src/cmd/link/internal/loadpe/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         "cmd/internal/bio"
  10         "cmd/internal/objabi"
  11         "cmd/internal/sys"
  12         "cmd/link/internal/sym"
  13         "debug/pe"
  14         "encoding/binary"
  15         "errors"
  16         "fmt"
  17         "io"
  18         "sort"
  19         "strings"
  20 )
  21
  22 const (
  23         // TODO: the Microsoft doco says IMAGE_SYM_DTYPE_ARRAY is 3 (same with IMAGE_SYM_DTYPE_POINTER and IMAGE_SYM_DTYPE_FUNCTION)
  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          = 0x10
  46         IMAGE_SYM_DTYPE_FUNCTION         = 0x20
  47         IMAGE_SYM_DTYPE_ARRAY            = 0x30
  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 )
 118
 119 // TODO(crawshaw): de-duplicate these symbols with cmd/internal/ld, ideally in debug/pe.
 120 const (
 121         IMAGE_SCN_CNT_CODE               = 0x00000020
 122         IMAGE_SCN_CNT_INITIALIZED_DATA   = 0x00000040
 123         IMAGE_SCN_CNT_UNINITIALIZED_DATA = 0x00000080
 124         IMAGE_SCN_MEM_DISCARDABLE        = 0x02000000
 125         IMAGE_SCN_MEM_EXECUTE            = 0x20000000
 126         IMAGE_SCN_MEM_READ               = 0x40000000
 127         IMAGE_SCN_MEM_WRITE              = 0x80000000
 128 )
 129
 130 // TODO(brainman): maybe just add ReadAt method to bio.Reader instead of creating peBiobuf
 131
 132 // peBiobuf makes bio.Reader look like io.ReaderAt.
 133 type peBiobuf bio.Reader
 134
 135 func (f *peBiobuf) ReadAt(p []byte, off int64) (int, error) {
 136         ret := ((*bio.Reader)(f)).Seek(off, 0)
 137         if ret < 0 {
 138                 return 0, errors.New("fail to seek")
 139         }
 140         n, err := f.Read(p)
 141         if err != nil {
 142                 return 0, err
 143         }
 144         return n, nil
 145 }
 146
 147 // Load loads the PE file pn from input.
 148 // Symbols are written into syms, and a slice of the text symbols is returned.
 149 // If an .rsrc section is found, its symbol is returned as rsrc.
 150 func Load(arch *sys.Arch, syms *sym.Symbols, input *bio.Reader, pkg string, length int64, pn string) (textp []*sym.Symbol, rsrc *sym.Symbol, err error) {
 151         localSymVersion := syms.IncVersion()
 152
 153         sectsyms := make(map[*pe.Section]*sym.Symbol)
 154         sectdata := make(map[*pe.Section][]byte)
 155
 156         // Some input files are archives containing multiple of
 157         // object files, and pe.NewFile seeks to the start of
 158         // input file and get confused. Create section reader
 159         // to stop pe.NewFile looking before current position.
 160         sr := io.NewSectionReader((*peBiobuf)(input), input.Offset(), 1<<63-1)
 161
 162         // TODO: replace pe.NewFile with pe.Load (grep for "add Load function" in debug/pe for details)
 163         f, err := pe.NewFile(sr)
 164         if err != nil {
 165                 return nil, nil, err
 166         }
 167         defer f.Close()
 168
 169         // TODO return error if found .cormeta
 170
 171         // create symbols for mapped sections
 172         for _, sect := range f.Sections {
 173                 if sect.Characteristics&IMAGE_SCN_MEM_DISCARDABLE != 0 {
 174                         continue
 175                 }
 176
 177                 if sect.Characteristics&(IMAGE_SCN_CNT_CODE|IMAGE_SCN_CNT_INITIALIZED_DATA|IMAGE_SCN_CNT_UNINITIALIZED_DATA) == 0 {
 178                         // This has been seen for .idata sections, which we
 179                         // want to ignore. See issues 5106 and 5273.
