1 //===-- ArchiveReader.cpp - Read LLVM archive files -------------*- C++ -*-===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // Builds up standard unix archive files (.a) containing LLVM bitcode.
12 //===----------------------------------------------------------------------===//
14 #include "ArchiveInternals.h"
15 #include "llvm/ADT/SmallPtrSet.h"
16 #include "llvm/Bitcode/ReaderWriter.h"
17 #include "llvm/Support/MemoryBuffer.h"
18 #include "llvm/Module.h"
23 /// Read a variable-bit-rate encoded unsigned integer
24 static inline unsigned readInteger(const char*&At, const char*End) {
31 Result |= (unsigned)((*At++) & 0x7F) << Shift;
33 } while (At[-1] & 0x80);
37 // Completely parse the Archive's symbol table and populate symTab member var.
39 Archive::parseSymbolTable(const void* data, unsigned size, std::string* error) {
40 const char* At = (const char*) data;
41 const char* End = At + size;
43 unsigned offset = readInteger(At, End);
46 *error = "Ran out of data reading vbr_uint for symtab offset!";
49 unsigned length = readInteger(At, End);
52 *error = "Ran out of data reading vbr_uint for symtab length!";
55 if (At + length > End) {
57 *error = "Malformed symbol table: length not consistent with size";
60 // we don't care if it can't be inserted (duplicate entry)
61 symTab.insert(std::make_pair(std::string(At, length), offset));
68 // This member parses an ArchiveMemberHeader that is presumed to be pointed to
69 // by At. The At pointer is updated to the byte just after the header, which
70 // can be variable in size.
72 Archive::parseMemberHeader(const char*& At, const char* End, std::string* error)
74 if (At + sizeof(ArchiveMemberHeader) >= End) {
76 *error = "Unexpected end of file";
80 // Cast archive member header
81 ArchiveMemberHeader* Hdr = (ArchiveMemberHeader*)At;
82 At += sizeof(ArchiveMemberHeader);
84 // Extract the size and determine if the file is
85 // compressed or not (negative length).
87 int MemberSize = atoi(Hdr->size);
89 flags |= ArchiveMember::CompressedFlag;
90 MemberSize = -MemberSize;
93 // Check the size of the member for sanity
94 if (At + MemberSize > End) {
96 *error = "invalid member length in archive file";
100 // Check the member signature
101 if (!Hdr->checkSignature()) {
103 *error = "invalid file member signature";
107 // Convert and check the member name
108 // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol
109 // table. The special name "//" and 14 blanks is for a string table, used
110 // for long file names. This library doesn't generate either of those but
111 // it will accept them. If the name starts with #1/ and the remainder is
112 // digits, then those digits specify the length of the name that is
113 // stored immediately following the header. The special name
114 // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bitcode.
115 // Anything else is a regular, short filename that is terminated with
118 std::string pathname;
119 switch (Hdr->name[0]) {
121 if (Hdr->name[1] == '1' && Hdr->name[2] == '/') {
122 if (isdigit(Hdr->name[3])) {
123 unsigned len = atoi(&Hdr->name[3]);
124 const char *nulp = (const char *)memchr(At, '\0', len);
125 pathname.assign(At, nulp != 0 ? (uintptr_t)(nulp - At) : len);
128 flags |= ArchiveMember::HasLongFilenameFlag;
131 *error = "invalid long filename";
134 } else if (Hdr->name[1] == '_' &&
135 (0 == memcmp(Hdr->name, ARFILE_LLVM_SYMTAB_NAME, 16))) {
136 // The member is using a long file name (>15 chars) format.
137 // This format is standard for 4.4BSD and Mac OSX operating
138 // systems. LLVM uses it similarly. In this format, the
139 // remainder of the name field (after #1/) specifies the
140 // length of the file name which occupy the first bytes of
141 // the member's data. The pathname already has the #1/ stripped.
