Archive.cpp revision dce4a407a24b04eebc6a376f8e62b41aaa7b071f 
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/SmallString.h"
17#include "llvm/ADT/Twine.h"
18#include "llvm/Support/Endian.h"
19#include "llvm/Support/MemoryBuffer.h"
20
21using namespace llvm;
22using namespace object;
23
24static const char *const Magic = "!<arch>\n";
25
26void Archive::anchor() { }
27
28StringRef ArchiveMemberHeader::getName() const {
29  char EndCond;
30  if (Name[0] == '/' || Name[0] == '#')
31    EndCond = ' ';
32  else
33    EndCond = '/';
34  llvm::StringRef::size_type end =
35      llvm::StringRef(Name, sizeof(Name)).find(EndCond);
36  if (end == llvm::StringRef::npos)
37    end = sizeof(Name);
38  assert(end <= sizeof(Name) && end > 0);
39  // Don't include the EndCond if there is one.
40  return llvm::StringRef(Name, end);
41}
42
43uint32_t ArchiveMemberHeader::getSize() const {
44  uint32_t Ret;
45  if (llvm::StringRef(Size, sizeof(Size)).rtrim(" ").getAsInteger(10, Ret))
46    llvm_unreachable("Size is not a decimal number.");
47  return Ret;
48}
49
50sys::fs::perms ArchiveMemberHeader::getAccessMode() const {
51  unsigned Ret;
52  if (StringRef(AccessMode, sizeof(AccessMode)).rtrim(" ").getAsInteger(8, Ret))
53    llvm_unreachable("Access mode is not an octal number.");
54  return static_cast<sys::fs::perms>(Ret);
55}
56
57sys::TimeValue ArchiveMemberHeader::getLastModified() const {
58  unsigned Seconds;
59  if (StringRef(LastModified, sizeof(LastModified)).rtrim(" ")
60          .getAsInteger(10, Seconds))
61    llvm_unreachable("Last modified time not a decimal number.");
62
63  sys::TimeValue Ret;
64  Ret.fromEpochTime(Seconds);
65  return Ret;
66}
67
68unsigned ArchiveMemberHeader::getUID() const {
69  unsigned Ret;
70  if (StringRef(UID, sizeof(UID)).rtrim(" ").getAsInteger(10, Ret))
71    llvm_unreachable("UID time not a decimal number.");
72  return Ret;
73}
74
75unsigned ArchiveMemberHeader::getGID() const {
76  unsigned Ret;
77  if (StringRef(GID, sizeof(GID)).rtrim(" ").getAsInteger(10, Ret))
78    llvm_unreachable("GID time not a decimal number.");
79  return Ret;
80}
81
82Archive::Child::Child(const Archive *Parent, const char *Start)
83    : Parent(Parent) {
84  if (!Start)
85    return;
86
87  const ArchiveMemberHeader *Header =
88      reinterpret_cast<const ArchiveMemberHeader *>(Start);
89  Data = StringRef(Start, sizeof(ArchiveMemberHeader) + Header->getSize());
90
91  // Setup StartOfFile and PaddingBytes.
92  StartOfFile = sizeof(ArchiveMemberHeader);
93  // Don't include attached name.
94  StringRef Name = Header->getName();
95  if (Name.startswith("#1/")) {
96    uint64_t NameSize;
97    if (Name.substr(3).rtrim(" ").getAsInteger(10, NameSize))
98      llvm_unreachable("Long name length is not an integer");
99    StartOfFile += NameSize;
100  }
101}
102
103Archive::Child Archive::Child::getNext() const {
104  size_t SpaceToSkip = Data.size();
105  // If it's odd, add 1 to make it even.
106  if (SpaceToSkip & 1)
107    ++SpaceToSkip;
108
109  const char *NextLoc = Data.data() + SpaceToSkip;
110
111  // Check to see if this is past the end of the archive.
