Training courses

Kernel and Embedded Linux

Bootlin training courses

Embedded Linux, kernel,
Yocto Project, Buildroot, real-time,
graphics, boot time, debugging...

Bootlin logo

Elixir Cross Referencer

   1
   2
   3
   4
   5
   6
   7
   8
   9
  10
  11
  12
  13
  14
  15
  16
  17
  18
  19
  20
  21
  22
  23
  24
  25
  26
  27
  28
  29
  30
  31
  32
  33
  34
  35
  36
  37
  38
  39
  40
  41
  42
  43
  44
  45
  46
  47
  48
  49
  50
  51
  52
  53
  54
  55
  56
  57
  58
  59
  60
  61
  62
  63
  64
  65
  66
  67
  68
  69
  70
  71
  72
  73
  74
  75
  76
  77
  78
  79
  80
  81
  82
  83
  84
  85
  86
  87
  88
  89
  90
  91
  92
  93
  94
  95
  96
  97
  98
  99
 100
 101
 102
 103
 104
 105
 106
 107
 108
 109
 110
 111
 112
 113
 114
 115
 116
 117
 118
 119
 120
 121
 122
 123
 124
 125
 126
 127
 128
 129
 130
 131
 132
 133
 134
 135
 136
 137
 138
 139
 140
 141
 142
 143
 144
 145
 146
 147
 148
 149
 150
 151
 152
 153
 154
 155
 156
 157
 158
 159
 160
 161
 162
 163
 164
 165
 166
 167
 168
 169
 170
 171
 172
 173
 174
 175
 176
 177
 178
 179
 180
 181
 182
 183
 184
 185
 186
 187
 188
 189
 190
 191
 192
 193
 194
 195
 196
 197
 198
 199
 200
 201
 202
 203
 204
 205
 206
 207
 208
 209
 210
 211
 212
 213
 214
 215
 216
 217
 218
 219
 220
 221
 222
 223
 224
 225
 226
 227
 228
 229
 230
 231
 232
 233
 234
 235
 236
 237
 238
 239
 240
 241
 242
 243
 244
 245
 246
 247
 248
 249
 250
 251
 252
 253
 254
 255
 256
 257
 258
 259
 260
 261
 262
 263
 264
 265
 266
 267
 268
 269
 270
 271
 272
 273
 274
 275
 276
 277
 278
 279
 280
 281
 282
 283
 284
 285
 286
 287
 288
 289
 290
 291
 292
 293
 294
 295
 296
 297
 298
 299
 300
 301
 302
 303
 304
 305
 306
 307
 308
 309
 310
 311
 312
 313
 314
 315
 316
 317
 318
 319
 320
 321
 322
 323
 324
 325
 326
 327
 328
 329
 330
 331
 332
 333
 334
 335
 336
 337
 338
 339
 340
 341
 342
 343
 344
 345
 346
 347
 348
 349
 350
 351
 352
 353
 354
 355
 356
 357
 358
 359
 360
 361
 362
 363
 364
 365
 366
 367
 368
 369
 370
 371
 372
 373
 374
 375
 376
 377
 378
 379
 380
 381
 382
 383
 384
 385
 386
 387
 388
 389
 390
 391
 392
 393
 394
 395
 396
 397
 398
 399
 400
 401
 402
 403
 404
 405
 406
 407
 408
 409
 410
 411
 412
 413
 414
 415
 416
 417
 418
 419
 420
 421
 422
 423
 424
 425
 426
 427
 428
 429
 430
 431
 432
 433
 434
 435
 436
 437
 438
 439
 440
 441
 442
 443
 444
 445
 446
 447
 448
 449
 450
 451
 452
 453
 454
 455
 456
 457
 458
 459
 460
 461
 462
 463
 464
 465
 466
 467
 468
 469
 470
 471
 472
 473
 474
 475
 476
 477
 478
 479
 480
 481
 482
 483
 484
 485
 486
 487
 488
 489
 490
