Disassembler.cpp revision 8c0a1608d582b5cd277d3a55fb92324e48386bc9 
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
17#include "llvm/ADT/OwningPtr.h"
18#include "llvm/ADT/Triple.h"
19#include "llvm/MC/MCAsmInfo.h"
20#include "llvm/MC/MCDisassembler.h"
21#include "llvm/MC/MCInst.h"
22#include "llvm/MC/MCInstPrinter.h"
23#include "llvm/Target/TargetRegistry.h"
24#include "llvm/Support/MemoryBuffer.h"
25#include "llvm/Support/MemoryObject.h"
26#include "llvm/Support/raw_ostream.h"
27#include "llvm/Support/SourceMgr.h"
28
29#include "llvm-c/EnhancedDisassembly.h"
30
31using namespace llvm;
32
33typedef std::vector<std::pair<unsigned char, const char*> > ByteArrayTy;
34
35namespace {
36class VectorMemoryObject : public MemoryObject {
37private:
38  const ByteArrayTy &Bytes;
39public:
40  VectorMemoryObject(const ByteArrayTy &bytes) : Bytes(bytes) {}
41
42  uint64_t getBase() const { return 0; }
43  uint64_t getExtent() const { return Bytes.size(); }
44
45  int readByte(uint64_t Addr, uint8_t *Byte) const {
46    if (Addr > getExtent())
47      return -1;
48    *Byte = Bytes[Addr].first;
49    return 0;
50  }
51};
52}
53
54static bool PrintInsts(const MCDisassembler &DisAsm,
55                       MCInstPrinter &Printer, const ByteArrayTy &Bytes,
56                       SourceMgr &SM) {
57  // Wrap the vector in a MemoryObject.
58  VectorMemoryObject memoryObject(Bytes);
59
60  // Disassemble it to strings.
61  uint64_t Size;
62  uint64_t Index;
63
64  for (Index = 0; Index < Bytes.size(); Index += Size) {
65    MCInst Inst;
66
67    if (DisAsm.getInstruction(Inst, Size, memoryObject, Index,
68                               /*REMOVE*/ nulls())) {
69      Printer.printInst(&Inst, outs());
70      outs() << "\n";
71    } else {
72      SM.PrintMessage(SMLoc::getFromPointer(Bytes[Index].second),
73                      "invalid instruction encoding", "warning");
74      if (Size == 0)
75        Size = 1; // skip illegible bytes
76    }
77  }
78
79  return false;
80}
81
82static bool ByteArrayFromString(ByteArrayTy &ByteArray,
83                                StringRef &Str,
84                                SourceMgr &SM) {
85  while (!Str.empty()) {
86    // Strip horizontal whitespace.
87    if (size_t Pos = Str.find_first_not_of(" \t\r")) {
88      Str = Str.substr(Pos);
89      continue;
90    }
91
92    // If this is the end of a line or start of a comment, remove the rest of
93    // the line.
94    if (Str[0] == '\n' || Str[0] == '#') {
95      // Strip to the end of line if we already processed any bytes on this
96      // line.  This strips the comment and/or the \n.
97      if (Str[0] == '\n') {
98        Str = Str.substr(1);
99      } else {
100        Str = Str.substr(Str.find_first_of('\n'));
101        if (!Str.empty())
102          Str = Str.substr(1);
103      }
104      continue;
105    }
106
107    // Get the current token.
108    size_t Next = Str.find_first_of(" \t\n\r#");
109    StringRef Value = Str.substr(0, Next);
110
111    // Convert to a byte and add to the byte vector.
112    unsigned ByteVal;
113    if (Value.getAsInteger(0, ByteVal) || ByteVal > 255) {
114      // If we have an error, print it and skip to the end of line.
115      SM.PrintMessage(SMLoc::getFromPointer(Value.data()),
116                      "invalid input token", "error");
117      Str = Str.substr(Str.find('\n'));
118      ByteArray.clear();
119      continue;
120    }
121
122    ByteArray.push_back(std::make_pair((unsigned char)ByteVal, Value.data()));
123    Str = Str.substr(Next);
124  }
125
126  return false;
127}
128
129int Disassembler::disassemble(const Target &T, const std::string &Triple,
130                              MemoryBuffer &Buffer) {
131  // Set up disassembler.
132  OwningPtr<const MCAsmInfo> AsmInfo(T.createAsmInfo(Triple));
133
134  if (!AsmInfo) {
135    errs() << "error: no assembly info for target " << Triple << "\n";
136    return -1;
137  }
138
139  OwningPtr<const MCDisassembler> DisAsm(T.createMCDisassembler());
140  if (!DisAsm) {
141    errs() << "error: no disassembler for target " << Triple << "\n";
142    return -1;
143  }
144
145  OwningPtr<MCInstPrinter> IP(T.createMCInstPrinter(0, *AsmInfo));
146  if (!IP) {
147    errs() << "error: no instruction printer for target " << Triple << '\n';
148    return -1;
149  }
150
151  bool ErrorOccurred = false;
152
153  SourceMgr SM;
154  SM.AddNewSourceBuffer(&Buffer, SMLoc());
155
156  // Convert the input to a vector for disassembly.
