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 "llvm/ADT/OwningPtr.h"
  17 #include "llvm/ADT/SmallVector.h"
  18 #include "llvm/MC/MCAsmInfo.h"
  19 #include "llvm/MC/MCContext.h"
  20 #include "llvm/MC/MCDisassembler.h"
  21 #include "llvm/MC/MCExpr.h"
  22 #include "llvm/MC/MCInst.h"
  23 #include "llvm/MC/MCInstPrinter.h"
  24 #include "llvm/MC/MCStreamer.h"
  25 #include "llvm/MC/MCParser/AsmLexer.h"
  26 #include "llvm/MC/MCParser/AsmParser.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
  39 #include "EDDisassembler.h"
  40 #include "EDInst.h"
  41
  42 #include "../../lib/Target/X86/X86GenEDInfo.inc"
  43
  44 using namespace llvm;
  45
  46 bool EDDisassembler::sInitialized = false;
  47 EDDisassembler::DisassemblerMap_t EDDisassembler::sDisassemblers;
  48
  49 struct InfoMap {
  50   Triple::ArchType Arch;
  51   const char *String;
  52   const InstInfo *Info;
  53 };
  54
  55 static struct InfoMap infomap[] = {
  56   { Triple::x86,          "i386-unknown-unknown",   instInfoX86 },
  57   { Triple::x86_64,       "x86_64-unknown-unknown", instInfoX86 },
  58   { Triple::InvalidArch,  NULL,                     NULL        }
  59 };
  60
  61 /// infoFromArch - Returns the InfoMap corresponding to a given architecture,
  62 ///   or NULL if there is an error
  63 ///
  64 /// @arg arch - The Triple::ArchType for the desired architecture
  65 static const InfoMap *infoFromArch(Triple::ArchType arch) {
  66   unsigned int infoIndex;
  67
  68   for (infoIndex = 0; infomap[infoIndex].String != NULL; ++infoIndex) {
  69     if(arch == infomap[infoIndex].Arch)
  70       return &infomap[infoIndex];
  71   }
  72
  73   return NULL;
  74 }
  75
  76 /// getLLVMSyntaxVariant - gets the constant to use to get an assembly printer
  77 ///   for the desired assembly syntax, suitable for passing to
  78 ///   Target::createMCInstPrinter()
  79 ///
  80 /// @arg arch   - The target architecture
  81 /// @arg syntax - The assembly syntax in sd form
  82 static int getLLVMSyntaxVariant(Triple::ArchType arch,
  83                                 EDAssemblySyntax_t syntax) {
  84   switch (syntax) {
  85   default:
  86     return -1;
  87   // Mappings below from X86AsmPrinter.cpp
  88   case kEDAssemblySyntaxX86ATT:
  89     if (arch == Triple::x86 || arch == Triple::x86_64)
  90       return 0;
  91     else
  92       return -1;
  93   case kEDAssemblySyntaxX86Intel:
  94     if (arch == Triple::x86 || arch == Triple::x86_64)
  95       return 1;
  96     else
  97       return -1;
  98   }
  99 }
 100
 101 #define BRINGUP_TARGET(tgt)           \
 102   LLVMInitialize##tgt##TargetInfo();  \
 103   LLVMInitialize##tgt##Target();      \
 104   LLVMInitialize##tgt##AsmPrinter();  \
 105   LLVMInitialize##tgt##AsmParser();   \
 106   LLVMInitialize##tgt##Disassembler();
 107
 108 void EDDisassembler::initialize() {
 109   if (sInitialized)
 110     return;
 111
 112   sInitialized = true;
 113
 114   BRINGUP_TARGET(X86)
 115 }
 116
 117 #undef BRINGUP_TARGET
 118
 119 EDDisassembler *EDDisassembler::getDisassembler(Triple::ArchType arch,
 120                                                 EDAssemblySyntax_t 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   }
 130   else {
 131     EDDisassembler* sdd = new EDDisassembler(key);
 132     if(!sdd->valid()) {
 133       delete sdd;
 134       return NULL;
 135     }
 136
 137     sDisassemblers[key] = sdd;
 138
 139     return sdd;
 140   }
 141
 142   return NULL;
 143 }
 144
 145 EDDisassembler *EDDisassembler::getDisassembler(StringRef str,
 146                                                 EDAssemblySyntax_t syntax) {
 147   Triple triple(str);
 148
 149   return getDisassembler(triple.getArch(), syntax);
 150 }
 151
 152 EDDisassembler::EDDisassembler(CPUKey &key) :
 153   Valid(false), ErrorString(), ErrorStream(ErrorString), Key(key) {
 154   const InfoMap *infoMap = infoFromArch(key.Arch);
 155
 156   if (!infoMap)
 157     return;
 158
 159   const char *triple = infoMap->String;
 160
 161   int syntaxVariant = getLLVMSyntaxVariant(key.Arch, key.Syntax);
 162
 163   if (syntaxVariant < 0)
 164     return;
 165
 166   std::string tripleString(triple);
 167   std::string errorString;
 168
 169   Tgt = TargetRegistry::lookupTarget(tripleString,
 170                                      errorString);
 171
 172   if (!Tgt)
 173     return;
 174
 175   std::string featureString;
 176
 177   OwningPtr<const TargetMachine>
 178     targetMachine(Tgt->createTargetMachine(tripleString,
 179                                            featureString));
 180
 181   const TargetRegisterInfo *registerInfo = targetMachine->getRegisterInfo();
 182
 183   if (!registerInfo)
 184     return;
 185
 186   AsmInfo.reset(Tgt->createAsmInfo(tripleString));
 187
 188   if (!AsmInfo)
 189     return;
 190
 191   Disassembler.reset(Tgt->createMCDisassembler());
 192
 193   if (!Disassembler)
 194     return;
 195
 196   InstString.reset(new std::string);
 197   InstStream.reset(new raw_string_ostream(*InstString));
 198
 199   InstPrinter.reset(Tgt->createMCInstPrinter(syntaxVariant,
 200                                                 *AsmInfo,
 201                                                 *InstStream));
 202
 203   if (!InstPrinter)
 204     return;
 205
 206   GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
 207   SpecificAsmLexer.reset(Tgt->createAsmLexer(*AsmInfo));
 208   SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
 209
 210   InstInfos = infoMap->Info;
 211
 212   initMaps(*targetMachine->getRegisterInfo());
 213
 214   Valid = true;
 215 }
 216
 217 EDDisassembler::~EDDisassembler() {
 218   if(!valid())
 219     return;
 220 }
 221
 222 namespace {
 223   /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
 224   ///   as provided by the sd interface.  See MemoryObject.
