ldpe.go

     1  // Copyright 2010 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 loadpe implements a PE/COFF file reader.
     6  package loadpe
     7  
     8  import (
     9  	"bytes"
    10  	"cmd/internal/bio"
    11  	"cmd/internal/objabi"
    12  	"cmd/internal/sys"
    13  	"cmd/link/internal/loader"
    14  	"cmd/link/internal/sym"
    15  	"debug/pe"
    16  	"encoding/binary"
    17  	"errors"
    18  	"fmt"
    19  	"io"
    20  	"strings"
    21  )
    22  
    23  const (
    24  	IMAGE_SYM_UNDEFINED              = 0
    25  	IMAGE_SYM_ABSOLUTE               = -1
    26  	IMAGE_SYM_DEBUG                  = -2
    27  	IMAGE_SYM_TYPE_NULL              = 0
    28  	IMAGE_SYM_TYPE_VOID              = 1
    29  	IMAGE_SYM_TYPE_CHAR              = 2
    30  	IMAGE_SYM_TYPE_SHORT             = 3
    31  	IMAGE_SYM_TYPE_INT               = 4
    32  	IMAGE_SYM_TYPE_LONG              = 5
    33  	IMAGE_SYM_TYPE_FLOAT             = 6
    34  	IMAGE_SYM_TYPE_DOUBLE            = 7
    35  	IMAGE_SYM_TYPE_STRUCT            = 8
    36  	IMAGE_SYM_TYPE_UNION             = 9
    37  	IMAGE_SYM_TYPE_ENUM              = 10
    38  	IMAGE_SYM_TYPE_MOE               = 11
    39  	IMAGE_SYM_TYPE_BYTE              = 12
    40  	IMAGE_SYM_TYPE_WORD              = 13
    41  	IMAGE_SYM_TYPE_UINT              = 14
    42  	IMAGE_SYM_TYPE_DWORD             = 15
    43  	IMAGE_SYM_TYPE_PCODE             = 32768
    44  	IMAGE_SYM_DTYPE_NULL             = 0
    45  	IMAGE_SYM_DTYPE_POINTER          = 1
    46  	IMAGE_SYM_DTYPE_FUNCTION         = 2
    47  	IMAGE_SYM_DTYPE_ARRAY            = 3
    48  	IMAGE_SYM_CLASS_END_OF_FUNCTION  = -1
    49  	IMAGE_SYM_CLASS_NULL             = 0
    50  	IMAGE_SYM_CLASS_AUTOMATIC        = 1
    51  	IMAGE_SYM_CLASS_EXTERNAL         = 2
    52  	IMAGE_SYM_CLASS_STATIC           = 3
    53  	IMAGE_SYM_CLASS_REGISTER         = 4
    54  	IMAGE_SYM_CLASS_EXTERNAL_DEF     = 5
    55  	IMAGE_SYM_CLASS_LABEL            = 6
    56  	IMAGE_SYM_CLASS_UNDEFINED_LABEL  = 7
    57  	IMAGE_SYM_CLASS_MEMBER_OF_STRUCT = 8
    58  	IMAGE_SYM_CLASS_ARGUMENT         = 9
    59  	IMAGE_SYM_CLASS_STRUCT_TAG       = 10
    60  	IMAGE_SYM_CLASS_MEMBER_OF_UNION  = 11
    61  	IMAGE_SYM_CLASS_UNION_TAG        = 12
    62  	IMAGE_SYM_CLASS_TYPE_DEFINITION  = 13
    63  	IMAGE_SYM_CLASS_UNDEFINED_STATIC = 14
    64  	IMAGE_SYM_CLASS_ENUM_TAG         = 15
    65  	IMAGE_SYM_CLASS_MEMBER_OF_ENUM   = 16
    66  	IMAGE_SYM_CLASS_REGISTER_PARAM   = 17
    67  	IMAGE_SYM_CLASS_BIT_FIELD        = 18
    68  	IMAGE_SYM_CLASS_FAR_EXTERNAL     = 68 /* Not in PECOFF v8 spec */
    69  	IMAGE_SYM_CLASS_BLOCK            = 100
    70  	IMAGE_SYM_CLASS_FUNCTION         = 101
    71  	IMAGE_SYM_CLASS_END_OF_STRUCT    = 102
    72  	IMAGE_SYM_CLASS_FILE             = 103
    73  	IMAGE_SYM_CLASS_SECTION          = 104
    74  	IMAGE_SYM_CLASS_WEAK_EXTERNAL    = 105
    75  	IMAGE_SYM_CLASS_CLR_TOKEN        = 107
    76  	IMAGE_REL_I386_ABSOLUTE          = 0x0000
    77  	IMAGE_REL_I386_DIR16             = 0x0001
    78  	IMAGE_REL_I386_REL16             = 0x0002
    79  	IMAGE_REL_I386_DIR32             = 0x0006
    80  	IMAGE_REL_I386_DIR32NB           = 0x0007
    81  	IMAGE_REL_I386_SEG12             = 0x0009
    82  	IMAGE_REL_I386_SECTION           = 0x000A
    83  	IMAGE_REL_I386_SECREL            = 0x000B
    84  	IMAGE_REL_I386_TOKEN             = 0x000C
    85  	IMAGE_REL_I386_SECREL7           = 0x000D
    86  	IMAGE_REL_I386_REL32             = 0x0014
    87  	IMAGE_REL_AMD64_ABSOLUTE         = 0x0000
    88  	IMAGE_REL_AMD64_ADDR64           = 0x0001
