1 //===- Disassembler.cpp - Disassembler for hex strings --------------------===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // This class implements the disassembler of strings of bytes written in
11 // hexadecimal, from standard input or from a file.
13 //===----------------------------------------------------------------------===//
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/Support/MemoryBuffer.h"
28 #include "llvm/Support/MemoryObject.h"
29 #include "llvm/Support/raw_ostream.h"
30 #include "llvm/Support/SourceMgr.h"
33 typedef std::vector<std::pair<unsigned char, const char*> > ByteArrayTy;
36 class VectorMemoryObject : public MemoryObject {
38 const ByteArrayTy &Bytes;
40 VectorMemoryObject(const ByteArrayTy &bytes) : Bytes(bytes) {}
42 uint64_t getBase() const { return 0; }
43 uint64_t getExtent() const { return Bytes.size(); }
45 int readByte(uint64_t Addr, uint8_t *Byte) const {
46 if (Addr > getExtent())
48 *Byte = Bytes[Addr].first;
54 static bool PrintInsts(const MCDisassembler &DisAsm,
55 MCInstPrinter &Printer, const ByteArrayTy &Bytes,
57 // Wrap the vector in a MemoryObject.
58 VectorMemoryObject memoryObject(Bytes);
60 // Disassemble it to strings.
64 for (Index = 0; Index < Bytes.size(); Index += Size) {
67 if (DisAsm.getInstruction(Inst, Size, memoryObject, Index,
68 /*REMOVE*/ nulls())) {
69 Printer.printInst(&Inst, outs());
72 SM.PrintMessage(SMLoc::getFromPointer(Bytes[Index].second),
73 "invalid instruction encoding", "warning");
75 Size = 1; // skip illegible bytes
82 static bool ByteArrayFromString(ByteArrayTy &ByteArray,
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);
92 // If this is the end of a line or start of a comment, remove the rest of
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.
100 Str = Str.substr(Str.find_first_of('\n'));
107 // Get the current token.
108 size_t Next = Str.find_first_of(" \t\n\r#");
109 StringRef Value = Str.substr(0, Next);
111 // Convert to a byte and add to the byte vector.
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'));
122 ByteArray.push_back(std::make_pair((unsigned char)ByteVal, Value.data()));
123 Str = Str.substr(Next);
129 int Disassembler::disassemble(const Target &T, const std::string &Triple,
130 MemoryBuffer &Buffer) {
131 // Set up disassembler.
132 OwningPtr<const MCAsmInfo> AsmInfo(T.createAsmInfo(Triple));
135 errs() << "error: no assembly info for target " << Triple << "\n";
139 OwningPtr<const MCDisassembler> DisAsm(T.createMCDisassembler());
141 errs() << "error: no disassembler for target " << Triple << "\n";
145 int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
146 OwningPtr<MCInstPrinter> IP(T.createMCInstPrinter(AsmPrinterVariant,
149 errs() << "error: no instruction printer for target " << Triple << '\n';
153 bool ErrorOccurred = false;
156 SM.AddNewSourceBuffer(&Buffer, SMLoc());
158 // Convert the input to a vector for disassembly.
159 ByteArrayTy ByteArray;
160 StringRef Str = Buffer.getBuffer();
162 ErrorOccurred |= ByteArrayFromString(ByteArray, Str, SM);
164 if (!ByteArray.empty())
165 ErrorOccurred |= PrintInsts(*DisAsm, *IP, ByteArray, SM);
167 return ErrorOccurred;
170 static int byteArrayReader(uint8_t *B, uint64_t A, void *Arg) {
171 ByteArrayTy &ByteArray = *((ByteArrayTy*)Arg);
173 if (A >= ByteArray.size())
176 *B = ByteArray[A].first;
181 static int verboseEvaluator(uint64_t *V, unsigned R, void *Arg) {
182 EDDisassembler &disassembler = *((EDDisassembler *)Arg);
184 if (const char *regName = disassembler.nameWithRegisterID(R))
185 outs() << "[" << regName << "/" << R << "]";
187 if (disassembler.registerIsStackPointer(R))
189 if (disassembler.registerIsProgramCounter(R))
196 int Disassembler::disassembleEnhanced(const std::string &TS,
197 MemoryBuffer &Buffer) {
198 ByteArrayTy ByteArray;
199 StringRef Str = Buffer.getBuffer();
202 SM.AddNewSourceBuffer(&Buffer, SMLoc());
204 if (ByteArrayFromString(ByteArray, Str, SM)) {
209 EDDisassembler::AssemblySyntax AS;
211 switch (T.getArch()) {
213 errs() << "error: no default assembly syntax for " << TS.c_str() << "\n";
217 AS = EDDisassembler::kEDAssemblySyntaxARMUAL;
221 AS = EDDisassembler::kEDAssemblySyntaxX86ATT;
225 EDDisassembler::initialize();
226 EDDisassembler *disassembler =
227 EDDisassembler::getDisassembler(TS.c_str(), AS);
229 if (disassembler == 0) {
230 errs() << "error: couldn't get disassembler for " << TS << '\n';
235 disassembler->createInst(byteArrayReader, 0, &ByteArray);
238 errs() << "error: Didn't get an instruction\n";
242 unsigned numTokens = inst->numTokens();
243 if ((int)numTokens < 0) {
244 errs() << "error: couldn't count the instruction's tokens\n";
248 for (unsigned tokenIndex = 0; tokenIndex != numTokens; ++tokenIndex) {
251 if (inst->getToken(token, tokenIndex)) {
252 errs() << "error: Couldn't get token\n";
257 if (token->getString(buf)) {
258 errs() << "error: Couldn't get string for token\n";
263 int operandIndex = token->operandID();
265 if (operandIndex >= 0)
266 outs() << operandIndex << "-";
268 switch (token->type()) {
269 default: outs() << "?"; break;
270 case EDToken::kTokenWhitespace: outs() << "w"; break;
271 case EDToken::kTokenPunctuation: outs() << "p"; break;
272 case EDToken::kTokenOpcode: outs() << "o"; break;
273 case EDToken::kTokenLiteral: outs() << "l"; break;
274 case EDToken::kTokenRegister: outs() << "r"; break;
277 outs() << ":" << buf;
279 if (token->type() == EDToken::kTokenLiteral) {
281 if (token->literalSign())
283 uint64_t absoluteValue;
284 if (token->literalAbsoluteValue(absoluteValue)) {
285 errs() << "error: Couldn't get the value of a literal token\n";
288 outs() << absoluteValue;
289 } else if (token->type() == EDToken::kTokenRegister) {
292 if (token->registerID(regID)) {
293 errs() << "error: Couldn't get the ID of a register token\n";
296 outs() << "r" << regID;
304 if (inst->isBranch())
309 unsigned numOperands = inst->numOperands();
311 if ((int)numOperands < 0) {
312 errs() << "error: Couldn't count operands\n";
316 for (unsigned operandIndex = 0; operandIndex != numOperands; ++operandIndex) {
317 outs() << operandIndex << ":";
320 if (inst->getOperand(operand, operandIndex)) {
321 errs() << "error: couldn't get operand\n";
325 uint64_t evaluatedResult;
326 evaluatedResult = operand->evaluate(evaluatedResult, verboseEvaluator,
328 outs() << "=" << evaluatedResult << " ";