Chris Lattner | b257d24 | 2009-12-22 22:50:29 +0000 | [diff] [blame] | 1 | //===- Disassembler.cpp - Disassembler for hex strings --------------------===// |
Sean Callanan | 7e64550 | 2009-12-17 01:49:59 +0000 | [diff] [blame] | 2 | // |
Chandler Carruth | 2946cd7 | 2019-01-19 08:50:56 +0000 | [diff] [blame] | 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 |
Sean Callanan | 7e64550 | 2009-12-17 01:49:59 +0000 | [diff] [blame] | 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 | |
Chris Lattner | b257d24 | 2009-12-22 22:50:29 +0000 | [diff] [blame] | 14 | #include "Disassembler.h" |
Chandler Carruth | 4d88a1c | 2012-12-04 10:44:52 +0000 | [diff] [blame] | 15 | #include "llvm/ADT/Triple.h" |
Lang Hames | a1bc0f5 | 2014-04-15 04:40:56 +0000 | [diff] [blame] | 16 | #include "llvm/MC/MCAsmInfo.h" |
| 17 | #include "llvm/MC/MCContext.h" |
Benjamin Kramer | f57c197 | 2016-01-26 16:44:37 +0000 | [diff] [blame] | 18 | #include "llvm/MC/MCDisassembler/MCDisassembler.h" |
Sean Callanan | 7e64550 | 2009-12-17 01:49:59 +0000 | [diff] [blame] | 19 | #include "llvm/MC/MCInst.h" |
Thomas Lively | 2cb2707 | 2019-10-15 18:28:22 +0000 | [diff] [blame] | 20 | #include "llvm/MC/MCObjectFileInfo.h" |
Lang Hames | a1bc0f5 | 2014-04-15 04:40:56 +0000 | [diff] [blame] | 21 | #include "llvm/MC/MCRegisterInfo.h" |
Richard Barton | def81b9 | 2012-04-16 11:32:10 +0000 | [diff] [blame] | 22 | #include "llvm/MC/MCStreamer.h" |
James Molloy | 4c493e8 | 2011-09-07 17:24:38 +0000 | [diff] [blame] | 23 | #include "llvm/MC/MCSubtargetInfo.h" |
Sean Callanan | 7e64550 | 2009-12-17 01:49:59 +0000 | [diff] [blame] | 24 | #include "llvm/Support/MemoryBuffer.h" |
Chris Lattner | 8879918 | 2009-12-22 06:45:48 +0000 | [diff] [blame] | 25 | #include "llvm/Support/SourceMgr.h" |
Evan Cheng | 2bb4035 | 2011-08-24 18:08:43 +0000 | [diff] [blame] | 26 | #include "llvm/Support/TargetRegistry.h" |
| 27 | #include "llvm/Support/raw_ostream.h" |
Richard Barton | def81b9 | 2012-04-16 11:32:10 +0000 | [diff] [blame] | 28 | |
Sean Callanan | 7e64550 | 2009-12-17 01:49:59 +0000 | [diff] [blame] | 29 | using namespace llvm; |
| 30 | |
Rafael Espindola | 45a753a | 2014-11-07 17:59:05 +0000 | [diff] [blame] | 31 | typedef std::pair<std::vector<unsigned char>, std::vector<const char *>> |
| 32 | ByteArrayTy; |
Sean Callanan | 7e64550 | 2009-12-17 01:49:59 +0000 | [diff] [blame] | 33 | |
Daniel Dunbar | d9d5b31 | 2010-03-20 22:36:35 +0000 | [diff] [blame] | 34 | static bool PrintInsts(const MCDisassembler &DisAsm, |
Richard Barton | def81b9 | 2012-04-16 11:32:10 +0000 | [diff] [blame] | 35 | const ByteArrayTy &Bytes, |
| 36 | SourceMgr &SM, raw_ostream &Out, |
David Woodhouse | e6c13e4 | 2014-01-28 23:12:42 +0000 | [diff] [blame] | 37 | MCStreamer &Streamer, bool InAtomicBlock, |
| 38 | const MCSubtargetInfo &STI) { |
