1//===-- MachOUtils.h - Mach-o specific helpers for dsymutil  --------------===//
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#include "MachOUtils.h"
11#include "BinaryHolder.h"
12#include "DebugMap.h"
13#include "dsymutil.h"
14#include "NonRelocatableStringpool.h"
15#include "llvm/MC/MCSectionMachO.h"
16#include "llvm/MC/MCAsmLayout.h"
17#include "llvm/MC/MCSectionMachO.h"
18#include "llvm/MC/MCObjectStreamer.h"
19#include "llvm/MC/MCMachObjectWriter.h"
20#include "llvm/MC/MCStreamer.h"
21#include "llvm/Object/MachO.h"
22#include "llvm/Support/FileUtilities.h"
23#include "llvm/Support/Program.h"
24#include "llvm/Support/raw_ostream.h"
25
26namespace llvm {
27namespace dsymutil {
28namespace MachOUtils {
29
30std::string getArchName(StringRef Arch) {
31  if (Arch.startswith("thumb"))
32    return (llvm::Twine("arm") + Arch.drop_front(5)).str();
33  return Arch;
34}
35
36static bool runLipo(StringRef SDKPath, SmallVectorImpl<const char *> &Args) {
37  auto Path = sys::findProgramByName("lipo", makeArrayRef(SDKPath));
38  if (!Path)
39    Path = sys::findProgramByName("lipo");
40
41  if (!Path) {
42    errs() << "error: lipo: " << Path.getError().message() << "\n";
43    return false;
44  }
45
46  std::string ErrMsg;
47  int result =
48      sys::ExecuteAndWait(*Path, Args.data(), nullptr, nullptr, 0, 0, &ErrMsg);
49  if (result) {
50    errs() << "error: lipo: " << ErrMsg << "\n";
51    return false;
52  }
53
54  return true;
55}
56
57bool generateUniversalBinary(SmallVectorImpl<ArchAndFilename> &ArchFiles,
58                             StringRef OutputFileName,
59                             const LinkOptions &Options, StringRef SDKPath) {
60  // No need to merge one file into a universal fat binary. First, try
61  // to move it (rename) to the final location. If that fails because
62  // of cross-device link issues then copy and delete.
63  if (ArchFiles.size() == 1) {
64    StringRef From(ArchFiles.front().Path);
65    if (sys::fs::rename(From, OutputFileName)) {
66      if (std::error_code EC = sys::fs::copy_file(From, OutputFileName)) {
67        errs() << "error: while copying " << From << " to " << OutputFileName
68               << ": " << EC.message() << "\n";
69        return false;
70      }
71      sys::fs::remove(From);
72    }
73    return true;
74  }
75
76  SmallVector<const char *, 8> Args;
77  Args.push_back("lipo");
78  Args.push_back("-create");
79
80  for (auto &Thin : ArchFiles)
81    Args.push_back(Thin.Path.c_str());
82
83  // Align segments to match dsymutil-classic alignment
84  for (auto &Thin : ArchFiles) {
85    Thin.Arch = getArchName(Thin.Arch);
86    Args.push_back("-segalign");
87    Args.push_back(Thin.Arch.c_str());
88    Args.push_back("20");
89  }
90
91  Args.push_back("-output");
92  Args.push_back(OutputFileName.data());
93  Args.push_back(nullptr);
94
95  if (Options.Verbose) {
96    outs() << "Running lipo\n";
97    for (auto Arg : Args)
98      outs() << ' ' << ((Arg == nullptr) ? "\n" : Arg);
99  }
100
101  return Options.NoOutput ? true : runLipo(SDKPath, Args);
102}
103
104// Return a MachO::segment_command_64 that holds the same values as
105// the passed MachO::segment_command. We do that to avoid having to
106// duplicat the logic for 32bits and 64bits segments.
