1 //===-- ArchiveReader.cpp - Read LLVM archive files -------------*- C++ -*-===//
3 // The LLVM Compiler Infrastructure
5 // This file was developed by Reid Spencer and is distributed under the
6 // University of Illinois Open Source License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // Builds up standard unix archive files (.a) containing LLVM bytecode.
12 //===----------------------------------------------------------------------===//
14 #include "ArchiveInternals.h"
15 #include "llvm/Bytecode/Reader.h"
19 /// Read a variable-bit-rate encoded unsigned integer
20 inline unsigned readInteger(const char*&At, const char*End) {
26 throw std::string("Ran out of data reading vbr_uint!");
27 Result |= (unsigned)((*At++) & 0x7F) << Shift;
29 } while (At[-1] & 0x80);
33 // Completely parse the Archive's symbol table and populate symTab member var.
35 Archive::parseSymbolTable(const void* data, unsigned size) {
36 const char* At = (const char*) data;
37 const char* End = At + size;
39 unsigned offset = readInteger(At, End);
40 unsigned length = readInteger(At, End);
41 if (At + length > End)
42 throw std::string("malformed symbol table");
43 // we don't care if it can't be inserted (duplicate entry)
44 symTab.insert(std::make_pair(std::string(At, length), offset));
50 // This member parses an ArchiveMemberHeader that is presumed to be pointed to
51 // by At. The At pointer is updated to the byte just after the header, which
52 // can be variable in size.
54 Archive::parseMemberHeader(const char*& At, const char* End) {
55 assert(At + sizeof(ArchiveMemberHeader) < End && "Not enough data");
57 // Cast archive member header
58 ArchiveMemberHeader* Hdr = (ArchiveMemberHeader*)At;
59 At += sizeof(ArchiveMemberHeader);
61 // Instantiate the ArchiveMember to be filled
62 ArchiveMember* member = new ArchiveMember(this);
64 // Extract the size and determine if the file is
65 // compressed or not (negative length).
67 int MemberSize = atoi(Hdr->size);
69 flags |= ArchiveMember::CompressedFlag;
70 MemberSize = -MemberSize;
73 // Check the size of the member for sanity
74 if (At + MemberSize > End)
75 throw std::string("invalid member length in archive file");
77 // Check the member signature
78 if (!Hdr->checkSignature())
79 throw std::string("invalid file member signature");
81 // Convert and check the member name
82 // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol
83 // table. The special name "//" and 14 blanks is for a string table, used
84 // for long file names. This library doesn't generate either of those but
85 // it will accept them. If the name starts with #1/ and the remainder is
86 // digits, then those digits specify the length of the name that is
87 // stored immediately following the header. The special name
88 // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bytecode.
89 // Anything else is a regular, short filename that is terminated with
93 switch (Hdr->name[0]) {
95 if (Hdr->name[1] == '1' && Hdr->name[2] == '/') {
96 if (isdigit(Hdr->name[3])) {
97 unsigned len = atoi(&Hdr->name[3]);
98 pathname.assign(At, len);
101 flags |= ArchiveMember::HasLongFilenameFlag;
103 throw std::string("invalid long filename");
104 } else if (Hdr->name[1] == '_' &&
105 (0 == memcmp(Hdr->name, ARFILE_LLVM_SYMTAB_NAME, 16))) {
106 // The member is using a long file name (>15 chars) format.
107 // This format is standard for 4.4BSD and Mac OSX operating
108 // systems. LLVM uses it similarly. In this format, the
109 // remainder of the name field (after #1/) specifies the
110 // length of the file name which occupy the first bytes of
111 // the member's data. The pathname already has the #1/ stripped.
