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 "llvm/ADT/Triple.h"
17 #include "llvm/MC/MCAsmInfo.h"
18 #include "llvm/MC/MCContext.h"
19 #include "llvm/MC/MCDisassembler.h"
20 #include "llvm/MC/MCInst.h"
21 #include "llvm/MC/MCRegisterInfo.h"
22 #include "llvm/MC/MCStreamer.h"
23 #include "llvm/MC/MCSubtargetInfo.h"
24 #include "llvm/Support/MemoryBuffer.h"
25 #include "llvm/Support/StringRefMemoryObject.h"
26 #include "llvm/Support/SourceMgr.h"
27 #include "llvm/Support/TargetRegistry.h"
28 #include "llvm/Support/raw_ostream.h"
32 typedef std::pair<std::vector<unsigned char>, std::vector<const char *>>
35 static bool PrintInsts(const MCDisassembler &DisAsm,
36 const ByteArrayTy &Bytes,
37 SourceMgr &SM, raw_ostream &Out,
38 MCStreamer &Streamer, bool InAtomicBlock,
39 const MCSubtargetInfo &STI) {
40 // Wrap the vector in a MemoryObject.
41 StringRef Data((const char*)Bytes.first.data(), Bytes.first.size());
42 StringRefMemoryObject memoryObject(Data);
44 // Disassemble it to strings.
48 for (Index = 0; Index < Bytes.first.size(); Index += Size) {
51 MCDisassembler::DecodeStatus S;
52 S = DisAsm.getInstruction(Inst, Size, memoryObject, Index,
53 /*REMOVE*/ nulls(), nulls());
55 case MCDisassembler::Fail:
56 SM.PrintMessage(SMLoc::getFromPointer(Bytes.second[Index]),
57 SourceMgr::DK_Warning,
58 "invalid instruction encoding");
59 // Don't try to resynchronise the stream in a block
64 Size = 1; // skip illegible bytes
68 case MCDisassembler::SoftFail:
69 SM.PrintMessage(SMLoc::getFromPointer(Bytes.second[Index]),
70 SourceMgr::DK_Warning,
71 "potentially undefined instruction encoding");
74 case MCDisassembler::Success:
75 Streamer.EmitInstruction(Inst, STI);
83 static bool SkipToToken(StringRef &Str) {
84 while (!Str.empty() && Str.find_first_not_of(" \t\r\n#,") != 0) {
85 // Strip horizontal whitespace and commas.
86 if (size_t Pos = Str.find_first_not_of(" \t\r,")) {
87 Str = Str.substr(Pos);
90 // If this is the end of a line or start of a comment, remove the rest of
92 if (Str[0] == '\n' || Str[0] == '#') {
93 // Strip to the end of line if we already processed any bytes on this
94 // line. This strips the comment and/or the \n.
98 Str = Str.substr(Str.find_first_of('\n'));
110 static bool ByteArrayFromString(ByteArrayTy &ByteArray,
113 while (SkipToToken(Str)) {
114 // Handled by higher level
115 if (Str[0] == '[' || Str[0] == ']')
118 // Get the current token.
119 size_t Next = Str.find_first_of(" \t\n\r,#[]");
120 StringRef Value = Str.substr(0, Next);
122 // Convert to a byte and add to the byte vector.
124 if (Value.getAsInteger(0, ByteVal) || ByteVal > 255) {
125 // If we have an error, print it and skip to the end of line.
126 SM.PrintMessage(SMLoc::getFromPointer(Value.data()), SourceMgr::DK_Error,
127 "invalid input token");
128 Str = Str.substr(Str.find('\n'));
129 ByteArray.first.clear();
130 ByteArray.second.clear();
134 ByteArray.first.push_back(ByteVal);
135 ByteArray.second.push_back(Value.data());
136 Str = Str.substr(Next);
142 int Disassembler::disassemble(const Target &T,
143 const std::string &Triple,
144 MCSubtargetInfo &STI,
145 MCStreamer &Streamer,
146 MemoryBuffer &Buffer,
150 std::unique_ptr<const MCRegisterInfo> MRI(T.createMCRegInfo(Triple));
152 errs() << "error: no register info for target " << Triple << "\n";
156 std::unique_ptr<const MCAsmInfo> MAI(T.createMCAsmInfo(*MRI, Triple));
158 errs() << "error: no assembly info for target " << Triple << "\n";
162 // Set up the MCContext for creating symbols and MCExpr's.
163 MCContext Ctx(MAI.get(), MRI.get(), nullptr);
165 std::unique_ptr<const MCDisassembler> DisAsm(
166 T.createMCDisassembler(STI, Ctx));
168 errs() << "error: no disassembler for target " << Triple << "\n";
172 // Set up initial section manually here
173 Streamer.InitSections(false);
175 bool ErrorOccurred = false;
177 // Convert the input to a vector for disassembly.
178 ByteArrayTy ByteArray;
179 StringRef Str = Buffer.getBuffer();
180 bool InAtomicBlock = false;
182 while (SkipToToken(Str)) {
183 ByteArray.first.clear();
184 ByteArray.second.clear();
188 SM.PrintMessage(SMLoc::getFromPointer(Str.data()), SourceMgr::DK_Error,
189 "nested atomic blocks make no sense");
190 ErrorOccurred = true;
192 InAtomicBlock = true;
193 Str = Str.drop_front();
195 } else if (Str[0] == ']') {
196 if (!InAtomicBlock) {
197 SM.PrintMessage(SMLoc::getFromPointer(Str.data()), SourceMgr::DK_Error,
198 "attempt to close atomic block without opening");
199 ErrorOccurred = true;
201 InAtomicBlock = false;
202 Str = Str.drop_front();
206 // It's a real token, get the bytes and emit them
207 ErrorOccurred |= ByteArrayFromString(ByteArray, Str, SM);
209 if (!ByteArray.first.empty())
210 ErrorOccurred |= PrintInsts(*DisAsm, ByteArray, SM, Out, Streamer,
215 SM.PrintMessage(SMLoc::getFromPointer(Str.data()), SourceMgr::DK_Error,
216 "unclosed atomic block");
217 ErrorOccurred = true;
220 return ErrorOccurred;