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 //===----------------------------------------------------------------------===//
15 #include "ArchiveInternals.h"
16 #include "llvm/ADT/OwningPtr.h"
17 #include "llvm/ADT/SmallPtrSet.h"
18 #include "llvm/Bitcode/ReaderWriter.h"
19 #include "llvm/IR/Module.h"
20 #include "llvm/Support/FileSystem.h"
21 #include "llvm/Support/MemoryBuffer.h"
26 /// Read a variable-bit-rate encoded unsigned integer
27 static inline unsigned readInteger(const char*&At, const char*End) {
34 Result |= (unsigned)((*At++) & 0x7F) << Shift;
36 } while (At[-1] & 0x80);
40 // This member parses an ArchiveMemberHeader that is presumed to be pointed to
41 // by At. The At pointer is updated to the byte just after the header, which
42 // can be variable in size.
44 Archive::parseMemberHeader(const char*& At, const char* End, std::string* error)
46 if (At + sizeof(ArchiveMemberHeader) >= End) {
48 *error = "Unexpected end of file";
52 // Cast archive member header
53 const ArchiveMemberHeader* Hdr = (const ArchiveMemberHeader*)At;
54 At += sizeof(ArchiveMemberHeader);
57 int MemberSize = atoi(Hdr->size);
58 assert(MemberSize >= 0);
60 // Check the size of the member for sanity
61 if (At + MemberSize > End) {
63 *error = "invalid member length in archive file";
67 // Check the member signature
68 if (!Hdr->checkSignature()) {
70 *error = "invalid file member signature";
74 // Convert and check the member name
75 // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol
76 // table. The special name "//" and 14 blanks is for a string table, used
77 // for long file names. This library doesn't generate either of those but
78 // it will accept them. If the name starts with #1/ and the remainder is
79 // digits, then those digits specify the length of the name that is
80 // stored immediately following the header. Anything else is a regular, short
81 // filename that is terminated with a '/' and blanks.
84 switch (Hdr->name[0]) {
86 if (Hdr->name[1] == '1' && Hdr->name[2] == '/') {
87 if (isdigit(Hdr->name[3])) {
88 unsigned len = atoi(&Hdr->name[3]);
89 const char *nulp = (const char *)memchr(At, '\0', len);
90 pathname.assign(At, nulp != 0 ? (uintptr_t)(nulp - At) : len);
93 flags |= ArchiveMember::HasLongFilenameFlag;
96 *error = "invalid long filename";
102 if (Hdr->name[1]== '/') {
103 if (0 == memcmp(Hdr->name, ARFILE_STRTAB_NAME, 16)) {
104 pathname.assign(ARFILE_STRTAB_NAME);
105 flags |= ArchiveMember::StringTableFlag;
108 *error = "invalid string table name";
111 } else if (Hdr->name[1] == ' ') {
112 if (0 == memcmp(Hdr->name, ARFILE_SVR4_SYMTAB_NAME, 16)) {
113 pathname.assign(ARFILE_SVR4_SYMTAB_NAME);
114 flags |= ArchiveMember::SVR4SymbolTableFlag;
117 *error = "invalid SVR4 symbol table name";
120 } else if (isdigit(Hdr->name[1])) {
121 unsigned index = atoi(&Hdr->name[1]);
122 if (index < strtab.length()) {
123 const char* namep = strtab.c_str() + index;
124 const char* endp = strtab.c_str() + strtab.length();
125 const char* p = namep;
126 const char* last_p = p;
128 if (*p == '\n' && *last_p == '/') {
129 pathname.assign(namep, last_p - namep);
130 flags |= ArchiveMember::HasLongFilenameFlag;
138 *error = "missing name terminator in string table";
143 *error = "name index beyond string table";
149 if (Hdr->name[1] == '_' &&
150 (0 == memcmp(Hdr->name, ARFILE_BSD4_SYMTAB_NAME, 16))) {
151 pathname.assign(ARFILE_BSD4_SYMTAB_NAME);
152 flags |= ArchiveMember::BSD4SymbolTableFlag;
158 const char* slash = (const char*) memchr(Hdr->name, '/', 16);
160 slash = Hdr->name + 16;
161 pathname.assign(Hdr->name, slash - Hdr->name);
165 // Determine if this is a bitcode file
166 if (sys::fs::identify_magic(StringRef(At, 4)) ==
167 sys::fs::file_magic::bitcode)
168 flags |= ArchiveMember::BitcodeFlag;
170 flags &= ~ArchiveMember::BitcodeFlag;
172 // Instantiate the ArchiveMember to be filled
173 ArchiveMember* member = new ArchiveMember(this);
175 // Fill in fields of the ArchiveMember
176 member->parent = this;
177 member->path.set(pathname);
178 member->info.fileSize = MemberSize;
179 member->info.modTime.fromEpochTime(atoi(Hdr->date));
181 sscanf(Hdr->mode, "%o", &mode);
182 member->info.mode = mode;
183 member->info.user = atoi(Hdr->uid);
184 member->info.group = atoi(Hdr->gid);
185 member->flags = flags;
192 Archive::checkSignature(std::string* error) {
193 // Check the magic string at file's header
194 if (mapfile->getBufferSize() < 8 || memcmp(base, ARFILE_MAGIC, 8)) {
196 *error = "invalid signature for an archive file";
202 // This function loads the entire archive and fully populates its ilist with
203 // the members of the archive file. This is typically used in preparation for
204 // editing the contents of the archive.
