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/Bitcode/ReaderWriter.h"
16 #include "llvm/Support/MemoryBuffer.h"
17 #include "llvm/System/MappedFile.h"
18 #include "llvm/Module.h"
22 /// Read a variable-bit-rate encoded unsigned integer
23 inline unsigned readInteger(const char*&At, const char*End){
30 Result |= (unsigned)((*At++) & 0x7F) << Shift;
32 } while (At[-1] & 0x80);
36 // Completely parse the Archive's symbol table and populate symTab member var.
38 Archive::parseSymbolTable(const void* data, unsigned size, std::string* error) {
39 const char* At = (const char*) data;
40 const char* End = At + size;
42 unsigned offset = readInteger(At, End);
45 *error = "Ran out of data reading vbr_uint for symtab offset!";
48 unsigned length = readInteger(At, End);
51 *error = "Ran out of data reading vbr_uint for symtab length!";
54 if (At + length > End) {
56 *error = "Malformed symbol table: length not consistent with size";
59 // we don't care if it can't be inserted (duplicate entry)
60 symTab.insert(std::make_pair(std::string(At, length), offset));
67 // This member parses an ArchiveMemberHeader that is presumed to be pointed to
68 // by At. The At pointer is updated to the byte just after the header, which
69 // can be variable in size.
71 Archive::parseMemberHeader(const char*& At, const char* End, std::string* error)
73 if (At + sizeof(ArchiveMemberHeader) >= End) {
75 *error = "Unexpected end of file";
79 // Cast archive member header
80 ArchiveMemberHeader* Hdr = (ArchiveMemberHeader*)At;
81 At += sizeof(ArchiveMemberHeader);
83 // Extract the size and determine if the file is
84 // compressed or not (negative length).
86 int MemberSize = atoi(Hdr->size);
88 flags |= ArchiveMember::CompressedFlag;
89 MemberSize = -MemberSize;
92 // Check the size of the member for sanity
93 if (At + MemberSize > End) {
95 *error = "invalid member length in archive file";
99 // Check the member signature
100 if (!Hdr->checkSignature()) {
102 *error = "invalid file member signature";
106 // Convert and check the member name
107 // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol
108 // table. The special name "//" and 14 blanks is for a string table, used
109 // for long file names. This library doesn't generate either of those but
110 // it will accept them. If the name starts with #1/ and the remainder is
111 // digits, then those digits specify the length of the name that is
112 // stored immediately following the header. The special name
113 // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bitcode.
114 // Anything else is a regular, short filename that is terminated with
117 std::string pathname;
118 switch (Hdr->name[0]) {
120 if (Hdr->name[1] == '1' && Hdr->name[2] == '/') {
121 if (isdigit(Hdr->name[3])) {
122 unsigned len = atoi(&Hdr->name[3]);
123 pathname.assign(At, len);
126 flags |= ArchiveMember::HasLongFilenameFlag;
129 *error = "invalid long filename";
132 } else if (Hdr->name[1] == '_' &&
133 (0 == memcmp(Hdr->name, ARFILE_LLVM_SYMTAB_NAME, 16))) {
134 // The member is using a long file name (>15 chars) format.
135 // This format is standard for 4.4BSD and Mac OSX operating
136 // systems. LLVM uses it similarly. In this format, the
137 // remainder of the name field (after #1/) specifies the
138 // length of the file name which occupy the first bytes of
139 // the member's data. The pathname already has the #1/ stripped.