142 pathname.assign(ARFILE_LLVM_SYMTAB_NAME);
143 flags |= ArchiveMember::LLVMSymbolTableFlag;
147 if (Hdr->name[1]== '/') {
148 if (0 == memcmp(Hdr->name, ARFILE_STRTAB_NAME, 16)) {
149 pathname.assign(ARFILE_STRTAB_NAME);
150 flags |= ArchiveMember::StringTableFlag;
153 *error = "invalid string table name";
156 } else if (Hdr->name[1] == ' ') {
157 if (0 == memcmp(Hdr->name, ARFILE_SVR4_SYMTAB_NAME, 16)) {
158 pathname.assign(ARFILE_SVR4_SYMTAB_NAME);
159 flags |= ArchiveMember::SVR4SymbolTableFlag;
162 *error = "invalid SVR4 symbol table name";
165 } else if (isdigit(Hdr->name[1])) {
166 unsigned index = atoi(&Hdr->name[1]);
167 if (index < strtab.length()) {
168 const char* namep = strtab.c_str() + index;
169 const char* endp = strtab.c_str() + strtab.length();
170 const char* p = namep;
171 const char* last_p = p;
173 if (*p == '\n' && *last_p == '/') {
174 pathname.assign(namep, last_p - namep);
175 flags |= ArchiveMember::HasLongFilenameFlag;
183 *error = "missing name termiantor in string table";
188 *error = "name index beyond string table";
194 if (Hdr->name[1] == '_' &&
195 (0 == memcmp(Hdr->name, ARFILE_BSD4_SYMTAB_NAME, 16))) {
196 pathname.assign(ARFILE_BSD4_SYMTAB_NAME);
197 flags |= ArchiveMember::BSD4SymbolTableFlag;
203 char* slash = (char*) memchr(Hdr->name, '/', 16);
205 slash = Hdr->name + 16;
206 pathname.assign(Hdr->name, slash - Hdr->name);
210 // Determine if this is a bitcode file
211 switch (sys::IdentifyFileType(At, 4)) {
212 case sys::Bitcode_FileType:
213 flags |= ArchiveMember::BitcodeFlag;
216 flags &= ~ArchiveMember::BitcodeFlag;
220 // Instantiate the ArchiveMember to be filled
221 ArchiveMember* member = new ArchiveMember(this);
223 // Fill in fields of the ArchiveMember
224 member->parent = this;
225 member->path.set(pathname);
226 member->info.fileSize = MemberSize;
227 member->info.modTime.fromEpochTime(atoi(Hdr->date));
229 sscanf(Hdr->mode, "%o", &mode);
230 member->info.mode = mode;
231 member->info.user = atoi(Hdr->uid);
232 member->info.group = atoi(Hdr->gid);
233 member->flags = flags;
240 Archive::checkSignature(std::string* error) {
241 // Check the magic string at file's header
242 if (mapfile->getBufferSize() < 8 || memcmp(base, ARFILE_MAGIC, 8)) {
244 *error = "invalid signature for an archive file";
250 // This function loads the entire archive and fully populates its ilist with
251 // the members of the archive file. This is typically used in preparation for
252 // editing the contents of the archive.
254 Archive::loadArchive(std::string* error) {
259 const char *At = base;
260 const char *End = mapfile->getBufferEnd();
262 if (!checkSignature(error))
265 At += 8; // Skip the magic string.
267 bool seenSymbolTable = false;
268 bool foundFirstFile = false;
270 // parse the member header
271 const char* Save = At;
272 ArchiveMember* mbr = parseMemberHeader(At, End, error);
276 // check if this is the foreign symbol table
277 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
278 // We just save this but don't do anything special
279 // with it. It doesn't count as the "first file".
281 // What? Multiple foreign symbol tables? Just chuck it
282 // and retain the last one found.
286 At += mbr->getSize();
287 if ((intptr_t(At) & 1) == 1)
289 } else if (mbr->isStringTable()) {
290 // Simply suck the entire string table into a string
291 // variable. This will be used to get the names of the
292 // members that use the "/ddd" format for their names
293 // (SVR4 style long names).
294 strtab.assign(At, mbr->getSize());
295 At += mbr->getSize();
296 if ((intptr_t(At) & 1) == 1)
299 } else if (mbr->isLLVMSymbolTable()) {
300 // This is the LLVM symbol table for the archive. If we've seen it
301 // already, its an error. Otherwise, parse the symbol table and move on.
