1 //===- Disassembler.cpp - Disassembler for hex strings --------------------===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 //
9 // This class implements the disassembler of strings of bytes written in
10 // hexadecimal, from standard input or from a file.
11 //
12 //===----------------------------------------------------------------------===//
13
14 #include "Disassembler.h"
15 #include "llvm/ADT/StringExtras.h"
16 #include "llvm/MC/MCAsmInfo.h"
17 #include "llvm/MC/MCContext.h"
18 #include "llvm/MC/MCDisassembler/MCDisassembler.h"
19 #include "llvm/MC/MCInst.h"
20 #include "llvm/MC/MCObjectFileInfo.h"
21 #include "llvm/MC/MCRegisterInfo.h"
22 #include "llvm/MC/MCStreamer.h"
23 #include "llvm/MC/MCSubtargetInfo.h"
24 #include "llvm/MC/TargetRegistry.h"
25 #include "llvm/Support/MemoryBuffer.h"
26 #include "llvm/Support/SourceMgr.h"
27 #include "llvm/Support/raw_ostream.h"
28 #include "llvm/TargetParser/Triple.h"
29
30 using namespace llvm;
31
32 typedef std::pair<std::vector<unsigned char>, std::vector<const char *>>
33 ByteArrayTy;
34
PrintInsts(const MCDisassembler & DisAsm,const ByteArrayTy & Bytes,SourceMgr & SM,MCStreamer & Streamer,bool InAtomicBlock,const MCSubtargetInfo & STI)35 static bool PrintInsts(const MCDisassembler &DisAsm, const ByteArrayTy &Bytes,
36 SourceMgr &SM, MCStreamer &Streamer, bool InAtomicBlock,
37 const MCSubtargetInfo &STI) {
38 ArrayRef<uint8_t> Data(Bytes.first);
39
40 // Disassemble it to strings.
41 uint64_t Size;
42 uint64_t Index;
43
44 for (Index = 0; Index < Bytes.first.size(); Index += Size) {
45 MCInst Inst;
46
47 MCDisassembler::DecodeStatus S;
48 if (STI.getTargetTriple().getArch() == Triple::hexagon)
49 S = DisAsm.getInstructionBundle(Inst, Size, Data.slice(Index), Index,
50 nulls());
51 else
52 S = DisAsm.getInstruction(Inst, Size, Data.slice(Index), Index, nulls());
53 switch (S) {
54 case MCDisassembler::Fail:
55 SM.PrintMessage(SMLoc::getFromPointer(Bytes.second[Index]),
56 SourceMgr::DK_Warning,
57 "invalid instruction encoding");
58 // Don't try to resynchronise the stream in a block
59 if (InAtomicBlock)
60 return true;
61
62 if (Size == 0)
63 Size = 1; // skip illegible bytes
64
65 break;
66
67 case MCDisassembler::SoftFail:
68 SM.PrintMessage(SMLoc::getFromPointer(Bytes.second[Index]),
69 SourceMgr::DK_Warning,
70 "potentially undefined instruction encoding");
71 [[fallthrough]];
72
73 case MCDisassembler::Success:
74 Streamer.emitInstruction(Inst, STI);
75 break;
76 }
77 }
78
79 return false;
80 }
81
SkipToToken(StringRef & Str)82 static bool SkipToToken(StringRef &Str) {
83 for (;;) {
84 if (Str.empty())
85 return false;
86
87 // Strip horizontal whitespace and commas.
88 if (size_t Pos = Str.find_first_not_of(" \t\r\n,")) {
89 Str = Str.substr(Pos);
90 continue;
91 }
92
93 // If this is the start of a comment, remove the rest of the line.
94 if (Str[0] == '#') {
95 Str = Str.substr(Str.find_first_of('\n'));
96 continue;
97 }
98 return true;
99 }
100 }
101
byteArrayFromString(ByteArrayTy & ByteArray,StringRef & Str,SourceMgr & SM,bool HexBytes)102 static bool byteArrayFromString(ByteArrayTy &ByteArray, StringRef &Str,
103 SourceMgr &SM, bool HexBytes) {
104 while (SkipToToken(Str)) {
105 // Handled by higher level
106 if (Str[0] == '[' || Str[0] == ']')
107 return false;
108
109 // Get the current token.
