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 
35 using namespace llvm;
36 
37 NewArchiveMember::NewArchiveMember(MemoryBufferRef BufRef)
38     : Buf(MemoryBuffer::getMemBuffer(BufRef, false)) {}
39 
40 Expected<NewArchiveMember>
41 NewArchiveMember::getOldMember(const object::Archive::Child &OldMember,
42                                bool Deterministic) {
43   Expected<llvm::MemoryBufferRef> BufOrErr = OldMember.getMemoryBufferRef();
44   if (!BufOrErr)
45     return BufOrErr.takeError();
46 
47   NewArchiveMember M;
48   assert(M.IsNew == false);
49   M.Buf = MemoryBuffer::getMemBuffer(*BufOrErr, false);
50   if (!Deterministic) {
51     auto ModTimeOrErr = OldMember.getLastModified();
52     if (!ModTimeOrErr)
53       return ModTimeOrErr.takeError();
54     M.ModTime = ModTimeOrErr.get();
55     Expected<unsigned> UIDOrErr = OldMember.getUID();
56     if (!UIDOrErr)
57       return UIDOrErr.takeError();
58     M.UID = UIDOrErr.get();
59     Expected<unsigned> GIDOrErr = OldMember.getGID();
60     if (!GIDOrErr)
61       return GIDOrErr.takeError();
62     M.GID = GIDOrErr.get();
63     Expected<sys::fs::perms> AccessModeOrErr = OldMember.getAccessMode();
64     if (!AccessModeOrErr)
65       return AccessModeOrErr.takeError();
66     M.Perms = AccessModeOrErr.get();
67   }
68   return std::move(M);
69 }
70 
71 Expected<NewArchiveMember> NewArchiveMember::getFile(StringRef FileName,
72                                                      bool Deterministic) {
73   sys::fs::file_status Status;
74   int FD;
75   if (auto EC = sys::fs::openFileForRead(FileName, FD))
76     return errorCodeToError(EC);
77   assert(FD != -1);
78 
79   if (auto EC = sys::fs::status(FD, Status))
80     return errorCodeToError(EC);
81 
82   // Opening a directory doesn't make sense. Let it fail.
83   // Linux cannot open directories with open(2), although
84   // cygwin and *bsd can.
85   if (Status.type() == sys::fs::file_type::directory_file)
86     return errorCodeToError(make_error_code(errc::is_a_directory));
87 
88   ErrorOr<std::unique_ptr<MemoryBuffer>> MemberBufferOrErr =
89       MemoryBuffer::getOpenFile(FD, FileName, Status.getSize(), false);
90   if (!MemberBufferOrErr)
91     return errorCodeToError(MemberBufferOrErr.getError());
92 
93   if (close(FD) != 0)
94     return errorCodeToError(std::error_code(errno, std::generic_category()));
95 
96   NewArchiveMember M;
97   M.IsNew = true;
98   M.Buf = std::move(*MemberBufferOrErr);
99   if (!Deterministic) {
100     M.ModTime = std::chrono::time_point_cast<std::chrono::seconds>(
101         Status.getLastModificationTime());
102     M.UID = Status.getUser();
103     M.GID = Status.getGroup();
104     M.Perms = Status.permissions();
105   }
106   return std::move(M);
107 }
108 
109 template <typename T>
110 static void printWithSpacePadding(raw_fd_ostream &OS, T Data, unsigned Size,
111                                   bool MayTruncate = false) {
112   uint64_t OldPos = OS.tell();
113   OS << Data;
114   unsigned SizeSoFar = OS.tell() - OldPos;
115   if (Size > SizeSoFar) {
116     OS.indent(Size - SizeSoFar);
117   } else if (Size < SizeSoFar) {
118     assert(MayTruncate && "Data doesn't fit in Size");
119     // Some of the data this is used for (like UID) can be larger than the
120     // space available in the archive format. Truncate in that case.