206 Archive::loadArchive(std::string* error) {
211 const char *At = base;
212 const char *End = mapfile->getBufferEnd();
214 if (!checkSignature(error))
217 At += 8; // Skip the magic string.
219 bool foundFirstFile = false;
221 // parse the member header
222 const char* Save = At;
223 ArchiveMember* mbr = parseMemberHeader(At, End, error);
227 // check if this is the foreign symbol table
228 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
229 // We just save this but don't do anything special
230 // with it. It doesn't count as the "first file".
232 // What? Multiple foreign symbol tables? Just chuck it
233 // and retain the last one found.
237 At += mbr->getSize();
238 if ((intptr_t(At) & 1) == 1)
240 } else if (mbr->isStringTable()) {
241 // Simply suck the entire string table into a string
242 // variable. This will be used to get the names of the
243 // members that use the "/ddd" format for their names
244 // (SVR4 style long names).
245 strtab.assign(At, mbr->getSize());
246 At += mbr->getSize();
247 if ((intptr_t(At) & 1) == 1)
251 // This is just a regular file. If its the first one, save its offset.
252 // Otherwise just push it on the list and move on to the next file.
253 if (!foundFirstFile) {
254 firstFileOffset = Save - base;
255 foundFirstFile = true;
257 members.push_back(mbr);
258 At += mbr->getSize();
259 if ((intptr_t(At) & 1) == 1)
266 // Open and completely load the archive file.
268 Archive::OpenAndLoad(const sys::Path& File, LLVMContext& C,
269 std::string* ErrorMessage) {
270 OwningPtr<Archive> result ( new Archive(File, C));
271 if (result->mapToMemory(ErrorMessage))
273 if (!result->loadArchive(ErrorMessage))
275 return result.take();
278 // Get all the bitcode modules from the archive
280 Archive::getAllModules(std::vector<Module*>& Modules,
281 std::string* ErrMessage) {
283 for (iterator I=begin(), E=end(); I != E; ++I) {
284 if (I->isBitcode()) {
285 std::string FullMemberName = archPath.str() +
286 "(" + I->getPath().str() + ")";
287 MemoryBuffer *Buffer =
288 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
289 FullMemberName.c_str());
291 Module *M = ParseBitcodeFile(Buffer, Context, ErrMessage);
296 Modules.push_back(M);
302 // Load just the symbol table from the archive file
304 Archive::loadSymbolTable(std::string* ErrorMsg) {
309 const char *At = base;
310 const char *End = mapfile->getBufferEnd();
312 // Make sure we're dealing with an archive
313 if (!checkSignature(ErrorMsg))
316 At += 8; // Skip signature
318 // Parse the first file member header
319 const char* FirstFile = At;
320 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
324 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
325 // Skip the foreign symbol table, we don't do anything with it
326 At += mbr->getSize();
327 if ((intptr_t(At) & 1) == 1)
333 mbr = parseMemberHeader(At, End, ErrorMsg);
340 if (mbr->isStringTable()) {
341 // Process the string table entry
342 strtab.assign((const char*)mbr->getData(), mbr->getSize());
343 At += mbr->getSize();
344 if ((intptr_t(At) & 1) == 1)
349 mbr = parseMemberHeader(At, End, ErrorMsg);
356 // There's no symbol table in the file. We have to rebuild it from scratch
357 // because the intent of this method is to get the symbol table loaded so
358 // it can be searched efficiently.
359 // Add the member to the members list
360 members.push_back(mbr);
362 firstFileOffset = FirstFile - base;
366 // Open the archive and load just the symbol tables
367 Archive* Archive::OpenAndLoadSymbols(const sys::Path& File,
369 std::string* ErrorMessage) {
370 OwningPtr<Archive> result ( new Archive(File, C) );
371 if (result->mapToMemory(ErrorMessage))
373 if (!result->loadSymbolTable(ErrorMessage))
375 return result.take();
378 // Look up one symbol in the symbol table and return the module that defines
381 Archive::findModuleDefiningSymbol(const std::string& symbol,
382 std::string* ErrMsg) {
383 SymTabType::iterator SI = symTab.find(symbol);
384 if (SI == symTab.end())
387 // The symbol table was previously constructed assuming that the members were
388 // written without the symbol table header. Because VBR encoding is used, the
389 // values could not be adjusted to account for the offset of the symbol table
390 // because that could affect the size of the symbol table due to VBR encoding.