157  ByteArrayTy ByteArray;
158  StringRef Str = Buffer.getBuffer();
159
160  ErrorOccurred |= ByteArrayFromString(ByteArray, Str, SM);
161
162  if (!ByteArray.empty())
163    ErrorOccurred |= PrintInsts(*DisAsm, *IP, ByteArray, SM);
164
165  return ErrorOccurred;
166}
167
168static int byteArrayReader(uint8_t *B, uint64_t A, void *Arg) {
169  ByteArrayTy &ByteArray = *((ByteArrayTy*)Arg);
170
171  if (A >= ByteArray.size())
172    return -1;
173
174  *B = ByteArray[A].first;
175
176  return 0;
177}
178
179static int verboseEvaluator(uint64_t *V, unsigned R, void *Arg) {
180  EDDisassemblerRef &disassembler = *((EDDisassemblerRef*)Arg);
181
182  const char *regName;
183
184  if (!EDGetRegisterName(&regName,
185                        disassembler,
186                        R))
187    outs() << "[" << regName << "/" << R << "]";
188  if (EDRegisterIsStackPointer(disassembler, R))
189    outs() << "(sp)";
190  if (EDRegisterIsProgramCounter(disassembler, R))
191    outs() << "(pc)";
192
193  *V = 0;
194
195  return 0;
196}
197
198int Disassembler::disassembleEnhanced(const std::string &TS,
199                                      MemoryBuffer &Buffer) {
200  ByteArrayTy ByteArray;
201  StringRef Str = Buffer.getBuffer();
202  SourceMgr SM;
203
204  SM.AddNewSourceBuffer(&Buffer, SMLoc());
205
206  if (ByteArrayFromString(ByteArray, Str, SM)) {
207    return -1;
208  }
209
210  EDDisassemblerRef disassembler;
211
212  Triple T(TS);
213  EDAssemblySyntax_t AS;
214
215  switch (T.getArch()) {
216  default:
217    errs() << "error: no default assembly syntax for " << TS.c_str() << "\n";
218    return -1;
219  case Triple::arm:
220  case Triple::thumb:
221    AS = kEDAssemblySyntaxARMUAL;
222    break;
223  case Triple::x86:
224  case Triple::x86_64:
225    AS = kEDAssemblySyntaxX86ATT;
226    break;
227  }
228
229  if (EDGetDisassembler(&disassembler,
230                        TS.c_str(),
231                        AS)) {
232    errs() << "error: couldn't get disassembler for " << TS.c_str() << "\n";
233    return -1;
234  }
235
236  EDInstRef inst;
237
238  if (EDCreateInsts(&inst, 1, disassembler, byteArrayReader, 0,&ByteArray)
239      != 1) {
240    errs() << "error: Didn't get an instruction\n";
241    return -1;
242  }
243
244  int numTokens = EDNumTokens(inst);
245
246  if (numTokens < 0) {
247    errs() << "error: Couldn't count the instruction's tokens\n";
248    return -1;
249  }
250
251  int tokenIndex;
252
253  for (tokenIndex = 0; tokenIndex < numTokens; ++tokenIndex) {
254    EDTokenRef token;
255
256    if (EDGetToken(&token, inst, tokenIndex)) {
257      errs() << "error: Couldn't get token\n";
258      return -1;
259    }
260
261    const char *buf;
262
263    if (EDGetTokenString(&buf, token)) {
264      errs() << "error: Couldn't get string for token\n";
265      return -1;
266    }
267
268    outs() << "[";
269
270    int operandIndex = EDOperandIndexForToken(token);
271
272    if (operandIndex >= 0)
273      outs() << operandIndex << "-";
274
275    if (EDTokenIsWhitespace(token)) {
276      outs() << "w";
277    } else if (EDTokenIsPunctuation(token)) {
278      outs() << "p";
279    } else if (EDTokenIsOpcode(token)) {
280      outs() << "o";
281    } else if (EDTokenIsLiteral(token)) {
282      outs() << "l";
283    } else if (EDTokenIsRegister(token)) {
284      outs() << "r";
285    } else {
286      outs() << "?";
287    }
288
289    outs() << ":" << buf;
290
291    if (EDTokenIsLiteral(token)) {
292      outs() << "=";
293      if (EDTokenIsNegativeLiteral(token))
294        outs() << "-";
295      uint64_t absoluteValue;
296      if (EDLiteralTokenAbsoluteValue(&absoluteValue, token)) {
297        errs() << "error: Couldn't get the value of a literal token\n";
298        return -1;
299      }
300      outs() << absoluteValue;
301    } else if (EDTokenIsRegister(token)) {
302      outs() << "=";
303      unsigned regID;
304      if (EDRegisterTokenValue(&regID, token)) {
305        errs() << "error: Couldn't get the ID of a register token\n";
306        return -1;
307      }
308      outs() << "r" << regID;
309    }
310
311    outs() << "]";
312  }
313
314  outs() << " ";
315
316  int numOperands = EDNumOperands(inst);
317
318  if (numOperands < 0) {
319    errs() << "error: Couldn't count operands\n";
320    return -1;
321  }
322
323  int operandIndex;
324
325  for (operandIndex = 0; operandIndex < numOperands; ++operandIndex) {
326    outs() << operandIndex << ":";
327
328    EDOperandRef operand;
329
330    if (EDGetOperand(&operand,
331                     inst,
332                     operandIndex)) {
333      errs() << "error: Couldn't get operand\n";
334      return -1;
335    }
336
337    uint64_t evaluatedResult;
338
339    EDEvaluateOperand(&evaluatedResult,
340                      operand,
341                      verboseEvaluator,
342                      &disassembler);
343
344    outs() << " ";
345  }
346
347  outs() << "\n";
348
349  return 0;
350}
351
352