Initial support for writing thin archives.
[oota-llvm.git] / lib / Object / ArchiveWriter.cpp
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 NewArchiveIterator::NewArchiveIterator(object::Archive::child_iterator I,
38                                        StringRef Name)
39     : IsNewMember(false), Name(Name), OldI(I) {}
40
41 NewArchiveIterator::NewArchiveIterator(StringRef NewFilename, StringRef Name)
42     : IsNewMember(true), Name(Name), NewFilename(NewFilename) {}
43
44 StringRef NewArchiveIterator::getName() const { return Name; }
45
46 bool NewArchiveIterator::isNewMember() const { return IsNewMember; }
47
48 object::Archive::child_iterator NewArchiveIterator::getOld() const {
49   assert(!IsNewMember);
50   return OldI;
51 }
52
53 StringRef NewArchiveIterator::getNew() const {
54   assert(IsNewMember);
55   return NewFilename;
56 }
57
58 llvm::ErrorOr<int>
59 NewArchiveIterator::getFD(sys::fs::file_status &NewStatus) const {
60   assert(IsNewMember);
61   int NewFD;
62   if (auto EC = sys::fs::openFileForRead(NewFilename, NewFD))
63     return EC;
64   assert(NewFD != -1);
65
66   if (auto EC = sys::fs::status(NewFD, NewStatus))
67     return 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 (NewStatus.type() == sys::fs::file_type::directory_file)
73     return make_error_code(errc::is_a_directory);
74
75   return NewFD;
76 }
77
78 template <typename T>
79 static void printWithSpacePadding(raw_fd_ostream &OS, T Data, unsigned Size,
80                                   bool MayTruncate = false) {
81   uint64_t OldPos = OS.tell();
82   OS << Data;
83   unsigned SizeSoFar = OS.tell() - OldPos;
84   if (Size > SizeSoFar) {
85     OS.indent(Size - SizeSoFar);
86   } else if (Size < SizeSoFar) {
87     assert(MayTruncate && "Data doesn't fit in Size");
88     // Some of the data this is used for (like UID) can be larger than the
89     // space available in the archive format. Truncate in that case.
90     OS.seek(OldPos + Size);
91   }
92 }
93
94 static void print32(raw_ostream &Out, object::Archive::Kind Kind,
95                     uint32_t Val) {
96   if (Kind == object::Archive::K_GNU)
97     support::endian::Writer<support::big>(Out).write(Val);
98   else
99     support::endian::Writer<support::little>(Out).write(Val);
100 }
101
102 static void printRestOfMemberHeader(raw_fd_ostream &Out,
103                                     const sys::TimeValue &ModTime, unsigned UID,
104                                     unsigned GID, unsigned Perms,
105                                     unsigned Size) {
106   printWithSpacePadding(Out, ModTime.toEpochTime(), 12);
107   printWithSpacePadding(Out, UID, 6, true);
108   printWithSpacePadding(Out, GID, 6, true);
109   printWithSpacePadding(Out, format("%o", Perms), 8);
110   printWithSpacePadding(Out, Size, 10);
111   Out << "`\n";
112 }
113
114 static void printGNUSmallMemberHeader(raw_fd_ostream &Out, StringRef Name,
115                                       const sys::TimeValue &ModTime,
116                                       unsigned UID, unsigned GID,
117                                       unsigned Perms, unsigned Size) {
118   printWithSpacePadding(Out, Twine(Name) + "/", 16);
119   printRestOfMemberHeader(Out, ModTime, UID, GID, Perms, Size);
120 }
121
122 static void printBSDMemberHeader(raw_fd_ostream &Out, StringRef Name,
123                                  const sys::TimeValue &ModTime, unsigned UID,
124                                  unsigned GID, unsigned Perms, unsigned Size) {
125   uint64_t PosAfterHeader = Out.tell() + 60 + Name.size();
126   // Pad so that even 64 bit object files are aligned.
