Martell Malone | 375dc90 | 2017-05-20 19:56:29 +0000 | [diff] [blame] | 1 | //===--- COFFModuleDefinition.cpp - Simple DEF parser ---------------------===// |
| 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 | // Windows-specific. |
| 11 | // A parser for the module-definition file (.def file). |
| 12 | // |
| 13 | // The format of module-definition files are described in this document: |
| 14 | // https://msdn.microsoft.com/en-us/library/28d6s79h.aspx |
| 15 | // |
| 16 | //===----------------------------------------------------------------------===// |
| 17 | |
| 18 | #include "llvm/Object/COFFModuleDefinition.h" |
| 19 | #include "llvm/ADT/StringRef.h" |
| 20 | #include "llvm/ADT/StringSwitch.h" |
| 21 | #include "llvm/Object/COFF.h" |
| 22 | #include "llvm/Object/COFFImportFile.h" |
| 23 | #include "llvm/Object/Error.h" |
| 24 | #include "llvm/Support/Error.h" |
| 25 | #include "llvm/Support/raw_ostream.h" |
| 26 | |
| 27 | using namespace llvm::COFF; |
| 28 | using namespace llvm; |
| 29 | |
| 30 | namespace llvm { |
| 31 | namespace object { |
| 32 | |
| 33 | enum Kind { |
| 34 | Unknown, |
| 35 | Eof, |
| 36 | Identifier, |
| 37 | Comma, |
| 38 | Equal, |
| 39 | KwBase, |
| 40 | KwConstant, |
| 41 | KwData, |
| 42 | KwExports, |
| 43 | KwHeapsize, |
| 44 | KwLibrary, |
| 45 | KwName, |
| 46 | KwNoname, |
| 47 | KwPrivate, |
| 48 | KwStacksize, |
| 49 | KwVersion, |
| 50 | }; |
| 51 | |
| 52 | struct Token { |
| 53 | explicit Token(Kind T = Unknown, StringRef S = "") : K(T), Value(S) {} |
| 54 | Kind K; |
| 55 | StringRef Value; |
| 56 | }; |
| 57 | |
| 58 | static bool isDecorated(StringRef Sym) { |
| 59 | return Sym.startswith("_") || Sym.startswith("@") || Sym.startswith("?"); |
| 60 | } |
| 61 | |
| 62 | static Error createError(const Twine &Err) { |
Martell Malone | 36af8f4 | 2017-05-20 21:54:15 +0000 | [diff] [blame] | 63 | return make_error<StringError>(StringRef(Err.str()), |
Martell Malone | 375dc90 | 2017-05-20 19:56:29 +0000 | [diff] [blame] | 64 | object_error::parse_failed); |
| 65 | } |
| 66 | |
| 67 | class Lexer { |
| 68 | public: |
| 69 | Lexer(StringRef S) : Buf(S) {} |
| 70 | |
| 71 | Token lex() { |
| 72 | Buf = Buf.trim(); |
| 73 | if (Buf.empty()) |
| 74 | return Token(Eof); |
| 75 | |
| 76 | switch (Buf[0]) { |
| 77 | case '\0': |
| 78 | return Token(Eof); |
| 79 | case ';': { |
| 80 | size_t End = Buf.find('\n'); |
| 81 | Buf = (End == Buf.npos) ? "" : Buf.drop_front(End); |
| 82 | return lex(); |
| 83 | } |
| 84 | case '=': |
| 85 | Buf = Buf.drop_front(); |
| 86 | return Token(Equal, "="); |
| 87 | case ',': |
| 88 | Buf = Buf.drop_front(); |
| 89 | return Token(Comma, ","); |
| 90 | case '"': { |
| 91 | StringRef S; |
| 92 | std::tie(S, Buf) = Buf.substr(1).split('"'); |
| 93 | return Token(Identifier, S); |
| 94 | } |
| 95 | default: { |
| 96 | size_t End = Buf.find_first_of("=,\r\n \t\v"); |
| 97 | StringRef Word = Buf.substr(0, End); |
| 98 | Kind K = llvm::StringSwitch<Kind>(Word) |
| 99 | .Case("BASE", KwBase) |
| 100 | .Case("CONSTANT", KwConstant) |
| 101 | .Case("DATA", KwData) |
| 102 | .Case("EXPORTS", KwExports) |
| 103 | .Case("HEAPSIZE", KwHeapsize) |
| 104 | .Case("LIBRARY", KwLibrary) |
| 105 | .Case("NAME", KwName) |