    89  	IMAGE_REL_AMD64_ADDR32           = 0x0002
    90  	IMAGE_REL_AMD64_ADDR32NB         = 0x0003
    91  	IMAGE_REL_AMD64_REL32            = 0x0004
    92  	IMAGE_REL_AMD64_REL32_1          = 0x0005
    93  	IMAGE_REL_AMD64_REL32_2          = 0x0006
    94  	IMAGE_REL_AMD64_REL32_3          = 0x0007
    95  	IMAGE_REL_AMD64_REL32_4          = 0x0008
    96  	IMAGE_REL_AMD64_REL32_5          = 0x0009
    97  	IMAGE_REL_AMD64_SECTION          = 0x000A
    98  	IMAGE_REL_AMD64_SECREL           = 0x000B
    99  	IMAGE_REL_AMD64_SECREL7          = 0x000C
   100  	IMAGE_REL_AMD64_TOKEN            = 0x000D
   101  	IMAGE_REL_AMD64_SREL32           = 0x000E
   102  	IMAGE_REL_AMD64_PAIR             = 0x000F
   103  	IMAGE_REL_AMD64_SSPAN32          = 0x0010
   104  	IMAGE_REL_ARM_ABSOLUTE           = 0x0000
   105  	IMAGE_REL_ARM_ADDR32             = 0x0001
   106  	IMAGE_REL_ARM_ADDR32NB           = 0x0002
   107  	IMAGE_REL_ARM_BRANCH24           = 0x0003
   108  	IMAGE_REL_ARM_BRANCH11           = 0x0004
   109  	IMAGE_REL_ARM_SECTION            = 0x000E
   110  	IMAGE_REL_ARM_SECREL             = 0x000F
   111  	IMAGE_REL_ARM_MOV32              = 0x0010
   112  	IMAGE_REL_THUMB_MOV32            = 0x0011
   113  	IMAGE_REL_THUMB_BRANCH20         = 0x0012
   114  	IMAGE_REL_THUMB_BRANCH24         = 0x0014
   115  	IMAGE_REL_THUMB_BLX23            = 0x0015
   116  	IMAGE_REL_ARM_PAIR               = 0x0016
   117  	IMAGE_REL_ARM64_ABSOLUTE         = 0x0000
   118  	IMAGE_REL_ARM64_ADDR32           = 0x0001
   119  	IMAGE_REL_ARM64_ADDR32NB         = 0x0002
   120  	IMAGE_REL_ARM64_BRANCH26         = 0x0003
   121  	IMAGE_REL_ARM64_PAGEBASE_REL21   = 0x0004
   122  	IMAGE_REL_ARM64_REL21            = 0x0005
   123  	IMAGE_REL_ARM64_PAGEOFFSET_12A   = 0x0006
   124  	IMAGE_REL_ARM64_PAGEOFFSET_12L   = 0x0007
   125  	IMAGE_REL_ARM64_SECREL           = 0x0008
   126  	IMAGE_REL_ARM64_SECREL_LOW12A    = 0x0009
   127  	IMAGE_REL_ARM64_SECREL_HIGH12A   = 0x000A
   128  	IMAGE_REL_ARM64_SECREL_LOW12L    = 0x000B
   129  	IMAGE_REL_ARM64_TOKEN            = 0x000C
   130  	IMAGE_REL_ARM64_SECTION          = 0x000D
   131  	IMAGE_REL_ARM64_ADDR64           = 0x000E
   132  	IMAGE_REL_ARM64_BRANCH19         = 0x000F
   133  	IMAGE_REL_ARM64_BRANCH14         = 0x0010
   134  	IMAGE_REL_ARM64_REL32            = 0x0011
   135  )
   136  
   137  const (
   138  	// When stored into the PLT value for a symbol, this token tells
   139  	// windynrelocsym to redirect direct references to this symbol to a stub
   140  	// that loads from the corresponding import symbol and then does
   141  	// a jump to the loaded value.
   142  	CreateImportStubPltToken = -2
   143  
   144  	// When stored into the GOT value for an import symbol __imp_X this
   145  	// token tells windynrelocsym to redirect references to the
   146  	// underlying DYNIMPORT symbol X.
   147  	RedirectToDynImportGotToken = -2
   148  )
   149  
   150  // TODO(brainman): maybe just add ReadAt method to bio.Reader instead of creating peBiobuf
   151  
   152  // peBiobuf makes bio.Reader look like io.ReaderAt.
   153  type peBiobuf bio.Reader
   154  
   155  func (f *peBiobuf) ReadAt(p []byte, off int64) (int, error) {
   156  	ret := ((*bio.Reader)(f)).MustSeek(off, 0)
   157  	if ret < 0 {
   158  		return 0, errors.New("fail to seek")
   159  	}
   160  	n, err := f.Read(p)
   161  	if err != nil {
   162  		return 0, err
   163  	}
   164  	return n, nil
   165  }
   166  
   167  // makeUpdater creates a loader.SymbolBuilder if one hasn't been created previously.
