1//===-- ArchiveReader.cpp - Read LLVM archive files -------------*- C++ -*-===//
2//
3// The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// Builds up standard unix archive files (.a) containing LLVM bitcode.
11//
12//===----------------------------------------------------------------------===//
13
14#include "ArchiveInternals.h"
15#include "llvm/Bitcode/ReaderWriter.h"
16#include "llvm/Support/MemoryBuffer.h"
17#include "llvm/Module.h"
18#include <cstdlib>
19#include <memory>
20using namespace llvm;
21
22/// Read a variable-bit-rate encoded unsigned integer
23static inline unsigned readInteger(const char*&At, const char*End) {
24 unsigned Shift = 0;
25 unsigned Result = 0;
26
27 do {
28 if (At == End)
29 return Result;
30 Result |= (unsigned)((*At++) & 0x7F) << Shift;
31 Shift += 7;
32 } while (At[-1] & 0x80);
33 return Result;
34}
35
36// Completely parse the Archive's symbol table and populate symTab member var.
37bool
38Archive::parseSymbolTable(const void* data, unsigned size, std::string* error) {
39 const char* At = (const char*) data;
40 const char* End = At + size;
41 while (At < End) {
42 unsigned offset = readInteger(At, End);
43 if (At == End) {
44 if (error)
45 *error = "Ran out of data reading vbr_uint for symtab offset!";
46 return false;
47 }
48 unsigned length = readInteger(At, End);
49 if (At == End) {
50 if (error)
51 *error = "Ran out of data reading vbr_uint for symtab length!";
52 return false;
53 }
54 if (At + length > End) {
55 if (error)
56 *error = "Malformed symbol table: length not consistent with size";
57 return false;
58 }
59 // we don't care if it can't be inserted (duplicate entry)
60 symTab.insert(std::make_pair(std::string(At, length), offset));
61 At += length;
62 }
63 symTabSize = size;
64 return true;
65}
66
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.
70ArchiveMember*
71Archive::parseMemberHeader(const char*& At, const char* End, std::string* error)
72{
73 if (At + sizeof(ArchiveMemberHeader) >= End) {
74 if (error)
75 *error = "Unexpected end of file";
76 return 0;
77 }
78
79 // Cast archive member header
80 ArchiveMemberHeader* Hdr = (ArchiveMemberHeader*)At;
81 At += sizeof(ArchiveMemberHeader);
82
83 // Extract the size and determine if the file is
84 // compressed or not (negative length).
85 int flags = 0;
86 int MemberSize = atoi(Hdr->size);
87 if (MemberSize < 0) {
88 flags |= ArchiveMember::CompressedFlag;
89 MemberSize = -MemberSize;
90 }
91
92 // Check the size of the member for sanity
93 if (At + MemberSize > End) {
94 if (error)
95 *error = "invalid member length in archive file";
96 return 0;
97 }
98
99 // Check the member signature
100 if (!Hdr->checkSignature()) {
101 if (error)
102 *error = "invalid file member signature";
103 return 0;
104 }
105
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
115 // a '/' and blanks.