302 if (seenSymbolTable) {
304 *error = "invalid archive: multiple symbol tables";
307 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), error))
309 seenSymbolTable = true;
310 At += mbr->getSize();
311 if ((intptr_t(At) & 1) == 1)
313 delete mbr; // We don't need this member in the list of members.
315 // This is just a regular file. If its the first one, save its offset.
316 // Otherwise just push it on the list and move on to the next file.
317 if (!foundFirstFile) {
318 firstFileOffset = Save - base;
319 foundFirstFile = true;
321 members.push_back(mbr);
322 At += mbr->getSize();
323 if ((intptr_t(At) & 1) == 1)
330 // Open and completely load the archive file.
332 Archive::OpenAndLoad(const sys::Path& file, LLVMContext& C,
333 std::string* ErrorMessage) {
334 std::auto_ptr<Archive> result ( new Archive(file, C));
335 if (result->mapToMemory(ErrorMessage))
337 if (!result->loadArchive(ErrorMessage))
339 return result.release();
342 // Get all the bitcode modules from the archive
344 Archive::getAllModules(std::vector<Module*>& Modules,
345 std::string* ErrMessage) {
347 for (iterator I=begin(), E=end(); I != E; ++I) {
348 if (I->isBitcode()) {
349 std::string FullMemberName = archPath.str() +
350 "(" + I->getPath().str() + ")";
351 MemoryBuffer *Buffer =
352 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
353 FullMemberName.c_str());
355 Module *M = ParseBitcodeFile(Buffer, Context, ErrMessage);
360 Modules.push_back(M);
366 // Load just the symbol table from the archive file
368 Archive::loadSymbolTable(std::string* ErrorMsg) {
373 const char *At = base;
374 const char *End = mapfile->getBufferEnd();
376 // Make sure we're dealing with an archive
377 if (!checkSignature(ErrorMsg))
380 At += 8; // Skip signature
382 // Parse the first file member header
383 const char* FirstFile = At;
384 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
388 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
389 // Skip the foreign symbol table, we don't do anything with it
390 At += mbr->getSize();
391 if ((intptr_t(At) & 1) == 1)
397 mbr = parseMemberHeader(At, End, ErrorMsg);
404 if (mbr->isStringTable()) {
405 // Process the string table entry
406 strtab.assign((const char*)mbr->getData(), mbr->getSize());
407 At += mbr->getSize();
408 if ((intptr_t(At) & 1) == 1)
413 mbr = parseMemberHeader(At, End, ErrorMsg);
420 // See if its the symbol table
421 if (mbr->isLLVMSymbolTable()) {
422 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), ErrorMsg)) {
427 At += mbr->getSize();
428 if ((intptr_t(At) & 1) == 1)
431 // Can't be any more symtab headers so just advance
434 // There's no symbol table in the file. We have to rebuild it from scratch
435 // because the intent of this method is to get the symbol table loaded so
436 // it can be searched efficiently.
437 // Add the member to the members list
438 members.push_back(mbr);
441 firstFileOffset = FirstFile - base;
445 // Open the archive and load just the symbol tables
446 Archive* Archive::OpenAndLoadSymbols(const sys::Path& file,
448 std::string* ErrorMessage) {
449 std::auto_ptr<Archive> result ( new Archive(file, C) );
450 if (result->mapToMemory(ErrorMessage))
452 if (!result->loadSymbolTable(ErrorMessage))
454 return result.release();
457 // Look up one symbol in the symbol table and return the module that defines
460 Archive::findModuleDefiningSymbol(const std::string& symbol,
461 std::string* ErrMsg) {
462 SymTabType::iterator SI = symTab.find(symbol);
463 if (SI == symTab.end())
466 // The symbol table was previously constructed assuming that the members were
467 // written without the symbol table header. Because VBR encoding is used, the
468 // values could not be adjusted to account for the offset of the symbol table
469 // because that could affect the size of the symbol table due to VBR encoding.
470 // We now have to account for this by adjusting the offset by the size of the
471 // symbol table and its header.
