1//===- ArchiveWriter.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 writeArchive function.
11//
12//===----------------------------------------------------------------------===//
13
14#include "llvm/Object/ArchiveWriter.h"
15#include "llvm/ADT/ArrayRef.h"
16#include "llvm/ADT/StringRef.h"
17#include "llvm/IR/LLVMContext.h"
18#include "llvm/Object/Archive.h"
19#include "llvm/Object/ObjectFile.h"
20#include "llvm/Object/SymbolicFile.h"
21#include "llvm/Support/EndianStream.h"
22#include "llvm/Support/Errc.h"
23#include "llvm/Support/ErrorHandling.h"
24#include "llvm/Support/Format.h"
25#include "llvm/Support/Path.h"
26#include "llvm/Support/ToolOutputFile.h"
27#include "llvm/Support/raw_ostream.h"
28
29#if !defined(_MSC_VER) && !defined(__MINGW32__)
30#include <unistd.h>
31#else
32#include <io.h>
33#endif
34
35using namespace llvm;
36
37NewArchiveMember::NewArchiveMember(MemoryBufferRef BufRef)
38    : Buf(MemoryBuffer::getMemBuffer(BufRef, false)) {}
39
40Expected<NewArchiveMember>
41NewArchiveMember::getOldMember(const object::Archive::Child &OldMember,
42                               bool Deterministic) {
43  ErrorOr<llvm::MemoryBufferRef> BufOrErr = OldMember.getMemoryBufferRef();
44  if (!BufOrErr)
45    return errorCodeToError(BufOrErr.getError());
46
47  NewArchiveMember M;
48  M.Buf = MemoryBuffer::getMemBuffer(*BufOrErr, false);
49  if (!Deterministic) {
50    M.ModTime = OldMember.getLastModified();
51    M.UID = OldMember.getUID();
52    M.GID = OldMember.getGID();
53    M.Perms = OldMember.getAccessMode();
54  }
55  return std::move(M);
56}
57
58Expected<NewArchiveMember> NewArchiveMember::getFile(StringRef FileName,
59                                                     bool Deterministic) {
60  sys::fs::file_status Status;
61  int FD;
62  if (auto EC = sys::fs::openFileForRead(FileName, FD))
63    return errorCodeToError(EC);
64  assert(FD != -1);
65
66  if (auto EC = sys::fs::status(FD, Status))
67    return errorCodeToError(EC);
68
69  // Opening a directory doesn't make sense. Let it fail.
70  // Linux cannot open directories with open(2), although
71  // cygwin and *bsd can.
72  if (Status.type() == sys::fs::file_type::directory_file)
73    return errorCodeToError(make_error_code(errc::is_a_directory));
74
75  ErrorOr<std::unique_ptr<MemoryBuffer>> MemberBufferOrErr =
76      MemoryBuffer::getOpenFile(FD, FileName, Status.getSize(), false);
77  if (!MemberBufferOrErr)
78    return errorCodeToError(MemberBufferOrErr.getError());
79
80  if (close(FD) != 0)
81    return errorCodeToError(std::error_code(errno, std::generic_category()));
82
83  NewArchiveMember M;
84  M.Buf = std::move(*MemberBufferOrErr);
85  if (!Deterministic) {
86    M.ModTime = Status.getLastModificationTime();
87    M.UID = Status.getUser();
88    M.GID = Status.getGroup();
89    M.Perms = Status.permissions();
90  }
91  return std::move(M);
92}
93
94template <typename T>
95static void printWithSpacePadding(raw_fd_ostream &OS, T Data, unsigned Size,
96                                  bool MayTruncate = false) {
97  uint64_t OldPos = OS.tell();
98  OS << Data;
99  unsigned SizeSoFar = OS.tell() - OldPos;
100  if (Size > SizeSoFar) {
101    OS.indent(Size - SizeSoFar);
102  } else if (Size < SizeSoFar) {
103    assert(MayTruncate && "Data doesn't fit in Size");
104    // Some of the data this is used for (like UID) can be larger than the
105    // space available in the archive format. Truncate in that case.
