Disassembler.cpp revision 83e3f67fb68d497b600da83a62f000fcce7868a9 
1//===- Disassembler.cpp - Disassembler for hex strings --------------------===//
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 class implements the disassembler of strings of bytes written in
11// hexadecimal, from standard input or from a file.
12//
13//===----------------------------------------------------------------------===//
14
15#include "Disassembler.h"
16#include "../../lib/MC/MCDisassembler/EDDisassembler.h"
17#include "../../lib/MC/MCDisassembler/EDInst.h"
18#include "../../lib/MC/MCDisassembler/EDOperand.h"
19#include "../../lib/MC/MCDisassembler/EDToken.h"
20#include "llvm/MC/MCAsmInfo.h"
21#include "llvm/MC/MCDisassembler.h"
22#include "llvm/MC/MCInst.h"
23#include "llvm/MC/MCInstPrinter.h"
24#include "llvm/Target/TargetRegistry.h"
25#include "llvm/ADT/OwningPtr.h"
26#include "llvm/ADT/Triple.h"
27#include "llvm/ADT/Twine.h"
28#include "llvm/Support/MemoryBuffer.h"
29#include "llvm/Support/MemoryObject.h"
30#include "llvm/Support/raw_ostream.h"
31#include "llvm/Support/SourceMgr.h"
32using namespace llvm;
33
34typedef std::vector<std::pair<unsigned char, const char*> > ByteArrayTy;
35
36namespace {
37class VectorMemoryObject : public MemoryObject {
38private:
39  const ByteArrayTy &Bytes;
40public:
41  VectorMemoryObject(const ByteArrayTy &bytes) : Bytes(bytes) {}
42
43  uint64_t getBase() const { return 0; }
44  uint64_t getExtent() const { return Bytes.size(); }
45
46  int readByte(uint64_t Addr, uint8_t *Byte) const {
47    if (Addr >= getExtent())
48      return -1;
49    *Byte = Bytes[Addr].first;
50    return 0;
51  }
52};
53}
54
55static bool PrintInsts(const MCDisassembler &DisAsm,
56                       MCInstPrinter &Printer, const ByteArrayTy &Bytes,
57                       SourceMgr &SM, raw_ostream &Out) {
58  // Wrap the vector in a MemoryObject.
59  VectorMemoryObject memoryObject(Bytes);
60
61  // Disassemble it to strings.
62  uint64_t Size;
63  uint64_t Index;
64
65  for (Index = 0; Index < Bytes.size(); Index += Size) {
66    MCInst Inst;
67
68    MCDisassembler::DecodeStatus S;
69    S = DisAsm.getInstruction(Inst, Size, memoryObject, Index,
70                              /*REMOVE*/ nulls());
71    switch (S) {
72    case MCDisassembler::Fail:
73      SM.PrintMessage(SMLoc::getFromPointer(Bytes[Index].second),
74                      "invalid instruction encoding", "warning");
75      if (Size == 0)
76        Size = 1; // skip illegible bytes
77      break;
78
79    case MCDisassembler::SoftFail:
80      SM.PrintMessage(SMLoc::getFromPointer(Bytes[Index].second),
81                      "potentially undefined instruction encoding", "warning");
82      // Fall through
83
84    case MCDisassembler::Success:
85      Printer.printInst(&Inst, Out);
86      Out << "\n";
87      break;
88    }
89  }
90
91  return false;
92}
93
94static bool ByteArrayFromString(ByteArrayTy &ByteArray,
95                                StringRef &Str,
96                                SourceMgr &SM) {
97  while (!Str.empty()) {
98    // Strip horizontal whitespace.
99    if (size_t Pos = Str.find_first_not_of(" \t\r")) {
100      Str = Str.substr(Pos);
101      continue;
102    }
103
104    // If this is the end of a line or start of a comment, remove the rest of
105    // the line.
106    if (Str[0] == '\n' || Str[0] == '#') {
107      // Strip to the end of line if we already processed any bytes on this
108      // line.  This strips the comment and/or the \n.