   168  // We use this to lazily make SymbolBuilders as we don't always need a builder, and creating them for all symbols might be an error.
   169  func makeUpdater(l *loader.Loader, bld *loader.SymbolBuilder, s loader.Sym) *loader.SymbolBuilder {
   170  	if bld != nil {
   171  		return bld
   172  	}
   173  	bld = l.MakeSymbolUpdater(s)
   174  	return bld
   175  }
   176  
   177  // peImportSymsState tracks the set of DLL import symbols we've seen
   178  // while reading host objects. We create a singleton instance of this
   179  // type, which will persist across multiple host objects.
   180  type peImportSymsState struct {
   181  
   182  	// Text and non-text sections read in by the host object loader.
   183  	secSyms []loader.Sym
   184  
   185  	// Loader and arch, for use in postprocessing.
   186  	l    *loader.Loader
   187  	arch *sys.Arch
   188  }
   189  
   190  var importSymsState *peImportSymsState
   191  
   192  func createImportSymsState(l *loader.Loader, arch *sys.Arch) {
   193  	if importSymsState != nil {
   194  		return
   195  	}
   196  	importSymsState = &peImportSymsState{
   197  		l:    l,
   198  		arch: arch,
   199  	}
   200  }
   201  
   202  // peLoaderState holds various bits of useful state information needed
   203  // while loading a single PE object file.
   204  type peLoaderState struct {
   205  	l               *loader.Loader
   206  	arch            *sys.Arch
   207  	f               *pe.File
   208  	pn              string
   209  	sectsyms        map[*pe.Section]loader.Sym
   210  	comdats         map[uint16]int64 // key is section index, val is size
   211  	sectdata        map[*pe.Section][]byte
   212  	localSymVersion int
   213  }
   214  
   215  // comdatDefinitions records the names of symbols for which we've
   216  // previously seen a definition in COMDAT. Key is symbol name, value
   217  // is symbol size (or -1 if we're using the "any" strategy).
   218  var comdatDefinitions map[string]int64
   219  
   220  // Symbols contains the symbols that can be loaded from a PE file.
   221  type Symbols struct {
   222  	Textp     []loader.Sym // text symbols
   223  	Resources []loader.Sym // .rsrc section or set of .rsrc$xx sections
   224  	PData     loader.Sym
   225  	XData     loader.Sym
   226  }
   227  
   228  // Load loads the PE file pn from input.
   229  // Symbols from the object file are created via the loader 'l'.
   230  func Load(l *loader.Loader, arch *sys.Arch, localSymVersion int, input *bio.Reader, pkg string, length int64, pn string) (*Symbols, error) {
   231  	state := &peLoaderState{
   232  		l:               l,
   233  		arch:            arch,
   234  		sectsyms:        make(map[*pe.Section]loader.Sym),
   235  		sectdata:        make(map[*pe.Section][]byte),
   236  		localSymVersion: localSymVersion,
   237  		pn:              pn,
   238  	}
   239  	createImportSymsState(state.l, state.arch)
   240  	if comdatDefinitions == nil {
   241  		comdatDefinitions = make(map[string]int64)
   242  	}
   243  
   244  	// Some input files are archives containing multiple of
   245  	// object files, and pe.NewFile seeks to the start of
   246  	// input file and get confused. Create section reader
   247  	// to stop pe.NewFile looking before current position.
   248  	sr := io.NewSectionReader((*peBiobuf)(input), input.Offset(), 1<<63-1)
   249  
   250  	// TODO: replace pe.NewFile with pe.Load (grep for "add Load function" in debug/pe for details)
   251  	f, err := pe.NewFile(sr)
   252  	if err != nil {
   253  		return nil, err
   254  	}
   255  	defer f.Close()
   256  	state.f = f
   257  
   258  	var ls Symbols
   259  
   260  	// TODO return error if found .cormeta
   261  
   262  	// create symbols for mapped sections
   263  	for _, sect := range f.Sections {
   264  		if sect.Characteristics&pe.IMAGE_SCN_MEM_DISCARDABLE != 0 {
   265  			continue
   266  		}
   267  
   268  		if sect.Characteristics&(pe.IMAGE_SCN_CNT_CODE|pe.IMAGE_SCN_CNT_INITIALIZED_DATA|pe.IMAGE_SCN_CNT_UNINITIALIZED_DATA) == 0 {
   269  			// This has been seen for .idata sections, which we
   270  			// want to ignore. See issues 5106 and 5273.
