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));
193 sscanf(Hdr->mode, "%o", &(member->info.mode));
194 member->info.user = atoi(Hdr->uid);
195 member->info.group = atoi(Hdr->gid);
196 member->flags = flags;
203 Archive::checkSignature() {
204 // Check the magic string at file's header
205 if (mapfile->size() < 8 || memcmp(base, ARFILE_MAGIC, 8))
206 throw std::string("invalid signature for an archive file");
209 // This function loads the entire archive and fully populates its ilist with
210 // the members of the archive file. This is typically used in preparation for
211 // editing the contents of the archive.
213 Archive::loadArchive() {
218 const char *At = base;
219 const char *End = base + mapfile->size();
222 At += 8; // Skip the magic string.
224 bool seenSymbolTable = false;
225 bool foundFirstFile = false;
227 // parse the member header
228 const char* Save = At;
229 ArchiveMember* mbr = parseMemberHeader(At, End);
231 // check if this is the foreign symbol table
232 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
233 // We just save this but don't do anything special
234 // with it. It doesn't count as the "first file".
236 // What? Multiple foreign symbol tables? Just chuck it
237 // and retain the last one found.
241 At += mbr->getSize();
242 if ((intptr_t(At) & 1) == 1)
244 } else if (mbr->isStringTable()) {
245 // Simply suck the entire string table into a string
246 // variable. This will be used to get the names of the
247 // members that use the "/ddd" format for their names
248 // (SVR4 style long names).
249 strtab.assign(At, mbr->getSize());
250 At += mbr->getSize();
251 if ((intptr_t(At) & 1) == 1)
254 } else if (mbr->isLLVMSymbolTable()) {
255 // This is the LLVM symbol table for the archive. If we've seen it
256 // already, its an error. Otherwise, parse the symbol table and move on.
258 throw std::string("invalid archive: multiple symbol tables");
259 parseSymbolTable(mbr->getData(), mbr->getSize());
260 seenSymbolTable = true;
261 At += mbr->getSize();
262 if ((intptr_t(At) & 1) == 1)
264 delete mbr; // We don't need this member in the list of members.
266 // This is just a regular file. If its the first one, save its offset.
267 // Otherwise just push it on the list and move on to the next file.
268 if (!foundFirstFile) {
269 firstFileOffset = Save - base;
270 foundFirstFile = true;
272 members.push_back(mbr);
273 At += mbr->getSize();
274 if ((intptr_t(At) & 1) == 1)
280 // Open and completely load the archive file.
282 Archive::OpenAndLoad(const sys::Path& file, std::string* ErrorMessage) {
284 std::auto_ptr<Archive> result ( new Archive(file, true));
285 result->loadArchive();
286 return result.release();
287 } catch (const std::string& msg) {
295 // Get all the bytecode modules from the archive
297 Archive::getAllModules(std::vector<Module*>& Modules, std::string* ErrMessage) {
299 for (iterator I=begin(), E=end(); I != E; ++I) {
300 if (I->isBytecode() || I->isCompressedBytecode()) {
301 std::string FullMemberName = archPath.toString() +
302 "(" + I->getPath().toString() + ")";
303 Module* M = ParseBytecodeBuffer((const unsigned char*)I->getData(),
304 I->getSize(), FullMemberName, ErrMessage);
308 Modules.push_back(M);
314 // Load just the symbol table from the archive file
316 Archive::loadSymbolTable() {
321 const char *At = base;
322 const char *End = base + mapfile->size();
324 // Make sure we're dealing with an archive
327 At += 8; // Skip signature
329 // Parse the first file member header
330 const char* FirstFile = At;
331 ArchiveMember* mbr = parseMemberHeader(At, End);
333 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
334 // Skip the foreign symbol table, we don't do anything with it
335 At += mbr->getSize();
336 if ((intptr_t(At) & 1) == 1)
342 mbr = parseMemberHeader(At, End);
345 if (mbr->isStringTable()) {
346 // Process the string table entry
347 strtab.assign((const char*)mbr->getData(), mbr->getSize());
348 At += mbr->getSize();
349 if ((intptr_t(At) & 1) == 1)
354 mbr = parseMemberHeader(At, End);
357 // See if its the symbol table
358 if (mbr->isLLVMSymbolTable()) {
359 parseSymbolTable(mbr->getData(), mbr->getSize());
360 At += mbr->getSize();
361 if ((intptr_t(At) & 1) == 1)
365 // There's no symbol table in the file. We have to rebuild it from scratch
366 // because the intent of this method is to get the symbol table loaded so
367 // it can be searched efficiently.