112  if (NextLoc >= Parent->Data->getBufferEnd())
113    return Child(Parent, nullptr);
114
115  return Child(Parent, NextLoc);
116}
117
118error_code Archive::Child::getName(StringRef &Result) const {
119  StringRef name = getRawName();
120  // Check if it's a special name.
121  if (name[0] == '/') {
122    if (name.size() == 1) { // Linker member.
123      Result = name;
124      return object_error::success;
125    }
126    if (name.size() == 2 && name[1] == '/') { // String table.
127      Result = name;
128      return object_error::success;
129    }
130    // It's a long name.
131    // Get the offset.
132    std::size_t offset;
133    if (name.substr(1).rtrim(" ").getAsInteger(10, offset))
134      llvm_unreachable("Long name offset is not an integer");
135    const char *addr = Parent->StringTable->Data.begin()
136                       + sizeof(ArchiveMemberHeader)
137                       + offset;
138    // Verify it.
139    if (Parent->StringTable == Parent->child_end()
140        || addr < (Parent->StringTable->Data.begin()
141                   + sizeof(ArchiveMemberHeader))
142        || addr > (Parent->StringTable->Data.begin()
143                   + sizeof(ArchiveMemberHeader)
144                   + Parent->StringTable->getSize()))
145      return object_error::parse_failed;
146
147    // GNU long file names end with a /.
148    if (Parent->kind() == K_GNU) {
149      StringRef::size_type End = StringRef(addr).find('/');
150      Result = StringRef(addr, End);
151    } else {
152      Result = addr;
153    }
154    return object_error::success;
155  } else if (name.startswith("#1/")) {
156    uint64_t name_size;
157    if (name.substr(3).rtrim(" ").getAsInteger(10, name_size))
158      llvm_unreachable("Long name length is not an ingeter");
159    Result = Data.substr(sizeof(ArchiveMemberHeader), name_size)
160        .rtrim(StringRef("\0", 1));
161    return object_error::success;
162  }
163  // It's a simple name.
164  if (name[name.size() - 1] == '/')
165    Result = name.substr(0, name.size() - 1);
166  else
167    Result = name;
168  return object_error::success;
169}
170
171error_code Archive::Child::getMemoryBuffer(std::unique_ptr<MemoryBuffer> &Result,
172                                           bool FullPath) const {
173  StringRef Name;
174  if (error_code ec = getName(Name))
175    return ec;
176  SmallString<128> Path;
177  Result.reset(MemoryBuffer::getMemBuffer(
178      getBuffer(), FullPath ? (Twine(Parent->getFileName()) + "(" + Name + ")")
179                                  .toStringRef(Path)
180                            : Name,
181      false));
182  return error_code::success();
183}
184
185error_code Archive::Child::getAsBinary(std::unique_ptr<Binary> &Result,
186                                       LLVMContext *Context) const {
187  std::unique_ptr<Binary> ret;
188  std::unique_ptr<MemoryBuffer> Buff;
189  if (error_code ec = getMemoryBuffer(Buff))
190    return ec;
191  ErrorOr<Binary *> BinaryOrErr = createBinary(Buff.release(), Context);
192  if (error_code EC = BinaryOrErr.getError())
193    return EC;
194  Result.reset(BinaryOrErr.get());
195  return object_error::success;
196}
197
198ErrorOr<Archive*> Archive::create(MemoryBuffer *Source) {
199  error_code EC;
200  std::unique_ptr<Archive> Ret(new Archive(Source, EC));
201  if (EC)
202    return EC;
203  return Ret.release();
204}
205
206Archive::Archive(MemoryBuffer *source, error_code &ec)
207  : Binary(Binary::ID_Archive, source), SymbolTable(child_end()) {
208  // Check for sufficient magic.
209  assert(source);
210  if (source->getBufferSize() < 8 ||
211      StringRef(source->getBufferStart(), 8) != Magic) {
212    ec = object_error::invalid_file_type;
213    return;
214  }
215
216  // Get the special members.