110 size_t Next = Str.find_first_of(" \t\n\r,#[]");
111 StringRef Value = Str.substr(0, Next);
112
113 // Convert to a byte and add to the byte vector.
114 unsigned ByteVal;
115 if (HexBytes) {
116 if (Next < 2) {
117 SM.PrintMessage(SMLoc::getFromPointer(Value.data()),
118 SourceMgr::DK_Error, "expected two hex digits");
119 Str = Str.substr(Next);
120 return true;
121 }
122 Next = 2;
123 unsigned C0 = hexDigitValue(Value[0]);
124 unsigned C1 = hexDigitValue(Value[1]);
125 if (C0 == -1u || C1 == -1u) {
126 SM.PrintMessage(SMLoc::getFromPointer(Value.data()),
127 SourceMgr::DK_Error, "invalid input token");
128 Str = Str.substr(Next);
129 return true;
130 }
131 ByteVal = C0 * 16 + C1;
132 } else if (Value.getAsInteger(0, ByteVal) || ByteVal > 255) {
133 // If we have an error, print it and skip to the end of line.
134 SM.PrintMessage(SMLoc::getFromPointer(Value.data()), SourceMgr::DK_Error,
135 "invalid input token");
136 Str = Str.substr(Str.find('\n'));
137 ByteArray.first.clear();
138 ByteArray.second.clear();
139 continue;
140 }
141
142 ByteArray.first.push_back(ByteVal);
143 ByteArray.second.push_back(Value.data());
144 Str = Str.substr(Next);
145 }
146
147 return false;
148 }
149
disassemble(const Target & T,const std::string & Triple,MCSubtargetInfo & STI,MCStreamer & Streamer,MemoryBuffer & Buffer,SourceMgr & SM,MCContext & Ctx,const MCTargetOptions & MCOptions,bool HexBytes)150 int Disassembler::disassemble(const Target &T, const std::string &Triple,
151 MCSubtargetInfo &STI, MCStreamer &Streamer,
152 MemoryBuffer &Buffer, SourceMgr &SM,
153 MCContext &Ctx, const MCTargetOptions &MCOptions,
154 bool HexBytes) {
155 std::unique_ptr<const MCRegisterInfo> MRI(T.createMCRegInfo(Triple));
156 if (!MRI) {
157 errs() << "error: no register info for target " << Triple << "\n";
158 return -1;
159 }
160
161 std::unique_ptr<const MCAsmInfo> MAI(
162 T.createMCAsmInfo(*MRI, Triple, MCOptions));
163 if (!MAI) {
164 errs() << "error: no assembly info for target " << Triple << "\n";
165 return -1;
166 }
167
168 std::unique_ptr<const MCDisassembler> DisAsm(
169 T.createMCDisassembler(STI, Ctx));
170 if (!DisAsm) {
171 errs() << "error: no disassembler for target " << Triple << "\n";
172 return -1;
173 }
174
175 bool ErrorOccurred = false;
176
177 // Convert the input to a vector for disassembly.
178 ByteArrayTy ByteArray;
179 StringRef Str = Buffer.getBuffer();
180 bool InAtomicBlock = false;
181
182 while (SkipToToken(Str)) {
183 ByteArray.first.clear();
184 ByteArray.second.clear();
185
186 if (Str[0] == '[') {
187 if (InAtomicBlock) {
188 SM.PrintMessage(SMLoc::getFromPointer(Str.data()), SourceMgr::DK_Error,
189 "nested atomic blocks make no sense");
190 ErrorOccurred = true;
191 }
192 InAtomicBlock = true;
193 Str = Str.drop_front();
194 continue;
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;
200 }
201 InAtomicBlock = false;
202 Str = Str.drop_front();
203 continue;
204 }
205
206 // It's a real token, get the bytes and emit them
207 ErrorOccurred |= byteArrayFromString(ByteArray, Str, SM, HexBytes);
208
209 if (!ByteArray.first.empty())
210 ErrorOccurred |=
211 PrintInsts(*DisAsm, ByteArray, SM, Streamer, InAtomicBlock, STI);
212 }
213
214 if (InAtomicBlock) {
215 SM.PrintMessage(SMLoc::getFromPointer(Str.data()), SourceMgr::DK_Error,
216 "unclosed atomic block");
217 ErrorOccurred = true;
218 }
219
220 return ErrorOccurred;
221 }
222