140 pathname.assign(ARFILE_LLVM_SYMTAB_NAME);
141 flags |= ArchiveMember::LLVMSymbolTableFlag;
145 if (Hdr->name[1]== '/') {
146 if (0 == memcmp(Hdr->name, ARFILE_STRTAB_NAME, 16)) {
147 pathname.assign(ARFILE_STRTAB_NAME);
148 flags |= ArchiveMember::StringTableFlag;
151 *error = "invalid string table name";
154 } else if (Hdr->name[1] == ' ') {
155 if (0 == memcmp(Hdr->name, ARFILE_SVR4_SYMTAB_NAME, 16)) {
156 pathname.assign(ARFILE_SVR4_SYMTAB_NAME);
157 flags |= ArchiveMember::SVR4SymbolTableFlag;
160 *error = "invalid SVR4 symbol table name";
163 } else if (isdigit(Hdr->name[1])) {
164 unsigned index = atoi(&Hdr->name[1]);
165 if (index < strtab.length()) {
166 const char* namep = strtab.c_str() + index;
167 const char* endp = strtab.c_str() + strtab.length();
168 const char* p = namep;
169 const char* last_p = p;
171 if (*p == '\n' && *last_p == '/') {
172 pathname.assign(namep, last_p - namep);
173 flags |= ArchiveMember::HasLongFilenameFlag;
181 *error = "missing name termiantor in string table";
186 *error = "name index beyond string table";
192 if (Hdr->name[1] == '_' &&
193 (0 == memcmp(Hdr->name, ARFILE_BSD4_SYMTAB_NAME, 16))) {
194 pathname.assign(ARFILE_BSD4_SYMTAB_NAME);
195 flags |= ArchiveMember::BSD4SymbolTableFlag;
201 char* slash = (char*) memchr(Hdr->name, '/', 16);
203 slash = Hdr->name + 16;
204 pathname.assign(Hdr->name, slash - Hdr->name);
208 // Determine if this is a bitcode file
209 switch (sys::IdentifyFileType(At, 4)) {
210 case sys::Bitcode_FileType:
211 flags |= ArchiveMember::BitcodeFlag;
214 flags &= ~ArchiveMember::BitcodeFlag;
218 // Instantiate the ArchiveMember to be filled
219 ArchiveMember* member = new ArchiveMember(this);
221 // 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->size() < 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 = base + mapfile->size();
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, std::string* ErrorMessage)
334 std::auto_ptr<Archive> result ( new Archive(file));
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, std::string* ErrMessage) {
346 for (iterator I=begin(), E=end(); I != E; ++I) {
347 if (I->isBitcode()) {
348 std::string FullMemberName = archPath.toString() +
349 "(" + I->getPath().toString() + ")";
350 MemoryBuffer *Buffer =
351 MemoryBuffer::getNewMemBuffer(I->getSize(), FullMemberName.c_str());
352 memcpy((char*)Buffer->getBufferStart(), I->getData(), I->getSize());
354 Module *M = ParseBitcodeFile(Buffer, ErrMessage);
359 Modules.push_back(M);
365 // Load just the symbol table from the archive file
367 Archive::loadSymbolTable(std::string* ErrorMsg) {
372 const char *At = base;
373 const char *End = base + mapfile->size();
375 // Make sure we're dealing with an archive
376 if (!checkSignature(ErrorMsg))
379 At += 8; // Skip signature
381 // Parse the first file member header
382 const char* FirstFile = At;
383 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
387 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
388 // Skip the foreign symbol table, we don't do anything with it
389 At += mbr->getSize();
390 if ((intptr_t(At) & 1) == 1)
396 mbr = parseMemberHeader(At, End, ErrorMsg);
403 if (mbr->isStringTable()) {
404 // Process the string table entry
405 strtab.assign((const char*)mbr->getData(), mbr->getSize());
406 At += mbr->getSize();
407 if ((intptr_t(At) & 1) == 1)
412 mbr = parseMemberHeader(At, End, ErrorMsg);
419 // See if its the symbol table
420 if (mbr->isLLVMSymbolTable()) {
421 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), ErrorMsg)) {
426 At += mbr->getSize();
427 if ((intptr_t(At) & 1) == 1)
430 // Can't be any more symtab headers so just advance
433 // There's no symbol table in the file. We have to rebuild it from scratch
434 // because the intent of this method is to get the symbol table loaded so
435 // it can be searched efficiently.
436 // Add the member to the members list
437 members.push_back(mbr);
440 firstFileOffset = FirstFile - base;
444 // Open the archive and load just the symbol tables
446 Archive::OpenAndLoadSymbols(const sys::Path& file, std::string* ErrorMessage) {
447 std::auto_ptr<Archive> result ( new Archive(file) );
448 if (result->mapToMemory(ErrorMessage))
450 if (!result->loadSymbolTable(ErrorMessage))
452 return result.release();
455 // Look up one symbol in the symbol table and return a ModuleProvider for the
456 // module that defines that symbol.
458 Archive::findModuleDefiningSymbol(const std::string& symbol,
459 std::string* ErrMsg) {
460 SymTabType::iterator SI = symTab.find(symbol);
461 if (SI == symTab.end())
464 // The symbol table was previously constructed assuming that the members were
465 // written without the symbol table header. Because VBR encoding is used, the
466 // values could not be adjusted to account for the offset of the symbol table
467 // because that could affect the size of the symbol table due to VBR encoding.
468 // We now have to account for this by adjusting the offset by the size of the
469 // symbol table and its header.