   271  			continue
   272  		}
   273  
   274  		name := fmt.Sprintf("%s(%s)", pkg, sect.Name)
   275  		s := state.l.LookupOrCreateCgoExport(name, localSymVersion)
   276  		bld := l.MakeSymbolUpdater(s)
   277  
   278  		switch sect.Characteristics & (pe.IMAGE_SCN_CNT_UNINITIALIZED_DATA | pe.IMAGE_SCN_CNT_INITIALIZED_DATA | pe.IMAGE_SCN_MEM_READ | pe.IMAGE_SCN_MEM_WRITE | pe.IMAGE_SCN_CNT_CODE | pe.IMAGE_SCN_MEM_EXECUTE) {
   279  		case pe.IMAGE_SCN_CNT_INITIALIZED_DATA | pe.IMAGE_SCN_MEM_READ: //.rdata
   280  			if issehsect(arch, sect) {
   281  				bld.SetType(sym.SSEHSECT)
   282  				bld.SetAlign(4)
   283  			} else {
   284  				bld.SetType(sym.SRODATA)
   285  			}
   286  
   287  		case pe.IMAGE_SCN_CNT_UNINITIALIZED_DATA | pe.IMAGE_SCN_MEM_READ | pe.IMAGE_SCN_MEM_WRITE: //.bss
   288  			bld.SetType(sym.SNOPTRBSS)
   289  
   290  		case pe.IMAGE_SCN_CNT_INITIALIZED_DATA | pe.IMAGE_SCN_MEM_READ | pe.IMAGE_SCN_MEM_WRITE: //.data
   291  			bld.SetType(sym.SNOPTRDATA)
   292  
   293  		case pe.IMAGE_SCN_CNT_CODE | pe.IMAGE_SCN_MEM_EXECUTE | pe.IMAGE_SCN_MEM_READ: //.text
   294  			bld.SetType(sym.STEXT)
   295  
   296  		default:
   297  			return nil, fmt.Errorf("unexpected flags %#06x for PE section %s", sect.Characteristics, sect.Name)
   298  		}
   299  
   300  		if bld.Type() != sym.SNOPTRBSS {
   301  			data, err := sect.Data()
   302  			if err != nil {
   303  				return nil, err
   304  			}
   305  			state.sectdata[sect] = data
   306  			bld.SetData(data)
   307  		}
   308  		bld.SetSize(int64(sect.Size))
   309  		state.sectsyms[sect] = s
   310  		if sect.Name == ".rsrc" || strings.HasPrefix(sect.Name, ".rsrc$") {
   311  			ls.Resources = append(ls.Resources, s)
   312  		} else if bld.Type() == sym.SSEHSECT {
   313  			if sect.Name == ".pdata" {
   314  				ls.PData = s
   315  			} else if sect.Name == ".xdata" {
   316  				ls.XData = s
   317  			}
   318  		}
   319  	}
   320  
   321  	// Make a prepass over the symbols to collect info about COMDAT symbols.
   322  	if err := state.preprocessSymbols(); err != nil {
   323  		return nil, err
   324  	}
   325  
   326  	// load relocations
   327  	for _, rsect := range f.Sections {
   328  		if _, found := state.sectsyms[rsect]; !found {
   329  			continue
   330  		}
   331  		if rsect.NumberOfRelocations == 0 {
   332  			continue
   333  		}
   334  		if rsect.Characteristics&pe.IMAGE_SCN_MEM_DISCARDABLE != 0 {
   335  			continue
   336  		}
   337  		if rsect.Characteristics&(pe.IMAGE_SCN_CNT_CODE|pe.IMAGE_SCN_CNT_INITIALIZED_DATA|pe.IMAGE_SCN_CNT_UNINITIALIZED_DATA) == 0 {
   338  			// This has been seen for .idata sections, which we
   339  			// want to ignore. See issues 5106 and 5273.
   340  			continue
   341  		}
   342  
   343  		splitResources := strings.HasPrefix(rsect.Name, ".rsrc$")
   344  		issehsect := issehsect(arch, rsect)
   345  		sb := l.MakeSymbolUpdater(state.sectsyms[rsect])
   346  		for j, r := range rsect.Relocs {
   347  			if int(r.SymbolTableIndex) >= len(f.COFFSymbols) {
   348  				return nil, fmt.Errorf("relocation number %d symbol index idx=%d cannot be large then number of symbols %d", j, r.SymbolTableIndex, len(f.COFFSymbols))
   349  			}
   350  			pesym := &f.COFFSymbols[r.SymbolTableIndex]
   351  			_, gosym, err := state.readpesym(pesym)
   352  			if err != nil {
   353  				return nil, err
   354  			}
   355  			if gosym == 0 {
   356  				name, err := pesym.FullName(f.StringTable)
   357  				if err != nil {
   358  					name = string(pesym.Name[:])
   359  				}
   360  				return nil, fmt.Errorf("reloc of invalid sym %s idx=%d type=%d", name, r.SymbolTableIndex, pesym.Type)
   361  			}
   362  
   363  			rSym := gosym
   364  			rSize := uint8(4)
   365  			rOff := int32(r.VirtualAddress)
   366  			var rAdd int64
   367  			var rType objabi.RelocType
   368  			switch arch.Family {
   369  			default:
   370  				return nil, fmt.Errorf("%s: unsupported arch %v", pn, arch.Family)
   371  			case sys.I386, sys.AMD64:
   372  				switch r.Type {
   373  				default:
   374  					return nil, fmt.Errorf("%s: %v: unknown relocation type %v", pn, state.sectsyms[rsect], r.Type)
   375  
   376  				case IMAGE_REL_I386_REL32, IMAGE_REL_AMD64_REL32,
   377  					IMAGE_REL_AMD64_ADDR32, // R_X86_64_PC32