 180                         continue
 181                 }
 182
 183                 name := fmt.Sprintf("%s(%s)", pkg, sect.Name)
 184                 s := syms.Lookup(name, localSymVersion)
 185
 186                 switch sect.Characteristics & (IMAGE_SCN_CNT_UNINITIALIZED_DATA | IMAGE_SCN_CNT_INITIALIZED_DATA | IMAGE_SCN_MEM_READ | IMAGE_SCN_MEM_WRITE | IMAGE_SCN_CNT_CODE | IMAGE_SCN_MEM_EXECUTE) {
 187                 case IMAGE_SCN_CNT_INITIALIZED_DATA | IMAGE_SCN_MEM_READ: //.rdata
 188                         s.Type = sym.SRODATA
 189
 190                 case IMAGE_SCN_CNT_UNINITIALIZED_DATA | IMAGE_SCN_MEM_READ | IMAGE_SCN_MEM_WRITE: //.bss
 191                         s.Type = sym.SNOPTRBSS
 192
 193                 case IMAGE_SCN_CNT_INITIALIZED_DATA | IMAGE_SCN_MEM_READ | IMAGE_SCN_MEM_WRITE: //.data
 194                         s.Type = sym.SNOPTRDATA
 195
 196                 case IMAGE_SCN_CNT_CODE | IMAGE_SCN_MEM_EXECUTE | IMAGE_SCN_MEM_READ: //.text
 197                         s.Type = sym.STEXT
 198
 199                 default:
 200                         return nil, nil, fmt.Errorf("unexpected flags %#06x for PE section %s", sect.Characteristics, sect.Name)
 201                 }
 202
 203                 if s.Type != sym.SNOPTRBSS {
 204                         data, err := sect.Data()
 205                         if err != nil {
 206                                 return nil, nil, err
 207                         }
 208                         sectdata[sect] = data
 209                         s.P = data
 210                 }
 211                 s.Size = int64(sect.Size)
 212                 sectsyms[sect] = s
 213                 if sect.Name == ".rsrc" {
 214                         rsrc = s
 215                 }
 216         }
 217
 218         // load relocations
 219         for _, rsect := range f.Sections {
 220                 if _, found := sectsyms[rsect]; !found {
 221                         continue
 222                 }
 223                 if rsect.NumberOfRelocations == 0 {
 224                         continue
 225                 }
 226                 if rsect.Characteristics&IMAGE_SCN_MEM_DISCARDABLE != 0 {
 227                         continue
 228                 }
 229                 if rsect.Characteristics&(IMAGE_SCN_CNT_CODE|IMAGE_SCN_CNT_INITIALIZED_DATA|IMAGE_SCN_CNT_UNINITIALIZED_DATA) == 0 {
 230                         // This has been seen for .idata sections, which we
 231                         // want to ignore. See issues 5106 and 5273.
 232                         continue
 233                 }
 234
 235                 rs := make([]sym.Reloc, rsect.NumberOfRelocations)
 236                 for j, r := range rsect.Relocs {
 237                         rp := &rs[j]
 238                         if int(r.SymbolTableIndex) >= len(f.COFFSymbols) {
 239                                 return nil, nil, fmt.Errorf("relocation number %d symbol index idx=%d cannot be large then number of symbols %d", j, r.SymbolTableIndex, len(f.COFFSymbols))
 240                         }
 241                         pesym := &f.COFFSymbols[r.SymbolTableIndex]
 242                         gosym, err := readpesym(arch, syms, f, pesym, sectsyms, localSymVersion)
 243                         if err != nil {
 244                                 return nil, nil, err
 245                         }
 246                         if gosym == nil {
 247                                 name, err := pesym.FullName(f.StringTable)
 248                                 if err != nil {
 249                                         name = string(pesym.Name[:])
 250                                 }
 251                                 return nil, nil, fmt.Errorf("reloc of invalid sym %s idx=%d type=%d", name, r.SymbolTableIndex, pesym.Type)
 252                         }
 253
 254                         rp.Sym = gosym
 255                         rp.Siz = 4
 256                         rp.Off = int32(r.VirtualAddress)
 257                         switch arch.Family {
 258                         default:
 259                                 return nil, nil, fmt.Errorf("%s: unsupported arch %v", pn, arch.Family)
 260                         case sys.I386, sys.AMD64:
 261                                 switch r.Type {
 262                                 default:
 263                                         return nil, nil, fmt.Errorf("%s: %v: unknown relocation type %v", pn, sectsyms[rsect], r.Type)
 264
 265                                 case IMAGE_REL_I386_REL32, IMAGE_REL_AMD64_REL32,
