Archive.cpp revision 36b56886974eae4f9c5ebc96befd3e7bfe5de338 
1//===- Archive.cpp - ar File Format implementation --------------*- 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// This file defines the ArchiveObjectFile class.
11//
12//===----------------------------------------------------------------------===//
13
14#include "llvm/Object/Archive.h"
15#include "llvm/ADT/APInt.h"
16#include "llvm/ADT/OwningPtr.h"
17#include "llvm/ADT/SmallString.h"
18#include "llvm/ADT/Twine.h"
19#include "llvm/Support/Endian.h"
20#include "llvm/Support/MemoryBuffer.h"
21
22using namespace llvm;
23using namespace object;
24
25static const char *const Magic = "!<arch>\n";
26
27void Archive::anchor() { }
28
29StringRef ArchiveMemberHeader::getName() const {
30  char EndCond;
31  if (Name[0] == '/' || Name[0] == '#')
32    EndCond = ' ';
33  else
34    EndCond = '/';
35  llvm::StringRef::size_type end =
36      llvm::StringRef(Name, sizeof(Name)).find(EndCond);
37  if (end == llvm::StringRef::npos)
38    end = sizeof(Name);
39  assert(end <= sizeof(Name) && end > 0);
40  // Don't include the EndCond if there is one.
41  return llvm::StringRef(Name, end);
42}
43
44uint32_t ArchiveMemberHeader::getSize() const {
45  uint32_t Ret;
46  if (llvm::StringRef(Size, sizeof(Size)).rtrim(" ").getAsInteger(10, Ret))
47    llvm_unreachable("Size is not a decimal number.");
48  return Ret;
49}
50
51sys::fs::perms ArchiveMemberHeader::getAccessMode() const {
52  unsigned Ret;
53  if (StringRef(AccessMode, sizeof(AccessMode)).rtrim(" ").getAsInteger(8, Ret))
54    llvm_unreachable("Access mode is not an octal number.");
55  return static_cast<sys::fs::perms>(Ret);
56}
57
58sys::TimeValue ArchiveMemberHeader::getLastModified() const {
59  unsigned Seconds;
60  if (StringRef(LastModified, sizeof(LastModified)).rtrim(" ")
61          .getAsInteger(10, Seconds))
62    llvm_unreachable("Last modified time not a decimal number.");
63
64  sys::TimeValue Ret;
65  Ret.fromEpochTime(Seconds);
66  return Ret;
67}
68
69unsigned ArchiveMemberHeader::getUID() const {
70  unsigned Ret;
71  if (StringRef(UID, sizeof(UID)).rtrim(" ").getAsInteger(10, Ret))
72    llvm_unreachable("UID time not a decimal number.");
73  return Ret;
74}
75
76unsigned ArchiveMemberHeader::getGID() const {
77  unsigned Ret;
78  if (StringRef(GID, sizeof(GID)).rtrim(" ").getAsInteger(10, Ret))
79    llvm_unreachable("GID time not a decimal number.");
80  return Ret;
81}
82
83Archive::Child::Child(const Archive *Parent, const char *Start)
84    : Parent(Parent) {
85  if (!Start)
86    return;
87
88  const ArchiveMemberHeader *Header =
89      reinterpret_cast<const ArchiveMemberHeader *>(Start);
90  Data = StringRef(Start, sizeof(ArchiveMemberHeader) + Header->getSize());
91
92  // Setup StartOfFile and PaddingBytes.
93  StartOfFile = sizeof(ArchiveMemberHeader);
94  // Don't include attached name.
95  StringRef Name = Header->getName();
96  if (Name.startswith("#1/")) {
97    uint64_t NameSize;
98    if (Name.substr(3).rtrim(" ").getAsInteger(10, NameSize))
99      llvm_unreachable("Long name length is not an integer");
100    StartOfFile += NameSize;
101  }
102}
103
104Archive::Child Archive::Child::getNext() const {
105  size_t SpaceToSkip = Data.size();
106  // If it's odd, add 1 to make it even.
107  if (SpaceToSkip & 1)
108    ++SpaceToSkip;
109
110  const char *NextLoc = Data.data() + SpaceToSkip;
111
112  // Check to see if this is past the end of the archive.