127   unsigned Pad = OffsetToAlignment(PosAfterHeader, 8);
128   unsigned NameWithPadding = Name.size() + Pad;
129   printWithSpacePadding(Out, Twine("#1/") + Twine(NameWithPadding), 16);
130   printRestOfMemberHeader(Out, ModTime, UID, GID, Perms,
131                           NameWithPadding + Size);
132   Out << Name;
133   assert(PosAfterHeader == Out.tell());
134   while (Pad--)
135     Out.write(uint8_t(0));
136 }
137
138 static bool useStringTable(bool Thin, StringRef Name) {
139   return Thin || Name.size() >= 16;
140 }
141
142 static void
143 printMemberHeader(raw_fd_ostream &Out, object::Archive::Kind Kind, bool Thin,
144                   StringRef Name,
145                   std::vector<unsigned>::iterator &StringMapIndexIter,
146                   const sys::TimeValue &ModTime, unsigned UID, unsigned GID,
147                   unsigned Perms, unsigned Size) {
148   if (Kind == object::Archive::K_BSD)
149     return printBSDMemberHeader(Out, Name, ModTime, UID, GID, Perms, Size);
150   if (!useStringTable(Thin, Name))
151     return printGNUSmallMemberHeader(Out, Name, ModTime, UID, GID, Perms, Size);
152   Out << '/';
153   printWithSpacePadding(Out, *StringMapIndexIter++, 15);
154   printRestOfMemberHeader(Out, ModTime, UID, GID, Perms, Size);
155 }
156
157 static void writeStringTable(raw_fd_ostream &Out,
158                              ArrayRef<NewArchiveIterator> Members,
159                              std::vector<unsigned> &StringMapIndexes,
160                              bool Thin) {
161   unsigned StartOffset = 0;
162   for (const NewArchiveIterator &I : Members) {
163     StringRef Name = I.getName();
164     if (!useStringTable(Thin, Name))
165       continue;
166     if (StartOffset == 0) {
167       printWithSpacePadding(Out, "//", 58);
168       Out << "`\n";
169       StartOffset = Out.tell();
170     }
171     StringMapIndexes.push_back(Out.tell() - StartOffset);
172     Out << Name << "/\n";
173   }
174   if (StartOffset == 0)
175     return;
176   if (Out.tell() % 2)
177     Out << '\n';
178   int Pos = Out.tell();
179   Out.seek(StartOffset - 12);
180   printWithSpacePadding(Out, Pos - StartOffset, 10);
181   Out.seek(Pos);
182 }
183
184 static sys::TimeValue now(bool Deterministic) {
185   if (!Deterministic)
186     return sys::TimeValue::now();
187   sys::TimeValue TV;
188   TV.fromEpochTime(0);
189   return TV;
190 }
191
192 // Returns the offset of the first reference to a member offset.
193 static ErrorOr<unsigned>
194 writeSymbolTable(raw_fd_ostream &Out, object::Archive::Kind Kind,
195                  ArrayRef<NewArchiveIterator> Members,
196                  ArrayRef<MemoryBufferRef> Buffers,
197                  std::vector<unsigned> &MemberOffsetRefs, bool Deterministic) {
198   unsigned HeaderStartOffset = 0;
199   unsigned BodyStartOffset = 0;
200   SmallString<128> NameBuf;
201   raw_svector_ostream NameOS(NameBuf);
202   LLVMContext Context;
203   for (unsigned MemberNum = 0, N = Members.size(); MemberNum < N; ++MemberNum) {
204     MemoryBufferRef MemberBuffer = Buffers[MemberNum];
205     ErrorOr<std::unique_ptr<object::SymbolicFile>> ObjOrErr =
206         object::SymbolicFile::createSymbolicFile(
207             MemberBuffer, sys::fs::file_magic::unknown, &Context);
208     if (!ObjOrErr)
209       continue;  // FIXME: check only for "not an object file" errors.
