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"
24 /// Read a variable-bit-rate encoded unsigned integer
25 static inline unsigned readInteger(const char*&At, const char*End) {
32 Result |= (unsigned)((*At++) & 0x7F) << Shift;
34 } while (At[-1] & 0x80);
38 // Completely parse the Archive's symbol table and populate symTab member var.
40 Archive::parseSymbolTable(const void* data, unsigned size, std::string* error) {
41 const char* At = (const char*) data;
42 const char* End = At + size;
44 unsigned offset = readInteger(At, End);
47 *error = "Ran out of data reading vbr_uint for symtab offset!";
50 unsigned length = readInteger(At, End);
53 *error = "Ran out of data reading vbr_uint for symtab length!";
56 if (At + length > End) {
58 *error = "Malformed symbol table: length not consistent with size";
61 // we don't care if it can't be inserted (duplicate entry)
62 symTab.insert(std::make_pair(std::string(At, length), offset));
69 // This member parses an ArchiveMemberHeader that is presumed to be pointed to
70 // by At. The At pointer is updated to the byte just after the header, which
71 // can be variable in size.
73 Archive::parseMemberHeader(const char*& At, const char* End, std::string* error)
75 if (At + sizeof(ArchiveMemberHeader) >= End) {
77 *error = "Unexpected end of file";
81 // Cast archive member header
82 const ArchiveMemberHeader* Hdr = (const ArchiveMemberHeader*)At;
83 At += sizeof(ArchiveMemberHeader);
86 int MemberSize = atoi(Hdr->size);
87 assert(MemberSize >= 0);
89 // Check the size of the member for sanity
90 if (At + MemberSize > End) {
92 *error = "invalid member length in archive file";
96 // Check the member signature
97 if (!Hdr->checkSignature()) {
99 *error = "invalid file member signature";
103 // Convert and check the member name
104 // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol
105 // table. The special name "//" and 14 blanks is for a string table, used
106 // for long file names. This library doesn't generate either of those but
107 // it will accept them. If the name starts with #1/ and the remainder is
108 // digits, then those digits specify the length of the name that is
109 // stored immediately following the header. The special name
110 // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bitcode.
111 // Anything else is a regular, short filename that is terminated with
114 std::string pathname;
115 switch (Hdr->name[0]) {
117 if (Hdr->name[1] == '1' && Hdr->name[2] == '/') {
118 if (isdigit(Hdr->name[3])) {
119 unsigned len = atoi(&Hdr->name[3]);
120 const char *nulp = (const char *)memchr(At, '\0', len);
121 pathname.assign(At, nulp != 0 ? (uintptr_t)(nulp - At) : len);
124 flags |= ArchiveMember::HasLongFilenameFlag;
127 *error = "invalid long filename";
130 } else if (Hdr->name[1] == '_' &&
131 (0 == memcmp(Hdr->name, ARFILE_LLVM_SYMTAB_NAME, 16))) {
132 // The member is using a long file name (>15 chars) format.
133 // This format is standard for 4.4BSD and Mac OSX operating
134 // systems. LLVM uses it similarly. In this format, the
135 // remainder of the name field (after #1/) specifies the
136 // length of the file name which occupy the first bytes of
137 // the member's data. The pathname already has the #1/ stripped.