113  if (NextLoc >= Parent->Data->getBufferEnd())
114    return Child(Parent, NULL);
115
116  return Child(Parent, NextLoc);
117}
118
119error_code Archive::Child::getName(StringRef &Result) const {
120  StringRef name = getRawName();
121  // Check if it's a special name.
122  if (name[0] == '/') {
123    if (name.size() == 1) { // Linker member.
124      Result = name;
125      return object_error::success;
126    }
127    if (name.size() == 2 && name[1] == '/') { // String table.
128      Result = name;
129      return object_error::success;
130    }
131    // It's a long name.
132    // Get the offset.
133    std::size_t offset;
134    if (name.substr(1).rtrim(" ").getAsInteger(10, offset))
135      llvm_unreachable("Long name offset is not an integer");
136    const char *addr = Parent->StringTable->Data.begin()
137                       + sizeof(ArchiveMemberHeader)
138                       + offset;
139    // Verify it.
140    if (Parent->StringTable == Parent->child_end()
141        || addr < (Parent->StringTable->Data.begin()
142                   + sizeof(ArchiveMemberHeader))
143        || addr > (Parent->StringTable->Data.begin()
144                   + sizeof(ArchiveMemberHeader)
145                   + Parent->StringTable->getSize()))
146      return object_error::parse_failed;
147
148    // GNU long file names end with a /.
149    if (Parent->kind() == K_GNU) {
150      StringRef::size_type End = StringRef(addr).find('/');
151      Result = StringRef(addr, End);
152    } else {
153      Result = addr;
154    }
155    return object_error::success;
156  } else if (name.startswith("#1/")) {
157    uint64_t name_size;
158    if (name.substr(3).rtrim(" ").getAsInteger(10, name_size))
159      llvm_unreachable("Long name length is not an ingeter");
160    Result = Data.substr(sizeof(ArchiveMemberHeader), name_size)
161        .rtrim(StringRef("\0", 1));
162    return object_error::success;
163  }
164  // It's a simple name.
165  if (name[name.size() - 1] == '/')
166    Result = name.substr(0, name.size() - 1);
167  else
168    Result = name;
169  return object_error::success;
170}
171
172error_code Archive::Child::getMemoryBuffer(std::unique_ptr<MemoryBuffer> &Result,
173                                           bool FullPath) const {
174  StringRef Name;
175  if (error_code ec = getName(Name))
176    return ec;
177  SmallString<128> Path;
178  Result.reset(MemoryBuffer::getMemBuffer(
179      getBuffer(), FullPath ? (Twine(Parent->getFileName()) + "(" + Name + ")")
180                                  .toStringRef(Path)
181                            : Name,
182      false));
183  return error_code::success();
184}
185
186error_code Archive::Child::getMemoryBuffer(OwningPtr<MemoryBuffer> &Result,
187                                           bool FullPath) const {
188  std::unique_ptr<MemoryBuffer> MB;
189  error_code ec = getMemoryBuffer(MB, FullPath);
190  Result = std::move(MB);
191  return ec;
192}
193
194error_code Archive::Child::getAsBinary(std::unique_ptr<Binary> &Result,
195                                       LLVMContext *Context) const {
196  std::unique_ptr<Binary> ret;
197  std::unique_ptr<MemoryBuffer> Buff;
198  if (error_code ec = getMemoryBuffer(Buff))
199    return ec;
200  ErrorOr<Binary *> BinaryOrErr = createBinary(Buff.release(), Context);
201  if (error_code EC = BinaryOrErr.getError())
202    return EC;
203  Result.reset(BinaryOrErr.get());
204  return object_error::success;
205}
206
207error_code Archive::Child::getAsBinary(OwningPtr<Binary> &Result,
208                                       LLVMContext *Context) const {
209  std::unique_ptr<Binary> B;
210  error_code ec = getAsBinary(B, Context);
211  Result = std::move(B);
212  return ec;
213}
214
215ErrorOr<Archive*> Archive::create(MemoryBuffer *Source) {
216  error_code EC;
217  std::unique_ptr<Archive> Ret(new Archive(Source, EC));
218  if (EC)
219    return EC;
220  return Ret.release();
221}
222
223Archive::Archive(MemoryBuffer *source, error_code &ec)
224  : Binary(Binary::ID_Archive, source), SymbolTable(child_end()) {
225  // Check for sufficient magic.