112 pathname.assign(ARFILE_LLVM_SYMTAB_NAME);
113 flags |= ArchiveMember::LLVMSymbolTableFlag;
117 if (Hdr->name[1]== '/') {
118 if (0 == memcmp(Hdr->name, ARFILE_STRTAB_NAME, 16)) {
119 pathname.assign(ARFILE_STRTAB_NAME);
120 flags |= ArchiveMember::StringTableFlag;
122 throw std::string("invalid string table name");
124 } else if (Hdr->name[1] == ' ') {
125 if (0 == memcmp(Hdr->name, ARFILE_SVR4_SYMTAB_NAME, 16)) {
126 pathname.assign(ARFILE_SVR4_SYMTAB_NAME);
127 flags |= ArchiveMember::SVR4SymbolTableFlag;
129 throw std::string("invalid SVR4 symbol table name");
131 } else if (isdigit(Hdr->name[1])) {
132 unsigned index = atoi(&Hdr->name[1]);
133 if (index < strtab.length()) {
134 const char* namep = strtab.c_str() + index;
135 const char* endp = strtab.c_str() + strtab.length();
136 const char* p = namep;
137 const char* last_p = p;
139 if (*p == '\n' && *last_p == '/') {
140 pathname.assign(namep, last_p - namep);
141 flags |= ArchiveMember::HasLongFilenameFlag;
148 throw std::string("missing name termiantor in string table");
150 throw std::string("name index beyond string table");
155 if (Hdr->name[1] == '_' &&
156 (0 == memcmp(Hdr->name, ARFILE_BSD4_SYMTAB_NAME, 16))) {
157 pathname.assign(ARFILE_BSD4_SYMTAB_NAME);
158 flags |= ArchiveMember::BSD4SymbolTableFlag;
164 char* slash = (char*) memchr(Hdr->name, '/', 16);
166 slash = Hdr->name + 16;
167 pathname.assign(Hdr->name, slash - Hdr->name);
171 // Determine if this is a bytecode file
172 switch (sys::IdentifyFileType(At, 4)) {
173 case sys::BytecodeFileType:
174 flags |= ArchiveMember::BytecodeFlag;
176 case sys::CompressedBytecodeFileType:
177 flags |= ArchiveMember::CompressedBytecodeFlag;
178 flags &= ~ArchiveMember::CompressedFlag;
181 flags &= ~(ArchiveMember::BytecodeFlag|
182 ArchiveMember::CompressedBytecodeFlag);
186 // Fill in fields of the ArchiveMember
189 member->parent = this;
190 member->path.setFile(pathname);
191 member->info.fileSize = MemberSize;
192 member->info.modTime.fromEpochTime(atoi(Hdr->date));
194 sscanf(Hdr->mode, "%o", &mode);
195 member->info.mode = mode;
196 member->info.user = atoi(Hdr->uid);
197 member->info.group = atoi(Hdr->gid);
198 member->flags = flags;
205 Archive::checkSignature() {
206 // Check the magic string at file's header
207 if (mapfile->size() < 8 || memcmp(base, ARFILE_MAGIC, 8))
208 throw std::string("invalid signature for an archive file");
211 // This function loads the entire archive and fully populates its ilist with
212 // the members of the archive file. This is typically used in preparation for
213 // editing the contents of the archive.
215 Archive::loadArchive() {
220 const char *At = base;
221 const char *End = base + mapfile->size();
224 At += 8; // Skip the magic string.
226 bool seenSymbolTable = false;
227 bool foundFirstFile = false;
229 // parse the member header
230 const char* Save = At;
231 ArchiveMember* mbr = parseMemberHeader(At, End);
233 // check if this is the foreign symbol table
234 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
235 // We just save this but don't do anything special
236 // with it. It doesn't count as the "first file".
238 // What? Multiple foreign symbol tables? Just chuck it
239 // and retain the last one found.
243 At += mbr->getSize();
244 if ((intptr_t(At) & 1) == 1)
246 } else if (mbr->isStringTable()) {
247 // Simply suck the entire string table into a string
248 // variable. This will be used to get the names of the
249 // members that use the "/ddd" format for their names
250 // (SVR4 style long names).