106    OS.seek(OldPos + Size);
107  }
108}
109
110static void print32(raw_ostream &Out, object::Archive::Kind Kind,
111                    uint32_t Val) {
112  if (Kind == object::Archive::K_GNU)
113    support::endian::Writer<support::big>(Out).write(Val);
114  else
115    support::endian::Writer<support::little>(Out).write(Val);
116}
117
118static void printRestOfMemberHeader(raw_fd_ostream &Out,
119                                    const sys::TimeValue &ModTime, unsigned UID,
120                                    unsigned GID, unsigned Perms,
121                                    unsigned Size) {
122  printWithSpacePadding(Out, ModTime.toEpochTime(), 12);
123  printWithSpacePadding(Out, UID, 6, true);
124  printWithSpacePadding(Out, GID, 6, true);
125  printWithSpacePadding(Out, format("%o", Perms), 8);
126  printWithSpacePadding(Out, Size, 10);
127  Out << "`\n";
128}
129
130static void printGNUSmallMemberHeader(raw_fd_ostream &Out, StringRef Name,
131                                      const sys::TimeValue &ModTime,
132                                      unsigned UID, unsigned GID,
133                                      unsigned Perms, unsigned Size) {
134  printWithSpacePadding(Out, Twine(Name) + "/", 16);
135  printRestOfMemberHeader(Out, ModTime, UID, GID, Perms, Size);
136}
137
138static void printBSDMemberHeader(raw_fd_ostream &Out, StringRef Name,
139                                 const sys::TimeValue &ModTime, unsigned UID,
140                                 unsigned GID, unsigned Perms, unsigned Size) {
141  uint64_t PosAfterHeader = Out.tell() + 60 + Name.size();
142  // Pad so that even 64 bit object files are aligned.
143  unsigned Pad = OffsetToAlignment(PosAfterHeader, 8);
144  unsigned NameWithPadding = Name.size() + Pad;
145  printWithSpacePadding(Out, Twine("#1/") + Twine(NameWithPadding), 16);
146  printRestOfMemberHeader(Out, ModTime, UID, GID, Perms,
147                          NameWithPadding + Size);
148  Out << Name;
149  assert(PosAfterHeader == Out.tell());
150  while (Pad--)
151    Out.write(uint8_t(0));
152}
153
154static bool useStringTable(bool Thin, StringRef Name) {
155  return Thin || Name.size() >= 16;
156}
157
158static void
159printMemberHeader(raw_fd_ostream &Out, object::Archive::Kind Kind, bool Thin,
160                  StringRef Name,
161                  std::vector<unsigned>::iterator &StringMapIndexIter,
162                  const sys::TimeValue &ModTime, unsigned UID, unsigned GID,
163                  unsigned Perms, unsigned Size) {
164  if (Kind == object::Archive::K_BSD)
165    return printBSDMemberHeader(Out, Name, ModTime, UID, GID, Perms, Size);
166  if (!useStringTable(Thin, Name))
167    return printGNUSmallMemberHeader(Out, Name, ModTime, UID, GID, Perms, Size);
168  Out << '/';
169  printWithSpacePadding(Out, *StringMapIndexIter++, 15);
170  printRestOfMemberHeader(Out, ModTime, UID, GID, Perms, Size);
171}
172
173// Compute the relative path from From to To.
174static std::string computeRelativePath(StringRef From, StringRef To) {
175  if (sys::path::is_absolute(From) || sys::path::is_absolute(To))
176    return To;
177
178  StringRef DirFrom = sys::path::parent_path(From);
179  auto FromI = sys::path::begin(DirFrom);
180  auto ToI = sys::path::begin(To);
181  while (*FromI == *ToI) {
182    ++FromI;
183    ++ToI;
184  }
185
186  SmallString<128> Relative;
187  for (auto FromE = sys::path::end(DirFrom); FromI != FromE; ++FromI)
188    sys::path::append(Relative, "..");
189
190  for (auto ToE = sys::path::end(To); ToI != ToE; ++ToI)
191    sys::path::append(Relative, *ToI);
192
193  return Relative.str();
194}
195
196static void writeStringTable(raw_fd_ostream &Out, StringRef ArcName,
197                             ArrayRef<NewArchiveMember> Members,
198                             std::vector<unsigned> &StringMapIndexes,
199                             bool Thin) {
200  unsigned StartOffset = 0;
201  for (const NewArchiveMember &M : Members) {
202    StringRef Path = M.Buf->getBufferIdentifier();
203    StringRef Name = sys::path::filename(Path);
204    if (!useStringTable(Thin, Name))
205      continue;
206    if (StartOffset == 0) {
207      printWithSpacePadding(Out, "//", 58);
208      Out << "`\n";
209      StartOffset = Out.tell();
210    }
211    StringMapIndexes.push_back(Out.tell() - StartOffset);
212
213    if (Thin)
214      Out << computeRelativePath(ArcName, Path);
215    else
216      Out << Name;
217
218    Out << "/\n";
219  }
220  if (StartOffset == 0)
221    return;
222  if (Out.tell() % 2)
223    Out << '\n';
224  int Pos = Out.tell();
225  Out.seek(StartOffset - 12);
226  printWithSpacePadding(Out, Pos - StartOffset, 10);
227  Out.seek(Pos);
228}
229
230static sys::TimeValue now(bool Deterministic) {
231  if (!Deterministic)
232    return sys::TimeValue::now();
233  sys::TimeValue TV;
234  TV.fromEpochTime(0);
235  return TV;
236}
237
238// Returns the offset of the first reference to a member offset.