107struct MachO::segment_command_64 adaptFrom32bits(MachO::segment_command Seg) {
108  MachO::segment_command_64 Seg64;
109  Seg64.cmd = Seg.cmd;
110  Seg64.cmdsize = Seg.cmdsize;
111  memcpy(Seg64.segname, Seg.segname, sizeof(Seg.segname));
112  Seg64.vmaddr = Seg.vmaddr;
113  Seg64.vmsize = Seg.vmsize;
114  Seg64.fileoff = Seg.fileoff;
115  Seg64.filesize = Seg.filesize;
116  Seg64.maxprot = Seg.maxprot;
117  Seg64.initprot = Seg.initprot;
118  Seg64.nsects = Seg.nsects;
119  Seg64.flags = Seg.flags;
120  return Seg64;
121}
122
123// Iterate on all \a Obj segments, and apply \a Handler to them.
124template <typename FunctionTy>
125static void iterateOnSegments(const object::MachOObjectFile &Obj,
126                              FunctionTy Handler) {
127  for (const auto &LCI : Obj.load_commands()) {
128    MachO::segment_command_64 Segment;
129    if (LCI.C.cmd == MachO::LC_SEGMENT)
130      Segment = adaptFrom32bits(Obj.getSegmentLoadCommand(LCI));
131    else if (LCI.C.cmd == MachO::LC_SEGMENT_64)
132      Segment = Obj.getSegment64LoadCommand(LCI);
133    else
134      continue;
135
136    Handler(Segment);
137  }
138}
139
140// Transfer the symbols described by \a NList to \a NewSymtab which is
141// just the raw contents of the symbol table for the dSYM companion file.
142// \returns whether the symbol was tranfered or not.
143template <typename NListTy>
144static bool transferSymbol(NListTy NList, bool IsLittleEndian,
145                           StringRef Strings, SmallVectorImpl<char> &NewSymtab,
146                           NonRelocatableStringpool &NewStrings,
147                           bool &InDebugNote) {
148  // Do not transfer undefined symbols, we want real addresses.
149  if ((NList.n_type & MachO::N_TYPE) == MachO::N_UNDF)
150    return false;
151
152  StringRef Name = StringRef(Strings.begin() + NList.n_strx);
153  if (InDebugNote) {
154    InDebugNote =
155        (NList.n_type != MachO::N_SO) || (!Name.empty() && Name[0] != '\0');
156    return false;
157  } else if (NList.n_type == MachO::N_SO) {
158    InDebugNote = true;
159    return false;
160  }
161
162  // FIXME: The + 1 is here to mimic dsymutil-classic that has 2 empty
163  // strings at the start of the generated string table (There is
164  // corresponding code in the string table emission).
165  NList.n_strx = NewStrings.getStringOffset(Name) + 1;
166  if (IsLittleEndian != sys::IsLittleEndianHost)
167    MachO::swapStruct(NList);
168
169  NewSymtab.append(reinterpret_cast<char *>(&NList),
170                   reinterpret_cast<char *>(&NList + 1));
171  return true;
172}
173
174// Wrapper around transferSymbol to transfer all of \a Obj symbols
175// to \a NewSymtab. This function does not write in the output file.
176// \returns the number of symbols in \a NewSymtab.