   378  					IMAGE_REL_AMD64_ADDR32NB:
   379  					if r.Type == IMAGE_REL_AMD64_ADDR32NB {
   380  						rType = objabi.R_PEIMAGEOFF
   381  					} else {
   382  						rType = objabi.R_PCREL
   383  					}
   384  
   385  					rAdd = int64(int32(binary.LittleEndian.Uint32(state.sectdata[rsect][rOff:])))
   386  
   387  				case IMAGE_REL_I386_DIR32NB, IMAGE_REL_I386_DIR32:
   388  					if r.Type == IMAGE_REL_I386_DIR32NB {
   389  						rType = objabi.R_PEIMAGEOFF
   390  					} else {
   391  						rType = objabi.R_ADDR
   392  					}
   393  
   394  					// load addend from image
   395  					rAdd = int64(int32(binary.LittleEndian.Uint32(state.sectdata[rsect][rOff:])))
   396  
   397  				case IMAGE_REL_AMD64_ADDR64: // R_X86_64_64
   398  					rSize = 8
   399  
   400  					rType = objabi.R_ADDR
   401  
   402  					// load addend from image
   403  					rAdd = int64(binary.LittleEndian.Uint64(state.sectdata[rsect][rOff:]))
   404  				}
   405  
   406  			case sys.ARM:
   407  				switch r.Type {
   408  				default:
   409  					return nil, fmt.Errorf("%s: %v: unknown ARM relocation type %v", pn, state.sectsyms[rsect], r.Type)
   410  
   411  				case IMAGE_REL_ARM_SECREL:
   412  					rType = objabi.R_PCREL
   413  
   414  					rAdd = int64(int32(binary.LittleEndian.Uint32(state.sectdata[rsect][rOff:])))
   415  
   416  				case IMAGE_REL_ARM_ADDR32, IMAGE_REL_ARM_ADDR32NB:
   417  					if r.Type == IMAGE_REL_ARM_ADDR32NB {
   418  						rType = objabi.R_PEIMAGEOFF
   419  					} else {
   420  						rType = objabi.R_ADDR
   421  					}
   422  
   423  					rAdd = int64(int32(binary.LittleEndian.Uint32(state.sectdata[rsect][rOff:])))
   424  
   425  				case IMAGE_REL_ARM_BRANCH24:
   426  					rType = objabi.R_CALLARM
   427  
   428  					rAdd = int64(int32(binary.LittleEndian.Uint32(state.sectdata[rsect][rOff:])))
   429  				}
   430  
   431  			case sys.ARM64:
   432  				switch r.Type {
   433  				default:
   434  					return nil, fmt.Errorf("%s: %v: unknown ARM64 relocation type %v", pn, state.sectsyms[rsect], r.Type)
   435  
   436  				case IMAGE_REL_ARM64_ADDR32, IMAGE_REL_ARM64_ADDR32NB:
   437  					if r.Type == IMAGE_REL_ARM64_ADDR32NB {
   438  						rType = objabi.R_PEIMAGEOFF
   439  					} else {
   440  						rType = objabi.R_ADDR
   441  					}
   442  
   443  					rAdd = int64(int32(binary.LittleEndian.Uint32(state.sectdata[rsect][rOff:])))
   444  				}
   445  			}
   446  
   447  			// ld -r could generate multiple section symbols for the
   448  			// same section but with different values, we have to take
   449  			// that into account, or in the case of split resources,
   450  			// the section and its symbols are split into two sections.
   451  			if issect(pesym) || splitResources {
   452  				rAdd += int64(pesym.Value)
   453  			}
   454  			if issehsect {
   455  				// .pdata and .xdata sections can contain records
   456  				// associated to functions that won't be used in
   457  				// the final binary, in which case the relocation
   458  				// target symbol won't be reachable.
   459  				rType |= objabi.R_WEAK
   460  			}
   461  
   462  			rel, _ := sb.AddRel(rType)
   463  			rel.SetOff(rOff)
   464  			rel.SetSiz(rSize)
   465  			rel.SetSym(rSym)
   466  			rel.SetAdd(rAdd)
   467  
   468  		}
   469  
   470  		sb.SortRelocs()
   471  	}
   472  
   473  	// enter sub-symbols into symbol table.
   474  	for i, numaux := 0, 0; i < len(f.COFFSymbols); i += numaux + 1 {
   475  		pesym := &f.COFFSymbols[i]
   476  
   477  		numaux = int(pesym.NumberOfAuxSymbols)
   478  
   479  		name, err := pesym.FullName(f.StringTable)
   480  		if err != nil {
   481  			return nil, err
   482  		}
   483  		if name == "" {
   484  			continue
   485  		}
   486  		if issect(pesym) {
   487  			continue
   488  		}
   489  		if int(pesym.SectionNumber) > len(f.Sections) {
   490  			continue
   491  		}
   492  		if pesym.SectionNumber == IMAGE_SYM_DEBUG {
   493  			continue
   494  		}
   495  		if pesym.SectionNumber == IMAGE_SYM_ABSOLUTE && bytes.Equal(pesym.Name[:], []byte("@feat.00")) {
   496  			// The PE documentation says that, on x86 platforms, the absolute symbol named @feat.00
   497  			// is used to indicate that the COFF object supports SEH.