239static ErrorOr<unsigned>
240writeSymbolTable(raw_fd_ostream &Out, object::Archive::Kind Kind,
241                 ArrayRef<NewArchiveMember> Members,
242                 std::vector<unsigned> &MemberOffsetRefs, bool Deterministic) {
243  unsigned HeaderStartOffset = 0;
244  unsigned BodyStartOffset = 0;
245  SmallString<128> NameBuf;
246  raw_svector_ostream NameOS(NameBuf);
247  LLVMContext Context;
248  for (unsigned MemberNum = 0, N = Members.size(); MemberNum < N; ++MemberNum) {
249    MemoryBufferRef MemberBuffer = Members[MemberNum].Buf->getMemBufferRef();
250    Expected<std::unique_ptr<object::SymbolicFile>> ObjOrErr =
251        object::SymbolicFile::createSymbolicFile(
252            MemberBuffer, sys::fs::file_magic::unknown, &Context);
253    if (!ObjOrErr) {
254      // FIXME: check only for "not an object file" errors.
255      consumeError(ObjOrErr.takeError());
256      continue;
257    }
258    object::SymbolicFile &Obj = *ObjOrErr.get();
259
260    if (!HeaderStartOffset) {
261      HeaderStartOffset = Out.tell();
262      if (Kind == object::Archive::K_GNU)
263        printGNUSmallMemberHeader(Out, "", now(Deterministic), 0, 0, 0, 0);
264      else
265        printBSDMemberHeader(Out, "__.SYMDEF", now(Deterministic), 0, 0, 0, 0);
266      BodyStartOffset = Out.tell();
267      print32(Out, Kind, 0); // number of entries or bytes
268    }
269
270    for (const object::BasicSymbolRef &S : Obj.symbols()) {
271      uint32_t Symflags = S.getFlags();
272      if (Symflags & object::SymbolRef::SF_FormatSpecific)
273        continue;
274      if (!(Symflags & object::SymbolRef::SF_Global))
275        continue;
276      if (Symflags & object::SymbolRef::SF_Undefined)
277        continue;
278
279      unsigned NameOffset = NameOS.tell();
280      if (auto EC = S.printName(NameOS))
281        return EC;
282      NameOS << '\0';
283      MemberOffsetRefs.push_back(MemberNum);
284      if (Kind == object::Archive::K_BSD)
285        print32(Out, Kind, NameOffset);
286      print32(Out, Kind, 0); // member offset
287    }
288  }
289
290  if (HeaderStartOffset == 0)
291    return 0;
292
293  StringRef StringTable = NameOS.str();
294  if (Kind == object::Archive::K_BSD)
295    print32(Out, Kind, StringTable.size()); // byte count of the string table
296  Out << StringTable;
297
298  // ld64 requires the next member header to start at an offset that is
299  // 4 bytes aligned.
300  unsigned Pad = OffsetToAlignment(Out.tell(), 4);
301  while (Pad--)
302    Out.write(uint8_t(0));
303
304  // Patch up the size of the symbol table now that we know how big it is.
305  unsigned Pos = Out.tell();
306  const unsigned MemberHeaderSize = 60;
307  Out.seek(HeaderStartOffset + 48); // offset of the size field.
308  printWithSpacePadding(Out, Pos - MemberHeaderSize - HeaderStartOffset, 10);
309
310  // Patch up the number of symbols.