177static unsigned transferSymbols(const object::MachOObjectFile &Obj,
178                                SmallVectorImpl<char> &NewSymtab,
179                                NonRelocatableStringpool &NewStrings) {
180  unsigned Syms = 0;
181  StringRef Strings = Obj.getStringTableData();
182  bool IsLittleEndian = Obj.isLittleEndian();
183  bool InDebugNote = false;
184
185  if (Obj.is64Bit()) {
186    for (const object::SymbolRef &Symbol : Obj.symbols()) {
187      object::DataRefImpl DRI = Symbol.getRawDataRefImpl();
188      if (transferSymbol(Obj.getSymbol64TableEntry(DRI), IsLittleEndian,
189                         Strings, NewSymtab, NewStrings, InDebugNote))
190        ++Syms;
191    }
192  } else {
193    for (const object::SymbolRef &Symbol : Obj.symbols()) {
194      object::DataRefImpl DRI = Symbol.getRawDataRefImpl();
195      if (transferSymbol(Obj.getSymbolTableEntry(DRI), IsLittleEndian, Strings,
196                         NewSymtab, NewStrings, InDebugNote))
197        ++Syms;
198    }
199  }
200  return Syms;
201}
202
203static MachO::section
204getSection(const object::MachOObjectFile &Obj,
205           const MachO::segment_command &Seg,
206           const object::MachOObjectFile::LoadCommandInfo &LCI, unsigned Idx) {
207  return Obj.getSection(LCI, Idx);
208}
209
210static MachO::section_64
211getSection(const object::MachOObjectFile &Obj,
212           const MachO::segment_command_64 &Seg,
213           const object::MachOObjectFile::LoadCommandInfo &LCI, unsigned Idx) {
214  return Obj.getSection64(LCI, Idx);
215}
216
217// Transfer \a Segment from \a Obj to the output file. This calls into \a Writer
218// to write these load commands directly in the output file at the current
219// position.
220// The function also tries to find a hole in the address map to fit the __DWARF
221// segment of \a DwarfSegmentSize size. \a EndAddress is updated to point at the
222// highest segment address.
223// When the __LINKEDIT segment is transfered, its offset and size are set resp.
224// to \a LinkeditOffset and \a LinkeditSize.
225template <typename SegmentTy>
226static void transferSegmentAndSections(
227    const object::MachOObjectFile::LoadCommandInfo &LCI, SegmentTy Segment,
228    const object::MachOObjectFile &Obj, MCObjectWriter &Writer,
229    uint64_t LinkeditOffset, uint64_t LinkeditSize, uint64_t DwarfSegmentSize,
230    uint64_t &GapForDwarf, uint64_t &EndAddress) {
231  if (StringRef("__DWARF") == Segment.segname)
232    return;
233
234  Segment.fileoff = Segment.filesize = 0;
235
236  if (StringRef("__LINKEDIT") == Segment.segname) {
237    Segment.fileoff = LinkeditOffset;
238    Segment.filesize = LinkeditSize;
239  }
240
241  // Check if the end address of the last segment and our current
242  // start address leave a sufficient gap to store the __DWARF
243  // segment.
244  uint64_t PrevEndAddress = EndAddress;
245  EndAddress = alignTo(EndAddress, 0x1000);
246  if (GapForDwarf == UINT64_MAX && Segment.vmaddr > EndAddress &&
247      Segment.vmaddr - EndAddress >= DwarfSegmentSize)
248    GapForDwarf = EndAddress;
249
250  // The segments are not necessarily sorted by their vmaddr.
251  EndAddress =
252      std::max<uint64_t>(PrevEndAddress, Segment.vmaddr + Segment.vmsize);
253  unsigned nsects = Segment.nsects;
254  if (Obj.isLittleEndian() != sys::IsLittleEndianHost)
255    MachO::swapStruct(Segment);
256  Writer.writeBytes(
257      StringRef(reinterpret_cast<char *>(&Segment), sizeof(Segment)));
258  for (unsigned i = 0; i < nsects; ++i) {
259    auto Sect = getSection(Obj, Segment, LCI, i);
260    Sect.offset = Sect.reloff = Sect.nreloc = 0;
261    if (Obj.isLittleEndian() != sys::IsLittleEndianHost)
262      MachO::swapStruct(Sect);
263    Writer.writeBytes(StringRef(reinterpret_cast<char *>(&Sect), sizeof(Sect)));
264  }
265}
266
267// Write the __DWARF segment load command to the output file.