Rafael Espindola | 7fc5b87 | 2014-11-12 02:04:27 +0000 | [diff] [blame] | 39 | ArrayRef<uint8_t> Data(Bytes.first.data(), Bytes.first.size()); |
Jim Grosbach | 112a2de | 2011-05-09 20:05:25 +0000 | [diff] [blame] | 40 | |
Sean Callanan | 6a6f9cc | 2010-02-03 03:46:41 +0000 | [diff] [blame] | 41 | // Disassemble it to strings. |
Chris Lattner | dd0c01b | 2009-12-22 06:56:51 +0000 | [diff] [blame] | 42 | uint64_t Size; |
Sean Callanan | 6a6f9cc | 2010-02-03 03:46:41 +0000 | [diff] [blame] | 43 | uint64_t Index; |
Jim Grosbach | 112a2de | 2011-05-09 20:05:25 +0000 | [diff] [blame] | 44 | |
Rafael Espindola | 45a753a | 2014-11-07 17:59:05 +0000 | [diff] [blame] | 45 | for (Index = 0; Index < Bytes.first.size(); Index += Size) { |
Sean Callanan | 6a6f9cc | 2010-02-03 03:46:41 +0000 | [diff] [blame] | 46 | MCInst Inst; |
Jim Grosbach | 112a2de | 2011-05-09 20:05:25 +0000 | [diff] [blame] | 47 | |
Owen Anderson | a4043c4 | 2011-08-17 17:44:15 +0000 | [diff] [blame] | 48 | MCDisassembler::DecodeStatus S; |
Fangrui Song | 6fdd6a7 | 2020-01-11 12:36:13 -0800 | [diff] [blame] | 49 | S = DisAsm.getInstruction(Inst, Size, Data.slice(Index), Index, nulls()); |
Owen Anderson | a4043c4 | 2011-08-17 17:44:15 +0000 | [diff] [blame] | 50 | switch (S) { |
| 51 | case MCDisassembler::Fail: |
Rafael Espindola | 45a753a | 2014-11-07 17:59:05 +0000 | [diff] [blame] | 52 | SM.PrintMessage(SMLoc::getFromPointer(Bytes.second[Index]), |
Chris Lattner | 03b80a4 | 2011-10-16 05:43:57 +0000 | [diff] [blame] | 53 | SourceMgr::DK_Warning, |
| 54 | "invalid instruction encoding"); |
Tim Northover | 48cf6cc | 2013-07-19 10:05:04 +0000 | [diff] [blame] | 55 | // Don't try to resynchronise the stream in a block |
| 56 | if (InAtomicBlock) |
| 57 | return true; |
| 58 | |
Sean Callanan | 6a6f9cc | 2010-02-03 03:46:41 +0000 | [diff] [blame] | 59 | if (Size == 0) |
| 60 | Size = 1; // skip illegible bytes |
Tim Northover | 48cf6cc | 2013-07-19 10:05:04 +0000 | [diff] [blame] | 61 | |
Owen Anderson | a4043c4 | 2011-08-17 17:44:15 +0000 | [diff] [blame] | 62 | break; |
| 63 | |
| 64 | case MCDisassembler::SoftFail: |
Rafael Espindola | 45a753a | 2014-11-07 17:59:05 +0000 | [diff] [blame] | 65 | SM.PrintMessage(SMLoc::getFromPointer(Bytes.second[Index]), |
Chris Lattner | 03b80a4 | 2011-10-16 05:43:57 +0000 | [diff] [blame] | 66 | SourceMgr::DK_Warning, |
| 67 | "potentially undefined instruction encoding"); |
Justin Bogner | cd1d5aa | 2016-08-17 20:30:52 +0000 | [diff] [blame] | 68 | LLVM_FALLTHROUGH; |
Owen Anderson | a4043c4 | 2011-08-17 17:44:15 +0000 | [diff] [blame] | 69 | |
| 70 | case MCDisassembler::Success: |
Fangrui Song | bcd24b2 | 2020-02-13 21:58:16 -0800 | [diff] [blame^] | 71 | Streamer.emitInstruction(Inst, STI); |
Owen Anderson | a4043c4 | 2011-08-17 17:44:15 +0000 | [diff] [blame] | 72 | break; |