   498  			// Go doesn't support SEH on windows/386, so we can ignore this symbol.
   499  			// See https://learn.microsoft.com/en-us/windows/win32/debug/pe-format#the-sxdata-section
   500  			continue
   501  		}
   502  		var sect *pe.Section
   503  		if pesym.SectionNumber > 0 {
   504  			sect = f.Sections[pesym.SectionNumber-1]
   505  			if _, found := state.sectsyms[sect]; !found {
   506  				continue
   507  			}
   508  		}
   509  
   510  		bld, s, err := state.readpesym(pesym)
   511  		if err != nil {
   512  			return nil, err
   513  		}
   514  
   515  		if pesym.SectionNumber == 0 { // extern
   516  			if l.SymType(s) == sym.SXREF && pesym.Value > 0 { // global data
   517  				bld = makeUpdater(l, bld, s)
   518  				bld.SetType(sym.SNOPTRDATA)
   519  				bld.SetSize(int64(pesym.Value))
   520  			}
   521  
   522  			continue
   523  		} else if pesym.SectionNumber > 0 && int(pesym.SectionNumber) <= len(f.Sections) {
   524  			sect = f.Sections[pesym.SectionNumber-1]
   525  			if _, found := state.sectsyms[sect]; !found {
   526  				return nil, fmt.Errorf("%s: %v: missing sect.sym", pn, s)
   527  			}
   528  		} else {
   529  			return nil, fmt.Errorf("%s: %v: sectnum < 0!", pn, s)
   530  		}
   531  
   532  		if sect == nil {
   533  			return nil, nil
   534  		}
   535  
   536  		// Check for COMDAT symbol.
   537  		if sz, ok1 := state.comdats[uint16(pesym.SectionNumber-1)]; ok1 {
   538  			if psz, ok2 := comdatDefinitions[l.SymName(s)]; ok2 {
   539  				if sz == psz {
   540  					//  OK to discard, we've seen an instance
   541  					// already.
   542  					continue
   543  				}
   544  			}
   545  		}
   546  		if l.OuterSym(s) != 0 {
   547  			if l.AttrDuplicateOK(s) {
   548  				continue
   549  			}
   550  			outerName := l.SymName(l.OuterSym(s))
   551  			sectName := l.SymName(state.sectsyms[sect])
   552  			return nil, fmt.Errorf("%s: duplicate symbol reference: %s in both %s and %s", pn, l.SymName(s), outerName, sectName)
   553  		}
   554  
   555  		bld = makeUpdater(l, bld, s)
   556  		sectsym := state.sectsyms[sect]
   557  		bld.SetType(l.SymType(sectsym))
   558  		l.AddInteriorSym(sectsym, s)
   559  		bld.SetValue(int64(pesym.Value))
   560  		bld.SetSize(4)
   561  		if l.SymType(sectsym) == sym.STEXT {
   562  			if bld.External() && !bld.DuplicateOK() {
   563  				return nil, fmt.Errorf("%s: duplicate symbol definition", l.SymName(s))
   564  			}
   565  			bld.SetExternal(true)
   566  		}
   567  		if sz, ok := state.comdats[uint16(pesym.SectionNumber-1)]; ok {
   568  			// This is a COMDAT definition. Record that we're picking
   569  			// this instance so that we can ignore future defs.
   570  			if _, ok := comdatDefinitions[l.SymName(s)]; ok {
   571  				return nil, fmt.Errorf("internal error: preexisting COMDAT definition for %q", name)
   572  			}
   573  			comdatDefinitions[l.SymName(s)] = sz
   574  		}
   575  	}
   576  
   577  	// Sort outer lists by address, adding to textp.
   578  	// This keeps textp in increasing address order.
   579  	for _, sect := range f.Sections {
   580  		s := state.sectsyms[sect]
   581  		if s == 0 {
   582  			continue
   583  		}
   584  		l.SortSub(s)
   585  		importSymsState.secSyms = append(importSymsState.secSyms, s)
   586  		if l.SymType(s) == sym.STEXT {
   587  			for ; s != 0; s = l.SubSym(s) {
   588  				if l.AttrOnList(s) {
   589  					return nil, fmt.Errorf("symbol %s listed multiple times", l.SymName(s))
   590  				}
   591  				l.SetAttrOnList(s, true)
   592  				ls.Textp = append(ls.Textp, s)
   593  			}
   594  		}
   595  	}
   596  
   597  	if ls.PData != 0 {
   598  		processSEH(l, arch, ls.PData, ls.XData)
   599  	}
   600  
   601  	return &ls, nil
   602  }
   603  
   604  // PostProcessImports works to resolve inconsistencies with DLL import
   605  // symbols; it is needed when building with more "modern" C compilers
   606  // with internal linkage.
   607  //
   608  // Background: DLL import symbols are data (SNOPTRDATA) symbols whose
   609  // name is of the form "__imp_XXX", which contain a pointer/reference
   610  // to symbol XXX. It's possible to have import symbols for both data
   611  // symbols ("__imp__fmode") and text symbols ("__imp_CreateEventA").
   612  // In some case import symbols are just references to some external
   613  // thing, and in other cases we see actual definitions of import
   614  // symbols when reading host objects.
