COFFModuleDefinition.cpp 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368
  1. //===--- COFFModuleDefinition.cpp - Simple DEF parser ---------------------===//
  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. // Windows-specific.
  10. // A parser for the module-definition file (.def file).
  11. //
  12. // The format of module-definition files are described in this document:
  13. // https://msdn.microsoft.com/en-us/library/28d6s79h.aspx
  14. //
  15. //===----------------------------------------------------------------------===//
  16. #include "llvm/Object/COFFModuleDefinition.h"
  17. #include "llvm/ADT/StringRef.h"
  18. #include "llvm/ADT/StringSwitch.h"
  19. #include "llvm/Object/COFF.h"
  20. #include "llvm/Object/COFFImportFile.h"
  21. #include "llvm/Object/Error.h"
  22. #include "llvm/Support/Error.h"
  23. #include "llvm/Support/Path.h"
  24. #include "llvm/Support/raw_ostream.h"
  25. using namespace llvm::COFF;
  26. using namespace llvm;
  27. namespace llvm {
  28. namespace object {
  29. enum Kind {
  30. Unknown,
  31. Eof,
  32. Identifier,
  33. Comma,
  34. Equal,
  35. EqualEqual,
  36. KwBase,
  37. KwConstant,
  38. KwData,
  39. KwExports,
  40. KwHeapsize,
  41. KwLibrary,
  42. KwName,
  43. KwNoname,
  44. KwPrivate,
  45. KwStacksize,
  46. KwVersion,
  47. };
  48. struct Token {
  49. explicit Token(Kind T = Unknown, StringRef S = "") : K(T), Value(S) {}
  50. Kind K;
  51. StringRef Value;
  52. };
  53. static bool isDecorated(StringRef Sym, bool MingwDef) {
  54. // In def files, the symbols can either be listed decorated or undecorated.
  55. //
  56. // - For cdecl symbols, only the undecorated form is allowed.
  57. // - For fastcall and vectorcall symbols, both fully decorated or
  58. // undecorated forms can be present.
  59. // - For stdcall symbols in non-MinGW environments, the decorated form is
  60. // fully decorated with leading underscore and trailing stack argument
  61. // size - like "_Func@0".
  62. // - In MinGW def files, a decorated stdcall symbol does not include the
  63. // leading underscore though, like "Func@0".
  64. // This function controls whether a leading underscore should be added to
  65. // the given symbol name or not. For MinGW, treat a stdcall symbol name such
  66. // as "Func@0" as undecorated, i.e. a leading underscore must be added.
  67. // For non-MinGW, look for '@' in the whole string and consider "_Func@0"
  68. // as decorated, i.e. don't add any more leading underscores.
  69. // We can't check for a leading underscore here, since function names
  70. // themselves can start with an underscore, while a second one still needs
  71. // to be added.
  72. return Sym.startswith("@") || Sym.contains("@@") || Sym.startswith("?") ||
  73. (!MingwDef && Sym.contains('@'));
  74. }
  75. static Error createError(const Twine &Err) {
  76. return make_error<StringError>(StringRef(Err.str()),
  77. object_error::parse_failed);
  78. }
  79. class Lexer {
  80. public:
  81. Lexer(StringRef S) : Buf(S) {}
  82. Token lex() {
  83. Buf = Buf.trim();
  84. if (Buf.empty())
  85. return Token(Eof);
  86. switch (Buf[0]) {
  87. case '\0':
  88. return Token(Eof);
  89. case ';': {
  90. size_t End = Buf.find('\n');
  91. Buf = (End == Buf.npos) ? "" : Buf.drop_front(End);
  92. return lex();
  93. }
  94. case '=':
  95. Buf = Buf.drop_front();
  96. if (Buf.startswith("=")) {
  97. Buf = Buf.drop_front();
  98. return Token(EqualEqual, "==");
  99. }
  100. return Token(Equal, "=");
  101. case ',':
  102. Buf = Buf.drop_front();
  103. return Token(Comma, ",");
  104. case '"': {
  105. StringRef S;
  106. std::tie(S, Buf) = Buf.substr(1).split('"');
  107. return Token(Identifier, S);
  108. }
  109. default: {
  110. size_t End = Buf.find_first_of("=,;\r\n \t\v");
  111. StringRef Word = Buf.substr(0, End);