121     OS.seek(OldPos + Size);
122   }
123 }
124 
125 static bool isBSDLike(object::Archive::Kind Kind) {
126   switch (Kind) {
127   case object::Archive::K_GNU:
128     return false;
129   case object::Archive::K_BSD:
130   case object::Archive::K_DARWIN:
131     return true;
132   case object::Archive::K_MIPS64:
133   case object::Archive::K_DARWIN64:
134   case object::Archive::K_COFF:
135     break;
136   }
137   llvm_unreachable("not supported for writting");
138 }
139 
140 static void print32(raw_ostream &Out, object::Archive::Kind Kind,
141                     uint32_t Val) {
142   if (isBSDLike(Kind))
143     support::endian::Writer<support::little>(Out).write(Val);
144   else
145     support::endian::Writer<support::big>(Out).write(Val);
146 }
147 
148 static void printRestOfMemberHeader(
149     raw_fd_ostream &Out, const sys::TimePoint<std::chrono::seconds> &ModTime,
150     unsigned UID, unsigned GID, unsigned Perms, unsigned Size) {
151   printWithSpacePadding(Out, sys::toTimeT(ModTime), 12);
152   printWithSpacePadding(Out, UID, 6, true);
153   printWithSpacePadding(Out, GID, 6, true);
154   printWithSpacePadding(Out, format("%o", Perms), 8);
155   printWithSpacePadding(Out, Size, 10);
156   Out << "`\n";
157 }
158 
159 static void
160 printGNUSmallMemberHeader(raw_fd_ostream &Out, StringRef Name,
161                           const sys::TimePoint<std::chrono::seconds> &ModTime,
162                           unsigned UID, unsigned GID, unsigned Perms,
163                           unsigned Size) {
164   printWithSpacePadding(Out, Twine(Name) + "/", 16);
165   printRestOfMemberHeader(Out, ModTime, UID, GID, Perms, Size);
166 }
167 
168 static void
169 printBSDMemberHeader(raw_fd_ostream &Out, StringRef Name,
170                      const sys::TimePoint<std::chrono::seconds> &ModTime,
171                      unsigned UID, unsigned GID, unsigned Perms,
172                      unsigned Size) {
173   uint64_t PosAfterHeader = Out.tell() + 60 + Name.size();
174   // Pad so that even 64 bit object files are aligned.
175   unsigned Pad = OffsetToAlignment(PosAfterHeader, 8);
176   unsigned NameWithPadding = Name.size() + Pad;
177   printWithSpacePadding(Out, Twine("#1/") + Twine(NameWithPadding), 16);
178   printRestOfMemberHeader(Out, ModTime, UID, GID, Perms,
179                           NameWithPadding + Size);
180   Out << Name;
181   assert(PosAfterHeader == Out.tell());
182   while (Pad--)
183     Out.write(uint8_t(0));
184 }
185 
186 static bool useStringTable(bool Thin, StringRef Name) {
187   return Thin || Name.size() >= 16;
188 }
189 
190 static void
191 printMemberHeader(raw_fd_ostream &Out, object::Archive::Kind Kind, bool Thin,
192                   StringRef Name,
193                   std::vector<unsigned>::iterator &StringMapIndexIter,
194                   const sys::TimePoint<std::chrono::seconds> &ModTime,
195                   unsigned UID, unsigned GID, unsigned Perms, unsigned Size) {
196   if (isBSDLike(Kind))
197     return printBSDMemberHeader(Out, Name, ModTime, UID, GID, Perms, Size);
198   if (!useStringTable(Thin, Name))
199     return printGNUSmallMemberHeader(Out, Name, ModTime, UID, GID, Perms, Size);
200   Out << '/';
201   printWithSpacePadding(Out, *StringMapIndexIter++, 15);
202   printRestOfMemberHeader(Out, ModTime, UID, GID, Perms, Size);
203 }
204 
205 // Compute the relative path from From to To.
206 static std::string computeRelativePath(StringRef From, StringRef To) {
207   if (sys::path::is_absolute(From) || sys::path::is_absolute(To))
208     return To;
209 
210   StringRef DirFrom = sys::path::parent_path(From);
211   auto FromI = sys::path::begin(DirFrom);
212   auto ToI = sys::path::begin(To);
213   while (*FromI == *ToI) {
214     ++FromI;
215     ++ToI;
216   }
217 
218   SmallString<128> Relative;
219   for (auto FromE = sys::path::end(DirFrom); FromI != FromE; ++FromI)
220     sys::path::append(Relative, "..");
221 
222   for (auto ToE = sys::path::end(To); ToI != ToE; ++ToI)
223     sys::path::append(Relative, *ToI);
224 
225 #ifdef LLVM_ON_WIN32
226   // Replace backslashes with slashes so that the path is portable between *nix
227   // and Windows.