217  child_iterator i = child_begin(false);
218  child_iterator e = child_end();
219
220  if (i == e) {
221    ec = object_error::success;
222    return;
223  }
224
225  StringRef Name = i->getRawName();
226
227  // Below is the pattern that is used to figure out the archive format
228  // GNU archive format
229  //  First member : / (may exist, if it exists, points to the symbol table )
230  //  Second member : // (may exist, if it exists, points to the string table)
231  //  Note : The string table is used if the filename exceeds 15 characters
232  // BSD archive format
233  //  First member : __.SYMDEF or "__.SYMDEF SORTED" (the symbol table)
234  //  There is no string table, if the filename exceeds 15 characters or has a
235  //  embedded space, the filename has #1/<size>, The size represents the size
236  //  of the filename that needs to be read after the archive header
237  // COFF archive format
238  //  First member : /
239  //  Second member : / (provides a directory of symbols)
240  //  Third member : // (may exist, if it exists, contains the string table)
241  //  Note: Microsoft PE/COFF Spec 8.3 says that the third member is present
242  //  even if the string table is empty. However, lib.exe does not in fact
243  //  seem to create the third member if there's no member whose filename
244  //  exceeds 15 characters. So the third member is optional.
245
246  if (Name == "__.SYMDEF") {
247    Format = K_BSD;
248    SymbolTable = i;
249    ++i;
250    FirstRegular = i;
251    ec = object_error::success;
252    return;
253  }
254
255  if (Name.startswith("#1/")) {
256    Format = K_BSD;
257    // We know this is BSD, so getName will work since there is no string table.
258    ec = i->getName(Name);
259    if (ec)
260      return;
261    if (Name == "__.SYMDEF SORTED") {
262      SymbolTable = i;
263      ++i;
264    }
265    FirstRegular = i;
266    return;
267  }
268
269  if (Name == "/") {
270    SymbolTable = i;
271
272    ++i;
273    if (i == e) {
274      ec = object_error::parse_failed;
275      return;
276    }
277    Name = i->getRawName();
278  }
279
280  if (Name == "//") {
281    Format = K_GNU;
282    StringTable = i;
283    ++i;
284    FirstRegular = i;
285    ec = object_error::success;
286    return;
287  }
288
289  if (Name[0] != '/') {
290    Format = K_GNU;
291    FirstRegular = i;
292    ec = object_error::success;
293    return;
294  }
295
296  if (Name != "/") {
297    ec = object_error::parse_failed;
298    return;
299  }
300
301  Format = K_COFF;
302  SymbolTable = i;
303
304  ++i;
305  if (i == e) {
306    FirstRegular = i;
307    ec = object_error::success;
308    return;
309  }
310
311  Name = i->getRawName();
312
313  if (Name == "//") {
314    StringTable = i;
315    ++i;
316  }
317
318  FirstRegular = i;
319  ec = object_error::success;
320}
321
322Archive::child_iterator Archive::child_begin(bool SkipInternal) const {
323  if (Data->getBufferSize() == 8) // empty archive.
324    return child_end();
325
326  if (SkipInternal)
327    return FirstRegular;
328
329  const char *Loc = Data->getBufferStart() + strlen(Magic);
330  Child c(this, Loc);
331  return c;
332}
333
334Archive::child_iterator Archive::child_end() const {
335  return Child(this, nullptr);
336}
337
338error_code Archive::Symbol::getName(StringRef &Result) const {
339  Result = StringRef(Parent->SymbolTable->getBuffer().begin() + StringIndex);
340  return object_error::success;
341}
342
343error_code Archive::Symbol::getMember(child_iterator &Result) const {
344  const char *Buf = Parent->SymbolTable->getBuffer().begin();
345  const char *Offsets = Buf + 4;
346  uint32_t Offset = 0;
347  if (Parent->kind() == K_GNU) {
348    Offset = *(reinterpret_cast<const support::ubig32_t*>(Offsets)
349               + SymbolIndex);
350  } else if (Parent->kind() == K_BSD) {
351    llvm_unreachable("BSD format is not supported");
352  } else {
353    uint32_t MemberCount = *reinterpret_cast<const support::ulittle32_t*>(Buf);
354
355    // Skip offsets.
