1146eb7a6SReid Kleckner //===--- COFFModuleDefinition.cpp - Simple DEF parser ---------------------===// 2146eb7a6SReid Kleckner // 3146eb7a6SReid Kleckner // The LLVM Compiler Infrastructure 4146eb7a6SReid Kleckner // 5146eb7a6SReid Kleckner // This file is distributed under the University of Illinois Open Source 6146eb7a6SReid Kleckner // License. See LICENSE.TXT for details. 7146eb7a6SReid Kleckner // 8146eb7a6SReid Kleckner //===----------------------------------------------------------------------===// 9146eb7a6SReid Kleckner // 10146eb7a6SReid Kleckner // Windows-specific. 11146eb7a6SReid Kleckner // A parser for the module-definition file (.def file). 12146eb7a6SReid Kleckner // 13146eb7a6SReid Kleckner // The format of module-definition files are described in this document: 14146eb7a6SReid Kleckner // https://msdn.microsoft.com/en-us/library/28d6s79h.aspx 15146eb7a6SReid Kleckner // 16146eb7a6SReid Kleckner //===----------------------------------------------------------------------===// 17146eb7a6SReid Kleckner 18146eb7a6SReid Kleckner #include "llvm/Object/COFFModuleDefinition.h" 19146eb7a6SReid Kleckner #include "llvm/ADT/StringRef.h" 20146eb7a6SReid Kleckner #include "llvm/ADT/StringSwitch.h" 21146eb7a6SReid Kleckner #include "llvm/Object/COFF.h" 22146eb7a6SReid Kleckner #include "llvm/Object/COFFImportFile.h" 23146eb7a6SReid Kleckner #include "llvm/Object/Error.h" 24146eb7a6SReid Kleckner #include "llvm/Support/Error.h" 25146eb7a6SReid Kleckner #include "llvm/Support/raw_ostream.h" 26146eb7a6SReid Kleckner 27146eb7a6SReid Kleckner using namespace llvm::COFF; 28146eb7a6SReid Kleckner using namespace llvm; 29146eb7a6SReid Kleckner 30146eb7a6SReid Kleckner namespace llvm { 31146eb7a6SReid Kleckner namespace object { 32146eb7a6SReid Kleckner 33146eb7a6SReid Kleckner enum Kind { 34146eb7a6SReid Kleckner Unknown, 35146eb7a6SReid Kleckner Eof, 36146eb7a6SReid Kleckner Identifier, 37146eb7a6SReid Kleckner Comma, 38146eb7a6SReid Kleckner Equal, 39146eb7a6SReid Kleckner KwBase, 40146eb7a6SReid Kleckner KwConstant, 41146eb7a6SReid Kleckner KwData, 42146eb7a6SReid Kleckner KwExports, 43146eb7a6SReid Kleckner KwHeapsize, 44146eb7a6SReid Kleckner KwLibrary, 45146eb7a6SReid Kleckner KwName, 46146eb7a6SReid Kleckner KwNoname, 47146eb7a6SReid Kleckner KwPrivate, 48146eb7a6SReid Kleckner KwStacksize, 49146eb7a6SReid Kleckner KwVersion, 50146eb7a6SReid Kleckner }; 51146eb7a6SReid Kleckner 52146eb7a6SReid Kleckner struct Token { 53146eb7a6SReid Kleckner explicit Token(Kind T = Unknown, StringRef S = "") : K(T), Value(S) {} 54146eb7a6SReid Kleckner Kind K; 55146eb7a6SReid Kleckner StringRef Value; 56146eb7a6SReid Kleckner }; 57146eb7a6SReid Kleckner 58*1079ef8dSMartell Malone static bool isDecorated(StringRef Sym, bool MingwDef) { 59*1079ef8dSMartell Malone // mingw does not prepend "_". 