391 // We now have to account for this by adjusting the offset by the size of the
392 // symbol table and its header.
393 unsigned fileOffset =
394 SI->second + // offset in symbol-table-less file
395 firstFileOffset; // add offset to first "real" file in archive
397 // See if the module is already loaded
398 ModuleMap::iterator MI = modules.find(fileOffset);
399 if (MI != modules.end())
400 return MI->second.first;
402 // Module hasn't been loaded yet, we need to load it
403 const char* modptr = base + fileOffset;
404 ArchiveMember* mbr = parseMemberHeader(modptr, mapfile->getBufferEnd(),
409 // Now, load the bitcode module to get the Module.
410 std::string FullMemberName = archPath.str() + "(" +
411 mbr->getPath().str() + ")";
412 MemoryBuffer *Buffer =
413 MemoryBuffer::getMemBufferCopy(StringRef(mbr->getData(), mbr->getSize()),
414 FullMemberName.c_str());
416 Module *m = getLazyBitcodeModule(Buffer, Context, ErrMsg);
420 modules.insert(std::make_pair(fileOffset, std::make_pair(m, mbr)));
425 // Look up multiple symbols in the symbol table and return a set of
426 // Modules that define those symbols.
428 Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
429 SmallVectorImpl<Module*>& result,
430 std::string* error) {
431 if (!mapfile || !base) {
433 *error = "Empty archive invalid for finding modules defining symbols";
437 if (symTab.empty()) {
438 // We don't have a symbol table, so we must build it now but lets also
439 // make sure that we populate the modules table as we do this to ensure
440 // that we don't load them twice when findModuleDefiningSymbol is called
443 // Get a pointer to the first file
444 const char* At = base + firstFileOffset;
445 const char* End = mapfile->getBufferEnd();
448 // Compute the offset to be put in the symbol table
449 unsigned offset = At - base - firstFileOffset;
451 // Parse the file's header
452 ArchiveMember* mbr = parseMemberHeader(At, End, error);
456 // If it contains symbols
457 if (mbr->isBitcode()) {
459 std::vector<std::string> symbols;
460 std::string FullMemberName = archPath.str() + "(" +
461 mbr->getPath().str() + ")";
463 GetBitcodeSymbols(At, mbr->getSize(), FullMemberName, Context,
467 // Insert the module's symbols into the symbol table
468 for (std::vector<std::string>::iterator I = symbols.begin(),
469 E=symbols.end(); I != E; ++I ) {
470 symTab.insert(std::make_pair(*I, offset));
472 // Insert the Module and the ArchiveMember into the table of
474 modules.insert(std::make_pair(offset, std::make_pair(M, mbr)));
477 *error = "Can't parse bitcode member: " +
478 mbr->getPath().str() + ": " + *error;
484 // Go to the next file location
485 At += mbr->getSize();
486 if ((intptr_t(At) & 1) == 1)
491 // At this point we have a valid symbol table (one way or another) so we
492 // just use it to quickly find the symbols requested.
494 SmallPtrSet<Module*, 16> Added;
495 for (std::set<std::string>::iterator I=symbols.begin(),
497 E=symbols.end(); I != E; I = Next) {
498 // Increment Next before we invalidate it.
501 // See if this symbol exists
502 Module* m = findModuleDefiningSymbol(*I,error);
505 bool NewMember = Added.insert(m);
509 // The symbol exists, insert the Module into our result.
512 // Remove the symbol now that its been resolved.
518 bool Archive::isBitcodeArchive() {
519 // Make sure the symTab has been loaded. In most cases this should have been
520 // done when the archive was constructed, but still, this is just in case.
522 if (!loadSymbolTable(0))
525 // Now that we know it's been loaded, return true
527 if (symTab.size()) return true;
529 // We still can't be sure it isn't a bitcode archive
533 std::vector<Module *> Modules;
534 std::string ErrorMessage;
536 // Scan the archive, trying to load a bitcode member. We only load one to
537 // see if this works.
538 for (iterator I = begin(), E = end(); I != E; ++I) {
542 std::string FullMemberName =
543 archPath.str() + "(" + I->getPath().str() + ")";
545 MemoryBuffer *Buffer =
546 MemoryBuffer::getMemBufferCopy(StringRef(I->getData(), I->getSize()),
547 FullMemberName.c_str());
548 Module *M = ParseBitcodeFile(Buffer, Context);
551 return false; // Couldn't parse bitcode, not a bitcode archive.