251 strtab.assign(At, mbr->getSize());
252 At += mbr->getSize();
253 if ((intptr_t(At) & 1) == 1)
256 } else if (mbr->isLLVMSymbolTable()) {
257 // This is the LLVM symbol table for the archive. If we've seen it
258 // already, its an error. Otherwise, parse the symbol table and move on.
260 throw std::string("invalid archive: multiple symbol tables");
261 parseSymbolTable(mbr->getData(), mbr->getSize());
262 seenSymbolTable = true;
263 At += mbr->getSize();
264 if ((intptr_t(At) & 1) == 1)
266 delete mbr; // We don't need this member in the list of members.
268 // This is just a regular file. If its the first one, save its offset.
269 // Otherwise just push it on the list and move on to the next file.
270 if (!foundFirstFile) {
271 firstFileOffset = Save - base;
272 foundFirstFile = true;
274 members.push_back(mbr);
275 At += mbr->getSize();
276 if ((intptr_t(At) & 1) == 1)
282 // Open and completely load the archive file.
284 Archive::OpenAndLoad(const sys::Path& file, std::string* ErrorMessage) {
286 std::auto_ptr<Archive> result ( new Archive(file, true));
287 result->loadArchive();
288 return result.release();
289 } catch (const std::string& msg) {
297 // Get all the bytecode modules from the archive
299 Archive::getAllModules(std::vector<Module*>& Modules, std::string* ErrMessage) {
301 for (iterator I=begin(), E=end(); I != E; ++I) {
302 if (I->isBytecode() || I->isCompressedBytecode()) {
303 std::string FullMemberName = archPath.toString() +
304 "(" + I->getPath().toString() + ")";
305 Module* M = ParseBytecodeBuffer((const unsigned char*)I->getData(),
306 I->getSize(), FullMemberName, ErrMessage);
310 Modules.push_back(M);
316 // Load just the symbol table from the archive file
318 Archive::loadSymbolTable() {
323 const char *At = base;
324 const char *End = base + mapfile->size();
326 // Make sure we're dealing with an archive
329 At += 8; // Skip signature
331 // Parse the first file member header
332 const char* FirstFile = At;
333 ArchiveMember* mbr = parseMemberHeader(At, End);
335 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
336 // Skip the foreign symbol table, we don't do anything with it
337 At += mbr->getSize();
338 if ((intptr_t(At) & 1) == 1)
344 mbr = parseMemberHeader(At, End);
347 if (mbr->isStringTable()) {
348 // Process the string table entry
349 strtab.assign((const char*)mbr->getData(), mbr->getSize());
350 At += mbr->getSize();
351 if ((intptr_t(At) & 1) == 1)
356 mbr = parseMemberHeader(At, End);
359 // See if its the symbol table
360 if (mbr->isLLVMSymbolTable()) {
361 parseSymbolTable(mbr->getData(), mbr->getSize());
362 At += mbr->getSize();
363 if ((intptr_t(At) & 1) == 1)
367 // There's no symbol table in the file. We have to rebuild it from scratch
368 // because the intent of this method is to get the symbol table loaded so
369 // it can be searched efficiently.
370 // Add the member to the members list
371 members.push_back(mbr);
374 firstFileOffset = FirstFile - base;
377 // Open the archive and load just the symbol tables
379 Archive::OpenAndLoadSymbols(const sys::Path& file, std::string* ErrorMessage) {
381 std::auto_ptr<Archive> result ( new Archive(file, true) );
382 result->loadSymbolTable();
383 return result.release();
384 } catch (const std::string& msg) {
392 // Look up one symbol in the symbol table and return a ModuleProvider for the
393 // module that defines that symbol.
395 Archive::findModuleDefiningSymbol(const std::string& symbol) {
396 SymTabType::iterator SI = symTab.find(symbol);
397 if (SI == symTab.end())
400 // The symbol table was previously constructed assuming that the members were
401 // written without the symbol table header. Because VBR encoding is used, the
402 // values could not be adjusted to account for the offset of the symbol table
403 // because that could affect the size of the symbol table due to VBR encoding.