116
117 std::string pathname;
118 switch (Hdr->name[0]) {
119 case '#':
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);
124 At += len;
125 MemberSize -= len;
126 flags |= ArchiveMember::HasLongFilenameFlag;
127 } else {
128 if (error)
129 *error = "invalid long filename";
130 return 0;
131 }
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;
142 }
143 break;
144 case '/':
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;
149 } else {
150 if (error)
151 *error = "invalid string table name";
152 return 0;
153 }
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;
158 } else {
159 if (error)
160 *error = "invalid SVR4 symbol table name";
161 return 0;
162 }
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;
170 while (p < endp) {
171 if (*p == '\n' && *last_p == '/') {
172 pathname.assign(namep, last_p - namep);
173 flags |= ArchiveMember::HasLongFilenameFlag;
174 break;
175 }
176 last_p = p;
177 p++;
178 }
179 if (p >= endp) {
180 if (error)
181 *error = "missing name termiantor in string table";
182 return 0;
183 }
184 } else {
185 if (error)
186 *error = "name index beyond string table";
187 return 0;
188 }
189 }
190 break;
191 case '_':
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;
196 break;
197 }
198 /* FALL THROUGH */
199
200 default:
201 char* slash = (char*) memchr(Hdr->name, '/', 16);
202 if (slash == 0)
203 slash = Hdr->name + 16;
204 pathname.assign(Hdr->name, slash - Hdr->name);
205 break;
206 }
207
208 // Determine if this is a bitcode file
209 switch (sys::IdentifyFileType(At, 4)) {
210 case sys::Bitcode_FileType:
211 flags |= ArchiveMember::BitcodeFlag;
212 break;
213 default:
214 flags &= ~ArchiveMember::BitcodeFlag;
215 break;
216 }
217
218 // Instantiate the ArchiveMember to be filled
219 ArchiveMember* member = new ArchiveMember(this);
220
221 // Fill in fields of the ArchiveMember
222 member->parent = this;
223 member->path.set(pathname);
224 member->info.fileSize = MemberSize;
225 member->info.modTime.fromEpochTime(atoi(Hdr->date));
226 unsigned int mode;
227 sscanf(Hdr->mode, "%o", &mode);
228 member->info.mode = mode;
229 member->info.user = atoi(Hdr->uid);
230 member->info.group = atoi(Hdr->gid);
231 member->flags = flags;
232 member->data = At;
233
234 return member;
235}
236
237bool
238Archive::checkSignature(std::string* error) {
239 // Check the magic string at file's header
240 if (mapfile->getBufferSize() < 8 || memcmp(base, ARFILE_MAGIC, 8)) {
241 if (error)
242 *error = "invalid signature for an archive file";
243 return false;
244 }
245 return true;
246}
247
248// This function loads the entire archive and fully populates its ilist with
249// the members of the archive file. This is typically used in preparation for
250// editing the contents of the archive.
251bool
252Archive::loadArchive(std::string* error) {
253
254 // Set up parsing
255 members.clear();
256 symTab.clear();
257 const char *At = base;
258 const char *End = mapfile->getBufferEnd();
259
260 if (!checkSignature(error))
261 return false;
262
263 At += 8; // Skip the magic string.
264
265 bool seenSymbolTable = false;
266 bool foundFirstFile = false;
267 while (At < End) {
268 // parse the member header
269 const char* Save = At;
270 ArchiveMember* mbr = parseMemberHeader(At, End, error);
271 if (!mbr)
272 return false;
273
274 // check if this is the foreign symbol table
275 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
276 // We just save this but don't do anything special
277 // with it. It doesn't count as the "first file".
278 if (foreignST) {
279 // What? Multiple foreign symbol tables? Just chuck it
280 // and retain the last one found.
281 delete foreignST;
282 }
283 foreignST = mbr;
284 At += mbr->getSize();
285 if ((intptr_t(At) & 1) == 1)
286 At++;
287 } else if (mbr->isStringTable()) {
288 // Simply suck the entire string table into a string
289 // variable. This will be used to get the names of the
290 // members that use the "/ddd" format for their names
291 // (SVR4 style long names).
292 strtab.assign(At, mbr->getSize());
293 At += mbr->getSize();
294 if ((intptr_t(At) & 1) == 1)
295 At++;
296 delete mbr;
297 } else if (mbr->isLLVMSymbolTable()) {
298 // This is the LLVM symbol table for the archive. If we've seen it
299 // already, its an error. Otherwise, parse the symbol table and move on.
300 if (seenSymbolTable) {
301 if (error)
302 *error = "invalid archive: multiple symbol tables";
303 return false;
304 }
305 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), error))
306 return false;
307 seenSymbolTable = true;
308 At += mbr->getSize();
309 if ((intptr_t(At) & 1) == 1)
310 At++;
311 delete mbr; // We don't need this member in the list of members.
312 } else {
313 // This is just a regular file. If its the first one, save its offset.
314 // Otherwise just push it on the list and move on to the next file.
315 if (!foundFirstFile) {
316 firstFileOffset = Save - base;
317 foundFirstFile = true;
318 }
319 members.push_back(mbr);
320 At += mbr->getSize();
321 if ((intptr_t(At) & 1) == 1)
322 At++;
323 }
324 }
325 return true;
326}
327
328// Open and completely load the archive file.