 225   class EDMemoryObject : public llvm::MemoryObject {
 226   private:
 227     EDByteReaderCallback Callback;
 228     void *Arg;
 229   public:
 230     EDMemoryObject(EDByteReaderCallback callback,
 231                    void *arg) : Callback(callback), Arg(arg) { }
 232     ~EDMemoryObject() { }
 233     uint64_t getBase() const { return 0x0; }
 234     uint64_t getExtent() const { return (uint64_t)-1; }
 235     int readByte(uint64_t address, uint8_t *ptr) const {
 236       if(!Callback)
 237         return -1;
 238
 239       if(Callback(ptr, address, Arg))
 240         return -1;
 241
 242       return 0;
 243     }
 244   };
 245 }
 246
 247 EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
 248                                    uint64_t address,
 249                                    void *arg) {
 250   EDMemoryObject memoryObject(byteReader, arg);
 251
 252   MCInst* inst = new MCInst;
 253   uint64_t byteSize;
 254
 255   if (!Disassembler->getInstruction(*inst,
 256                                     byteSize,
 257                                     memoryObject,
 258                                     address,
 259                                     ErrorStream)) {
 260     delete inst;
 261     return NULL;
 262   }
 263   else {
 264     const InstInfo *thisInstInfo = &InstInfos[inst->getOpcode()];
 265
 266     EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
 267     return sdInst;
 268   }
 269 }
 270
 271 void EDDisassembler::initMaps(const TargetRegisterInfo &registerInfo) {
 272   unsigned numRegisters = registerInfo.getNumRegs();
 273   unsigned registerIndex;
 274
 275   for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
 276     const char* registerName = registerInfo.get(registerIndex).Name;
 277
 278     RegVec.push_back(registerName);
 279     RegRMap[registerName] = registerIndex;
 280   }
 281
 282   if (Key.Arch == Triple::x86 ||
 283       Key.Arch == 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   }
 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,
 318                               MCInst& inst) {
 319   PrinterMutex.acquire();
 320
 321   InstPrinter->printInst(&inst);
 322   InstStream->flush();
 323   str = *InstString;
 324   InstString->clear();
 325
 326   PrinterMutex.release();
 327
 328   return 0;
 329 }
 330
 331 int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
 332                               SmallVectorImpl<AsmToken> &tokens,
 333                               const std::string &str) {
 334   int ret = 0;
 335
 336   const char *cStr = str.c_str();
 337   MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
 338
 339   StringRef instName;
 340   SMLoc instLoc;
 341
 342   SourceMgr sourceMgr;
 343   sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
 344   MCContext context;
 345   OwningPtr<MCStreamer> streamer
 346     (createNullStreamer(context));
 347   AsmParser genericParser(sourceMgr, context, *streamer, *AsmInfo);
 348   OwningPtr<TargetAsmParser> specificParser
 349     (Tgt->createAsmParser(genericParser));
 350
 351   AsmToken OpcodeToken = genericParser.Lex();
 352
 353   if(OpcodeToken.is(AsmToken::Identifier)) {
 354     instName = OpcodeToken.getString();
 355     instLoc = OpcodeToken.getLoc();
 356     if (specificParser->ParseInstruction(instName, instLoc, operands))
 357       ret = -1;
 358   }
 359   else {
 360     ret = -1;
 361   }
 362
 363   ParserMutex.acquire();
 364
 365   if (!ret) {
 366     GenericAsmLexer->setBuffer(buf);
 367
 368     while (SpecificAsmLexer->Lex(),
 369            SpecificAsmLexer->isNot(AsmToken::Eof) &&
 370            SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
 371       if (SpecificAsmLexer->is(AsmToken::Error)) {
 372         ret = -1;
 373         break;
 374       }
 375       tokens.push_back(SpecificAsmLexer->getTok());
 376     }
 377   }
 378
 379   ParserMutex.release();
 380
 381   return ret;
 382 }
 383
 384 int EDDisassembler::llvmSyntaxVariant() const {
 385   return LLVMSyntaxVariant;
 386 }