 491
 492
 493
 494
 495
 496
 497
 498
 499
 500
 501
 502
 503
 504
 505
 506
 507
 508
 509
 510
 511
 512
 513
 514
 515
 516
 517
 518
 519
 520
 521
 522
 523
 524
 525
 526
 527
 528
 529
 530
 531
 532
 533
 534
 535
 536
 537
 538
 539
 540
 541
 542
 543
 544
 545
 546
 547
 548
 549
 550
 551
 552
 553
 554
 555
 556
 557
 558
 559
 560
 561
 562
 563
 564
 565
 566
 567
 568
 569
 570
 571
 572
 573
 574
 575
 576
 577
 578
 579
 580
 581
 582
 583
 584
 585
 586
 587
 588
 589
 590
 591
 592
 593
 594
 595
 596
 597
 598
 599
 600
 601
 602
 603
 604
 605
 606
 607
 608
 609
 610
 611
 612
 613
 614
 615
 616
 617
 618
 619
 620
 621
 622
 623
 624
 625
 626
 627
 628
 629
 630
 631
 632
 633
 634
 635
 636
 637
 638
 639
 640
 641
 642
 643
 644
 645
 646
 647
 648
 649
 650
 651
 652
 653
 654
 655
 656
 657
 658
 659
 660
 661
 662
 663
 664
 665
 666
 667
 668
 669
 670
 671
 672
 673
 674
 675
 676
 677
 678
 679
 680
 681
 682
 683
 684
 685
 686
 687
 688
 689
 690
 691
 692
 693
 694
 695
 696
 697
 698
 699
 700
 701
 702
 703
 704
 705
 706
 707
 708
 709
 710
 711
 712
 713
 714
 715
 716
 717
 718
 719
 720
 721
 722
 723
 724
 725
 726
 727
 728
 729
 730
 731
 732
 733
 734
 735
 736
 737
 738
 739
 740
 741
 742
 743
 744
 745
 746
 747
 748
 749
 750
 751
 752
 753
 754
 755
 756
 757
 758
 759
 760
 761
 762
 763
 764
 765
 766
 767
 768
 769
 770
 771
 772
 773
 774
 775
 776
 777
 778
 779
 780
 781
 782
 783
 784
 785
 786
 787
 788
 789
 790
 791
 792
 793
 794
 795
 796
 797
 798
 799
 800
 801
 802
 803
 804
 805
 806
 807
 808
 809
 810
 811
 812
 813
 814
 815
 816
 817
 818
 819
 820
 821
 822
 823
 824
 825
 826
 827
 828
 829
 830
 831
 832
 833
 834
 835
 836
 837
 838
 839
 840
 841
 842
 843
 844
 845
 846
 847
 848
 849
 850
 851
 852
 853
 854
 855
 856
 857
 858
 859
 860
 861
 862
 863
 864
 865
 866
 867
 868
 869
 870
 871
 872
 873
 874
 875
 876
 877
 878
 879
 880
 881
 882
 883
 884
 885
 886
 887
 888
 889
 890
 891
 892
 893
 894
 895
 896
 897
 898
 899
 900
 901
 902
 903
 904
 905
 906
 907
 908
 909
 910
 911
 912
 913
 914
 915
 916
 917
 918
 919
 920
 921
 922
 923
 924
 925
 926
 927
 928
 929
 930
 931
 932
 933
 934
 935
 936
 937
 938
 939
 940
 941
 942
 943
 944
 945
 946
 947
 948
 949
 950
 951
 952
 953
 954
 955
 956
 957
 958
 959
 960
 961
 962
 963
 964
 965
 966
 967
 968
 969
 970
 971
 972
 973
 974
 975
 976
 977
 978
 979
 980
 981
 982
 983
 984
 985
 986
 987
 988
 989
 990
 991
 992
 993
 994
 995
 996
 997
 998
 999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
// ehframe.cc -- handle exception frame sections for gold