Sean Callanan | 6a6f9cc | 2010-02-03 03:46:41 +0000 | [diff] [blame] | 73 | } |
Chris Lattner | dd0c01b | 2009-12-22 06:56:51 +0000 | [diff] [blame] | 74 | } |
Jim Grosbach | 112a2de | 2011-05-09 20:05:25 +0000 | [diff] [blame] | 75 | |
Chris Lattner | dd0c01b | 2009-12-22 06:56:51 +0000 | [diff] [blame] | 76 | return false; |
Sean Callanan | 7e64550 | 2009-12-17 01:49:59 +0000 | [diff] [blame] | 77 | } |
| 78 | |
Tim Northover | 48cf6cc | 2013-07-19 10:05:04 +0000 | [diff] [blame] | 79 | static bool SkipToToken(StringRef &Str) { |
Colin LeMahieu | eb4675f | 2014-11-11 21:03:09 +0000 | [diff] [blame] | 80 | for (;;) { |
| 81 | if (Str.empty()) |
| 82 | return false; |
Jim Grosbach | 112a2de | 2011-05-09 20:05:25 +0000 | [diff] [blame] | 83 | |
Colin LeMahieu | eb4675f | 2014-11-11 21:03:09 +0000 | [diff] [blame] | 84 | // Strip horizontal whitespace and commas. |
| 85 | if (size_t Pos = Str.find_first_not_of(" \t\r\n,")) { |
| 86 | Str = Str.substr(Pos); |
Sean Callanan | 2d03d3a | 2010-04-12 19:43:00 +0000 | [diff] [blame] | 87 | continue; |
| 88 | } |
Tim Northover | 48cf6cc | 2013-07-19 10:05:04 +0000 | [diff] [blame] | 89 | |
Colin LeMahieu | eb4675f | 2014-11-11 21:03:09 +0000 | [diff] [blame] | 90 | // If this is the start of a comment, remove the rest of the line. |
| 91 | if (Str[0] == '#') { |
| 92 | Str = Str.substr(Str.find_first_of('\n')); |
| 93 | continue; |
| 94 | } |
| 95 | return true; |
| 96 | } |
Tim Northover | 48cf6cc | 2013-07-19 10:05:04 +0000 | [diff] [blame] | 97 | } |
| 98 | |
| 99 | |
| 100 | static bool ByteArrayFromString(ByteArrayTy &ByteArray, |
| 101 | StringRef &Str, |
| 102 | SourceMgr &SM) { |
| 103 | while (SkipToToken(Str)) { |
| 104 | // Handled by higher level |
| 105 | if (Str[0] == '[' || Str[0] == ']') |
| 106 | return false; |
Jim Grosbach | 112a2de | 2011-05-09 20:05:25 +0000 | [diff] [blame] | 107 | |
Sean Callanan | 2d03d3a | 2010-04-12 19:43:00 +0000 | [diff] [blame] | 108 | // Get the current token. |
Tim Northover | 48cf6cc | 2013-07-19 10:05:04 +0000 | [diff] [blame] | 109 | size_t Next = Str.find_first_of(" \t\n\r,#[]"); |
Sean Callanan | 2d03d3a | 2010-04-12 19:43:00 +0000 | [diff] [blame] | 110 | StringRef Value = Str.substr(0, Next); |
Jim Grosbach | 112a2de | 2011-05-09 20:05:25 +0000 | [diff] [blame] | 111 | |
Sean Callanan | 2d03d3a | 2010-04-12 19:43:00 +0000 | [diff] [blame] | 112 | // Convert to a byte and add to the byte vector. |
| 113 | unsigned ByteVal; |
| 114 | if (Value.getAsInteger(0, ByteVal) || ByteVal > 255) { |
| 115 | // If we have an error, print it and skip to the end of line. |
Chris Lattner | 03b80a4 | 2011-10-16 05:43:57 +0000 | [diff] [blame] | 116 | SM.PrintMessage(SMLoc::getFromPointer(Value.data()), SourceMgr::DK_Error, |
| 117 | "invalid input token"); |
