220     uint64_t getBase() const { return 0x0; }
 221     uint64_t getExtent() const { return (uint64_t)-1; }
 222     int readByte(uint64_t address, uint8_t *ptr) const {
 223       if (!Callback)
 224         return -1;
 225
 226       if (Callback(ptr, address, Arg))
 227         return -1;
 228
 229       return 0;
 230     }
 231   };
 232 }
 233
 234 EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
 235                                    uint64_t address,
 236                                    void *arg) {
 237   EDMemoryObject memoryObject(byteReader, arg);
 238
 239   MCInst* inst = new MCInst;
 240   uint64_t byteSize;
 241
 242   if (!Disassembler->getInstruction(*inst,
 243                                     byteSize,
 244                                     memoryObject,
 245                                     address,
 246                                     ErrorStream)) {
 247     delete inst;
 248     return NULL;
 249   } else {
 250     const llvm::EDInstInfo *thisInstInfo = NULL;
 251
 252     if (InstInfos) {
 253       thisInstInfo = &InstInfos[inst->getOpcode()];
 254     }
 255
 256     EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
 257     return sdInst;
 258   }
 259 }
 260
 261 void EDDisassembler::initMaps(const MCRegisterInfo &registerInfo) {
 262   unsigned numRegisters = registerInfo.getNumRegs();
 263   unsigned registerIndex;
 264
 265   for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
 266     const char* registerName = registerInfo.get(registerIndex).Name;
 267
 268     RegVec.push_back(registerName);
 269     RegRMap[registerName] = registerIndex;
 270   }
 271
 272   switch (Key.Arch) {
 273   default:
 274     break;
 275   case Triple::x86:
 276   case Triple::x86_64:
 277     stackPointers.insert(registerIDWithName("SP"));
 278     stackPointers.insert(registerIDWithName("ESP"));
 279     stackPointers.insert(registerIDWithName("RSP"));
 280
 281     programCounters.insert(registerIDWithName("IP"));
 282     programCounters.insert(registerIDWithName("EIP"));
 283     programCounters.insert(registerIDWithName("RIP"));
 284     break;
 285   case Triple::arm:
 286   case Triple::thumb:
 287     stackPointers.insert(registerIDWithName("SP"));
 288
 289     programCounters.insert(registerIDWithName("PC"));
 290     break;
 291   }
 292 }
 293
 294 const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
 295   if (registerID >= RegVec.size())
 296     return NULL;
 297   else
 298     return RegVec[registerID].c_str();
 299 }
 300
 301 unsigned EDDisassembler::registerIDWithName(const char *name) const {
 302   regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
 303   if (iter == RegRMap.end())
 304     return 0;
 305   else
 306     return (*iter).second;
 307 }
 308
 309 bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
 310   return (stackPointers.find(registerID) != stackPointers.end());
 311 }
 312
 313 bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
 314   return (programCounters.find(registerID) != programCounters.end());
 315 }
 316
 317 int EDDisassembler::printInst(std::string &str, MCInst &inst) {
 318   PrinterMutex.acquire();
 319
 320   InstPrinter->printInst(&inst, *InstStream);
 321   InstStream->flush();
 322   str = *InstString;
 323   InstString->clear();
 324
 325   PrinterMutex.release();
 326
 327   return 0;
 328 }
 329
 330 static void diag_handler(const SMDiagnostic &diag,
 331                          void *context)
 332 {
 333   if (context) {
 334     EDDisassembler *disassembler = static_cast<EDDisassembler*>(context);
 335     diag.Print("", disassembler->ErrorStream);
 336   }
 337 }
 338
 339 int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
 340                               SmallVectorImpl<AsmToken> &tokens,
 341                               const std::string &str) {
 342   int ret = 0;
 343
 344   switch (Key.Arch) {
 345   default:
 346     return -1;
 347   case Triple::x86:
 348   case Triple::x86_64:
 349   case Triple::arm:
 350   case Triple::thumb:
 351     break;
 352   }
 353
 354   const char *cStr = str.c_str();
 355   MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
 356
 357   StringRef instName;
 358   SMLoc instLoc;
 359
 360   SourceMgr sourceMgr;
 361   sourceMgr.setDiagHandler(diag_handler, static_cast<void*>(this));
 362   sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
 363   MCContext context(*AsmInfo, *MRI, NULL);
 364   OwningPtr<MCStreamer> streamer(createNullStreamer(context));
 365   OwningPtr<MCAsmParser> genericParser(createMCAsmParser(sourceMgr,
 366                                                          context, *streamer,
 367                                                          *AsmInfo));
 368
 369   StringRef triple = tripleFromArch(Key.Arch);
 370   OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(triple, "", ""));
 371   OwningPtr<MCTargetAsmParser>
 372     TargetParser(Tgt->createMCAsmParser(*STI, *genericParser));
 373
 374   AsmToken OpcodeToken = genericParser->Lex();
 375   AsmToken NextToken = genericParser->Lex();  // consume next token, because specificParser expects us to
 376
 377   if (OpcodeToken.is(AsmToken::Identifier)) {
 378     instName = OpcodeToken.getString();
 379     instLoc = OpcodeToken.getLoc();
 380
 381     if (NextToken.isNot(AsmToken::Eof) &&
 382         TargetParser->ParseInstruction(instName, instLoc, operands))
 383       ret = -1;
 384   } else {
 385     ret = -1;
 386   }
 387
 388   ParserMutex.acquire();
 389
 390   if (!ret) {
 391     GenericAsmLexer->setBuffer(buf);
 392
 393     while (SpecificAsmLexer->Lex(),
 394            SpecificAsmLexer->isNot(AsmToken::Eof) &&
 395            SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
 396       if (SpecificAsmLexer->is(AsmToken::Error)) {
 397         ret = -1;
 398         break;
 399       }
 400       tokens.push_back(SpecificAsmLexer->getTok());
 401     }
 402   }
 403
 404   ParserMutex.release();
 405
 406   return ret;
 407 }
 408
 409 int EDDisassembler::llvmSyntaxVariant() const {
 410   return LLVMSyntaxVariant;
 411 }