 266                                         IMAGE_REL_AMD64_ADDR32, // R_X86_64_PC32
 267                                         IMAGE_REL_AMD64_ADDR32NB:
 268                                         rp.Type = objabi.R_PCREL
 269
 270                                         rp.Add = int64(int32(binary.LittleEndian.Uint32(sectdata[rsect][rp.Off:])))
 271
 272                                 case IMAGE_REL_I386_DIR32NB, IMAGE_REL_I386_DIR32:
 273                                         rp.Type = objabi.R_ADDR
 274
 275                                         // load addend from image
 276                                         rp.Add = int64(int32(binary.LittleEndian.Uint32(sectdata[rsect][rp.Off:])))
 277
 278                                 case IMAGE_REL_AMD64_ADDR64: // R_X86_64_64
 279                                         rp.Siz = 8
 280
 281                                         rp.Type = objabi.R_ADDR
 282
 283                                         // load addend from image
 284                                         rp.Add = int64(binary.LittleEndian.Uint64(sectdata[rsect][rp.Off:]))
 285                                 }
 286
 287                         case sys.ARM:
 288                                 switch r.Type {
 289                                 default:
 290                                         return nil, nil, fmt.Errorf("%s: %v: unknown ARM relocation type %v", pn, sectsyms[rsect], r.Type)
 291
 292                                 case IMAGE_REL_ARM_SECREL:
 293                                         rp.Type = objabi.R_PCREL
 294
 295                                         rp.Add = int64(int32(binary.LittleEndian.Uint32(sectdata[rsect][rp.Off:])))
 296
 297                                 case IMAGE_REL_ARM_ADDR32:
 298                                         rp.Type = objabi.R_ADDR
 299
 300                                         rp.Add = int64(int32(binary.LittleEndian.Uint32(sectdata[rsect][rp.Off:])))
 301
 302                                 case IMAGE_REL_ARM_BRANCH24:
 303                                         rp.Type = objabi.R_CALLARM
 304
 305                                         rp.Add = int64(int32(binary.LittleEndian.Uint32(sectdata[rsect][rp.Off:])))
 306                                 }
 307                         }
 308
 309                         // ld -r could generate multiple section symbols for the
 310                         // same section but with different values, we have to take
 311                         // that into account
 312                         if issect(pesym) {
 313                                 rp.Add += int64(pesym.Value)
 314                         }
 315                 }
 316
 317                 sort.Sort(sym.RelocByOff(rs[:rsect.NumberOfRelocations]))
 318
 319                 s := sectsyms[rsect]
 320                 s.R = rs
 321                 s.R = s.R[:rsect.NumberOfRelocations]
 322         }
 323
 324         // enter sub-symbols into symbol table.
 325         for i, numaux := 0, 0; i < len(f.COFFSymbols); i += numaux + 1 {
 326                 pesym := &f.COFFSymbols[i]
 327
 328                 numaux = int(pesym.NumberOfAuxSymbols)
 329
 330                 name, err := pesym.FullName(f.StringTable)
 331                 if err != nil {
 332                         return nil, nil, err
 333                 }
 334                 if name == "" {
 335                         continue
 336                 }
 337                 if issect(pesym) {
 338                         continue
 339                 }
 340                 if int(pesym.SectionNumber) > len(f.Sections) {
 341                         continue
 342                 }
 343                 if pesym.SectionNumber == IMAGE_SYM_DEBUG {
 344                         continue
 345                 }
 346                 var sect *pe.Section
 347                 if pesym.SectionNumber > 0 {
 348                         sect = f.Sections[pesym.SectionNumber-1]
 349                         if _, found := sectsyms[sect]; !found {
 350                                 continue
 351                         }
 352                 }
 353
 354                 s, err := readpesym(arch, syms, f, pesym, sectsyms, localSymVersion)
 355                 if err != nil {
 356                         return nil, nil, err
 357                 }
 358
 359                 if pesym.SectionNumber == 0 { // extern
 360                         if s.Type == sym.SDYNIMPORT {
 361                                 s.SetPlt(-2) // flag for dynimport in PE object files.