268static void createDwarfSegment(uint64_t VMAddr, uint64_t FileOffset,
269                               uint64_t FileSize, unsigned NumSections,
270                               MCAsmLayout &Layout, MachObjectWriter &Writer) {
271  Writer.writeSegmentLoadCommand("__DWARF", NumSections, VMAddr,
272                                 alignTo(FileSize, 0x1000), FileOffset,
273                                 FileSize, /* MaxProt */ 7,
274                                 /* InitProt =*/3);
275
276  for (unsigned int i = 0, n = Layout.getSectionOrder().size(); i != n; ++i) {
277    MCSection *Sec = Layout.getSectionOrder()[i];
278    if (Sec->begin() == Sec->end() || !Layout.getSectionFileSize(Sec))
279      continue;
280
281    unsigned Align = Sec->getAlignment();
282    if (Align > 1) {
283      VMAddr = alignTo(VMAddr, Align);
284      FileOffset = alignTo(FileOffset, Align);
285    }
286    Writer.writeSection(Layout, *Sec, VMAddr, FileOffset, 0, 0, 0);
287
288    FileOffset += Layout.getSectionAddressSize(Sec);
289    VMAddr += Layout.getSectionAddressSize(Sec);
290  }
291}
292
293static bool isExecutable(const object::MachOObjectFile &Obj) {
294  if (Obj.is64Bit())
295    return Obj.getHeader64().filetype != MachO::MH_OBJECT;
296  else
297    return Obj.getHeader().filetype != MachO::MH_OBJECT;
298}
299
300static bool hasLinkEditSegment(const object::MachOObjectFile &Obj) {
301  bool HasLinkEditSegment = false;
302  iterateOnSegments(Obj, [&](const MachO::segment_command_64 &Segment) {
303    if (StringRef("__LINKEDIT") == Segment.segname)
304      HasLinkEditSegment = true;
305  });
306  return HasLinkEditSegment;
307}
308
309static unsigned segmentLoadCommandSize(bool Is64Bit, unsigned NumSections) {
310  if (Is64Bit)
311    return sizeof(MachO::segment_command_64) +
312           NumSections * sizeof(MachO::section_64);
313
314  return sizeof(MachO::segment_command) + NumSections * sizeof(MachO::section);
315}
316
317// Stream a dSYM companion binary file corresponding to the binary referenced
318// by \a DM to \a OutFile. The passed \a MS MCStreamer is setup to write to
319// \a OutFile and it must be using a MachObjectWriter object to do so.
320bool generateDsymCompanion(const DebugMap &DM, MCStreamer &MS,
321                           raw_fd_ostream &OutFile) {
322  auto &ObjectStreamer = static_cast<MCObjectStreamer &>(MS);
323  MCAssembler &MCAsm = ObjectStreamer.getAssembler();
324  auto &Writer = static_cast<MachObjectWriter &>(MCAsm.getWriter());
325  MCAsmLayout Layout(MCAsm);
326
327  MCAsm.layout(Layout);
328
329  BinaryHolder InputBinaryHolder(false);
330  auto ErrOrObjs = InputBinaryHolder.GetObjectFiles(DM.getBinaryPath());
331  if (auto Error = ErrOrObjs.getError())
332    return error(Twine("opening ") + DM.getBinaryPath() + ": " +
333                     Error.message(),
334                 "output file streaming");
335
336  auto ErrOrInputBinary =
337      InputBinaryHolder.GetAs<object::MachOObjectFile>(DM.getTriple());
338  if (auto Error = ErrOrInputBinary.getError())
339    return error(Twine("opening ") + DM.getBinaryPath() + ": " +
340                     Error.message(),
341                 "output file streaming");
342  auto &InputBinary = *ErrOrInputBinary;
343
344  bool Is64Bit = Writer.is64Bit();
345  MachO::symtab_command SymtabCmd = InputBinary.getSymtabLoadCommand();
346
347  // Get UUID.
348  MachO::uuid_command UUIDCmd;
349  memset(&UUIDCmd, 0, sizeof(UUIDCmd));
350  UUIDCmd.cmd = MachO::LC_UUID;
351  UUIDCmd.cmdsize = sizeof(MachO::uuid_command);
352  for (auto &LCI : InputBinary.load_commands()) {
353    if (LCI.C.cmd == MachO::LC_UUID) {
354      UUIDCmd = InputBinary.getUuidCommand(LCI);
355      break;
356    }
357  }
358
359  // Compute the number of load commands we will need.