// Copyright (C) 2006-2020 Free Software Foundation, Inc.
// Written by Ian Lance Taylor <iant@google.com>.

// This file is part of gold.

// This program is free software; you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation; either version 3 of the License, or
// (at your option) any later version.

// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
// GNU General Public License for more details.

// You should have received a copy of the GNU General Public License
// along with this program; if not, write to the Free Software
// Foundation, Inc., 51 Franklin Street - Fifth Floor, Boston,
// MA 02110-1301, USA.

#include "gold.h"

#include <cstring>
#include <algorithm>

#include "elfcpp.h"
#include "dwarf.h"
#include "symtab.h"
#include "reloc.h"
#include "ehframe.h"

namespace gold
{

// This file handles generation of the exception frame header that
// gcc's runtime support libraries use to find unwind information at
// runtime.  This file also handles discarding duplicate exception
// frame information.

// The exception frame header starts with four bytes:

// 0: The version number, currently 1.

// 1: The encoding of the pointer to the exception frames.  This can
//    be any DWARF unwind encoding (DW_EH_PE_*).  It is normally a 4
//    byte PC relative offset (DW_EH_PE_pcrel | DW_EH_PE_sdata4).

// 2: The encoding of the count of the number of FDE pointers in the
//    lookup table.  This can be any DWARF unwind encoding, and in
//    particular can be DW_EH_PE_omit if the count is omitted.  It is
//    normally a 4 byte unsigned count (DW_EH_PE_udata4).

// 3: The encoding of the lookup table entries.  Currently gcc's
//    libraries will only support DW_EH_PE_datarel | DW_EH_PE_sdata4,
//    which means that the values are 4 byte offsets from the start of
//    the table.

// The exception frame header is followed by a pointer to the contents
// of the exception frame section (.eh_frame).  This pointer is
// encoded as specified in the byte at offset 1 of the header (i.e.,
// it is normally a 4 byte PC relative offset).

// If there is a lookup table, this is followed by the count of the
// number of FDE pointers, encoded as specified in the byte at offset
// 2 of the header (i.e., normally a 4 byte unsigned integer).

// This is followed by the table, which should start at an 4-byte
// aligned address in memory.  Each entry in the table is 8 bytes.
// Each entry represents an FDE.  The first four bytes of each entry
// are an offset to the starting PC for the FDE.  The last four bytes
// of each entry are an offset to the FDE data.  The offsets are from
// the start of the exception frame header information.  The entries
// are in sorted order by starting PC.

const int eh_frame_hdr_size = 4;

// Construct the exception frame header.

Eh_frame_hdr::Eh_frame_hdr(Output_section* eh_frame_section,
			   const Eh_frame* eh_frame_data)
  : Output_section_data(4),
    eh_frame_section_(eh_frame_section),
    eh_frame_data_(eh_frame_data),
    fde_offsets_(),
    any_unrecognized_eh_frame_sections_(false)
{
}

// Set the size of the exception frame header.

void
Eh_frame_hdr::set_final_data_size()
{
  unsigned int data_size = eh_frame_hdr_size + 4;
  if (!this->any_unrecognized_eh_frame_sections_)
    {
      unsigned int fde_count = this->eh_frame_data_->fde_count();
      if (fde_count != 0)
	data_size += 4 + 8 * fde_count;
      this->fde_offsets_.reserve(fde_count);
    }
  this->set_data_size(data_size);
}

// Write the data to the file.

void
Eh_frame_hdr::do_write(Output_file* of)
{
  switch (parameters->size_and_endianness())
    {
#ifdef HAVE_TARGET_32_LITTLE
    case Parameters::TARGET_32_LITTLE:
      this->do_sized_write<32, false>(of);
      break;
#endif
#ifdef HAVE_TARGET_32_BIG
    case Parameters::TARGET_32_BIG:
      this->do_sized_write<32, true>(of);
      break;
#endif
#ifdef HAVE_TARGET_64_LITTLE
    case Parameters::TARGET_64_LITTLE:
      this->do_sized_write<64, false>(of);
      break;
#endif
#ifdef HAVE_TARGET_64_BIG
    case Parameters::TARGET_64_BIG:
      this->do_sized_write<64, true>(of);
      break;
#endif
    default:
      gold_unreachable();
    }
}

// Write the data to the file with the right endianness.

template<int size, bool big_endian>
void
Eh_frame_hdr::do_sized_write(Output_file* of)
{
  const off_t off = this->offset();
  const off_t oview_size = this->data_size();
  unsigned char* const oview = of->get_output_view(off, oview_size);

  // Version number.
  oview[0] = 1;