60*1079ef8dSMartell Malone return (!MingwDef && Sym.startswith("_")) || Sym.startswith("@") || 61*1079ef8dSMartell Malone Sym.startswith("?"); 62146eb7a6SReid Kleckner } 63146eb7a6SReid Kleckner 64146eb7a6SReid Kleckner static Error createError(const Twine &Err) { 65146eb7a6SReid Kleckner return make_error<StringError>(StringRef(Err.str()), 66146eb7a6SReid Kleckner object_error::parse_failed); 67146eb7a6SReid Kleckner } 68146eb7a6SReid Kleckner 69146eb7a6SReid Kleckner class Lexer { 70146eb7a6SReid Kleckner public: 71146eb7a6SReid Kleckner Lexer(StringRef S) : Buf(S) {} 72146eb7a6SReid Kleckner 73146eb7a6SReid Kleckner Token lex() { 74146eb7a6SReid Kleckner Buf = Buf.trim(); 75146eb7a6SReid Kleckner if (Buf.empty()) 76146eb7a6SReid Kleckner return Token(Eof); 77146eb7a6SReid Kleckner 78146eb7a6SReid Kleckner switch (Buf[0]) { 79146eb7a6SReid Kleckner case '\0': 80146eb7a6SReid Kleckner return Token(Eof); 81146eb7a6SReid Kleckner case ';': { 82146eb7a6SReid Kleckner size_t End = Buf.find('\n'); 83146eb7a6SReid Kleckner Buf = (End == Buf.npos) ? "" : Buf.drop_front(End); 84146eb7a6SReid Kleckner return lex(); 85146eb7a6SReid Kleckner } 86146eb7a6SReid Kleckner case '=': 87146eb7a6SReid Kleckner Buf = Buf.drop_front(); 88*1079ef8dSMartell Malone // GNU dlltool accepts both = and ==. 89*1079ef8dSMartell Malone if (Buf.startswith("=")) 90*1079ef8dSMartell Malone Buf = Buf.drop_front(); 91146eb7a6SReid Kleckner return Token(Equal, "="); 92146eb7a6SReid Kleckner case ',': 93146eb7a6SReid Kleckner Buf = Buf.drop_front(); 94146eb7a6SReid Kleckner return Token(Comma, ","); 95146eb7a6SReid Kleckner case '"': { 96146eb7a6SReid Kleckner StringRef S; 97146eb7a6SReid Kleckner std::tie(S, Buf) = Buf.substr(1).split('"'); 98146eb7a6SReid Kleckner return Token(Identifier, S); 99146eb7a6SReid Kleckner } 100146eb7a6SReid Kleckner default: { 101146eb7a6SReid Kleckner size_t End = Buf.find_first_of("=,\r\n \t\v"); 102146eb7a6SReid Kleckner StringRef Word = Buf.substr(0, End); 103146eb7a6SReid Kleckner Kind K = llvm::StringSwitch<Kind>(Word) 104146eb7a6SReid Kleckner .Case("BASE", KwBase) 105146eb7a6SReid Kleckner .Case("CONSTANT", KwConstant) 106146eb7a6SReid Kleckner .Case("DATA", KwData) 107146eb7a6SReid Kleckner .Case("EXPORTS", KwExports) 108146eb7a6SReid Kleckner .Case("HEAPSIZE", KwHeapsize) 109146eb7a6SReid Kleckner .Case("LIBRARY", KwLibrary) 110146eb7a6SReid Kleckner .Case("NAME", KwName) 111146eb7a6SReid Kleckner .Case("NONAME", KwNoname) 112146eb7a6SReid Kleckner .Case("PRIVATE", KwPrivate) 113146eb7a6SReid Kleckner .Case("STACKSIZE", KwStacksize) 114146eb7a6SReid Kleckner .Case("VERSION", KwVersion) 115146eb7a6SReid Kleckner .Default(Identifier); 116146eb7a6SReid Kleckner Buf = (End == Buf.npos) ? "" : Buf.drop_front(End); 117146eb7a6SReid Kleckner return Token(K, Word); 118146eb7a6SReid Kleckner } 119146eb7a6SReid Kleckner } 120146eb7a6SReid Kleckner } 121146eb7a6SReid Kleckner 122146eb7a6SReid Kleckner private: 123146eb7a6SReid Kleckner StringRef Buf; 124146eb7a6SReid Kleckner }; 125146eb7a6SReid Kleckner 126146eb7a6SReid Kleckner class Parser { 127146eb7a6SReid Kleckner public: 128*1079ef8dSMartell Malone explicit Parser(StringRef S, MachineTypes M, bool B) 129*1079ef8dSMartell Malone : Lex(S), Machine(M), MingwDef(B) {} 130146eb7a6SReid