329Archive*
330Archive::OpenAndLoad(const sys::Path& file, std::string* ErrorMessage)
331{
332 std::auto_ptr result ( new Archive(file));
333 if (result->mapToMemory(ErrorMessage))
334 return 0;
335 if (!result->loadArchive(ErrorMessage))
336 return 0;
337 return result.release();
338}
339
340// Get all the bitcode modules from the archive
341bool
342Archive::getAllModules(std::vector<Module*>& Modules, std::string* ErrMessage) {
343
344 for (iterator I=begin(), E=end(); I != E; ++I) {
345 if (I->isBitcode()) {
346 std::string FullMemberName = archPath.toString() +
347 "(" + I->getPath().toString() + ")";
348 MemoryBuffer *Buffer =
349 MemoryBuffer::getNewMemBuffer(I->getSize(), FullMemberName.c_str());
350 memcpy((char*)Buffer->getBufferStart(), I->getData(), I->getSize());
351
352 Module *M = ParseBitcodeFile(Buffer, ErrMessage);
353 delete Buffer;
354 if (!M)
355 return true;
356
357 Modules.push_back(M);
358 }
359 }
360 return false;
361}
362
363// Load just the symbol table from the archive file
364bool
365Archive::loadSymbolTable(std::string* ErrorMsg) {
366
367 // Set up parsing
368 members.clear();
369 symTab.clear();
370 const char *At = base;
371 const char *End = mapfile->getBufferEnd();
372
373 // Make sure we're dealing with an archive
374 if (!checkSignature(ErrorMsg))
375 return false;
376
377 At += 8; // Skip signature
378
379 // Parse the first file member header
380 const char* FirstFile = At;
381 ArchiveMember* mbr = parseMemberHeader(At, End, ErrorMsg);
382 if (!mbr)
383 return false;
384
385 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) {
386 // Skip the foreign symbol table, we don't do anything with it
387 At += mbr->getSize();
388 if ((intptr_t(At) & 1) == 1)
389 At++;
390 delete mbr;
391
392 // Read the next one
393 FirstFile = At;
394 mbr = parseMemberHeader(At, End, ErrorMsg);
395 if (!mbr) {
396 delete mbr;
397 return false;
398 }
399 }
400
401 if (mbr->isStringTable()) {
402 // Process the string table entry
403 strtab.assign((const char*)mbr->getData(), mbr->getSize());
404 At += mbr->getSize();
405 if ((intptr_t(At) & 1) == 1)
406 At++;
407 delete mbr;
408 // Get the next one
409 FirstFile = At;
410 mbr = parseMemberHeader(At, End, ErrorMsg);
411 if (!mbr) {
412 delete mbr;
413 return false;
414 }
415 }
416
417 // See if its the symbol table
418 if (mbr->isLLVMSymbolTable()) {
419 if (!parseSymbolTable(mbr->getData(), mbr->getSize(), ErrorMsg)) {
420 delete mbr;
421 return false;
422 }
423
424 At += mbr->getSize();
425 if ((intptr_t(At) & 1) == 1)
426 At++;
427 delete mbr;
428 // Can't be any more symtab headers so just advance
429 FirstFile = At;
430 } else {
431 // There's no symbol table in the file. We have to rebuild it from scratch
432 // because the intent of this method is to get the symbol table loaded so
433 // it can be searched efficiently.
434 // Add the member to the members list
435 members.push_back(mbr);
436 }
437
438 firstFileOffset = FirstFile - base;
439 return true;
440}
441
442// Open the archive and load just the symbol tables
443Archive*
444Archive::OpenAndLoadSymbols(const sys::Path& file, std::string* ErrorMessage) {
445 std::auto_ptr<Archive> result ( new Archive(file) );
446 if (result->mapToMemory(ErrorMessage))
447 return 0;
448 if (!result->loadSymbolTable(ErrorMessage))
449 return 0;
450 return result.release();
451}
452
453// Look up one symbol in the symbol table and return a ModuleProvider for the
454// module that defines that symbol.