138 pathname.assign(ARFILE_LLVM_SYMTAB_NAME);
139 flags |= ArchiveMember::LLVMSymbolTableFlag;
143 if (Hdr->name[1]== '/') {
144 if (0 == memcmp(Hdr->name, ARFILE_STRTAB_NAME, 16)) {
145 pathname.assign(ARFILE_STRTAB_NAME);
146 flags |= ArchiveMember::StringTableFlag;
149 *error = "invalid string table name";
152 } else if (Hdr->name[1] == ' ') {
153 if (0 == memcmp(Hdr->name, ARFILE_SVR4_SYMTAB_NAME, 16)) {
154 pathname.assign(ARFILE_SVR4_SYMTAB_NAME);
155 flags |= ArchiveMember::SVR4SymbolTableFlag;
158 *error = "invalid SVR4 symbol table name";
161 } else if (isdigit(Hdr->name[1])) {
162 unsigned index = atoi(&Hdr->name[1]);
163 if (index < strtab.length()) {
164 const char* namep = strtab.c_str() + index;
165 const char* endp = strtab.c_str() + strtab.length();
166 const char* p = namep;
167 const char* last_p = p;
169 if (*p == '\n' && *last_p == '/') {
170 pathname.assign(namep, last_p - namep);
171 flags |= ArchiveMember::HasLongFilenameFlag;
179 *error = "missing name termiantor in string table";
184 *error = "name index beyond string table";
190 if (Hdr->name[1] == '_' &&
191 (0 == memcmp(Hdr->name, ARFILE_BSD4_SYMTAB_NAME, 16))) {
192 pathname.assign(ARFILE_BSD4_SYMTAB_NAME);
193 flags |= ArchiveMember::BSD4SymbolTableFlag;
199 const char* slash = (const char*) memchr(Hdr->name, '/', 16);
201 slash = Hdr->name + 16;
202 pathname.assign(Hdr->name, slash - Hdr->name);
206 // Determine if this is a bitcode file
207 switch (sys::IdentifyFileType(At, 4)) {
208 case sys::Bitcode_FileType:
209 flags |= ArchiveMember::BitcodeFlag;
212 flags &= ~ArchiveMember::BitcodeFlag;
216 // Instantiate the ArchiveMember to be filled
217 ArchiveMember* member = new ArchiveMember(this);
219 // Fill in fields of the ArchiveMember
220 member->parent = this;
221 member->path.set(pathname);
222 member->info.fileSize = MemberSize;
223 member->info.modTime.fromEpochTime(atoi(Hdr->date));
225 sscanf(Hdr->mode, "%o", &mode);
226 member->info.mode = mode;
227 member->info.user = atoi(Hdr->uid);
228 member->info.group = atoi(Hdr->gid);
229 member->flags = flags;
236 Archive::checkSignature(std::string* error) {
237 // Check the magic string at file's header
238 if (mapfile->getBufferSize() < 8 || memcmp(base, ARFILE_MAGIC, 8)) {
240 *error = "invalid signature for an archive file";
246 // This function loads the entire archive and fully populates its ilist with
247 // the members of the archive file. This is typically used in preparation for
248 // editing the contents of the archive.
250 Archive::loadArchive(std::string* error) {
255 const char *At = base;
256 const char *End = mapfile->getBufferEnd();
258 if (!checkSignature(error))
261 At += 8; // Skip the magic string.
263 bool seenSymbolTable = false;
264 bool foundFirstFile = false;
266 // parse the member header
267 const char* Save = At;
268 ArchiveMember* mbr = parseMemberHeader(At, End, error);
272 // check if this is the foreign symbol table
273 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
274 // We just save this but don't do anything special
275 // with it. It doesn't count as the "first file".
277 // What? Multiple foreign symbol tables? Just chuck it
278 // and retain the last one found.
282 At += mbr->getSize();
283 if ((intptr_t(At) & 1) == 1)
285 } else if (mbr->isStringTable()) {
286 // Simply suck the entire string table into a string
287 // variable. This will be used to get the names of the
288 // members that use the "/ddd" format for their names
289 // (SVR4 style long names).
290 strtab.assign(At, mbr->getSize());
291 At += mbr->getSize();
292 if ((intptr_t(At) & 1) == 1)
295 } else if (mbr->isLLVMSymbolTable()) {
296 // This is the LLVM symbol table for the archive. If we've seen it
297 // already, its an error. Otherwise, parse the symbol table and move on.
298 if (seenSymbolTable) {
300 *error = "invalid archive: multiple symbol tables";
303 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), error))
305 seenSymbolTable = true;
306 At += mbr->getSize();
307 if ((intptr_t(At) & 1) == 1)
309 delete mbr; // We don't need this member in the list of members.