Kleckner 131146eb7a6SReid Kleckner Expected<COFFModuleDefinition> parse() { 132146eb7a6SReid Kleckner do { 133146eb7a6SReid Kleckner if (Error Err = parseOne()) 134146eb7a6SReid Kleckner return std::move(Err); 135146eb7a6SReid Kleckner } while (Tok.K != Eof); 136146eb7a6SReid Kleckner return Info; 137146eb7a6SReid Kleckner } 138146eb7a6SReid Kleckner 139146eb7a6SReid Kleckner private: 140146eb7a6SReid Kleckner void read() { 141146eb7a6SReid Kleckner if (Stack.empty()) { 142146eb7a6SReid Kleckner Tok = Lex.lex(); 143146eb7a6SReid Kleckner return; 144146eb7a6SReid Kleckner } 145146eb7a6SReid Kleckner Tok = Stack.back(); 146146eb7a6SReid Kleckner Stack.pop_back(); 147146eb7a6SReid Kleckner } 148146eb7a6SReid Kleckner 149146eb7a6SReid Kleckner Error readAsInt(uint64_t *I) { 150146eb7a6SReid Kleckner read(); 151146eb7a6SReid Kleckner if (Tok.K != Identifier || Tok.Value.getAsInteger(10, *I)) 152146eb7a6SReid Kleckner return createError("integer expected"); 153146eb7a6SReid Kleckner return Error::success(); 154146eb7a6SReid Kleckner } 155146eb7a6SReid Kleckner 156146eb7a6SReid Kleckner Error expect(Kind Expected, StringRef Msg) { 157146eb7a6SReid Kleckner read(); 158146eb7a6SReid Kleckner if (Tok.K != Expected) 159146eb7a6SReid Kleckner return createError(Msg); 160146eb7a6SReid Kleckner return Error::success(); 161146eb7a6SReid Kleckner } 162146eb7a6SReid Kleckner 163146eb7a6SReid Kleckner void unget() { Stack.push_back(Tok); } 164146eb7a6SReid Kleckner 165146eb7a6SReid Kleckner Error parseOne() { 166146eb7a6SReid Kleckner read(); 167146eb7a6SReid Kleckner switch (Tok.K) { 168146eb7a6SReid Kleckner case Eof: 169146eb7a6SReid Kleckner return Error::success(); 170146eb7a6SReid Kleckner case KwExports: 171146eb7a6SReid Kleckner for (;;) { 172146eb7a6SReid Kleckner read(); 173146eb7a6SReid Kleckner if (Tok.K != Identifier) { 174146eb7a6SReid Kleckner unget(); 175146eb7a6SReid Kleckner return Error::success(); 176146eb7a6SReid Kleckner } 177146eb7a6SReid Kleckner if (Error Err = parseExport()) 178146eb7a6SReid Kleckner return Err; 179146eb7a6SReid Kleckner } 180146eb7a6SReid Kleckner case KwHeapsize: 181146eb7a6SReid Kleckner return parseNumbers(&Info.HeapReserve, &Info.HeapCommit); 182146eb7a6SReid Kleckner case KwStacksize: 183146eb7a6SReid Kleckner return parseNumbers(&Info.StackReserve, &Info.StackCommit); 184146eb7a6SReid Kleckner case KwLibrary: 185146eb7a6SReid Kleckner case KwName: { 186146eb7a6SReid Kleckner bool IsDll = Tok.K == KwLibrary; // Check before parseName. 187146eb7a6SReid Kleckner std::string Name; 188146eb7a6SReid Kleckner if (Error Err = parseName(&Name, &Info.ImageBase)) 189146eb7a6SReid Kleckner return Err; 190146eb7a6SReid Kleckner // Append the appropriate file extension if not already present. 191146eb7a6SReid Kleckner StringRef Ext = IsDll ? ".dll" : ".exe"; 192146eb7a6SReid Kleckner if (!StringRef(Name).endswith_lower(Ext)) 193146eb7a6SReid Kleckner Name += Ext; 194146eb7a6SReid Kleckner 195146eb7a6SReid Kleckner // Set the output file, but don't override /out if it was already passed. 