455ModuleProvider*
456Archive::findModuleDefiningSymbol(const std::string& symbol,
457 std::string* ErrMsg) {
458 SymTabType::iterator SI = symTab.find(symbol);
459 if (SI == symTab.end())
460 return 0;
461
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
471
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;
476
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(),
480 ErrMsg);
481 if (!mbr)
482 return 0;
483
484 // Now, load the bitcode module to get the ModuleProvider
485 std::string FullMemberName = archPath.toString() + "(" +
486 mbr->getPath().toString() + ")";
487 MemoryBuffer *Buffer =MemoryBuffer::getNewMemBuffer(mbr->getSize(),
488 FullMemberName.c_str());
489 memcpy((char*)Buffer->getBufferStart(), mbr->getData(), mbr->getSize());
490
491 ModuleProvider *mp = getBitcodeModuleProvider(Buffer, ErrMsg);
492 if (!mp)
493 return 0;
494
495 modules.insert(std::make_pair(fileOffset, std::make_pair(mp, mbr)));
496
497 return mp;
498}
499
500// Look up multiple symbols in the symbol table and return a set of
501// ModuleProviders that define those symbols.
502bool
503Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
504 std::set<ModuleProvider*>& result,
505 std::string* error) {
506 if (!mapfile || !base) {
507 if (error)
508 *error = "Empty archive invalid for finding modules defining symbols";
509 return false;
510 }
511
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
516 // below.
517
518 // Get a pointer to the first file
519 const char* At = base + firstFileOffset;
520 const char* End = mapfile->getBufferEnd();
521
522 while ( At < End) {
523 // Compute the offset to be put in the symbol table
524 unsigned offset = At - base - firstFileOffset;
525
526 // Parse the file's header
527 ArchiveMember* mbr = parseMemberHeader(At, End, error);
528 if (!mbr)
529 return false;
530
531 // If it contains symbols
532 if (mbr->isBitcode()) {
533 // Get the symbols
534 std::vector<std::string> symbols;
535 std::string FullMemberName = archPath.toString() + "(" +
536 mbr->getPath().toString() + ")";
537 ModuleProvider* MP =
538 GetBitcodeSymbols((const unsigned char*)At, mbr->getSize(),
539 FullMemberName, symbols, error);
540
541 if (MP) {
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));
546 }
547 // Insert the ModuleProvider and the ArchiveMember into the table of
548 // modules.
549 modules.insert(std::make_pair(offset, std::make_pair(MP, mbr)));
550 } else {
551 if (error)
552 *error = "Can't parse bitcode member: " +
553 mbr->getPath().toString() + ": " + *error;
554 delete mbr;
555 return false;
556 }
557 }
558
559 // Go to the next file location
560 At += mbr->getSize();
561 if ((intptr_t(At) & 1) == 1)
562 At++;
563 }
564 }
565
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.
568
569 for (std::set<std::string>::iterator I=symbols.begin(),
570 E=symbols.end(); I != E;) {
571 // See if this symbol exists
572 ModuleProvider* mp = findModuleDefiningSymbol(*I,error);
573 if (mp) {
574 // The symbol exists, insert the ModuleProvider into our result,
575 // duplicates wil be ignored
576 result.insert(mp);
577
578 // Remove the symbol now that its been resolved, being careful to
579 // post-increment the iterator.
580 symbols.erase(I++);
581 } else {
582 ++I;
583 }
584 }
585 return true;
586}
587
588bool Archive::isBitcodeArchive() {
589 // Make sure the symTab has been loaded. In most cases this should have been
590 // done when the archive was constructed, but still, this is just in case.
591 if (symTab.empty())
592 if (!loadSymbolTable(0))
593 return false;
594
595 // Now that we know it's been loaded, return true
596 // if it has a size
597 if (symTab.size()) return true;
598
599 // We still can't be sure it isn't a bitcode archive
600 if (!loadArchive(0))
601 return false;
602
603 std::vector<Module *> Modules;
604 std::string ErrorMessage;
605
606 // Scan the archive, trying to load a bitcode member. We only load one to
607 // see if this works.
608 for (iterator I = begin(), E = end(); I != E; ++I) {
609 if (!I->isBitcode())
610 continue;
611
612 std::string FullMemberName =
613 archPath.toString() + "(" + I->getPath().toString() + ")";
614
615 MemoryBuffer *Buffer =
616 MemoryBuffer::getNewMemBuffer(I->getSize(), FullMemberName.c_str());
617 memcpy((char*)Buffer->getBufferStart(), I->getData(), I->getSize());
618 Module *M = ParseBitcodeFile(Buffer);
619 delete Buffer;
620 if (!M)
621 return false; // Couldn't parse bitcode, not a bitcode archive.
622 delete M;
623 return true;
624 }
625
626 return false;
627}