472 unsigned fileOffset =
473 SI->second + // offset in symbol-table-less file
474 firstFileOffset; // add offset to first "real" file in archive
476 // See if the module is already loaded
477 ModuleMap::iterator MI = modules.find(fileOffset);
478 if (MI != modules.end())
479 return MI->second.first;
481 // Module hasn't been loaded yet, we need to load it
482 const char* modptr = base + fileOffset;
483 ArchiveMember* mbr = parseMemberHeader(modptr, mapfile->getBufferEnd(),
488 // Now, load the bitcode module to get the Module.
489 std::string FullMemberName = archPath.str() + "(" +
490 mbr->getPath().str() + ")";
491 MemoryBuffer *Buffer =
492 MemoryBuffer::getMemBufferCopy(StringRef(mbr->getData(), mbr->getSize()),
493 FullMemberName.c_str());
495 Module *m = getLazyBitcodeModule(Buffer, Context, ErrMsg);
499 modules.insert(std::make_pair(fileOffset, std::make_pair(m, mbr)));
504 // Look up multiple symbols in the symbol table and return a set of
505 // Modules that define those symbols.
507 Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
508 SmallVectorImpl<Module*>& result,
509 std::string* error) {
510 if (!mapfile || !base) {
512 *error = "Empty archive invalid for finding modules defining symbols";
516 if (symTab.empty()) {
517 // We don't have a symbol table, so we must build it now but lets also
518 // make sure that we populate the modules table as we do this to ensure
519 // that we don't load them twice when findModuleDefiningSymbol is called
522 // Get a pointer to the first file
523 const char* At = base + firstFileOffset;
524 const char* End = mapfile->getBufferEnd();
527 // Compute the offset to be put in the symbol table
528 unsigned offset = At - base - firstFileOffset;
530 // Parse the file's header
531 ArchiveMember* mbr = parseMemberHeader(At, End, error);
535 // If it contains symbols
536 if (mbr->isBitcode()) {
538 std::vector<std::string> symbols;
539 std::string FullMemberName = archPath.str() + "(" +
540 mbr->getPath().str() + ")";
542 GetBitcodeSymbols(At, mbr->getSize(), FullMemberName, Context,
546 // Insert the module's symbols into the symbol table
547 for (std::vector<std::string>::iterator I = symbols.begin(),
548 E=symbols.end(); I != E; ++I ) {
549 symTab.insert(std::make_pair(*I, offset));
551 // Insert the Module and the ArchiveMember into the table of
553 modules.insert(std::make_pair(offset, std::make_pair(M, mbr)));
556 *error = "Can't parse bitcode member: " +
557 mbr->getPath().str() + ": " + *error;
563 // Go to the next file location
564 At += mbr->getSize();
565 if ((intptr_t(At) & 1) == 1)
570 // At this point we have a valid symbol table (one way or another) so we
571 // just use it to quickly find the symbols requested.
573 SmallPtrSet<Module*, 16> Added;
574 for (std::set<std::string>::iterator I=symbols.begin(),
576 E=symbols.end(); I != E; I = Next) {
577 // Increment Next before we invalidate it.
580 // See if this symbol exists
581 Module* m = findModuleDefiningSymbol(*I,error);
584 bool NewMember = Added.insert(m);
588 // The symbol exists, insert the Module into our result.
591 // Remove the symbol now that its been resolved.
597 bool Archive::isBitcodeArchive() {
598 // Make sure the symTab has been loaded. In most cases this should have been
599 // done when the archive was constructed, but still, this is just in case.
601 if (!loadSymbolTable(0))
604 // Now that we know it's been loaded, return true
606 if (symTab.size()) return true;
608 // We still can't be sure it isn't a bitcode archive
612 std::vector<Module *> Modules;
613 std::string ErrorMessage;
615 // Scan the archive, trying to load a bitcode member. We only load one to
616 // see if this works.
617 for (iterator I = begin(), E = end(); I != E; ++I) {
621 std::string FullMemberName =
622 archPath.str() + "(" + I->getPath().str() + ")";
624 MemoryBuffer *Buffer =
625 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
626 FullMemberName.c_str());
627 Module *M = ParseBitcodeFile(Buffer, Context);
630 return false; // Couldn't parse bitcode, not a bitcode archive.