  112. Kind K = llvm::StringSwitch<Kind>(Word)
  113. .Case("BASE", KwBase)
  114. .Case("CONSTANT", KwConstant)
  115. .Case("DATA", KwData)
  116. .Case("EXPORTS", KwExports)
  117. .Case("HEAPSIZE", KwHeapsize)
  118. .Case("LIBRARY", KwLibrary)
  119. .Case("NAME", KwName)
  120. .Case("NONAME", KwNoname)
  121. .Case("PRIVATE", KwPrivate)
  122. .Case("STACKSIZE", KwStacksize)
  123. .Case("VERSION", KwVersion)
  124. .Default(Identifier);
  125. Buf = (End == Buf.npos) ? "" : Buf.drop_front(End);
  126. return Token(K, Word);
  127. }
  128. }
  129. }
  130. private:
  131. StringRef Buf;
  132. };
  133. class Parser {
  134. public:
  135. explicit Parser(StringRef S, MachineTypes M, bool B)
  136. : Lex(S), Machine(M), MingwDef(B) {}
  137. Expected<COFFModuleDefinition> parse() {
  138. do {
  139. if (Error Err = parseOne())
  140. return std::move(Err);
  141. } while (Tok.K != Eof);
  142. return Info;
  143. }
  144. private:
  145. void read() {
  146. if (Stack.empty()) {
  147. Tok = Lex.lex();
  148. return;
  149. }
  150. Tok = Stack.back();
  151. Stack.pop_back();
  152. }
  153. Error readAsInt(uint64_t *I) {
  154. read();
  155. if (Tok.K != Identifier || Tok.Value.getAsInteger(10, *I))
  156. return createError("integer expected");
  157. return Error::success();
  158. }
  159. Error expect(Kind Expected, StringRef Msg) {
  160. read();
  161. if (Tok.K != Expected)
  162. return createError(Msg);
  163. return Error::success();
  164. }
  165. void unget() { Stack.push_back(Tok); }
  166. Error parseOne() {
  167. read();
  168. switch (Tok.K) {
  169. case Eof:
  170. return Error::success();
  171. case KwExports:
  172. for (;;) {
  173. read();
  174. if (Tok.K != Identifier) {
  175. unget();
  176. return Error::success();
  177. }
  178. if (Error Err = parseExport())
  179. return Err;
  180. }
  181. case KwHeapsize:
  182. return parseNumbers(&Info.HeapReserve, &Info.HeapCommit);
  183. case KwStacksize:
  184. return parseNumbers(&Info.StackReserve, &Info.StackCommit);
  185. case KwLibrary:
  186. case KwName: {
  187. bool IsDll = Tok.K == KwLibrary; // Check before parseName.
  188. std::string Name;
  189. if (Error Err = parseName(&Name, &Info.ImageBase))
  190. return Err;
  191. Info.ImportName = Name;
  192. // Set the output file, but don't override /out if it was already passed.
  193. if (Info.OutputFile.empty()) {
  194. Info.OutputFile = Name;
  195. // Append the appropriate file extension if not already present.
  196. if (!sys::path::has_extension(Name))
  197. Info.OutputFile += IsDll ? ".dll" : ".exe";
  198. }
  199. return Error::success();
  200. }
  201. case KwVersion:
  202. return parseVersion(&Info.MajorImageVersion, &Info.MinorImageVersion);
  203. default:
  204. return createError("unknown directive: " + Tok.Value);
  205. }
  206. }
  207. Error parseExport() {
  208. COFFShortExport E;
  209. E.Name = std::string(Tok.Value);
  210. read();
  211. if (Tok.K == Equal) {
  212. read();
  213. if (Tok.K != Identifier)
  214. return createError("identifier expected, but got " + Tok.Value);
  215. E.ExtName = E.Name;
  216. E.Name = std::string(Tok.Value);
  217. } else {
  218. unget();
  219. }
  220. if (Machine == IMAGE_FILE_MACHINE_I386) {
  221. if (!isDecorated(E.Name, MingwDef))
  222. E.Name = (std::string("_").append(E.Name));
  223. if (!E.ExtName.empty() && !isDecorated(E.ExtName, MingwDef))
  224. E.ExtName = (std::string("_").append(E.ExtName));
  225. }
  226. for (;;) {
  227. read();
  228. if (Tok.K == Identifier && Tok.Value[0] == '@') {
  229. if (Tok.Value == "@") {
  230. // "foo @ 10"
  231. read();
  232. Tok.Value.getAsInteger(10, E.Ordinal);
  233. } else if (Tok.Value.drop_front().getAsInteger(10, E.Ordinal)) {
  234. // "foo \n @bar" - Not an ordinal modifier at all, but the next
  235. // export (fastcall decorated) - complete the current one.