Sean Callanan | 2d03d3a | 2010-04-12 19:43:00 +0000 | [diff] [blame] | 118 | Str = Str.substr(Str.find('\n')); |
Rafael Espindola | 45a753a | 2014-11-07 17:59:05 +0000 | [diff] [blame] | 119 | ByteArray.first.clear(); |
| 120 | ByteArray.second.clear(); |
Sean Callanan | 2d03d3a | 2010-04-12 19:43:00 +0000 | [diff] [blame] | 121 | continue; |
| 122 | } |
Jim Grosbach | 112a2de | 2011-05-09 20:05:25 +0000 | [diff] [blame] | 123 | |
Rafael Espindola | 45a753a | 2014-11-07 17:59:05 +0000 | [diff] [blame] | 124 | ByteArray.first.push_back(ByteVal); |
| 125 | ByteArray.second.push_back(Value.data()); |
Sean Callanan | 2d03d3a | 2010-04-12 19:43:00 +0000 | [diff] [blame] | 126 | Str = Str.substr(Next); |
| 127 | } |
Jim Grosbach | 112a2de | 2011-05-09 20:05:25 +0000 | [diff] [blame] | 128 | |
Sean Callanan | 2d03d3a | 2010-04-12 19:43:00 +0000 | [diff] [blame] | 129 | return false; |
| 130 | } |
| 131 | |
Thomas Lively | 2cb2707 | 2019-10-15 18:28:22 +0000 | [diff] [blame] | 132 | int Disassembler::disassemble(const Target &T, const std::string &Triple, |
| 133 | MCSubtargetInfo &STI, MCStreamer &Streamer, |
| 134 | MemoryBuffer &Buffer, SourceMgr &SM, |
Mirko Brkusanin | 4b63ca1 | 2019-10-23 12:24:35 +0200 | [diff] [blame] | 135 | MCContext &Ctx, raw_ostream &Out, |
| 136 | const MCTargetOptions &MCOptions) { |
Lang Hames | a1bc0f5 | 2014-04-15 04:40:56 +0000 | [diff] [blame] | 137 | |
| 138 | std::unique_ptr<const MCRegisterInfo> MRI(T.createMCRegInfo(Triple)); |
| 139 | if (!MRI) { |
| 140 | errs() << "error: no register info for target " << Triple << "\n"; |
| 141 | return -1; |
| 142 | } |
| 143 | |
Mirko Brkusanin | 4b63ca1 | 2019-10-23 12:24:35 +0200 | [diff] [blame] | 144 | std::unique_ptr<const MCAsmInfo> MAI( |
| 145 | T.createMCAsmInfo(*MRI, Triple, MCOptions)); |
Lang Hames | a1bc0f5 | 2014-04-15 04:40:56 +0000 | [diff] [blame] | 146 | if (!MAI) { |
| 147 | errs() << "error: no assembly info for target " << Triple << "\n"; |
| 148 | return -1; |
| 149 | } |
| 150 | |
Lang Hames | a1bc0f5 | 2014-04-15 04:40:56 +0000 | [diff] [blame] | 151 | std::unique_ptr<const MCDisassembler> DisAsm( |
| 152 | T.createMCDisassembler(STI, Ctx)); |
Chris Lattner | 72cbaa2 | 2009-12-22 06:24:00 +0000 | [diff] [blame] | 153 | if (!DisAsm) { |
| 154 | errs() << "error: no disassembler for target " << Triple << "\n"; |
Sean Callanan | 7e64550 | 2009-12-17 01:49:59 +0000 | [diff] [blame] | 155 | return -1; |
| 156 | } |
Jim Grosbach | 112a2de | 2011-05-09 20:05:25 +0000 | [diff] [blame] | 157 | |
Richard Barton | def81b9 | 2012-04-16 11:32:10 +0000 | [diff] [blame] | 158 | // Set up initial section manually here |
Rafael Espindola | 7b61ddf | 2014-10-15 16:12:52 +0000 | [diff] [blame] | 159 | Streamer.InitSections(false); |
Jim Grosbach | 112a2de | 2011-05-09 20:05:25 +0000 | [diff] [blame] | 160 | |