470 unsigned fileOffset =
471 SI->second + // offset in symbol-table-less file
472 firstFileOffset; // add offset to first "real" file in archive
474 // See if the module is already loaded
475 ModuleMap::iterator MI = modules.find(fileOffset);
476 if (MI != modules.end())
477 return MI->second.first;
479 // Module hasn't been loaded yet, we need to load it
480 const char* modptr = base + fileOffset;
481 ArchiveMember* mbr = parseMemberHeader(modptr, base + mapfile->size(),ErrMsg);
485 // Now, load the bitcode module to get the ModuleProvider
486 std::string FullMemberName = archPath.toString() + "(" +
487 mbr->getPath().toString() + ")";
488 MemoryBuffer *Buffer =MemoryBuffer::getNewMemBuffer(mbr->getSize(),
489 FullMemberName.c_str());
490 memcpy((char*)Buffer->getBufferStart(), mbr->getData(), mbr->getSize());
492 ModuleProvider *mp = getBitcodeModuleProvider(Buffer, ErrMsg);
496 modules.insert(std::make_pair(fileOffset, std::make_pair(mp, mbr)));
501 // Look up multiple symbols in the symbol table and return a set of
502 // ModuleProviders that define those symbols.
504 Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
505 std::set<ModuleProvider*>& result,
506 std::string* error) {
507 if (!mapfile || !base) {
509 *error = "Empty archive invalid for finding modules defining symbols";
513 if (symTab.empty()) {
514 // We don't have a symbol table, so we must build it now but lets also
515 // make sure that we populate the modules table as we do this to ensure
516 // that we don't load them twice when findModuleDefiningSymbol is called
519 // Get a pointer to the first file
520 const char* At = ((const char*)base) + firstFileOffset;
521 const char* End = ((const char*)base) + mapfile->size();
524 // Compute the offset to be put in the symbol table
525 unsigned offset = At - base - firstFileOffset;
527 // Parse the file's header
528 ArchiveMember* mbr = parseMemberHeader(At, End, error);
532 // If it contains symbols
533 if (mbr->isBitcode()) {
535 std::vector<std::string> symbols;
536 std::string FullMemberName = archPath.toString() + "(" +
537 mbr->getPath().toString() + ")";
539 GetBitcodeSymbols((const unsigned char*)At, mbr->getSize(),
540 FullMemberName, symbols, error);
543 // Insert the module's symbols into the symbol table
544 for (std::vector<std::string>::iterator I = symbols.begin(),
545 E=symbols.end(); I != E; ++I ) {
546 symTab.insert(std::make_pair(*I, offset));
548 // Insert the ModuleProvider and the ArchiveMember into the table of
550 modules.insert(std::make_pair(offset, std::make_pair(MP, mbr)));
553 *error = "Can't parse bitcode member: " +
554 mbr->getPath().toString() + ": " + *error;
560 // Go to the next file location
561 At += mbr->getSize();
562 if ((intptr_t(At) & 1) == 1)
567 // At this point we have a valid symbol table (one way or another) so we
568 // just use it to quickly find the symbols requested.
570 for (std::set<std::string>::iterator I=symbols.begin(),
571 E=symbols.end(); I != E;) {
572 // See if this symbol exists
573 ModuleProvider* mp = findModuleDefiningSymbol(*I,error);
575 // The symbol exists, insert the ModuleProvider into our result,
576 // duplicates wil be ignored
579 // Remove the symbol now that its been resolved, being careful to
580 // post-increment the iterator.
589 bool Archive::isBitcodeArchive() {
590 // Make sure the symTab has been loaded. In most cases this should have been
591 // done when the archive was constructed, but still, this is just in case.
593 if (!loadSymbolTable(0))
596 // Now that we know it's been loaded, return true
598 if (symTab.size()) return true;
600 // We still can't be sure it isn't a bitcode archive
604 std::vector<Module *> Modules;
605 std::string ErrorMessage;
607 // Scan the archive, trying to load a bitcode member. We only load one to
608 // see if this works.
609 for (iterator I = begin(), E = end(); I != E; ++I) {
613 std::string FullMemberName =
614 archPath.toString() + "(" + I->getPath().toString() + ")";
616 MemoryBuffer *Buffer =
617 MemoryBuffer::getNewMemBuffer(I->getSize(), FullMemberName.c_str());
618 memcpy((char*)Buffer->getBufferStart(), I->getData(), I->getSize());
619 Module *M = ParseBitcodeFile(Buffer);
622 return false; // Couldn't parse bitcode, not a bitcode archive.