 362                         }
 363                         if s.Type == sym.SXREF && pesym.Value > 0 { // global data
 364                                 s.Type = sym.SNOPTRDATA
 365                                 s.Size = int64(pesym.Value)
 366                         }
 367
 368                         continue
 369                 } else if pesym.SectionNumber > 0 && int(pesym.SectionNumber) <= len(f.Sections) {
 370                         sect = f.Sections[pesym.SectionNumber-1]
 371                         if _, found := sectsyms[sect]; !found {
 372                                 return nil, nil, fmt.Errorf("%s: %v: missing sect.sym", pn, s)
 373                         }
 374                 } else {
 375                         return nil, nil, fmt.Errorf("%s: %v: sectnum < 0!", pn, s)
 376                 }
 377
 378                 if sect == nil {
 379                         return nil, rsrc, nil
 380                 }
 381
 382                 if s.Outer != nil {
 383                         if s.Attr.DuplicateOK() {
 384                                 continue
 385                         }
 386                         return nil, nil, fmt.Errorf("%s: duplicate symbol reference: %s in both %s and %s", pn, s.Name, s.Outer.Name, sectsyms[sect].Name)
 387                 }
 388
 389                 sectsym := sectsyms[sect]
 390                 s.Sub = sectsym.Sub
 391                 sectsym.Sub = s
 392                 s.Type = sectsym.Type
 393                 s.Attr |= sym.AttrSubSymbol
 394                 s.Value = int64(pesym.Value)
 395                 s.Size = 4
 396                 s.Outer = sectsym
 397                 if sectsym.Type == sym.STEXT {
 398                         if s.Attr.External() && !s.Attr.DuplicateOK() {
 399                                 return nil, nil, fmt.Errorf("%s: duplicate symbol definition", s.Name)
 400                         }
 401                         s.Attr |= sym.AttrExternal
 402                 }
 403         }
 404
 405         // Sort outer lists by address, adding to textp.
 406         // This keeps textp in increasing address order.
 407         for _, sect := range f.Sections {
 408                 s := sectsyms[sect]
 409                 if s == nil {
 410                         continue
 411                 }
 412                 if s.Sub != nil {
 413                         s.Sub = sym.SortSub(s.Sub)
 414                 }
 415                 if s.Type == sym.STEXT {
 416                         if s.Attr.OnList() {
 417                                 return nil, nil, fmt.Errorf("symbol %s listed multiple times", s.Name)
 418                         }
 419                         s.Attr |= sym.AttrOnList
 420                         textp = append(textp, s)
 421                         for s = s.Sub; s != nil; s = s.Sub {
 422                                 if s.Attr.OnList() {
 423                                         return nil, nil, fmt.Errorf("symbol %s listed multiple times", s.Name)
 424                                 }
 425                                 s.Attr |= sym.AttrOnList
 426                                 textp = append(textp, s)
 427                         }
 428                 }
 429         }
 430
 431         return textp, rsrc, nil
 432 }
 433
 434 func issect(s *pe.COFFSymbol) bool {
 435         return s.StorageClass == IMAGE_SYM_CLASS_STATIC && s.Type == 0 && s.Name[0] == '.'
 436 }
 437
 438 func readpesym(arch *sys.Arch, syms *sym.Symbols, f *pe.File, pesym *pe.COFFSymbol, sectsyms map[*pe.Section]*sym.Symbol, localSymVersion int) (*sym.Symbol, error) {
 439         symname, err := pesym.FullName(f.StringTable)
 440         if err != nil {
 441                 return nil, err
 442         }
 443         var name string
 444         if issect(pesym) {
 445                 name = sectsyms[f.Sections[pesym.SectionNumber-1]].Name
 446         } else {
 447                 name = symname
 448                 name = strings.TrimPrefix(name, "__imp_") // __imp_Name => Name
 449                 if arch.Family == sys.I386 && name[0] == '_' {
 450                         name = name[1:] // _Name => Name
 451                 }
 452         }
 453
 454         // remove last @XXX
 455         if i := strings.LastIndex(name, "@"); i >= 0 {
 456                 name = name[:i]
 457         }
 458
 459         var s *sym.Symbol
 460         switch pesym.Type {
 461         default:
 462                 return nil, fmt.Errorf("%s: invalid symbol type %d", symname, pesym.Type)
 463
 464         case IMAGE_SYM_DTYPE_FUNCTION, IMAGE_SYM_DTYPE_NULL:
 465                 switch pesym.StorageClass {
 466                 case IMAGE_SYM_CLASS_EXTERNAL: //global
 467                         s = syms.Lookup(name, 0)
 468
 469                 case IMAGE_SYM_CLASS_NULL, IMAGE_SYM_CLASS_STATIC, IMAGE_SYM_CLASS_LABEL:
 470                         s = syms.Lookup(name, localSymVersion)
 471                         s.Attr |= sym.AttrDuplicateOK
 472
 473                 default:
 474                         return nil, fmt.Errorf("%s: invalid symbol binding %d", symname, pesym.StorageClass)
 475                 }
 476         }
 477
 478         if s != nil && s.Type == 0 && (pesym.StorageClass != IMAGE_SYM_CLASS_STATIC || pesym.Value != 0) {
 479                 s.Type = sym.SXREF
 480         }
 481         if strings.HasPrefix(symname, "__imp_") {
 482                 s.SetGot(-2) // flag for __imp_
 483         }
 484
 485         return s, nil
 486 }