226  assert(source);
227  if (source->getBufferSize() < 8 ||
228      StringRef(source->getBufferStart(), 8) != Magic) {
229    ec = object_error::invalid_file_type;
230    return;
231  }
232
233  // Get the special members.
234  child_iterator i = child_begin(false);
235  child_iterator e = child_end();
236
237  if (i == e) {
238    ec = object_error::success;
239    return;
240  }
241
242  StringRef Name = i->getRawName();
243
244  // Below is the pattern that is used to figure out the archive format
245  // GNU archive format
246  //  First member : / (may exist, if it exists, points to the symbol table )
247  //  Second member : // (may exist, if it exists, points to the string table)
248  //  Note : The string table is used if the filename exceeds 15 characters
249  // BSD archive format
250  //  First member : __.SYMDEF or "__.SYMDEF SORTED" (the symbol table)
251  //  There is no string table, if the filename exceeds 15 characters or has a
252  //  embedded space, the filename has #1/<size>, The size represents the size
253  //  of the filename that needs to be read after the archive header
254  // COFF archive format
255  //  First member : /
256  //  Second member : / (provides a directory of symbols)
257  //  Third member : // (may exist, if it exists, contains the string table)
258  //  Note: Microsoft PE/COFF Spec 8.3 says that the third member is present
259  //  even if the string table is empty. However, lib.exe does not in fact
260  //  seem to create the third member if there's no member whose filename
261  //  exceeds 15 characters. So the third member is optional.
262
263  if (Name == "__.SYMDEF") {
264    Format = K_BSD;
265    SymbolTable = i;
266    ++i;
267    FirstRegular = i;
268    ec = object_error::success;
269    return;
270  }
271
272  if (Name.startswith("#1/")) {
273    Format = K_BSD;
274    // We know this is BSD, so getName will work since there is no string table.
275    ec = i->getName(Name);
276    if (ec)
277      return;
278    if (Name == "__.SYMDEF SORTED") {
279      SymbolTable = i;
280      ++i;
281    }
282    FirstRegular = i;
283    return;
284  }
285
286  if (Name == "/") {
287    SymbolTable = i;
288
289    ++i;
290    if (i == e) {
291      ec = object_error::parse_failed;
292      return;
293    }
294    Name = i->getRawName();
295  }
296
297  if (Name == "//") {
298    Format = K_GNU;
299    StringTable = i;
300    ++i;
301    FirstRegular = i;
302    ec = object_error::success;
303    return;
304  }
305
306  if (Name[0] != '/') {
307    Format = K_GNU;
308    FirstRegular = i;
309    ec = object_error::success;
310    return;
311  }
312
313  if (Name != "/") {
314    ec = object_error::parse_failed;
315    return;
316  }
317
318  Format = K_COFF;
319  SymbolTable = i;
320
321  ++i;
322  if (i == e) {
323    FirstRegular = i;
324    ec = object_error::success;
325    return;
326  }
327
328  Name = i->getRawName();
329
330  if (Name == "//") {
331    StringTable = i;
332    ++i;
333  }
334
335  FirstRegular = i;
336  ec = object_error::success;
337}
338
339Archive::child_iterator Archive::child_begin(bool SkipInternal) const {
340  if (Data->getBufferSize() == 8) // empty archive.
341    return child_end();
342
343  if (SkipInternal)
344    return FirstRegular;
345
346  const char *Loc = Data->getBufferStart() + strlen(Magic);
347  Child c(this, Loc);
348  return c;
349}
350
351Archive::child_iterator Archive::child_end() const {
352  return Child(this, NULL);
353}
354
355error_code Archive::Symbol::getName(StringRef &Result) const {
356  Result = StringRef(Parent->SymbolTable->getBuffer().begin() + StringIndex);
357  return object_error::success;
358}
359
360error_code Archive::Symbol::getMember(child_iterator &Result) const {
361  const char *Buf = Parent->SymbolTable->getBuffer().begin();
362  const char *Offsets = Buf + 4;
363  uint32_t Offset = 0;
364  if (Parent->kind() == K_GNU) {
365    Offset = *(reinterpret_cast<const support::ubig32_t*>(Offsets)
366               + SymbolIndex);
367  } else if (Parent->kind() == K_BSD) {
368    llvm_unreachable("BSD format is not supported");
369  } else {
370    uint32_t MemberCount = *reinterpret_cast<const support::ulittle32_t*>(Buf);
371
372    // Skip offsets.