404 // We now have to account for this by adjusting the offset by the size of the
405 // symbol table and its header.
406 unsigned fileOffset =
407 SI->second + // offset in symbol-table-less file
408 firstFileOffset; // add offset to first "real" file in archive
410 // See if the module is already loaded
411 ModuleMap::iterator MI = modules.find(fileOffset);
412 if (MI != modules.end())
413 return MI->second.first;
415 // Module hasn't been loaded yet, we need to load it
416 const char* modptr = base + fileOffset;
417 ArchiveMember* mbr = parseMemberHeader(modptr, base + mapfile->size());
419 // Now, load the bytecode module to get the ModuleProvider
420 std::string FullMemberName = archPath.toString() + "(" +
421 mbr->getPath().toString() + ")";
422 ModuleProvider* mp = getBytecodeBufferModuleProvider(
423 (const unsigned char*) mbr->getData(), mbr->getSize(),
426 modules.insert(std::make_pair(fileOffset, std::make_pair(mp, mbr)));
431 // Look up multiple symbols in the symbol table and return a set of
432 // ModuleProviders that define those symbols.
434 Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
435 std::set<ModuleProvider*>& result)
437 assert(mapfile && base && "Can't findModulesDefiningSymbols on new archive");
438 if (symTab.empty()) {
439 // We don't have a symbol table, so we must build it now but lets also
440 // make sure that we populate the modules table as we do this to ensure
441 // that we don't load them twice when findModuleDefiningSymbol is called
444 // Get a pointer to the first file
445 const char* At = ((const char*)base) + firstFileOffset;
446 const char* End = ((const char*)base) + mapfile->size();
449 // Compute the offset to be put in the symbol table
450 unsigned offset = At - base - firstFileOffset;
452 // Parse the file's header
453 ArchiveMember* mbr = parseMemberHeader(At, End);
455 // If it contains symbols
456 if (mbr->isBytecode() || mbr->isCompressedBytecode()) {
458 std::vector<std::string> symbols;
459 std::string FullMemberName = archPath.toString() + "(" +
460 mbr->getPath().toString() + ")";
461 ModuleProvider* MP = GetBytecodeSymbols((const unsigned char*)At,
462 mbr->getSize(), FullMemberName, symbols);
465 // Insert the module's symbols into the symbol table
466 for (std::vector<std::string>::iterator I = symbols.begin(),
467 E=symbols.end(); I != E; ++I ) {
468 symTab.insert(std::make_pair(*I, offset));
470 // Insert the ModuleProvider and the ArchiveMember into the table of
472 modules.insert(std::make_pair(offset, std::make_pair(MP, mbr)));
474 throw std::string("Can't parse bytecode member: ") +
475 mbr->getPath().toString();
479 // Go to the next file location
480 At += mbr->getSize();
481 if ((intptr_t(At) & 1) == 1)
486 // At this point we have a valid symbol table (one way or another) so we
487 // just use it to quickly find the symbols requested.
489 for (std::set<std::string>::iterator I=symbols.begin(),
490 E=symbols.end(); I != E;) {
491 // See if this symbol exists
492 ModuleProvider* mp = findModuleDefiningSymbol(*I);
494 // The symbol exists, insert the ModuleProvider into our result,
495 // duplicates wil be ignored
498 // Remove the symbol now that its been resolved, being careful to
499 // post-increment the iterator.
508 Archive::isBytecodeArchive()
510 //Make sure the symTab has been loaded...
511 //in most cases this should have been done
512 //when the archive was constructed, but still,
513 //this is just in case.
514 if ( !symTab.size() )
517 //Now that we know it's been loaded, return true
519 if ( symTab.size() ) return true;
521 //We still can't be sure it isn't a bytecode archive
524 std::vector<Module *> Modules;
525 std::string ErrorMessage;
527 //If getAllModules gives an error then this isn't a proper
529 if ( getAllModules( Modules, &ErrorMessage ) ) return false;
531 //Finally, if we find any bytecode modules then this is a proper
533 return Modules.size();