228   std::replace(Relative.begin(), Relative.end(), '\\', '/');
229 #endif
230 
231   return Relative.str();
232 }
233 
234 static void writeStringTable(raw_fd_ostream &Out, StringRef ArcName,
235                              ArrayRef<NewArchiveMember> Members,
236                              std::vector<unsigned> &StringMapIndexes,
237                              bool Thin) {
238   unsigned StartOffset = 0;
239   for (const NewArchiveMember &M : Members) {
240     StringRef Path = M.Buf->getBufferIdentifier();
241     StringRef Name = sys::path::filename(Path);
242     if (!useStringTable(Thin, Name))
243       continue;
244     if (StartOffset == 0) {
245       printWithSpacePadding(Out, "//", 58);
246       Out << "`\n";
247       StartOffset = Out.tell();
248     }
249     StringMapIndexes.push_back(Out.tell() - StartOffset);
250 
251     if (Thin) {
252       if (M.IsNew)
253         Out << computeRelativePath(ArcName, Path);
254       else
255         Out << M.Buf->getBufferIdentifier();
256     } else
257       Out << Name;
258 
259     Out << "/\n";
260   }
261   if (StartOffset == 0)
262     return;
263   if (Out.tell() % 2)
264     Out << '\n';
265   int Pos = Out.tell();
266   Out.seek(StartOffset - 12);
267   printWithSpacePadding(Out, Pos - StartOffset, 10);
268   Out.seek(Pos);
269 }
270 
271 static sys::TimePoint<std::chrono::seconds> now(bool Deterministic) {
272   using namespace std::chrono;
273 
274   if (!Deterministic)
275     return time_point_cast<seconds>(system_clock::now());
276   return sys::TimePoint<seconds>();
277 }
278 
279 // Returns the offset of the first reference to a member offset.
280 static ErrorOr<unsigned>
281 writeSymbolTable(raw_fd_ostream &Out, object::Archive::Kind Kind,
282                  ArrayRef<NewArchiveMember> Members,
283                  std::vector<unsigned> &MemberOffsetRefs, bool Deterministic) {
284   unsigned HeaderStartOffset = 0;
285   unsigned BodyStartOffset = 0;
286   SmallString<128> NameBuf;
287   raw_svector_ostream NameOS(NameBuf);
288   LLVMContext Context;
289   for (unsigned MemberNum = 0, N = Members.size(); MemberNum < N; ++MemberNum) {
290     MemoryBufferRef MemberBuffer = Members[MemberNum].Buf->getMemBufferRef();
291     Expected<std::unique_ptr<object::SymbolicFile>> ObjOrErr =
292         object::SymbolicFile::createSymbolicFile(
293             MemberBuffer, sys::fs::file_magic::unknown, &Context);
294     if (!ObjOrErr) {
295       // FIXME: check only for "not an object file" errors.
296       consumeError(ObjOrErr.takeError());
297       continue;
298     }
299     object::SymbolicFile &Obj = *ObjOrErr.get();
300 
301     if (!HeaderStartOffset) {
302       HeaderStartOffset = Out.tell();
303       if (isBSDLike(Kind))
304         printBSDMemberHeader(Out, "__.SYMDEF", now(Deterministic), 0, 0, 0, 0);
305       else
306         printGNUSmallMemberHeader(Out, "", now(Deterministic), 0, 0, 0, 0);
307       BodyStartOffset = Out.tell();
308       print32(Out, Kind, 0); // number of entries or bytes
309     }
310 
311     for (const object::BasicSymbolRef &S : Obj.symbols()) {
312       uint32_t Symflags = S.getFlags();
313       if (Symflags & object::SymbolRef::SF_FormatSpecific)
314         continue;
315       if (!(Symflags & object::SymbolRef::SF_Global))
316         continue;
317       if (Symflags & object::SymbolRef::SF_Undefined)
318         continue;
319 
320       unsigned NameOffset = NameOS.tell();
321       if (auto EC = S.printName(NameOS))
322         return EC;
323       NameOS << '\0';
324       MemberOffsetRefs.push_back(MemberNum);
325       if (isBSDLike(Kind))
326         print32(Out, Kind, NameOffset);
327       print32(Out, Kind, 0); // member offset
328     }
329   }
330 
331   if (HeaderStartOffset == 0)
332     return 0;
333 
334   // ld64 prefers the cctools type archive which pads its string table to a
335   // boundary of sizeof(int32_t).
336   if (isBSDLike(Kind))
337     for (unsigned P = OffsetToAlignment(NameOS.tell(), sizeof(int32_t)); P--;)
338       NameOS << '\0';
339 
340   StringRef StringTable = NameOS.str();
341   if (isBSDLike(Kind))
342     print32(Out, Kind, StringTable.size()); // byte count of the string table
343   Out << StringTable;
344 
345   // ld64 requires the next member header to start at an offset that is
346   // 4 bytes aligned.
347   unsigned Pad = OffsetToAlignment(Out.tell(), 4);
348   while (Pad--)
349     Out.write(uint8_t(0));
350 
351   // Patch up the size of the symbol table now that we know how big it is.
352   unsigned Pos = Out.tell();
353   const unsigned MemberHeaderSize = 60;
354   Out.seek(HeaderStartOffset + 48); // offset of the size field.
355   printWithSpacePadding(Out, Pos - MemberHeaderSize - HeaderStartOffset, 10);
356 
357   // Patch up the number of symbols.
