macho_combine_dwarf.go

     1  // Copyright 2015 The Go Authors. All rights reserved.
     2  // Use of this source code is governed by a BSD-style
     3  // license that can be found in the LICENSE file.
     4  
     5  package ld
     6  
     7  import (
     8  	"bytes"
     9  	"compress/zlib"
    10  	"debug/macho"
    11  	"encoding/binary"
    12  	"fmt"
    13  	"io"
    14  	"os"
    15  	"reflect"
    16  	"unsafe"
    17  )
    18  
    19  type loadCmd struct {
    20  	Cmd macho.LoadCmd
    21  	Len uint32
    22  }
    23  
    24  type dyldInfoCmd struct {
    25  	Cmd                      macho.LoadCmd
    26  	Len                      uint32
    27  	RebaseOff, RebaseLen     uint32
    28  	BindOff, BindLen         uint32
    29  	WeakBindOff, WeakBindLen uint32
    30  	LazyBindOff, LazyBindLen uint32
    31  	ExportOff, ExportLen     uint32
    32  }
    33  
    34  type linkEditDataCmd struct {
    35  	Cmd              macho.LoadCmd
    36  	Len              uint32
    37  	DataOff, DataLen uint32
    38  }
    39  
    40  type encryptionInfoCmd struct {
    41  	Cmd                macho.LoadCmd
    42  	Len                uint32
    43  	CryptOff, CryptLen uint32
    44  	CryptId            uint32
    45  }
    46  
    47  type uuidCmd struct {
    48  	Cmd  macho.LoadCmd
    49  	Len  uint32
    50  	Uuid [16]byte
    51  }
    52  
    53  type loadCmdReader struct {
    54  	offset, next int64
    55  	f            *os.File
    56  	order        binary.ByteOrder
    57  }
    58  
    59  func (r *loadCmdReader) Next() (loadCmd, error) {
    60  	var cmd loadCmd
    61  
    62  	r.offset = r.next
    63  	if _, err := r.f.Seek(r.offset, 0); err != nil {
    64  		return cmd, err
    65  	}
    66  	if err := binary.Read(r.f, r.order, &cmd); err != nil {
    67  		return cmd, err
    68  	}
    69  	r.next = r.offset + int64(cmd.Len)
    70  	return cmd, nil
    71  }
    72  
    73  func (r loadCmdReader) ReadAt(offset int64, data interface{}) error {
    74  	if _, err := r.f.Seek(r.offset+offset, 0); err != nil {
    75  		return err
    76  	}
    77  	return binary.Read(r.f, r.order, data)
    78  }
    79  
    80  func (r loadCmdReader) WriteAt(offset int64, data interface{}) error {
    81  	if _, err := r.f.Seek(r.offset+offset, 0); err != nil {
    82  		return err
    83  	}
    84  	return binary.Write(r.f, r.order, data)
    85  }
    86  
    87  // machoCombineDwarf merges dwarf info generated by dsymutil into a macho executable.
    88  //
    89  // With internal linking, DWARF is embedded into the executable, this lets us do the
    90  // same for external linking.
    91  // exef is the file of the executable with no DWARF. It must have enough room in the macho
    92  // header to add the DWARF sections. (Use ld's -headerpad option)
    93  // exem is the macho representation of exef.
    94  // dsym is the path to the macho file containing DWARF from dsymutil.
    95  // outexe is the path where the combined executable should be saved.
    96  func machoCombineDwarf(ctxt *Link, exef *os.File, exem *macho.File, dsym, outexe string) error {
    97  	dwarff, err := os.Open(dsym)
    98  	if err != nil {
    99  		return err
   100  	}
   101  	defer dwarff.Close()
   102  	outf, err := os.OpenFile(outexe, os.O_RDWR|os.O_CREATE|os.O_TRUNC, 0755)
   103  	if err != nil {
   104  		return err
   105  	}
   106  	defer outf.Close()
   107  	dwarfm, err := macho.NewFile(dwarff)
   108  	if err != nil {
   109  		return err
   110  	}
   111  	defer dwarfm.Close()
   112  
   113  	// The string table needs to be the last thing in the file
   114  	// for code signing to work. So we'll need to move the
   115  	// linkedit section, but all the others can be copied directly.