210     object::SymbolicFile &Obj = *ObjOrErr.get();
211
212     if (!HeaderStartOffset) {
213       HeaderStartOffset = Out.tell();
214       if (Kind == object::Archive::K_GNU)
215         printGNUSmallMemberHeader(Out, "", now(Deterministic), 0, 0, 0, 0);
216       else
217         printBSDMemberHeader(Out, "__.SYMDEF", now(Deterministic), 0, 0, 0, 0);
218       BodyStartOffset = Out.tell();
219       print32(Out, Kind, 0); // number of entries or bytes
220     }
221
222     for (const object::BasicSymbolRef &S : Obj.symbols()) {
223       uint32_t Symflags = S.getFlags();
224       if (Symflags & object::SymbolRef::SF_FormatSpecific)
225         continue;
226       if (!(Symflags & object::SymbolRef::SF_Global))
227         continue;
228       if (Symflags & object::SymbolRef::SF_Undefined)
229         continue;
230
231       unsigned NameOffset = NameOS.tell();
232       if (auto EC = S.printName(NameOS))
233         return EC;
234       NameOS << '\0';
235       MemberOffsetRefs.push_back(MemberNum);
236       if (Kind == object::Archive::K_BSD)
237         print32(Out, Kind, NameOffset);
238       print32(Out, Kind, 0); // member offset
239     }
240   }
241
242   if (HeaderStartOffset == 0)
243     return 0;
244
245   StringRef StringTable = NameOS.str();
246   if (Kind == object::Archive::K_BSD)
247     print32(Out, Kind, StringTable.size()); // byte count of the string table
248   Out << StringTable;
249
250   // ld64 requires the next member header to start at an offset that is
251   // 4 bytes aligned.
252   unsigned Pad = OffsetToAlignment(Out.tell(), 4);
253   while (Pad--)
254     Out.write(uint8_t(0));
255
256   // Patch up the size of the symbol table now that we know how big it is.
257   unsigned Pos = Out.tell();
258   const unsigned MemberHeaderSize = 60;
259   Out.seek(HeaderStartOffset + 48); // offset of the size field.
260   printWithSpacePadding(Out, Pos - MemberHeaderSize - HeaderStartOffset, 10);
261
262   // Patch up the number of symbols.
263   Out.seek(BodyStartOffset);
264   unsigned NumSyms = MemberOffsetRefs.size();
265   if (Kind == object::Archive::K_GNU)
266     print32(Out, Kind, NumSyms);
267   else
268     print32(Out, Kind, NumSyms * 8);
269
270   Out.seek(Pos);
271   return BodyStartOffset + 4;
272 }
273
274 std::pair<StringRef, std::error_code>
275 llvm::writeArchive(StringRef ArcName,
276                    std::vector<NewArchiveIterator> &NewMembers,
277                    bool WriteSymtab, object::Archive::Kind Kind,
278                    bool Deterministic, bool Thin) {
279   SmallString<128> TmpArchive;
280   int TmpArchiveFD;
281   if (auto EC = sys::fs::createUniqueFile(ArcName + ".temp-archive-%%%%%%%.a",
282                                           TmpArchiveFD, TmpArchive))
283     return std::make_pair(ArcName, EC);
284
285   tool_output_file Output(TmpArchive, TmpArchiveFD);
286   raw_fd_ostream &Out = Output.os();
287   if (Thin)
288     Out << "!<thin>\n";
289   else
290     Out << "!<arch>\n";
291
292   std::vector<unsigned> MemberOffsetRefs;
293
294   std::vector<std::unique_ptr<MemoryBuffer>> Buffers;
295   std::vector<MemoryBufferRef> Members;
296   std::vector<sys::fs::file_status> NewMemberStatus;
297
298   for (unsigned I = 0, N = NewMembers.size(); I < N; ++I) {
299     NewArchiveIterator &Member = NewMembers[I];
300     MemoryBufferRef MemberRef;
301
302     if (Member.isNewMember()) {
303       StringRef Filename = Member.getNew();
304       NewMemberStatus.resize(NewMemberStatus.size() + 1);
305       sys::fs::file_status &Status = NewMemberStatus.back();
306       ErrorOr<int> FD = Member.getFD(Status);
307       if (auto EC = FD.getError())
308         return std::make_pair(Filename, EC);
309       ErrorOr<std::unique_ptr<MemoryBuffer>> MemberBufferOrErr =