  236. unget();
  237. Info.Exports.push_back(E);
  238. return Error::success();
  239. }
  240. // "foo @10"
  241. read();
  242. if (Tok.K == KwNoname) {
  243. E.Noname = true;
  244. } else {
  245. unget();
  246. }
  247. continue;
  248. }
  249. if (Tok.K == KwData) {
  250. E.Data = true;
  251. continue;
  252. }
  253. if (Tok.K == KwConstant) {
  254. E.Constant = true;
  255. continue;
  256. }
  257. if (Tok.K == KwPrivate) {
  258. E.Private = true;
  259. continue;
  260. }
  261. if (Tok.K == EqualEqual) {
  262. read();
  263. E.AliasTarget = std::string(Tok.Value);
  264. if (Machine == IMAGE_FILE_MACHINE_I386 && !isDecorated(E.AliasTarget, MingwDef))
  265. E.AliasTarget = std::string("_").append(E.AliasTarget);
  266. continue;
  267. }
  268. unget();
  269. Info.Exports.push_back(E);
  270. return Error::success();
  271. }
  272. }
  273. // HEAPSIZE/STACKSIZE reserve[,commit]
  274. Error parseNumbers(uint64_t *Reserve, uint64_t *Commit) {
  275. if (Error Err = readAsInt(Reserve))
  276. return Err;
  277. read();
  278. if (Tok.K != Comma) {
  279. unget();
  280. Commit = nullptr;
  281. return Error::success();
  282. }
  283. if (Error Err = readAsInt(Commit))
  284. return Err;
  285. return Error::success();
  286. }
  287. // NAME outputPath [BASE=address]
  288. Error parseName(std::string *Out, uint64_t *Baseaddr) {
  289. read();
  290. if (Tok.K == Identifier) {
  291. *Out = std::string(Tok.Value);
  292. } else {
  293. *Out = "";
  294. unget();
  295. return Error::success();
  296. }
  297. read();
  298. if (Tok.K == KwBase) {
  299. if (Error Err = expect(Equal, "'=' expected"))
  300. return Err;
  301. if (Error Err = readAsInt(Baseaddr))
  302. return Err;
  303. } else {
  304. unget();
  305. *Baseaddr = 0;
  306. }
  307. return Error::success();
  308. }
  309. // VERSION major[.minor]
  310. Error parseVersion(uint32_t *Major, uint32_t *Minor) {
  311. read();
  312. if (Tok.K != Identifier)
  313. return createError("identifier expected, but got " + Tok.Value);
  314. StringRef V1, V2;
  315. std::tie(V1, V2) = Tok.Value.split('.');
  316. if (V1.getAsInteger(10, *Major))
  317. return createError("integer expected, but got " + Tok.Value);
  318. if (V2.empty())
  319. *Minor = 0;
  320. else if (V2.getAsInteger(10, *Minor))
  321. return createError("integer expected, but got " + Tok.Value);
  322. return Error::success();
  323. }
  324. Lexer Lex;
  325. Token Tok;
  326. std::vector<Token> Stack;
  327. MachineTypes Machine;
  328. COFFModuleDefinition Info;
  329. bool MingwDef;
  330. };
  331. Expected<COFFModuleDefinition> parseCOFFModuleDefinition(MemoryBufferRef MB,
  332. MachineTypes Machine,
  333. bool MingwDef) {
  334. return Parser(MB.getBuffer(), Machine, MingwDef).parse();
  335. }
  336. } // namespace object
  337. } // namespace llvm