   116  	linkseg := exem.Segment("__LINKEDIT")
   117  	if linkseg == nil {
   118  		return fmt.Errorf("missing __LINKEDIT segment")
   119  	}
   120  
   121  	if _, err := exef.Seek(0, 0); err != nil {
   122  		return err
   123  	}
   124  	if _, err := io.CopyN(outf, exef, int64(linkseg.Offset)); err != nil {
   125  		return err
   126  	}
   127  
   128  	realdwarf := dwarfm.Segment("__DWARF")
   129  	if realdwarf == nil {
   130  		return fmt.Errorf("missing __DWARF segment")
   131  	}
   132  
   133  	// Try to compress the DWARF sections. This includes some Apple
   134  	// proprietary sections like __apple_types.
   135  	compressedSects, compressedBytes, err := machoCompressSections(ctxt, dwarfm)
   136  	if err != nil {
   137  		return err
   138  	}
   139  
   140  	// Now copy the dwarf data into the output.
   141  	// Kernel requires all loaded segments to be page-aligned in the file,
   142  	// even though we mark this one as being 0 bytes of virtual address space.
   143  	dwarfstart := Rnd(int64(linkseg.Offset), *FlagRound)
   144  	if _, err := outf.Seek(dwarfstart, 0); err != nil {
   145  		return err
   146  	}
   147  
   148  	if _, err := dwarff.Seek(int64(realdwarf.Offset), 0); err != nil {
   149  		return err
   150  	}
   151  
   152  	// Write out the compressed sections, or the originals if we gave up
   153  	// on compressing them.
   154  	var dwarfsize uint64
   155  	if compressedBytes != nil {
   156  		dwarfsize = uint64(len(compressedBytes))
   157  		if _, err := outf.Write(compressedBytes); err != nil {
   158  			return err
   159  		}
   160  	} else {
   161  		if _, err := io.CopyN(outf, dwarff, int64(realdwarf.Filesz)); err != nil {
   162  			return err
   163  		}
   164  		dwarfsize = realdwarf.Filesz
   165  	}
   166  
   167  	// And finally the linkedit section.
   168  	if _, err := exef.Seek(int64(linkseg.Offset), 0); err != nil {
   169  		return err
   170  	}
   171  	linkstart := Rnd(dwarfstart+int64(dwarfsize), *FlagRound)
   172  	if _, err := outf.Seek(linkstart, 0); err != nil {
   173  		return err
   174  	}
   175  	if _, err := io.Copy(outf, exef); err != nil {
   176  		return err
   177  	}
   178  
   179  	// Now we need to update the headers.
   180  	textsect := exem.Section("__text")
   181  	if textsect == nil {
   182  		return fmt.Errorf("missing __text section")
   183  	}
   184  
   185  	cmdOffset := unsafe.Sizeof(exem.FileHeader)
   186  	if is64bit := exem.Magic == macho.Magic64; is64bit {
   187  		// mach_header_64 has one extra uint32.
   188  		cmdOffset += unsafe.Sizeof(exem.Magic)
   189  	}
   190  	dwarfCmdOffset := uint32(cmdOffset) + exem.FileHeader.Cmdsz
   191  	availablePadding := textsect.Offset - dwarfCmdOffset
   192  	if availablePadding < realdwarf.Len {
   193  		return fmt.Errorf("no room to add dwarf info. Need at least %d padding bytes, found %d", realdwarf.Len, availablePadding)
   194  	}
   195  	// First, copy the dwarf load command into the header. It will be
   196  	// updated later with new offsets and lengths as necessary.