  // Write out a 4 byte PC relative offset to the address of the
  // .eh_frame section.
  oview[1] = elfcpp::DW_EH_PE_pcrel | elfcpp::DW_EH_PE_sdata4;
  uint64_t eh_frame_address = this->eh_frame_section_->address();
  uint64_t eh_frame_hdr_address = this->address();
  uint64_t eh_frame_offset = (eh_frame_address -
			      (eh_frame_hdr_address + 4));
  elfcpp::Swap<32, big_endian>::writeval(oview + 4, eh_frame_offset);

  if (this->any_unrecognized_eh_frame_sections_
      || this->fde_offsets_.empty())
    {
      // There are no FDEs, or we didn't recognize the format of the
      // some of the .eh_frame sections, so we can't write out the
      // sorted table.
      oview[2] = elfcpp::DW_EH_PE_omit;
      oview[3] = elfcpp::DW_EH_PE_omit;

      gold_assert(oview_size == 8);
    }
  else
    {
      oview[2] = elfcpp::DW_EH_PE_udata4;
      oview[3] = elfcpp::DW_EH_PE_datarel | elfcpp::DW_EH_PE_sdata4;

      elfcpp::Swap<32, big_endian>::writeval(oview + 8,
					     this->fde_offsets_.size());

      // We have the offsets of the FDEs in the .eh_frame section.  We
      // couldn't easily get the PC values before, as they depend on
      // relocations which are, of course, target specific.  This code
      // is run after all those relocations have been applied to the
      // output file.  Here we read the output file again to find the
      // PC values.  Then we sort the list and write it out.

      Fde_addresses<size> fde_addresses(this->fde_offsets_.size());
      this->get_fde_addresses<size, big_endian>(of, &this->fde_offsets_,
						&fde_addresses);

      std::sort(fde_addresses.begin(), fde_addresses.end(),
		Fde_address_compare<size>());

      typename elfcpp::Elf_types<size>::Elf_Addr output_address;
      output_address = this->address();

      unsigned char* pfde = oview + 12;
      for (typename Fde_addresses<size>::iterator p = fde_addresses.begin();
	   p != fde_addresses.end();
	   ++p)
	{
	  elfcpp::Swap<32, big_endian>::writeval(pfde,
						 p->first - output_address);
	  elfcpp::Swap<32, big_endian>::writeval(pfde + 4,
						 p->second - output_address);
	  pfde += 8;
	}

      gold_assert(pfde - oview == oview_size);
    }

  of->write_output_view(off, oview_size, oview);
}

// Given the offset FDE_OFFSET of an FDE in the .eh_frame section, and
// the contents of the .eh_frame section EH_FRAME_CONTENTS, where the
// FDE's encoding is FDE_ENCODING, return the output address of the
// FDE's PC.

template<int size, bool big_endian>
typename elfcpp::Elf_types<size>::Elf_Addr
Eh_frame_hdr::get_fde_pc(
    typename elfcpp::Elf_types<size>::Elf_Addr eh_frame_address,
    const unsigned char* eh_frame_contents,
    section_offset_type fde_offset,
    unsigned char fde_encoding)
{
  // The FDE starts with a 4 byte length and a 4 byte offset to the
  // CIE.  The PC follows.
  const unsigned char* p = eh_frame_contents + fde_offset + 8;

  typename elfcpp::Elf_types<size>::Elf_Addr pc;
  bool is_signed = (fde_encoding & elfcpp::DW_EH_PE_signed) != 0;
  int pc_size = fde_encoding & 7;
  if (pc_size == elfcpp::DW_EH_PE_absptr)
    {
      if (size == 32)
	pc_size = elfcpp::DW_EH_PE_udata4;
      else if (size == 64)
	pc_size = elfcpp::DW_EH_PE_udata8;
      else
	gold_unreachable();
    }

  switch (pc_size)
    {
    case elfcpp::DW_EH_PE_udata2:
      pc = elfcpp::Swap<16, big_endian>::readval(p);
      if (is_signed)
	pc = (pc ^ 0x8000) - 0x8000;
      break;

    case elfcpp::DW_EH_PE_udata4:
      pc = elfcpp::Swap<32, big_endian>::readval(p);
      if (size > 32 && is_signed)
	pc = (pc ^ 0x80000000) - 0x80000000;
      break;

    case elfcpp::DW_EH_PE_udata8:
      gold_assert(size == 64);
      pc = elfcpp::Swap_unaligned<64, big_endian>::readval(p);
      break;