Chris Lattner | dd0c01b | 2009-12-22 06:56:51 +0000 | [diff] [blame] | 161 | bool ErrorOccurred = false; |
Jim Grosbach | 112a2de | 2011-05-09 20:05:25 +0000 | [diff] [blame] | 162 | |
Sean Callanan | 7e64550 | 2009-12-17 01:49:59 +0000 | [diff] [blame] | 163 | // Convert the input to a vector for disassembly. |
Chris Lattner | dd0c01b | 2009-12-22 06:56:51 +0000 | [diff] [blame] | 164 | ByteArrayTy ByteArray; |
Chris Lattner | 55ddc30 | 2010-04-09 04:24:20 +0000 | [diff] [blame] | 165 | StringRef Str = Buffer.getBuffer(); |
Tim Northover | 48cf6cc | 2013-07-19 10:05:04 +0000 | [diff] [blame] | 166 | bool InAtomicBlock = false; |
Jim Grosbach | 112a2de | 2011-05-09 20:05:25 +0000 | [diff] [blame] | 167 | |
Tim Northover | 48cf6cc | 2013-07-19 10:05:04 +0000 | [diff] [blame] | 168 | while (SkipToToken(Str)) { |
Rafael Espindola | 45a753a | 2014-11-07 17:59:05 +0000 | [diff] [blame] | 169 | ByteArray.first.clear(); |
| 170 | ByteArray.second.clear(); |
Jim Grosbach | 112a2de | 2011-05-09 20:05:25 +0000 | [diff] [blame] | 171 | |
Tim Northover | 48cf6cc | 2013-07-19 10:05:04 +0000 | [diff] [blame] | 172 | if (Str[0] == '[') { |
| 173 | if (InAtomicBlock) { |
| 174 | SM.PrintMessage(SMLoc::getFromPointer(Str.data()), SourceMgr::DK_Error, |
| 175 | "nested atomic blocks make no sense"); |
| 176 | ErrorOccurred = true; |
| 177 | } |
| 178 | InAtomicBlock = true; |
| 179 | Str = Str.drop_front(); |
| 180 | continue; |
| 181 | } else if (Str[0] == ']') { |
| 182 | if (!InAtomicBlock) { |
| 183 | SM.PrintMessage(SMLoc::getFromPointer(Str.data()), SourceMgr::DK_Error, |
| 184 | "attempt to close atomic block without opening"); |
| 185 | ErrorOccurred = true; |
| 186 | } |
| 187 | InAtomicBlock = false; |
| 188 | Str = Str.drop_front(); |
| 189 | continue; |
| 190 | } |
| 191 | |
| 192 | // It's a real token, get the bytes and emit them |
| 193 | ErrorOccurred |= ByteArrayFromString(ByteArray, Str, SM); |
| 194 | |
Rafael Espindola | 45a753a | 2014-11-07 17:59:05 +0000 | [diff] [blame] | 195 | if (!ByteArray.first.empty()) |
Tim Northover | 48cf6cc | 2013-07-19 10:05:04 +0000 | [diff] [blame] | 196 | ErrorOccurred |= PrintInsts(*DisAsm, ByteArray, SM, Out, Streamer, |
David Woodhouse | e6c13e4 | 2014-01-28 23:12:42 +0000 | [diff] [blame] | 197 | InAtomicBlock, STI); |
Tim Northover | 48cf6cc | 2013-07-19 10:05:04 +0000 | [diff] [blame] | 198 | } |
| 199 | |
| 200 | if (InAtomicBlock) { |
| 201 | SM.PrintMessage(SMLoc::getFromPointer(Str.data()), SourceMgr::DK_Error, |
| 202 | "unclosed atomic block"); |
| 203 | ErrorOccurred = true; |
| 204 | } |
Jim Grosbach | 112a2de | 2011-05-09 20:05:25 +0000 | [diff] [blame] | 205 | |
Chris Lattner | dd0c01b | 2009-12-22 06:56:51 +0000 | [diff] [blame] | 206 | return ErrorOccurred; |
Sean Callanan | 7e64550 | 2009-12-17 01:49:59 +0000 | [diff] [blame] | 207 | } |