   197  	if _, err := outf.Seek(int64(dwarfCmdOffset), 0); err != nil {
   198  		return err
   199  	}
   200  	if _, err := io.CopyN(outf, bytes.NewReader(realdwarf.Raw()), int64(realdwarf.Len)); err != nil {
   201  		return err
   202  	}
   203  	if _, err := outf.Seek(int64(unsafe.Offsetof(exem.FileHeader.Ncmd)), 0); err != nil {
   204  		return err
   205  	}
   206  	if err := binary.Write(outf, exem.ByteOrder, exem.Ncmd+1); err != nil {
   207  		return err
   208  	}
   209  	if err := binary.Write(outf, exem.ByteOrder, exem.Cmdsz+realdwarf.Len); err != nil {
   210  		return err
   211  	}
   212  
   213  	reader := loadCmdReader{next: int64(cmdOffset), f: outf, order: exem.ByteOrder}
   214  	for i := uint32(0); i < exem.Ncmd; i++ {
   215  		cmd, err := reader.Next()
   216  		if err != nil {
   217  			return err
   218  		}
   219  		linkoffset := uint64(linkstart) - linkseg.Offset
   220  		switch cmd.Cmd {
   221  		case macho.LoadCmdSegment64:
   222  			err = machoUpdateSegment(reader, linkseg, linkoffset)
   223  		case macho.LoadCmdSegment:
   224  			panic("unexpected 32-bit segment")
   225  		case LC_DYLD_INFO, LC_DYLD_INFO_ONLY:
   226  			err = machoUpdateLoadCommand(reader, linkseg, linkoffset, &dyldInfoCmd{}, "RebaseOff", "BindOff", "WeakBindOff", "LazyBindOff", "ExportOff")
   227  		case macho.LoadCmdSymtab:
   228  			err = machoUpdateLoadCommand(reader, linkseg, linkoffset, &macho.SymtabCmd{}, "Symoff", "Stroff")
   229  		case macho.LoadCmdDysymtab:
   230  			err = machoUpdateLoadCommand(reader, linkseg, linkoffset, &macho.DysymtabCmd{}, "Tocoffset", "Modtaboff", "Extrefsymoff", "Indirectsymoff", "Extreloff", "Locreloff")
   231  		case LC_CODE_SIGNATURE, LC_SEGMENT_SPLIT_INFO, LC_FUNCTION_STARTS, LC_DATA_IN_CODE, LC_DYLIB_CODE_SIGN_DRS,
   232  			LC_DYLD_EXPORTS_TRIE, LC_DYLD_CHAINED_FIXUPS:
   233  			err = machoUpdateLoadCommand(reader, linkseg, linkoffset, &linkEditDataCmd{}, "DataOff")
   234  		case LC_ENCRYPTION_INFO, LC_ENCRYPTION_INFO_64:
   235  			err = machoUpdateLoadCommand(reader, linkseg, linkoffset, &encryptionInfoCmd{}, "CryptOff")
   236  		case LC_UUID:
   237  			var u uuidCmd
   238  			err = reader.ReadAt(0, &u)
   239  			if err == nil {
   240  				copy(u.Uuid[:], uuidFromGoBuildId(*flagBuildid))
   241  				err = reader.WriteAt(0, &u)
   242  			}
   243  		case macho.LoadCmdDylib, macho.LoadCmdThread, macho.LoadCmdUnixThread,
   244  			LC_PREBOUND_DYLIB, LC_VERSION_MIN_MACOSX, LC_VERSION_MIN_IPHONEOS, LC_SOURCE_VERSION,
   245  			LC_MAIN, LC_LOAD_DYLINKER, LC_LOAD_WEAK_DYLIB, LC_REEXPORT_DYLIB, LC_RPATH, LC_ID_DYLIB,
   246  			LC_SYMSEG, LC_LOADFVMLIB, LC_IDFVMLIB, LC_IDENT, LC_FVMFILE, LC_PREPAGE, LC_ID_DYLINKER,
   247  			LC_ROUTINES, LC_SUB_FRAMEWORK, LC_SUB_UMBRELLA, LC_SUB_CLIENT, LC_SUB_LIBRARY, LC_TWOLEVEL_HINTS,
   248  			LC_PREBIND_CKSUM, LC_ROUTINES_64, LC_LAZY_LOAD_DYLIB, LC_LOAD_UPWARD_DYLIB, LC_DYLD_ENVIRONMENT,
   249  			LC_LINKER_OPTION, LC_LINKER_OPTIMIZATION_HINT, LC_VERSION_MIN_TVOS, LC_VERSION_MIN_WATCHOS,
   250  			LC_VERSION_NOTE, LC_BUILD_VERSION:
   251  			// Nothing to update
   252  		default:
   253  			err = fmt.Errorf("unknown load command 0x%x (%s)", int(cmd.Cmd), cmd.Cmd)
   254  		}
   255  		if err != nil {
   256  			return err
   257  		}
   258  	}
   259  	// Do the final update of the DWARF segment's load command.