    default:
      // All other cases were rejected in Eh_frame::read_cie.
      gold_unreachable();
    }

  switch (fde_encoding & 0x70)
    {
    case 0:
      break;

    case elfcpp::DW_EH_PE_pcrel:
      pc += eh_frame_address + fde_offset + 8;
      break;

    case elfcpp::DW_EH_PE_datarel:
      pc += parameters->target().ehframe_datarel_base();
      break;

    default:
      // If other cases arise, then we have to handle them, or we have
      // to reject them by returning false in Eh_frame::read_cie.
      gold_unreachable();
    }

  gold_assert((fde_encoding & elfcpp::DW_EH_PE_indirect) == 0);

  return pc;
}

// Given an array of FDE offsets in the .eh_frame section, return an
// array of offsets from the exception frame header to the FDE's
// output PC and to the output address of the FDE itself.  We get the
// FDE's PC by actually looking in the .eh_frame section we just wrote
// to the output file.

template<int size, bool big_endian>
void
Eh_frame_hdr::get_fde_addresses(Output_file* of,
				const Fde_offsets* fde_offsets,
				Fde_addresses<size>* fde_addresses)
{
  typename elfcpp::Elf_types<size>::Elf_Addr eh_frame_address;
  eh_frame_address = this->eh_frame_section_->address();
  off_t eh_frame_offset = this->eh_frame_section_->offset();
  off_t eh_frame_size = this->eh_frame_section_->data_size();
  const unsigned char* eh_frame_contents = of->get_input_view(eh_frame_offset,
							      eh_frame_size);

  for (Fde_offsets::const_iterator p = fde_offsets->begin();
       p != fde_offsets->end();
       ++p)
    {
      typename elfcpp::Elf_types<size>::Elf_Addr fde_pc;
      fde_pc = this->get_fde_pc<size, big_endian>(eh_frame_address,
						  eh_frame_contents,
						  p->first, p->second);
      fde_addresses->push_back(fde_pc, eh_frame_address + p->first);
    }

  of->free_input_view(eh_frame_offset, eh_frame_size, eh_frame_contents);
}

// Class Fde.

// Write the FDE to OVIEW starting at OFFSET.  CIE_OFFSET is the
// offset of the CIE in OVIEW.  OUTPUT_OFFSET is the offset of the
// Eh_frame section within the output section.  FDE_ENCODING is the
// encoding, from the CIE.  ADDRALIGN is the required alignment.
// ADDRESS is the virtual address of OVIEW.  Record the FDE pc for
// EH_FRAME_HDR.  Return the new offset.

template<int size, bool big_endian>
section_offset_type
Fde::write(unsigned char* oview, section_offset_type output_offset,
	   section_offset_type offset, uint64_t address, unsigned int addralign,
	   section_offset_type cie_offset, unsigned char fde_encoding,
	   Eh_frame_hdr* eh_frame_hdr)
{
  gold_assert((offset & (addralign - 1)) == 0);

  size_t length = this->contents_.length();

  // We add 8 when getting the aligned length to account for the
  // length word and the CIE offset.
  size_t aligned_full_length = align_address(length + 8, addralign);

  // Write the length of the FDE as a 32-bit word.  The length word
  // does not include the four bytes of the length word itself, but it
  // does include the offset to the CIE.
  elfcpp::Swap<32, big_endian>::writeval(oview + offset,
                                         aligned_full_length - 4);

  // Write the offset to the CIE as a 32-bit word.  This is the
  // difference between the address of the offset word itself and the
  // CIE address.
  elfcpp::Swap<32, big_endian>::writeval(oview + offset + 4,
					 offset + 4 - cie_offset);

  // Copy the rest of the FDE.  Note that this is run before
  // relocation processing is done on this section, so the relocations
  // will later be applied to the FDE data.
  memcpy(oview + offset + 8, this->contents_.data(), length);