373    Buf += sizeof(support::ulittle32_t)
374           + (MemberCount * sizeof(support::ulittle32_t));
375
376    uint32_t SymbolCount = *reinterpret_cast<const support::ulittle32_t*>(Buf);
377
378    if (SymbolIndex >= SymbolCount)
379      return object_error::parse_failed;
380
381    // Skip SymbolCount to get to the indices table.
382    const char *Indices = Buf + sizeof(support::ulittle32_t);
383
384    // Get the index of the offset in the file member offset table for this
385    // symbol.
386    uint16_t OffsetIndex =
387      *(reinterpret_cast<const support::ulittle16_t*>(Indices)
388        + SymbolIndex);
389    // Subtract 1 since OffsetIndex is 1 based.
390    --OffsetIndex;
391
392    if (OffsetIndex >= MemberCount)
393      return object_error::parse_failed;
394
395    Offset = *(reinterpret_cast<const support::ulittle32_t*>(Offsets)
396               + OffsetIndex);
397  }
398
399  const char *Loc = Parent->getData().begin() + Offset;
400  Result = Child(Parent, Loc);
401
402  return object_error::success;
403}
404
405Archive::Symbol Archive::Symbol::getNext() const {
406  Symbol t(*this);
407  // Go to one past next null.
408  t.StringIndex =
409      Parent->SymbolTable->getBuffer().find('\0', t.StringIndex) + 1;
410  ++t.SymbolIndex;
411  return t;
412}
413
414Archive::symbol_iterator Archive::symbol_begin() const {
415  if (!hasSymbolTable())
416    return symbol_iterator(Symbol(this, 0, 0));
417
418  const char *buf = SymbolTable->getBuffer().begin();
419  if (kind() == K_GNU) {
420    uint32_t symbol_count = 0;
421    symbol_count = *reinterpret_cast<const support::ubig32_t*>(buf);
422    buf += sizeof(uint32_t) + (symbol_count * (sizeof(uint32_t)));
423  } else if (kind() == K_BSD) {
424    llvm_unreachable("BSD archive format is not supported");
425  } else {
426    uint32_t member_count = 0;
427    uint32_t symbol_count = 0;
428    member_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
429    buf += 4 + (member_count * 4); // Skip offsets.
430    symbol_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
431    buf += 4 + (symbol_count * 2); // Skip indices.
432  }
433  uint32_t string_start_offset = buf - SymbolTable->getBuffer().begin();
434  return symbol_iterator(Symbol(this, 0, string_start_offset));
435}
436
437Archive::symbol_iterator Archive::symbol_end() const {
438  if (!hasSymbolTable())
439    return symbol_iterator(Symbol(this, 0, 0));
440
441  const char *buf = SymbolTable->getBuffer().begin();
442  uint32_t symbol_count = 0;
443  if (kind() == K_GNU) {
444    symbol_count = *reinterpret_cast<const support::ubig32_t*>(buf);
445  } else if (kind() == K_BSD) {
446    llvm_unreachable("BSD archive format is not supported");
447  } else {
448    uint32_t member_count = 0;
449    member_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
450    buf += 4 + (member_count * 4); // Skip offsets.
451    symbol_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
452  }
453  return symbol_iterator(
454    Symbol(this, symbol_count, 0));
455}
456
457Archive::child_iterator Archive::findSym(StringRef name) const {
458  Archive::symbol_iterator bs = symbol_begin();
459  Archive::symbol_iterator es = symbol_end();
460  Archive::child_iterator result;
461
462  StringRef symname;
463  for (; bs != es; ++bs) {
464    if (bs->getName(symname))
465        return child_end();
466    if (symname == name) {
467      if (bs->getMember(result))
468        return child_end();
469      return result;
470    }
471  }
472  return child_end();
473}
474
475bool Archive::hasSymbolTable() const {
476  return SymbolTable != child_end();
477}
478