310           MemoryBuffer::getOpenFile(FD.get(), Filename, Status.getSize(),
311                                     false);
312       if (auto EC = MemberBufferOrErr.getError())
313         return std::make_pair(Filename, EC);
314       if (close(FD.get()) != 0)
315         return std::make_pair(Filename,
316                               std::error_code(errno, std::generic_category()));
317       Buffers.push_back(std::move(MemberBufferOrErr.get()));
318       MemberRef = Buffers.back()->getMemBufferRef();
319     } else {
320       object::Archive::child_iterator OldMember = Member.getOld();
321       ErrorOr<MemoryBufferRef> MemberBufferOrErr =
322           OldMember->getMemoryBufferRef();
323       if (auto EC = MemberBufferOrErr.getError())
324         return std::make_pair("", EC);
325       MemberRef = MemberBufferOrErr.get();
326     }
327     Members.push_back(MemberRef);
328   }
329
330   unsigned MemberReferenceOffset = 0;
331   if (WriteSymtab) {
332     ErrorOr<unsigned> MemberReferenceOffsetOrErr = writeSymbolTable(
333         Out, Kind, NewMembers, Members, MemberOffsetRefs, Deterministic);
334     if (auto EC = MemberReferenceOffsetOrErr.getError())
335       return std::make_pair(ArcName, EC);
336     MemberReferenceOffset = MemberReferenceOffsetOrErr.get();
337   }
338
339   std::vector<unsigned> StringMapIndexes;
340   if (Kind != object::Archive::K_BSD)
341     writeStringTable(Out, NewMembers, StringMapIndexes, Thin);
342
343   unsigned MemberNum = 0;
344   unsigned NewMemberNum = 0;
345   std::vector<unsigned>::iterator StringMapIndexIter = StringMapIndexes.begin();
346   std::vector<unsigned> MemberOffset;
347   for (const NewArchiveIterator &I : NewMembers) {
348     MemoryBufferRef File = Members[MemberNum++];
349
350     unsigned Pos = Out.tell();
351     MemberOffset.push_back(Pos);
352
353     sys::TimeValue ModTime;
354     unsigned UID;
355     unsigned GID;
356     unsigned Perms;
357     if (Deterministic) {
358       ModTime.fromEpochTime(0);
359       UID = 0;
360       GID = 0;
361       Perms = 0644;
362     } else if (I.isNewMember()) {
363       const sys::fs::file_status &Status = NewMemberStatus[NewMemberNum];
364       ModTime = Status.getLastModificationTime();
365       UID = Status.getUser();
366       GID = Status.getGroup();
367       Perms = Status.permissions();
368     } else {
369       object::Archive::child_iterator OldMember = I.getOld();
370       ModTime = OldMember->getLastModified();
371       UID = OldMember->getUID();
372       GID = OldMember->getGID();
373       Perms = OldMember->getAccessMode();
374     }
375
376     if (I.isNewMember()) {
377       StringRef FileName = I.getNew();
378       const sys::fs::file_status &Status = NewMemberStatus[NewMemberNum++];
379       printMemberHeader(Out, Kind, Thin, sys::path::filename(FileName),
380                         StringMapIndexIter, ModTime, UID, GID, Perms,
381                         Status.getSize());
382     } else {
383       object::Archive::child_iterator OldMember = I.getOld();
384       printMemberHeader(Out, Kind, Thin, I.getName(), StringMapIndexIter,
385                         ModTime, UID, GID, Perms, OldMember->getSize());
386     }
387
388     if (!Thin)
389       Out << File.getBuffer();
390
391     if (Out.tell() % 2)
392       Out << '\n';
393   }
394
395   if (MemberReferenceOffset) {
396     Out.seek(MemberReferenceOffset);
397     for (unsigned MemberNum : MemberOffsetRefs) {
398       if (Kind == object::Archive::K_BSD)
399         Out.seek(Out.tell() + 4); // skip over the string offset
400       print32(Out, Kind, MemberOffset[MemberNum]);
401     }
402   }
403
404   Output.keep();
405   Out.close();
406   sys::fs::rename(TmpArchive, ArcName);
407   return std::make_pair("", std::error_code());
408 }