   615  //
   616  // Previous versions of the linker would in most cases immediately
   617  // "forward" import symbol references, e.g. treat a references to
   618  // "__imp_XXX" a references to "XXX", however this doesn't work well
   619  // with more modern compilers, where you can sometimes see import
   620  // symbols that are defs (as opposed to external refs).
   621  //
   622  // The main actions taken below are to search for references to
   623  // SDYNIMPORT symbols in host object text/data sections and flag the
   624  // symbols for later fixup. When we see a reference to an import
   625  // symbol __imp_XYZ where XYZ corresponds to some SDYNIMPORT symbol,
   626  // we flag the symbol (via GOT setting) so that it can be redirected
   627  // to XYZ later in windynrelocsym. When we see a direct reference to
   628  // an SDYNIMPORT symbol XYZ, we also flag the symbol (via PLT setting)
   629  // to indicated that the reference will need to be redirected to a
   630  // stub.
   631  func PostProcessImports() error {
   632  	ldr := importSymsState.l
   633  	arch := importSymsState.arch
   634  	keeprelocneeded := make(map[loader.Sym]loader.Sym)
   635  	for _, s := range importSymsState.secSyms {
   636  		isText := ldr.SymType(s) == sym.STEXT
   637  		relocs := ldr.Relocs(s)
   638  		for i := 0; i < relocs.Count(); i++ {
   639  			r := relocs.At(i)
   640  			rs := r.Sym()
   641  			if ldr.SymType(rs) == sym.SDYNIMPORT {
   642  				// Tag the symbol for later stub generation.
   643  				ldr.SetPlt(rs, CreateImportStubPltToken)
   644  				continue
   645  			}
   646  			isym, err := LookupBaseFromImport(rs, ldr, arch)
   647  			if err != nil {
   648  				return err
   649  			}
   650  			if isym == 0 {
   651  				continue
   652  			}
   653  			if ldr.SymType(isym) != sym.SDYNIMPORT {
   654  				continue
   655  			}
   656  			// For non-text symbols, forward the reference from __imp_X to
   657  			// X immediately.
   658  			if !isText {
   659  				r.SetSym(isym)
   660  				continue
   661  			}
   662  			// Flag this imp symbol to be processed later in windynrelocsym.
   663  			ldr.SetGot(rs, RedirectToDynImportGotToken)
   664  			// Consistency check: should be no PLT token here.
   665  			splt := ldr.SymPlt(rs)
   666  			if splt != -1 {
   667  				return fmt.Errorf("internal error: import symbol %q has invalid PLT setting %d", ldr.SymName(rs), splt)
   668  			}
   669  			// Flag for dummy relocation.
   670  			keeprelocneeded[rs] = isym
   671  		}
   672  	}
   673  	for k, v := range keeprelocneeded {
   674  		sb := ldr.MakeSymbolUpdater(k)
   675  		r, _ := sb.AddRel(objabi.R_KEEP)
   676  		r.SetSym(v)
   677  	}
   678  	importSymsState = nil
   679  	return nil
   680  }
   681  
   682  func issehsect(arch *sys.Arch, s *pe.Section) bool {
   683  	return arch.Family == sys.AMD64 && (s.Name == ".pdata" || s.Name == ".xdata")
   684  }
   685  
   686  func issect(s *pe.COFFSymbol) bool {
   687  	return s.StorageClass == IMAGE_SYM_CLASS_STATIC && s.Type == 0 && s.Name[0] == '.'
   688  }
   689  
   690  func (state *peLoaderState) readpesym(pesym *pe.COFFSymbol) (*loader.SymbolBuilder, loader.Sym, error) {
   691  	symname, err := pesym.FullName(state.f.StringTable)
   692  	if err != nil {
   693  		return nil, 0, err
   694  	}
   695  	var name string
   696  	if issect(pesym) {
   697  		name = state.l.SymName(state.sectsyms[state.f.Sections[pesym.SectionNumber-1]])
   698  	} else {
   699  		name = symname
   700  		// A note on the "_main" exclusion below: the main routine
   701  		// defined by the Go runtime is named "_main", not "main", so
   702  		// when reading references to _main from a host object we want
   703  		// to avoid rewriting "_main" to "main" in this specific
   704  		// instance. See #issuecomment-1143698749 on #35006 for more
   705  		// details on this problem.
   706  		if state.arch.Family == sys.I386 && name[0] == '_' && name != "_main" && !strings.HasPrefix(name, "__imp_") {
   707  			name = name[1:] // _Name => Name
   708  		}
   709  	}
   710  
   711  	// remove last @XXX
   712  	if i := strings.LastIndex(name, "@"); i >= 0 {
   713  		name = name[:i]
   714  	}
   715  
   716  	var s loader.Sym
   717  	var bld *loader.SymbolBuilder
   718  	// Microsoft's PE documentation is contradictory. It says that the symbol's complex type
   719  	// is stored in the pesym.Type most significant byte, but MSVC, LLVM, and mingw store it
   720  	// in the 4 high bits of the less significant byte.