311 // This is just a regular file. If its the first one, save its offset.
312 // Otherwise just push it on the list and move on to the next file.
313 if (!foundFirstFile) {
314 firstFileOffset = Save - base;
315 foundFirstFile = true;
317 members.push_back(mbr);
318 At += mbr->getSize();
319 if ((intptr_t(At) & 1) == 1)
326 // Open and completely load the archive file.
328 Archive::OpenAndLoad(const sys::Path& file, LLVMContext& C,
329 std::string* ErrorMessage) {
330 std::auto_ptr<Archive> result ( new Archive(file, C));
331 if (result->mapToMemory(ErrorMessage))
333 if (!result->loadArchive(ErrorMessage))
335 return result.release();
338 // Get all the bitcode modules from the archive
340 Archive::getAllModules(std::vector<Module*>& Modules,
341 std::string* ErrMessage) {
343 for (iterator I=begin(), E=end(); I != E; ++I) {
344 if (I->isBitcode()) {
345 std::string FullMemberName = archPath.str() +
346 "(" + I->getPath().str() + ")";
347 MemoryBuffer *Buffer =
348 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
349 FullMemberName.c_str());
351 Module *M = ParseBitcodeFile(Buffer, Context, ErrMessage);
356 Modules.push_back(M);
362 // Load just the symbol table from the archive file
364 Archive::loadSymbolTable(std::string* ErrorMsg) {
369 const char *At = base;
370 const char *End = mapfile->getBufferEnd();
372 // Make sure we're dealing with an archive
373 if (!checkSignature(ErrorMsg))
376 At += 8; // Skip signature
378 // Parse the first file member header
379 const char* FirstFile = At;
380 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
384 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
385 // Skip the foreign symbol table, we don't do anything with it
386 At += mbr->getSize();
387 if ((intptr_t(At) & 1) == 1)
393 mbr = parseMemberHeader(At, End, ErrorMsg);
400 if (mbr->isStringTable()) {
401 // Process the string table entry
402 strtab.assign((const char*)mbr->getData(), mbr->getSize());
403 At += mbr->getSize();
404 if ((intptr_t(At) & 1) == 1)
409 mbr = parseMemberHeader(At, End, ErrorMsg);
416 // See if its the symbol table
417 if (mbr->isLLVMSymbolTable()) {
418 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), ErrorMsg)) {
423 At += mbr->getSize();
424 if ((intptr_t(At) & 1) == 1)
427 // Can't be any more symtab headers so just advance
430 // There's no symbol table in the file. We have to rebuild it from scratch
431 // because the intent of this method is to get the symbol table loaded so
432 // it can be searched efficiently.
433 // Add the member to the members list
434 members.push_back(mbr);
437 firstFileOffset = FirstFile - base;
441 // Open the archive and load just the symbol tables
442 Archive* Archive::OpenAndLoadSymbols(const sys::Path& file,
444 std::string* ErrorMessage) {
445 std::auto_ptr<Archive> result ( new Archive(file, C) );
446 if (result->mapToMemory(ErrorMessage))
448 if (!result->loadSymbolTable(ErrorMessage))
450 return result.release();
453 // Look up one symbol in the symbol table and return the module that defines
456 Archive::findModuleDefiningSymbol(const std::string& symbol,
457 std::string* ErrMsg) {
458 SymTabType::iterator SI = symTab.find(symbol);
459 if (SI == symTab.end())
462 // The symbol table was previously constructed assuming that the members were
463 // written without the symbol table header. Because VBR encoding is used, the
464 // values could not be adjusted to account for the offset of the symbol table
465 // because that could affect the size of the symbol table due to VBR encoding.
466 // We now have to account for this by adjusting the offset by the size of the
467 // symbol table and its header.