196146eb7a6SReid Kleckner if (Info.OutputFile.empty()) 197146eb7a6SReid Kleckner Info.OutputFile = Name; 198146eb7a6SReid Kleckner return Error::success(); 199146eb7a6SReid Kleckner } 200146eb7a6SReid Kleckner case KwVersion: 201146eb7a6SReid Kleckner return parseVersion(&Info.MajorImageVersion, &Info.MinorImageVersion); 202146eb7a6SReid Kleckner default: 203146eb7a6SReid Kleckner return createError("unknown directive: " + Tok.Value); 204146eb7a6SReid Kleckner } 205146eb7a6SReid Kleckner } 206146eb7a6SReid Kleckner 207146eb7a6SReid Kleckner Error parseExport() { 208146eb7a6SReid Kleckner COFFShortExport E; 209146eb7a6SReid Kleckner E.Name = Tok.Value; 210146eb7a6SReid Kleckner read(); 211146eb7a6SReid Kleckner if (Tok.K == Equal) { 212146eb7a6SReid Kleckner read(); 213146eb7a6SReid Kleckner if (Tok.K != Identifier) 214146eb7a6SReid Kleckner return createError("identifier expected, but got " + Tok.Value); 215146eb7a6SReid Kleckner E.ExtName = E.Name; 216146eb7a6SReid Kleckner E.Name = Tok.Value; 217146eb7a6SReid Kleckner } else { 218146eb7a6SReid Kleckner unget(); 219146eb7a6SReid Kleckner } 220146eb7a6SReid Kleckner 221146eb7a6SReid Kleckner if (Machine == IMAGE_FILE_MACHINE_I386) { 222*1079ef8dSMartell Malone if (!isDecorated(E.Name, MingwDef)) 223146eb7a6SReid Kleckner E.Name = (std::string("_").append(E.Name)); 224*1079ef8dSMartell Malone if (!E.ExtName.empty() && !isDecorated(E.ExtName, MingwDef)) 225146eb7a6SReid Kleckner E.ExtName = (std::string("_").append(E.ExtName)); 226146eb7a6SReid Kleckner } 227146eb7a6SReid Kleckner 228146eb7a6SReid Kleckner for (;;) { 229146eb7a6SReid Kleckner read(); 230146eb7a6SReid Kleckner if (Tok.K == Identifier && Tok.Value[0] == '@') { 231146eb7a6SReid Kleckner Tok.Value.drop_front().getAsInteger(10, E.Ordinal); 232146eb7a6SReid Kleckner read(); 233146eb7a6SReid Kleckner if (Tok.K == KwNoname) { 234146eb7a6SReid Kleckner E.Noname = true; 235146eb7a6SReid Kleckner } else { 236146eb7a6SReid Kleckner unget(); 237146eb7a6SReid Kleckner } 238146eb7a6SReid Kleckner continue; 239146eb7a6SReid Kleckner } 240146eb7a6SReid Kleckner if (Tok.K == KwData) { 241146eb7a6SReid Kleckner E.Data = true; 242146eb7a6SReid Kleckner continue; 243146eb7a6SReid Kleckner } 244146eb7a6SReid Kleckner if (Tok.K == KwConstant) { 245146eb7a6SReid Kleckner E.Constant = true; 246146eb7a6SReid Kleckner continue; 247146eb7a6SReid Kleckner } 248146eb7a6SReid Kleckner if (Tok.K == KwPrivate) { 249146eb7a6SReid Kleckner E.Private = true; 250146eb7a6SReid Kleckner continue; 251146eb7a6SReid Kleckner } 252146eb7a6SReid Kleckner unget(); 253146eb7a6SReid Kleckner Info.Exports.push_back(E); 254146eb7a6SReid Kleckner return Error::success(); 255146eb7a6SReid Kleckner } 256146eb7a6SReid Kleckner } 257146eb7a6SReid Kleckner 258146eb7a6SReid Kleckner // HEAPSIZE/STACKSIZE reserve[,commit] 259146eb7a6SReid Kleckner Error parseNumbers(uint64_t *Reserve, uint64_t *Commit) { 260146eb7a6SReid Kleckner if (Error Err = readAsInt(Reserve)) 261146eb7a6SReid Kleckner return Err; 262146eb7a6SReid Kleckner read(); 263146eb7a6SReid