356    Buf += sizeof(support::ulittle32_t)
357           + (MemberCount * sizeof(support::ulittle32_t));
358
359    uint32_t SymbolCount = *reinterpret_cast<const support::ulittle32_t*>(Buf);
360
361    if (SymbolIndex >= SymbolCount)
362      return object_error::parse_failed;
363
364    // Skip SymbolCount to get to the indices table.
365    const char *Indices = Buf + sizeof(support::ulittle32_t);
366
367    // Get the index of the offset in the file member offset table for this
368    // symbol.
369    uint16_t OffsetIndex =
370      *(reinterpret_cast<const support::ulittle16_t*>(Indices)
371        + SymbolIndex);
372    // Subtract 1 since OffsetIndex is 1 based.
373    --OffsetIndex;
374
375    if (OffsetIndex >= MemberCount)
376      return object_error::parse_failed;
377
378    Offset = *(reinterpret_cast<const support::ulittle32_t*>(Offsets)
379               + OffsetIndex);
380  }
381
382  const char *Loc = Parent->getData().begin() + Offset;
383  Result = Child(Parent, Loc);
384
385  return object_error::success;
386}
387
388Archive::Symbol Archive::Symbol::getNext() const {
389  Symbol t(*this);
390  // Go to one past next null.
391  t.StringIndex =
392      Parent->SymbolTable->getBuffer().find('\0', t.StringIndex) + 1;
393  ++t.SymbolIndex;
394  return t;
395}
396
397Archive::symbol_iterator Archive::symbol_begin() const {
398  if (!hasSymbolTable())
399    return symbol_iterator(Symbol(this, 0, 0));
400
401  const char *buf = SymbolTable->getBuffer().begin();
402  if (kind() == K_GNU) {
403    uint32_t symbol_count = 0;
404    symbol_count = *reinterpret_cast<const support::ubig32_t*>(buf);
405    buf += sizeof(uint32_t) + (symbol_count * (sizeof(uint32_t)));
406  } else if (kind() == K_BSD) {
407    llvm_unreachable("BSD archive format is not supported");
408  } else {
409    uint32_t member_count = 0;
410    uint32_t symbol_count = 0;
411    member_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
412    buf += 4 + (member_count * 4); // Skip offsets.
413    symbol_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
414    buf += 4 + (symbol_count * 2); // Skip indices.
415  }
416  uint32_t string_start_offset = buf - SymbolTable->getBuffer().begin();
417  return symbol_iterator(Symbol(this, 0, string_start_offset));
418}
419
420Archive::symbol_iterator Archive::symbol_end() const {
421  if (!hasSymbolTable())
422    return symbol_iterator(Symbol(this, 0, 0));
423
424  const char *buf = SymbolTable->getBuffer().begin();
425  uint32_t symbol_count = 0;
426  if (kind() == K_GNU) {
427    symbol_count = *reinterpret_cast<const support::ubig32_t*>(buf);
428  } else if (kind() == K_BSD) {
429    llvm_unreachable("BSD archive format is not supported");
430  } else {
431    uint32_t member_count = 0;
432    member_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
433    buf += 4 + (member_count * 4); // Skip offsets.
434    symbol_count = *reinterpret_cast<const support::ulittle32_t*>(buf);
435  }
436  return symbol_iterator(
437    Symbol(this, symbol_count, 0));
438}
439
440Archive::child_iterator Archive::findSym(StringRef name) const {
441  Archive::symbol_iterator bs = symbol_begin();
442  Archive::symbol_iterator es = symbol_end();
443  Archive::child_iterator result;
444
445  StringRef symname;
446  for (; bs != es; ++bs) {
447    if (bs->getName(symname))
448        return child_end();
449    if (symname == name) {
450      if (bs->getMember(result))
451        return child_end();
452      return result;
453    }
454  }
455  return child_end();
456}
457
458bool Archive::hasSymbolTable() const {
459  return SymbolTable != child_end();
460}
461