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