360  unsigned LoadCommandSize = 0;
361  unsigned NumLoadCommands = 0;
362  // We will copy the UUID if there is one.
363  if (UUIDCmd.cmd != 0) {
364    ++NumLoadCommands;
365    LoadCommandSize += sizeof(MachO::uuid_command);
366  }
367
368  // If we have a valid symtab to copy, do it.
369  bool ShouldEmitSymtab =
370      isExecutable(InputBinary) && hasLinkEditSegment(InputBinary);
371  if (ShouldEmitSymtab) {
372    LoadCommandSize += sizeof(MachO::symtab_command);
373    ++NumLoadCommands;
374  }
375
376  unsigned HeaderSize =
377      Is64Bit ? sizeof(MachO::mach_header_64) : sizeof(MachO::mach_header);
378  // We will copy every segment that isn't __DWARF.
379  iterateOnSegments(InputBinary, [&](const MachO::segment_command_64 &Segment) {
380    if (StringRef("__DWARF") == Segment.segname)
381      return;
382
383    ++NumLoadCommands;
384    LoadCommandSize += segmentLoadCommandSize(Is64Bit, Segment.nsects);
385  });
386
387  // We will add our own brand new __DWARF segment if we have debug
388  // info.
389  unsigned NumDwarfSections = 0;
390  uint64_t DwarfSegmentSize = 0;
391
392  for (unsigned int i = 0, n = Layout.getSectionOrder().size(); i != n; ++i) {
393    MCSection *Sec = Layout.getSectionOrder()[i];
394    if (Sec->begin() == Sec->end())
395      continue;
396
397    if (uint64_t Size = Layout.getSectionFileSize(Sec)) {
398      DwarfSegmentSize = alignTo(DwarfSegmentSize, Sec->getAlignment());
399      DwarfSegmentSize += Size;
400      ++NumDwarfSections;
401    }
402  }
403
404  if (NumDwarfSections) {
405    ++NumLoadCommands;
406    LoadCommandSize += segmentLoadCommandSize(Is64Bit, NumDwarfSections);
407  }
408
409  SmallString<0> NewSymtab;
410  NonRelocatableStringpool NewStrings;
411  unsigned NListSize = Is64Bit ? sizeof(MachO::nlist_64) : sizeof(MachO::nlist);
412  unsigned NumSyms = 0;
413  uint64_t NewStringsSize = 0;
414  if (ShouldEmitSymtab) {
415    NewSymtab.reserve(SymtabCmd.nsyms * NListSize / 2);
416    NumSyms = transferSymbols(InputBinary, NewSymtab, NewStrings);
417    NewStringsSize = NewStrings.getSize() + 1;
418  }
419
420  uint64_t SymtabStart = LoadCommandSize;
421  SymtabStart += HeaderSize;
422  SymtabStart = alignTo(SymtabStart, 0x1000);
423
424  // We gathered all the information we need, start emitting the output file.
425  Writer.writeHeader(MachO::MH_DSYM, NumLoadCommands, LoadCommandSize, false);
426
427  // Write the load commands.
428  assert(OutFile.tell() == HeaderSize);
429  if (UUIDCmd.cmd != 0) {
430    Writer.write32(UUIDCmd.cmd);
431    Writer.write32(UUIDCmd.cmdsize);
432    Writer.writeBytes(
433        StringRef(reinterpret_cast<const char *>(UUIDCmd.uuid), 16));
434    assert(OutFile.tell() == HeaderSize + sizeof(UUIDCmd));
435  }
436
437  assert(SymtabCmd.cmd && "No symbol table.");
438  uint64_t StringStart = SymtabStart + NumSyms * NListSize;
439  if (ShouldEmitSymtab)
440    Writer.writeSymtabLoadCommand(SymtabStart, NumSyms, StringStart,
441                                  NewStringsSize);
442
443  uint64_t DwarfSegmentStart = StringStart + NewStringsSize;
444  DwarfSegmentStart = alignTo(DwarfSegmentStart, 0x1000);
445
446  // Write the load commands for the segments and sections we 'import' from
447  // the original binary.