468 unsigned fileOffset =
469 SI->second + // offset in symbol-table-less file
470 firstFileOffset; // add offset to first "real" file in archive
472 // See if the module is already loaded
473 ModuleMap::iterator MI = modules.find(fileOffset);
474 if (MI != modules.end())
475 return MI->second.first;
477 // Module hasn't been loaded yet, we need to load it
478 const char* modptr = base + fileOffset;
479 ArchiveMember* mbr = parseMemberHeader(modptr, mapfile->getBufferEnd(),
484 // Now, load the bitcode module to get the Module.
485 std::string FullMemberName = archPath.str() + "(" +
486 mbr->getPath().str() + ")";
487 MemoryBuffer *Buffer =
488 MemoryBuffer::getMemBufferCopy(StringRef(mbr->getData(), mbr->getSize()),
489 FullMemberName.c_str());
491 Module *m = getLazyBitcodeModule(Buffer, Context, ErrMsg);
495 modules.insert(std::make_pair(fileOffset, std::make_pair(m, mbr)));
500 // Look up multiple symbols in the symbol table and return a set of
501 // Modules that define those symbols.
503 Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
504 SmallVectorImpl<Module*>& result,
505 std::string* error) {
506 if (!mapfile || !base) {
508 *error = "Empty archive invalid for finding modules defining symbols";
512 if (symTab.empty()) {
513 // We don't have a symbol table, so we must build it now but lets also
514 // make sure that we populate the modules table as we do this to ensure
515 // that we don't load them twice when findModuleDefiningSymbol is called
518 // Get a pointer to the first file
519 const char* At = base + firstFileOffset;
520 const char* End = mapfile->getBufferEnd();
523 // Compute the offset to be put in the symbol table
524 unsigned offset = At - base - firstFileOffset;
526 // Parse the file's header
527 ArchiveMember* mbr = parseMemberHeader(At, End, error);
531 // If it contains symbols
532 if (mbr->isBitcode()) {
534 std::vector<std::string> symbols;
535 std::string FullMemberName = archPath.str() + "(" +
536 mbr->getPath().str() + ")";
538 GetBitcodeSymbols(At, mbr->getSize(), FullMemberName, Context,
542 // Insert the module's symbols into the symbol table
543 for (std::vector<std::string>::iterator I = symbols.begin(),
544 E=symbols.end(); I != E; ++I ) {
545 symTab.insert(std::make_pair(*I, offset));
547 // Insert the Module and the ArchiveMember into the table of
549 modules.insert(std::make_pair(offset, std::make_pair(M, mbr)));
552 *error = "Can't parse bitcode member: " +
553 mbr->getPath().str() + ": " + *error;
559 // Go to the next file location
560 At += mbr->getSize();
561 if ((intptr_t(At) & 1) == 1)
566 // At this point we have a valid symbol table (one way or another) so we
567 // just use it to quickly find the symbols requested.
569 SmallPtrSet<Module*, 16> Added;
570 for (std::set<std::string>::iterator I=symbols.begin(),
572 E=symbols.end(); I != E; I = Next) {
573 // Increment Next before we invalidate it.
576 // See if this symbol exists
577 Module* m = findModuleDefiningSymbol(*I,error);
580 bool NewMember = Added.insert(m);
584 // The symbol exists, insert the Module into our result.
587 // Remove the symbol now that its been resolved.
593 bool Archive::isBitcodeArchive() {
594 // Make sure the symTab has been loaded. In most cases this should have been
595 // done when the archive was constructed, but still, this is just in case.
597 if (!loadSymbolTable(0))
600 // Now that we know it's been loaded, return true
602 if (symTab.size()) return true;
604 // We still can't be sure it isn't a bitcode archive
608 std::vector<Module *> Modules;
609 std::string ErrorMessage;
611 // Scan the archive, trying to load a bitcode member. We only load one to
612 // see if this works.
613 for (iterator I = begin(), E = end(); I != E; ++I) {
617 std::string FullMemberName =
618 archPath.str() + "(" + I->getPath().str() + ")";
620 MemoryBuffer *Buffer =
621 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
622 FullMemberName.c_str());
623 Module *M = ParseBitcodeFile(Buffer, Context);
626 return false; // Couldn't parse bitcode, not a bitcode archive.