Kleckner if (Tok.K != Comma) { 264146eb7a6SReid Kleckner unget(); 265146eb7a6SReid Kleckner Commit = nullptr; 266146eb7a6SReid Kleckner return Error::success(); 267146eb7a6SReid Kleckner } 268146eb7a6SReid Kleckner if (Error Err = readAsInt(Commit)) 269146eb7a6SReid Kleckner return Err; 270146eb7a6SReid Kleckner return Error::success(); 271146eb7a6SReid Kleckner } 272146eb7a6SReid Kleckner 273146eb7a6SReid Kleckner // NAME outputPath [BASE=address] 274146eb7a6SReid Kleckner Error parseName(std::string *Out, uint64_t *Baseaddr) { 275146eb7a6SReid Kleckner read(); 276146eb7a6SReid Kleckner if (Tok.K == Identifier) { 277146eb7a6SReid Kleckner *Out = Tok.Value; 278146eb7a6SReid Kleckner } else { 279146eb7a6SReid Kleckner *Out = ""; 280146eb7a6SReid Kleckner unget(); 281146eb7a6SReid Kleckner return Error::success(); 282146eb7a6SReid Kleckner } 283146eb7a6SReid Kleckner read(); 284146eb7a6SReid Kleckner if (Tok.K == KwBase) { 285146eb7a6SReid Kleckner if (Error Err = expect(Equal, "'=' expected")) 286146eb7a6SReid Kleckner return Err; 287146eb7a6SReid Kleckner if (Error Err = readAsInt(Baseaddr)) 288146eb7a6SReid Kleckner return Err; 289146eb7a6SReid Kleckner } else { 290146eb7a6SReid Kleckner unget(); 291146eb7a6SReid Kleckner *Baseaddr = 0; 292146eb7a6SReid Kleckner } 293146eb7a6SReid Kleckner return Error::success(); 294146eb7a6SReid Kleckner } 295146eb7a6SReid Kleckner 296146eb7a6SReid Kleckner // VERSION major[.minor] 297146eb7a6SReid Kleckner Error parseVersion(uint32_t *Major, uint32_t *Minor) { 298146eb7a6SReid Kleckner read(); 299146eb7a6SReid Kleckner if (Tok.K != Identifier) 300146eb7a6SReid Kleckner return createError("identifier expected, but got " + Tok.Value); 301146eb7a6SReid Kleckner StringRef V1, V2; 302146eb7a6SReid Kleckner std::tie(V1, V2) = Tok.Value.split('.'); 303146eb7a6SReid Kleckner if (V1.getAsInteger(10, *Major)) 304146eb7a6SReid Kleckner return createError("integer expected, but got " + Tok.Value); 305146eb7a6SReid Kleckner if (V2.empty()) 306146eb7a6SReid Kleckner *Minor = 0; 307146eb7a6SReid Kleckner else if (V2.getAsInteger(10, *Minor)) 308146eb7a6SReid Kleckner return createError("integer expected, but got " + Tok.Value); 309146eb7a6SReid Kleckner return Error::success(); 310146eb7a6SReid Kleckner } 311146eb7a6SReid Kleckner 312146eb7a6SReid Kleckner Lexer Lex; 313146eb7a6SReid Kleckner Token Tok; 314146eb7a6SReid Kleckner std::vector<Token> Stack; 315146eb7a6SReid Kleckner MachineTypes Machine; 316146eb7a6SReid Kleckner COFFModuleDefinition Info; 317*1079ef8dSMartell Malone bool MingwDef; 318146eb7a6SReid Kleckner }; 319146eb7a6SReid Kleckner 320146eb7a6SReid Kleckner Expected<COFFModuleDefinition> parseCOFFModuleDefinition(MemoryBufferRef MB, 321*1079ef8dSMartell Malone MachineTypes Machine, 322*1079ef8dSMartell Malone bool MingwDef) { 323*1079ef8dSMartell Malone return Parser(MB.getBuffer(), Machine, MingwDef).parse(); 324146eb7a6SReid Kleckner } 325146eb7a6SReid Kleckner 326146eb7a6SReid Kleckner } // namespace object 327146eb7a6SReid Kleckner } // namespace llvm 328