448  uint64_t EndAddress = 0;
449  uint64_t GapForDwarf = UINT64_MAX;
450  for (auto &LCI : InputBinary.load_commands()) {
451    if (LCI.C.cmd == MachO::LC_SEGMENT)
452      transferSegmentAndSections(LCI, InputBinary.getSegmentLoadCommand(LCI),
453                                 InputBinary, Writer, SymtabStart,
454                                 StringStart + NewStringsSize - SymtabStart,
455                                 DwarfSegmentSize, GapForDwarf, EndAddress);
456    else if (LCI.C.cmd == MachO::LC_SEGMENT_64)
457      transferSegmentAndSections(LCI, InputBinary.getSegment64LoadCommand(LCI),
458                                 InputBinary, Writer, SymtabStart,
459                                 StringStart + NewStringsSize - SymtabStart,
460                                 DwarfSegmentSize, GapForDwarf, EndAddress);
461  }
462
463  uint64_t DwarfVMAddr = alignTo(EndAddress, 0x1000);
464  uint64_t DwarfVMMax = Is64Bit ? UINT64_MAX : UINT32_MAX;
465  if (DwarfVMAddr + DwarfSegmentSize > DwarfVMMax ||
466      DwarfVMAddr + DwarfSegmentSize < DwarfVMAddr /* Overflow */) {
467    // There is no room for the __DWARF segment at the end of the
468    // address space. Look trhough segments to find a gap.
469    DwarfVMAddr = GapForDwarf;
470    if (DwarfVMAddr == UINT64_MAX)
471      warn("not enough VM space for the __DWARF segment.",
472           "output file streaming");
473  }
474
475  // Write the load command for the __DWARF segment.
476  createDwarfSegment(DwarfVMAddr, DwarfSegmentStart, DwarfSegmentSize,
477                     NumDwarfSections, Layout, Writer);
478
479  assert(OutFile.tell() == LoadCommandSize + HeaderSize);
480  Writer.WriteZeros(SymtabStart - (LoadCommandSize + HeaderSize));
481  assert(OutFile.tell() == SymtabStart);
482
483  // Transfer symbols.
484  if (ShouldEmitSymtab) {
485    Writer.writeBytes(NewSymtab.str());
486    assert(OutFile.tell() == StringStart);
487
488    // Transfer string table.
489    // FIXME: The NonRelocatableStringpool starts with an empty string, but
490    // dsymutil-classic starts the reconstructed string table with 2 of these.
491    // Reproduce that behavior for now (there is corresponding code in
492    // transferSymbol).
493    Writer.WriteZeros(1);
494    typedef NonRelocatableStringpool::MapTy MapTy;
495    for (auto *Entry = NewStrings.getFirstEntry(); Entry;
496         Entry = static_cast<MapTy::MapEntryTy *>(Entry->getValue().second))
497      Writer.writeBytes(
498          StringRef(Entry->getKey().data(), Entry->getKey().size() + 1));
499  }
500
501  assert(OutFile.tell() == StringStart + NewStringsSize);
502
503  // Pad till the Dwarf segment start.
504  Writer.WriteZeros(DwarfSegmentStart - (StringStart + NewStringsSize));
505  assert(OutFile.tell() == DwarfSegmentStart);
506
507  // Emit the Dwarf sections contents.
508  for (const MCSection &Sec : MCAsm) {
509    if (Sec.begin() == Sec.end())
510      continue;
511
512    uint64_t Pos = OutFile.tell();
513    Writer.WriteZeros(alignTo(Pos, Sec.getAlignment()) - Pos);
514    MCAsm.writeSectionData(&Sec, Layout);
515  }
516
517  return true;
518}
519}
520}
521}
522