109      if (Str[0] == '\n') {
110        Str = Str.substr(1);
111      } else {
112        Str = Str.substr(Str.find_first_of('\n'));
113        if (!Str.empty())
114          Str = Str.substr(1);
115      }
116      continue;
117    }
118
119    // Get the current token.
120    size_t Next = Str.find_first_of(" \t\n\r#");
121    StringRef Value = Str.substr(0, Next);
122
123    // Convert to a byte and add to the byte vector.
124    unsigned ByteVal;
125    if (Value.getAsInteger(0, ByteVal) || ByteVal > 255) {
126      // If we have an error, print it and skip to the end of line.
127      SM.PrintMessage(SMLoc::getFromPointer(Value.data()),
128                      "invalid input token", "error");
129      Str = Str.substr(Str.find('\n'));
130      ByteArray.clear();
131      continue;
132    }
133
134    ByteArray.push_back(std::make_pair((unsigned char)ByteVal, Value.data()));
135    Str = Str.substr(Next);
136  }
137
138  return false;
139}
140
141int Disassembler::disassemble(const Target &T,
142                              const std::string &Triple,
143                              MemoryBuffer &Buffer,
144                              raw_ostream &Out) {
145  // Set up disassembler.
146  OwningPtr<const MCAsmInfo> AsmInfo(T.createMCAsmInfo(Triple));
147
148  if (!AsmInfo) {
149    errs() << "error: no assembly info for target " << Triple << "\n";
150    return -1;
151  }
152
153  OwningPtr<const MCDisassembler> DisAsm(T.createMCDisassembler());
154  if (!DisAsm) {
155    errs() << "error: no disassembler for target " << Triple << "\n";
156    return -1;
157  }
158
159  int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
160  OwningPtr<MCInstPrinter> IP(T.createMCInstPrinter(AsmPrinterVariant,
161                                                    *AsmInfo));
162  if (!IP) {
163    errs() << "error: no instruction printer for target " << Triple << '\n';
164    return -1;
165  }
166
167  bool ErrorOccurred = false;
168
169  SourceMgr SM;
170  SM.AddNewSourceBuffer(&Buffer, SMLoc());
171
172  // Convert the input to a vector for disassembly.
173  ByteArrayTy ByteArray;
174  StringRef Str = Buffer.getBuffer();
175
176  ErrorOccurred |= ByteArrayFromString(ByteArray, Str, SM);
177
178  if (!ByteArray.empty())
179    ErrorOccurred |= PrintInsts(*DisAsm, *IP, ByteArray, SM, Out);
180
181  return ErrorOccurred;
182}
183
184static int byteArrayReader(uint8_t *B, uint64_t A, void *Arg) {
185  ByteArrayTy &ByteArray = *((ByteArrayTy*)Arg);
186
187  if (A >= ByteArray.size())
188    return -1;
189
190  *B = ByteArray[A].first;
191
192  return 0;
193}
194
195static int verboseEvaluator(uint64_t *V, unsigned R, void *Arg) {
196  EDDisassembler &disassembler = *(EDDisassembler *)((void **)Arg)[0];
197  raw_ostream &Out = *(raw_ostream *)((void **)Arg)[1];
198
199  if (const char *regName = disassembler.nameWithRegisterID(R))
200    Out << "[" << regName << "/" << R << "]";
201
202  if (disassembler.registerIsStackPointer(R))
203    Out << "(sp)";
204  if (disassembler.registerIsProgramCounter(R))
205    Out << "(pc)";
206
207  *V = 0;
208  return 0;
209}
210
211int Disassembler::disassembleEnhanced(const std::string &TS,
212                                      MemoryBuffer &Buffer,
213                                      raw_ostream &Out) {
214  ByteArrayTy ByteArray;
215  StringRef Str = Buffer.getBuffer();
216  SourceMgr SM;
217
218  SM.AddNewSourceBuffer(&Buffer, SMLoc());
219
220  if (ByteArrayFromString(ByteArray, Str, SM)) {
221    return -1;
222  }
223
224  Triple T(TS);
225  EDDisassembler::AssemblySyntax AS;
226
227  switch (T.getArch()) {
228  default:
229    errs() << "error: no default assembly syntax for " << TS.c_str() << "\n";
230    return -1;