368 // Add the member to the members list
369 members.push_back(mbr);
372 firstFileOffset = FirstFile - base;
375 // Open the archive and load just the symbol tables
377 Archive::OpenAndLoadSymbols(const sys::Path& file, std::string* ErrorMessage) {
379 std::auto_ptr<Archive> result ( new Archive(file, true) );
380 result->loadSymbolTable();
381 return result.release();
382 } catch (const std::string& msg) {
390 // Look up one symbol in the symbol table and return a ModuleProvider for the
391 // module that defines that symbol.
393 Archive::findModuleDefiningSymbol(const std::string& symbol) {
394 SymTabType::iterator SI = symTab.find(symbol);
395 if (SI == symTab.end())
398 // The symbol table was previously constructed assuming that the members were
399 // written without the symbol table header. Because VBR encoding is used, the
400 // values could not be adjusted to account for the offset of the symbol table
401 // because that could affect the size of the symbol table due to VBR encoding.
402 // We now have to account for this by adjusting the offset by the size of the
403 // symbol table and its header.
404 unsigned fileOffset =
405 SI->second + // offset in symbol-table-less file
406 firstFileOffset; // add offset to first "real" file in archive
408 // See if the module is already loaded
409 ModuleMap::iterator MI = modules.find(fileOffset);
410 if (MI != modules.end())
411 return MI->second.first;
413 // Module hasn't been loaded yet, we need to load it
414 const char* modptr = base + fileOffset;
415 ArchiveMember* mbr = parseMemberHeader(modptr, base + mapfile->size());
417 // Now, load the bytecode module to get the ModuleProvider
418 std::string FullMemberName = archPath.toString() + "(" +
419 mbr->getPath().toString() + ")";
420 ModuleProvider* mp = getBytecodeBufferModuleProvider(
421 (const unsigned char*) mbr->getData(), mbr->getSize(),
424 modules.insert(std::make_pair(fileOffset, std::make_pair(mp, mbr)));
429 // Look up multiple symbols in the symbol table and return a set of
430 // ModuleProviders that define those symbols.
432 Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
433 std::set<ModuleProvider*>& result)
435 assert(mapfile && base && "Can't findModulesDefiningSymbols on new archive");
436 if (symTab.empty()) {
437 // We don't have a symbol table, so we must build it now but lets also
438 // make sure that we populate the modules table as we do this to ensure
439 // that we don't load them twice when findModuleDefiningSymbol is called
442 // Get a pointer to the first file
443 const char* At = ((const char*)base) + firstFileOffset;
444 const char* End = ((const char*)base) + mapfile->size();
447 // Compute the offset to be put in the symbol table
448 unsigned offset = At - base - firstFileOffset;
450 // Parse the file's header
451 ArchiveMember* mbr = parseMemberHeader(At, End);
453 // If it contains symbols
454 if (mbr->isBytecode() || mbr->isCompressedBytecode()) {
456 std::vector<std::string> symbols;
457 std::string FullMemberName = archPath.toString() + "(" +
458 mbr->getPath().toString() + ")";
459 ModuleProvider* MP = GetBytecodeSymbols((const unsigned char*)At,
460 mbr->getSize(), FullMemberName, symbols);
463 // Insert the module's symbols into the symbol table
464 for (std::vector<std::string>::iterator I = symbols.begin(),
465 E=symbols.end(); I != E; ++I ) {
466 symTab.insert(std::make_pair(*I, offset));
468 // Insert the ModuleProvider and the ArchiveMember into the table of
470 modules.insert(std::make_pair(offset, std::make_pair(MP, mbr)));
472 throw std::string("Can't parse bytecode member: ") +
473 mbr->getPath().toString();
477 // Go to the next file location
478 At += mbr->getSize();
479 if ((intptr_t(At) & 1) == 1)
484 // At this point we have a valid symbol table (one way or another) so we
485 // just use it to quickly find the symbols requested.
487 for (std::set<std::string>::iterator I=symbols.begin(),
488 E=symbols.end(); I != E;) {
489 // See if this symbol exists
490 ModuleProvider* mp = findModuleDefiningSymbol(*I);
492 // The symbol exists, insert the ModuleProvider into our result,
493 // duplicates wil be ignored
496 // Remove the symbol now that its been resolved, being careful to
497 // post-increment the iterator.