| 106 | .Case("NONAME", KwNoname) |
| 107 | .Case("PRIVATE", KwPrivate) |
| 108 | .Case("STACKSIZE", KwStacksize) |
| 109 | .Case("VERSION", KwVersion) |
| 110 | .Default(Identifier); |
| 111 | Buf = (End == Buf.npos) ? "" : Buf.drop_front(End); |
| 112 | return Token(K, Word); |
| 113 | } |
| 114 | } |
| 115 | } |
| 116 | |
| 117 | private: |
| 118 | StringRef Buf; |
| 119 | }; |
| 120 | |
| 121 | class Parser { |
| 122 | public: |
| 123 | explicit Parser(StringRef S, MachineTypes M) : Lex(S), Machine(M) {} |
| 124 | |
| 125 | Expected<COFFModuleDefinition> parse() { |
| 126 | do { |
| 127 | if (Error Err = parseOne()) |
| 128 | return std::move(Err); |
| 129 | } while (Tok.K != Eof); |
| 130 | return Info; |
| 131 | } |
| 132 | |
| 133 | private: |
| 134 | void read() { |
| 135 | if (Stack.empty()) { |
| 136 | Tok = Lex.lex(); |
| 137 | return; |
| 138 | } |
| 139 | Tok = Stack.back(); |
| 140 | Stack.pop_back(); |
| 141 | } |
| 142 | |
| 143 | Error readAsInt(uint64_t *I) { |
| 144 | read(); |
| 145 | if (Tok.K != Identifier || Tok.Value.getAsInteger(10, *I)) |
| 146 | return createError("integer expected"); |
| 147 | return Error::success(); |
| 148 | } |
| 149 | |
| 150 | Error expect(Kind Expected, StringRef Msg) { |
| 151 | read(); |
| 152 | if (Tok.K != Expected) |
| 153 | return createError(Msg); |
| 154 | return Error::success(); |
| 155 | } |
| 156 | |
| 157 | void unget() { Stack.push_back(Tok); } |
| 158 | |
| 159 | Error parseOne() { |
| 160 | read(); |
| 161 | switch (Tok.K) { |
| 162 | case Eof: |
| 163 | return Error::success(); |
| 164 | case KwExports: |
| 165 | for (;;) { |
| 166 | read(); |
| 167 | if (Tok.K != Identifier) { |
| 168 | unget(); |
| 169 | return Error::success(); |
| 170 | } |
| 171 | if (Error Err = parseExport()) |
| 172 | return Err; |
| 173 | } |
| 174 | case KwHeapsize: |
| 175 | return parseNumbers(&Info.HeapReserve, &Info.HeapCommit); |
| 176 | case KwStacksize: |
| 177 | return parseNumbers(&Info.StackReserve, &Info.StackCommit); |
| 178 | case KwLibrary: |
| 179 | case KwName: { |
| 180 | bool IsDll = Tok.K == KwLibrary; // Check before parseName. |
| 181 | std::string Name; |
| 182 | if (Error Err = parseName(&Name, &Info.ImageBase)) |
| 183 | return Err; |
| 184 | // Append the appropriate file extension if not already present. |
| 185 | StringRef Ext = IsDll ? ".dll" : ".exe"; |
| 186 | if (!StringRef(Name).endswith_lower(Ext)) |
| 187 | Name += Ext; |
| 188 | |
| 189 | // Set the output file, but don't override /out if it was already passed. |
| 190 | if (Info.OutputFile.empty()) |
| 191 | Info.OutputFile = Name; |
| 192 | return Error::success(); |
| 193 | } |
| 194 | case KwVersion: |
| 195 | return parseVersion(&Info.MajorImageVersion, &Info.MinorImageVersion); |
| 196 | default: |
| 197 | return createError("unknown directive: " + Tok.Value); |
| 198 | } |
| 199 | } |
| 200 | |
| 201 | Error parseExport() { |
| 202 | COFFShortExport E; |
| 203 | E.Name = Tok.Value; |
| 204 | read(); |
| 205 | if (Tok.K == Equal) { |
| 206 | read(); |
| 207 | if (Tok.K != Identifier) |
| 208 | return createError("identifier expected, but got " + Tok.Value); |
| 209 | E.ExtName = E.Name; |
| 210 | E.Name = Tok.Value; |
| 211 | } else { |