358   Out.seek(BodyStartOffset);
359   unsigned NumSyms = MemberOffsetRefs.size();
360   if (isBSDLike(Kind))
361     print32(Out, Kind, NumSyms * 8);
362   else
363     print32(Out, Kind, NumSyms);
364 
365   Out.seek(Pos);
366   return BodyStartOffset + 4;
367 }
368 
369 std::pair<StringRef, std::error_code>
370 llvm::writeArchive(StringRef ArcName,
371                    std::vector<NewArchiveMember> &NewMembers,
372                    bool WriteSymtab, object::Archive::Kind Kind,
373                    bool Deterministic, bool Thin,
374                    std::unique_ptr<MemoryBuffer> OldArchiveBuf) {
375   assert((!Thin || !isBSDLike(Kind)) && "Only the gnu format has a thin mode");
376   SmallString<128> TmpArchive;
377   int TmpArchiveFD;
378   if (auto EC = sys::fs::createUniqueFile(ArcName + ".temp-archive-%%%%%%%.a",
379                                           TmpArchiveFD, TmpArchive))
380     return std::make_pair(ArcName, EC);
381 
382   tool_output_file Output(TmpArchive, TmpArchiveFD);
383   raw_fd_ostream &Out = Output.os();
384   if (Thin)
385     Out << "!<thin>\n";
386   else
387     Out << "!<arch>\n";
388 
389   std::vector<unsigned> MemberOffsetRefs;
390 
391   std::vector<std::unique_ptr<MemoryBuffer>> Buffers;
392   std::vector<MemoryBufferRef> Members;
393   std::vector<sys::fs::file_status> NewMemberStatus;
394 
395   unsigned MemberReferenceOffset = 0;
396   if (WriteSymtab) {
397     ErrorOr<unsigned> MemberReferenceOffsetOrErr = writeSymbolTable(
398         Out, Kind, NewMembers, MemberOffsetRefs, Deterministic);
399     if (auto EC = MemberReferenceOffsetOrErr.getError())
400       return std::make_pair(ArcName, EC);
401     MemberReferenceOffset = MemberReferenceOffsetOrErr.get();
402   }
403 
404   std::vector<unsigned> StringMapIndexes;
405   if (!isBSDLike(Kind))
406     writeStringTable(Out, ArcName, NewMembers, StringMapIndexes, Thin);
407 
408   std::vector<unsigned>::iterator StringMapIndexIter = StringMapIndexes.begin();
409   std::vector<unsigned> MemberOffset;
410   for (const NewArchiveMember &M : NewMembers) {
411     MemoryBufferRef File = M.Buf->getMemBufferRef();
412     unsigned Padding = 0;
413 
414     unsigned Pos = Out.tell();
415     MemberOffset.push_back(Pos);
416 
417     // ld64 expects the members to be 8-byte aligned for 64-bit content and at
418     // least 4-byte aligned for 32-bit content.  Opt for the larger encoding
419     // uniformly.  This matches the behaviour with cctools and ensures that ld64
420     // is happy with archives that we generate.
421     if (Kind == object::Archive::K_DARWIN)
422       Padding = OffsetToAlignment(M.Buf->getBufferSize(), 8);
423 
424     printMemberHeader(Out, Kind, Thin,
425                       sys::path::filename(M.Buf->getBufferIdentifier()),
426                       StringMapIndexIter, M.ModTime, M.UID, M.GID, M.Perms,
427                       M.Buf->getBufferSize() + Padding);
428 
429     if (!Thin)
430       Out << File.getBuffer();
431 
432     while (Padding--)
433       Out << '\n';
434     if (Out.tell() % 2)
435       Out << '\n';
436   }
437 
438   if (MemberReferenceOffset) {
439     Out.seek(MemberReferenceOffset);
440     for (unsigned MemberNum : MemberOffsetRefs) {
441       if (isBSDLike(Kind))
442         Out.seek(Out.tell() + 4); // skip over the string offset
443       print32(Out, Kind, MemberOffset[MemberNum]);
444     }
445   }
446 
447   Output.keep();
448   Out.close();
449 
450   // At this point, we no longer need whatever backing memory
451   // was used to generate the NewMembers. On Windows, this buffer
452   // could be a mapped view of the file we want to replace (if
453   // we're updating an existing archive, say). In that case, the
454   // rename would still succeed, but it would leave behind a
455   // temporary file (actually the original file renamed) because
456   // a file cannot be deleted while there's a handle open on it,
457   // only renamed. So by freeing this buffer, this ensures that
458   // the last open handle on the destination file, if any, is
459   // closed before we attempt to rename.
460   OldArchiveBuf.reset();
461 
462   sys::fs::rename(TmpArchive, ArcName);
463   return std::make_pair("", std::error_code());
464 }
465