   260  	return machoUpdateDwarfHeader(&reader, compressedSects, dwarfsize, dwarfstart, realdwarf)
   261  }
   262  
   263  // machoCompressSections tries to compress the DWARF segments in dwarfm,
   264  // returning the updated sections and segment contents, nils if the sections
   265  // weren't compressed, or an error if there was a problem reading dwarfm.
   266  func machoCompressSections(ctxt *Link, dwarfm *macho.File) ([]*macho.Section, []byte, error) {
   267  	if !ctxt.compressDWARF {
   268  		return nil, nil, nil
   269  	}
   270  
   271  	dwarfseg := dwarfm.Segment("__DWARF")
   272  	var sects []*macho.Section
   273  	var buf bytes.Buffer
   274  
   275  	for _, sect := range dwarfm.Sections {
   276  		if sect.Seg != "__DWARF" {
   277  			continue
   278  		}
   279  
   280  		// As of writing, there are no relocations in dsymutil's output
   281  		// so there's no point in worrying about them. Bail out if that
   282  		// changes.
   283  		if sect.Nreloc != 0 {
   284  			return nil, nil, nil
   285  		}
   286  
   287  		data, err := sect.Data()
   288  		if err != nil {
   289  			return nil, nil, err
   290  		}
   291  
   292  		compressed, contents, err := machoCompressSection(data)
   293  		if err != nil {
   294  			return nil, nil, err
   295  		}
   296  
   297  		newSec := *sect
   298  		newSec.Offset = uint32(dwarfseg.Offset) + uint32(buf.Len())
   299  		newSec.Addr = dwarfseg.Addr + uint64(buf.Len())
   300  		if compressed {
   301  			newSec.Name = "__z" + sect.Name[2:]
   302  			newSec.Size = uint64(len(contents))
   303  		}
   304  		sects = append(sects, &newSec)
   305  		buf.Write(contents)
   306  	}
   307  	return sects, buf.Bytes(), nil
   308  }
   309  
   310  // machoCompressSection compresses secBytes if it results in less data.
   311  func machoCompressSection(sectBytes []byte) (compressed bool, contents []byte, err error) {
   312  	var buf bytes.Buffer
   313  	buf.WriteString("ZLIB")
   314  	var sizeBytes [8]byte
   315  	binary.BigEndian.PutUint64(sizeBytes[:], uint64(len(sectBytes)))
   316  	buf.Write(sizeBytes[:])
   317  
   318  	z := zlib.NewWriter(&buf)
   319  	if _, err := z.Write(sectBytes); err != nil {
   320  		return false, nil, err
   321  	}
   322  	if err := z.Close(); err != nil {
   323  		return false, nil, err
   324  	}
   325  	if buf.Len() >= len(sectBytes) {
   326  		return false, sectBytes, nil
   327  	}
   328  	return true, buf.Bytes(), nil
   329  }
   330  
   331  // machoUpdateSegment updates the load command for a moved segment.
   332  // Only the linkedit segment should move, and it should have 0 sections.
   333  func machoUpdateSegment(r loadCmdReader, linkseg *macho.Segment, linkoffset uint64) error {
   334  	var seg macho.Segment64
   335  	if err := r.ReadAt(0, &seg); err != nil {
   336  		return err
   337  	}
   338  
   339  	// Only the linkedit segment moved, anything before that is fine.
   340  	if seg.Offset < linkseg.Offset {
   341  		return nil
   342  	}
   343  	seg.Offset += linkoffset
   344  	if err := r.WriteAt(0, &seg); err != nil {
   345  		return err
   346  	}
   347  	// There shouldn't be any sections, but just to make sure...