311  Out.seek(BodyStartOffset);
312  unsigned NumSyms = MemberOffsetRefs.size();
313  if (Kind == object::Archive::K_GNU)
314    print32(Out, Kind, NumSyms);
315  else
316    print32(Out, Kind, NumSyms * 8);
317
318  Out.seek(Pos);
319  return BodyStartOffset + 4;
320}
321
322std::pair<StringRef, std::error_code>
323llvm::writeArchive(StringRef ArcName,
324                   std::vector<NewArchiveMember> &NewMembers,
325                   bool WriteSymtab, object::Archive::Kind Kind,
326                   bool Deterministic, bool Thin,
327                   std::unique_ptr<MemoryBuffer> OldArchiveBuf) {
328  assert((!Thin || Kind == object::Archive::K_GNU) &&
329         "Only the gnu format has a thin mode");
330  SmallString<128> TmpArchive;
331  int TmpArchiveFD;
332  if (auto EC = sys::fs::createUniqueFile(ArcName + ".temp-archive-%%%%%%%.a",
333                                          TmpArchiveFD, TmpArchive))
334    return std::make_pair(ArcName, EC);
335
336  tool_output_file Output(TmpArchive, TmpArchiveFD);
337  raw_fd_ostream &Out = Output.os();
338  if (Thin)
339    Out << "!<thin>\n";
340  else
341    Out << "!<arch>\n";
342
343  std::vector<unsigned> MemberOffsetRefs;
344
345  std::vector<std::unique_ptr<MemoryBuffer>> Buffers;
346  std::vector<MemoryBufferRef> Members;
347  std::vector<sys::fs::file_status> NewMemberStatus;
348
349  unsigned MemberReferenceOffset = 0;
350  if (WriteSymtab) {
351    ErrorOr<unsigned> MemberReferenceOffsetOrErr = writeSymbolTable(
352        Out, Kind, NewMembers, MemberOffsetRefs, Deterministic);
353    if (auto EC = MemberReferenceOffsetOrErr.getError())
354      return std::make_pair(ArcName, EC);
355    MemberReferenceOffset = MemberReferenceOffsetOrErr.get();
356  }
357
358  std::vector<unsigned> StringMapIndexes;
359  if (Kind != object::Archive::K_BSD)
360    writeStringTable(Out, ArcName, NewMembers, StringMapIndexes, Thin);
361
362  std::vector<unsigned>::iterator StringMapIndexIter = StringMapIndexes.begin();
363  std::vector<unsigned> MemberOffset;
364  for (const NewArchiveMember &M : NewMembers) {
365    MemoryBufferRef File = M.Buf->getMemBufferRef();
366
367    unsigned Pos = Out.tell();
368    MemberOffset.push_back(Pos);
369
370    printMemberHeader(Out, Kind, Thin,
371                      sys::path::filename(M.Buf->getBufferIdentifier()),
372                      StringMapIndexIter, M.ModTime, M.UID, M.GID, M.Perms,
373                      M.Buf->getBufferSize());
374
375    if (!Thin)
376      Out << File.getBuffer();
377
378    if (Out.tell() % 2)
379      Out << '\n';
380  }
381
382  if (MemberReferenceOffset) {
383    Out.seek(MemberReferenceOffset);
384    for (unsigned MemberNum : MemberOffsetRefs) {
385      if (Kind == object::Archive::K_BSD)
386        Out.seek(Out.tell() + 4); // skip over the string offset
387      print32(Out, Kind, MemberOffset[MemberNum]);
388    }
389  }
390
391  Output.keep();
392  Out.close();
393
394  // At this point, we no longer need whatever backing memory
395  // was used to generate the NewMembers. On Windows, this buffer
396  // could be a mapped view of the file we want to replace (if
397  // we're updating an existing archive, say). In that case, the
398  // rename would still succeed, but it would leave behind a
399  // temporary file (actually the original file renamed) because
400  // a file cannot be deleted while there's a handle open on it,
401  // only renamed. So by freeing this buffer, this ensures that
402  // the last open handle on the destination file, if any, is
403  // closed before we attempt to rename.
404  OldArchiveBuf.reset();
405
406  sys::fs::rename(TmpArchive, ArcName);
407  return std::make_pair("", std::error_code());
408}
409