231  case Triple::arm:
232  case Triple::thumb:
233    AS = EDDisassembler::kEDAssemblySyntaxARMUAL;
234    break;
235  case Triple::x86:
236  case Triple::x86_64:
237    AS = EDDisassembler::kEDAssemblySyntaxX86ATT;
238    break;
239  }
240
241  EDDisassembler::initialize();
242  OwningPtr<EDDisassembler>
243    disassembler(EDDisassembler::getDisassembler(TS.c_str(), AS));
244
245  if (disassembler == 0) {
246    errs() << "error: couldn't get disassembler for " << TS << '\n';
247    return -1;
248  }
249
250  while (ByteArray.size()) {
251    OwningPtr<EDInst>
252      inst(disassembler->createInst(byteArrayReader, 0, &ByteArray));
253
254    if (inst == 0) {
255      errs() << "error: Didn't get an instruction\n";
256      return -1;
257    }
258
259    ByteArray.erase (ByteArray.begin(), ByteArray.begin() + inst->byteSize());
260
261    unsigned numTokens = inst->numTokens();
262    if ((int)numTokens < 0) {
263      errs() << "error: couldn't count the instruction's tokens\n";
264      return -1;
265    }
266
267    for (unsigned tokenIndex = 0; tokenIndex != numTokens; ++tokenIndex) {
268      EDToken *token;
269
270      if (inst->getToken(token, tokenIndex)) {
271        errs() << "error: Couldn't get token\n";
272        return -1;
273      }
274
275      const char *buf;
276      if (token->getString(buf)) {
277        errs() << "error: Couldn't get string for token\n";
278        return -1;
279      }
280
281      Out << '[';
282      int operandIndex = token->operandID();
283
284      if (operandIndex >= 0)
285        Out << operandIndex << "-";
286
287      switch (token->type()) {
288      default: Out << "?"; break;
289      case EDToken::kTokenWhitespace: Out << "w"; break;
290      case EDToken::kTokenPunctuation: Out << "p"; break;
291      case EDToken::kTokenOpcode: Out << "o"; break;
292      case EDToken::kTokenLiteral: Out << "l"; break;
293      case EDToken::kTokenRegister: Out << "r"; break;
294      }
295
296      Out << ":" << buf;
297
298      if (token->type() == EDToken::kTokenLiteral) {
299        Out << "=";
300        if (token->literalSign())
301          Out << "-";
302        uint64_t absoluteValue;
303        if (token->literalAbsoluteValue(absoluteValue)) {
304          errs() << "error: Couldn't get the value of a literal token\n";
305          return -1;
306        }
307        Out << absoluteValue;
308      } else if (token->type() == EDToken::kTokenRegister) {
309        Out << "=";
310        unsigned regID;
311        if (token->registerID(regID)) {
312          errs() << "error: Couldn't get the ID of a register token\n";
313          return -1;
314        }
315        Out << "r" << regID;
316      }
317
318      Out << "]";
319    }
320
321    Out << " ";
322
323    if (inst->isBranch())
324      Out << "<br> ";
325    if (inst->isMove())
326      Out << "<mov> ";
327
328    unsigned numOperands = inst->numOperands();
329
330    if ((int)numOperands < 0) {
331      errs() << "error: Couldn't count operands\n";
332      return -1;
333    }
334
335    for (unsigned operandIndex = 0; operandIndex != numOperands;
336         ++operandIndex) {
337      Out << operandIndex << ":";
338
339      EDOperand *operand;
340      if (inst->getOperand(operand, operandIndex)) {
341        errs() << "error: couldn't get operand\n";
342        return -1;
343      }
344
345      uint64_t evaluatedResult;
346      void *Arg[] = { disassembler.get(), &Out };
347      if (operand->evaluate(evaluatedResult, verboseEvaluator, Arg)) {
348        errs() << "error: Couldn't evaluate an operand\n";
349        return -1;
350      }
351      Out << "=" << evaluatedResult << " ";
352    }
353
354    Out << '\n';
355  }
356
357  return 0;
358}
359
360