   348  	return machoUpdateSections(r, &seg, linkoffset, nil)
   349  }
   350  
   351  func machoUpdateSections(r loadCmdReader, seg *macho.Segment64, deltaOffset uint64, compressedSects []*macho.Section) error {
   352  	nsect := seg.Nsect
   353  	if nsect == 0 {
   354  		return nil
   355  	}
   356  	sectOffset := int64(unsafe.Sizeof(*seg))
   357  
   358  	var sect macho.Section64
   359  	sectSize := int64(unsafe.Sizeof(sect))
   360  	for i := uint32(0); i < nsect; i++ {
   361  		if err := r.ReadAt(sectOffset, &sect); err != nil {
   362  			return err
   363  		}
   364  		if compressedSects != nil {
   365  			cSect := compressedSects[i]
   366  			copy(sect.Name[:], cSect.Name)
   367  			sect.Size = cSect.Size
   368  			if cSect.Offset != 0 {
   369  				sect.Offset = cSect.Offset + uint32(deltaOffset)
   370  			}
   371  			if cSect.Addr != 0 {
   372  				sect.Addr = cSect.Addr
   373  			}
   374  		} else {
   375  			if sect.Offset != 0 {
   376  				sect.Offset += uint32(deltaOffset)
   377  			}
   378  			if sect.Reloff != 0 {
   379  				sect.Reloff += uint32(deltaOffset)
   380  			}
   381  		}
   382  		if err := r.WriteAt(sectOffset, &sect); err != nil {
   383  			return err
   384  		}
   385  		sectOffset += sectSize
   386  	}
   387  	return nil
   388  }
   389  
   390  // machoUpdateDwarfHeader updates the DWARF segment load command.
   391  func machoUpdateDwarfHeader(r *loadCmdReader, compressedSects []*macho.Section, dwarfsize uint64, dwarfstart int64, realdwarf *macho.Segment) error {
   392  	cmd, err := r.Next()
   393  	if err != nil {
   394  		return err
   395  	}
   396  	if cmd.Cmd != macho.LoadCmdSegment64 {
   397  		panic("not a Segment64")
   398  	}
   399  	var seg macho.Segment64
   400  	if err := r.ReadAt(0, &seg); err != nil {
   401  		return err
   402  	}
   403  	seg.Offset = uint64(dwarfstart)
   404  
   405  	if compressedSects != nil {
   406  		var segSize uint64
   407  		for _, newSect := range compressedSects {
   408  			segSize += newSect.Size
   409  		}
   410  		seg.Filesz = segSize
   411  	} else {
   412  		seg.Filesz = dwarfsize
   413  	}
   414  
   415  	// We want the DWARF segment to be considered non-loadable, so
   416  	// force vmaddr and vmsize to zero. In addition, set the initial
   417  	// protection to zero so as to make the dynamic loader happy,
   418  	// since otherwise it may complain that the vm size and file
   419  	// size don't match for the segment. See issues 21647 and 32673
   420  	// for more context. Also useful to refer to the Apple dynamic
   421  	// loader source, specifically ImageLoaderMachO::sniffLoadCommands
   422  	// in ImageLoaderMachO.cpp (various versions can be found online, see
   423  	// https://opensource.apple.com/source/dyld/dyld-519.2.2/src/ImageLoaderMachO.cpp.auto.html
   424  	// as one example).
   425  	seg.Addr = 0
   426  	seg.Memsz = 0
   427  	seg.Prot = 0
   428  
   429  	if err := r.WriteAt(0, &seg); err != nil {
   430  		return err
   431  	}
   432  	return machoUpdateSections(*r, &seg, uint64(dwarfstart)-realdwarf.Offset, compressedSects)
   433  }
   434  
   435  func machoUpdateLoadCommand(r loadCmdReader, linkseg *macho.Segment, linkoffset uint64, cmd interface{}, fields ...string) error {
   436  	if err := r.ReadAt(0, cmd); err != nil {
   437  		return err
   438  	}
   439  	value := reflect.Indirect(reflect.ValueOf(cmd))
   440  
   441  	for _, name := range fields {
   442  		field := value.FieldByName(name)
   443  		if fieldval := field.Uint(); fieldval >= linkseg.Offset {
   444  			field.SetUint(fieldval + linkoffset)
   445  		}
   446  	}
   447  	if err := r.WriteAt(0, cmd); err != nil {
   448  		return err
   449  	}
   450  	return nil
   451  }
   452
View as plain text