   721  	switch uint8(pesym.Type&0xf0) >> 4 {
   722  	default:
   723  		return nil, 0, fmt.Errorf("%s: invalid symbol type %d", symname, pesym.Type)
   724  
   725  	case IMAGE_SYM_DTYPE_FUNCTION, IMAGE_SYM_DTYPE_NULL:
   726  		switch pesym.StorageClass {
   727  		case IMAGE_SYM_CLASS_EXTERNAL: //global
   728  			s = state.l.LookupOrCreateCgoExport(name, 0)
   729  
   730  		case IMAGE_SYM_CLASS_NULL, IMAGE_SYM_CLASS_STATIC, IMAGE_SYM_CLASS_LABEL:
   731  			s = state.l.LookupOrCreateCgoExport(name, state.localSymVersion)
   732  			bld = makeUpdater(state.l, bld, s)
   733  			bld.SetDuplicateOK(true)
   734  
   735  		default:
   736  			return nil, 0, fmt.Errorf("%s: invalid symbol binding %d", symname, pesym.StorageClass)
   737  		}
   738  	}
   739  
   740  	if s != 0 && state.l.SymType(s) == 0 && (pesym.StorageClass != IMAGE_SYM_CLASS_STATIC || pesym.Value != 0) {
   741  		bld = makeUpdater(state.l, bld, s)
   742  		bld.SetType(sym.SXREF)
   743  	}
   744  
   745  	return bld, s, nil
   746  }
   747  
   748  // preprocessSymbols walks the COFF symbols for the PE file we're
   749  // reading and looks for cases where we have both a symbol definition
   750  // for "XXX" and an "__imp_XXX" symbol, recording these cases in a map
   751  // in the state struct. This information will be used in readpesym()
   752  // above to give such symbols special treatment. This function also
   753  // gathers information about COMDAT sections/symbols for later use
   754  // in readpesym().
   755  func (state *peLoaderState) preprocessSymbols() error {
   756  
   757  	// Locate comdat sections.
   758  	state.comdats = make(map[uint16]int64)
   759  	for i, s := range state.f.Sections {
   760  		if s.Characteristics&uint32(pe.IMAGE_SCN_LNK_COMDAT) != 0 {
   761  			state.comdats[uint16(i)] = int64(s.Size)
   762  		}
   763  	}
   764  
   765  	// Examine symbol defs.
   766  	for i, numaux := 0, 0; i < len(state.f.COFFSymbols); i += numaux + 1 {
   767  		pesym := &state.f.COFFSymbols[i]
   768  		numaux = int(pesym.NumberOfAuxSymbols)
   769  		if pesym.SectionNumber == 0 { // extern
   770  			continue
   771  		}
   772  		symname, err := pesym.FullName(state.f.StringTable)
   773  		if err != nil {
   774  			return err
   775  		}
   776  		if _, isc := state.comdats[uint16(pesym.SectionNumber-1)]; !isc {
   777  			continue
   778  		}
   779  		if pesym.StorageClass != uint8(IMAGE_SYM_CLASS_STATIC) {
   780  			continue
   781  		}
   782  		// This symbol corresponds to a COMDAT section. Read the
   783  		// aux data for it.
   784  		auxsymp, err := state.f.COFFSymbolReadSectionDefAux(i)
   785  		if err != nil {
   786  			return fmt.Errorf("unable to read aux info for section def symbol %d %s: pe.COFFSymbolReadComdatInfo returns %v", i, symname, err)
   787  		}
   788  		if auxsymp.Selection == pe.IMAGE_COMDAT_SELECT_SAME_SIZE {
   789  			// This is supported.
   790  		} else if auxsymp.Selection == pe.IMAGE_COMDAT_SELECT_ANY {
   791  			// Also supported.
   792  			state.comdats[uint16(pesym.SectionNumber-1)] = int64(-1)
   793  		} else {
   794  			// We don't support any of the other strategies at the
   795  			// moment. I suspect that we may need to also support
   796  			// "associative", we'll see.
   797  			return fmt.Errorf("internal error: unsupported COMDAT selection strategy found in path=%s sec=%d strategy=%d idx=%d, please file a bug", state.pn, auxsymp.SecNum, auxsymp.Selection, i)
   798  		}
   799  	}
   800  	return nil
   801  }
   802  
   803  // LookupBaseFromImport examines the symbol "s" to see if it
   804  // corresponds to an import symbol (name of the form "__imp_XYZ") and
   805  // if so, it looks up the underlying target of the import symbol and
   806  // returns it. An error is returned if the symbol is of the form
   807  // "__imp_XYZ" but no XYZ can be found.
   808  func LookupBaseFromImport(s loader.Sym, ldr *loader.Loader, arch *sys.Arch) (loader.Sym, error) {
   809  	sname := ldr.SymName(s)
   810  	if !strings.HasPrefix(sname, "__imp_") {
   811  		return 0, nil
   812  	}
   813  	basename := sname[len("__imp_"):]
   814  	if arch.Family == sys.I386 && basename[0] == '_' {
   815  		basename = basename[1:] // _Name => Name
   816  	}
   817  	isym := ldr.Lookup(basename, 0)
   818  	if isym == 0 {
   819  		return 0, fmt.Errorf("internal error: import symbol %q with no underlying sym", sname)
   820  	}
   821  	return isym, nil
   822  }
   823
View as plain text