| 212 | unget(); |
| 213 | } |
| 214 | |
| 215 | if (Machine == IMAGE_FILE_MACHINE_I386) { |
| 216 | if (!isDecorated(E.Name)) |
| 217 | E.Name = (std::string("_").append(E.Name)); |
| 218 | if (!E.ExtName.empty() && !isDecorated(E.ExtName)) |
| 219 | E.ExtName = (std::string("_").append(E.ExtName)); |
| 220 | } |
| 221 | |
| 222 | for (;;) { |
| 223 | read(); |
| 224 | if (Tok.K == Identifier && Tok.Value[0] == '@') { |
| 225 | Tok.Value.drop_front().getAsInteger(10, E.Ordinal); |
| 226 | read(); |
| 227 | if (Tok.K == KwNoname) { |
| 228 | E.Noname = true; |
| 229 | } else { |
| 230 | unget(); |
| 231 | } |
| 232 | continue; |
| 233 | } |
| 234 | if (Tok.K == KwData) { |
| 235 | E.Data = true; |
| 236 | continue; |
| 237 | } |
| 238 | if (Tok.K == KwConstant) { |
| 239 | E.Constant = true; |
| 240 | continue; |
| 241 | } |
| 242 | if (Tok.K == KwPrivate) { |
| 243 | E.Private = true; |
| 244 | continue; |
| 245 | } |
| 246 | unget(); |
| 247 | Info.Exports.push_back(E); |
| 248 | return Error::success(); |
| 249 | } |
| 250 | } |
| 251 | |
| 252 | // HEAPSIZE/STACKSIZE reserve[,commit] |
| 253 | Error parseNumbers(uint64_t *Reserve, uint64_t *Commit) { |
| 254 | if (Error Err = readAsInt(Reserve)) |
| 255 | return Err; |
| 256 | read(); |
| 257 | if (Tok.K != Comma) { |
| 258 | unget(); |
| 259 | Commit = nullptr; |
| 260 | return Error::success(); |
| 261 | } |
| 262 | if (Error Err = readAsInt(Commit)) |
| 263 | return Err; |
| 264 | return Error::success(); |
| 265 | } |
| 266 | |
| 267 | // NAME outputPath [BASE=address] |
| 268 | Error parseName(std::string *Out, uint64_t *Baseaddr) { |
| 269 | read(); |
| 270 | if (Tok.K == Identifier) { |
| 271 | *Out = Tok.Value; |
| 272 | } else { |
| 273 | *Out = ""; |
| 274 | unget(); |
| 275 | return Error::success(); |
| 276 | } |
| 277 | read(); |
| 278 | if (Tok.K == KwBase) { |
| 279 | if (Error Err = expect(Equal, "'=' expected")) |
| 280 | return Err; |
| 281 | if (Error Err = readAsInt(Baseaddr)) |
| 282 | return Err; |
| 283 | } else { |
| 284 | unget(); |
| 285 | *Baseaddr = 0; |
| 286 | } |
| 287 | return Error::success(); |
| 288 | } |
| 289 | |
| 290 | // VERSION major[.minor] |
| 291 | Error parseVersion(uint32_t *Major, uint32_t *Minor) { |
| 292 | read(); |
| 293 | if (Tok.K != Identifier) |
| 294 | return createError("identifier expected, but got " + Tok.Value); |
| 295 | StringRef V1, V2; |
| 296 | std::tie(V1, V2) = Tok.Value.split('.'); |
| 297 | if (V1.getAsInteger(10, *Major)) |
| 298 | return createError("integer expected, but got " + Tok.Value); |
| 299 | if (V2.empty()) |
| 300 | *Minor = 0; |
| 301 | else if (V2.getAsInteger(10, *Minor)) |
| 302 | return createError("integer expected, but got " + Tok.Value); |
| 303 | return Error::success(); |
| 304 | } |
| 305 | |
| 306 | Lexer Lex; |
| 307 | Token Tok; |
| 308 | std::vector<Token> Stack; |
| 309 | MachineTypes Machine; |
| 310 | COFFModuleDefinition Info; |
| 311 | }; |
| 312 | |
| 313 | Expected<COFFModuleDefinition> parseCOFFModuleDefinition(MemoryBufferRef MB, |
| 314 | MachineTypes Machine) { |
| 315 | return Parser(MB.getBuffer(), Machine).parse(); |
| 316 | } |
| 317 | |
| 318 | } // namespace object |
| 319 | } // namespace llvm |