  // If this FDE is associated with a PLT, fill in the PLT's address
  // and size.
  if (this->object_ == NULL)
    {
      gold_assert(memcmp(oview + offset + 8, "\0\0\0\0\0\0\0\0", 8) == 0);
      uint64_t paddress;
      off_t psize;
      parameters->target().plt_fde_location(this->u_.from_linker.plt,
					    oview + offset + 8,
					    &paddress, &psize);
      uint64_t poffset = paddress - (address + offset + 8);
      int32_t spoffset = static_cast<int32_t>(poffset);
      uint32_t upsize = static_cast<uint32_t>(psize);
      if (static_cast<uint64_t>(static_cast<int64_t>(spoffset)) != poffset
	  || static_cast<off_t>(upsize) != psize)
	gold_warning(_("overflow in PLT unwind data; "
		       "unwinding through PLT may fail"));
      elfcpp::Swap<32, big_endian>::writeval(oview + offset + 8, spoffset);
      elfcpp::Swap<32, big_endian>::writeval(oview + offset + 12, upsize);
    }

  if (aligned_full_length > length + 8)
    memset(oview + offset + length + 8, 0, aligned_full_length - (length + 8));

  // Tell the exception frame header about this FDE.
  if (eh_frame_hdr != NULL)
    eh_frame_hdr->record_fde(output_offset + offset, fde_encoding);

  return offset + aligned_full_length;
}

// Class Cie.

// Destructor.

Cie::~Cie()
{
  for (std::vector<Fde*>::iterator p = this->fdes_.begin();
       p != this->fdes_.end();
       ++p)
    delete *p;
}

// Set the output offset of a CIE.  Return the new output offset.

section_offset_type
Cie::set_output_offset(section_offset_type output_offset,
		       unsigned int addralign,
		       Output_section_data *output_data)
{
  size_t length = this->contents_.length();

  // Add 4 for length and 4 for zero CIE identifier tag.
  length += 8;

  if (this->object_ != NULL)
    {
      // Add a mapping so that relocations are applied correctly.
      this->object_->add_merge_mapping(output_data, this->shndx_,
                                       this->input_offset_, length,
                                       output_offset);
    }

  length = align_address(length, addralign);

  for (std::vector<Fde*>::const_iterator p = this->fdes_.begin();
       p != this->fdes_.end();
       ++p)
    {
      (*p)->add_mapping(output_offset + length, output_data);

      size_t fde_length = (*p)->length();
      fde_length = align_address(fde_length, addralign);
      length += fde_length;
    }

  return output_offset + length;
}

// Write the CIE to OVIEW starting at OFFSET.  OUTPUT_OFFSET is the
// offset of the Eh_frame section within the output section.  Round up
// the bytes to ADDRALIGN.  ADDRESS is the virtual address of OVIEW.
// EH_FRAME_HDR is the exception frame header for FDE recording.
// POST_FDES stashes FDEs created after mappings were done, for later
// writing.  Return the new offset.

template<int size, bool big_endian>
section_offset_type
Cie::write(unsigned char* oview, section_offset_type output_offset,
	   section_offset_type offset, uint64_t address,
	   unsigned int addralign, Eh_frame_hdr* eh_frame_hdr,
	   Post_fdes* post_fdes)
{
  gold_assert((offset & (addralign - 1)) == 0);

  section_offset_type cie_offset = offset;

  size_t length = this->contents_.length();

  // We add 8 when getting the aligned length to account for the
  // length word and the CIE tag.
  size_t aligned_full_length = align_address(length + 8, addralign);

  // Write the length of the CIE as a 32-bit word.  The length word
  // does not include the four bytes of the length word itself.
  elfcpp::Swap<32, big_endian>::writeval(oview + offset,
                                         aligned_full_length - 4);

  // Write the tag which marks this as a CIE: a 32-bit zero.
  elfcpp::Swap<32, big_endian>::writeval(oview + offset + 4, 0);

  // Write out the CIE data.
  memcpy(oview + offset + 8, this->contents_.data(), length);

  if (aligned_full_length > length + 8)
    memset(oview + offset + length + 8, 0, aligned_full_length - (length + 8));

  offset += aligned_full_length;

  // Write out the associated FDEs.
  unsigned char fde_encoding = this->fde_encoding_;
  for (std::vector<Fde*>::const_iterator p = this->fdes_.begin();
       p != this->fdes_.end();
       ++p)
    {
      if ((*p)->post_map())
	post_fdes->push_back(Post_fde(*p, cie_offset, fde_encoding));
      else
	offset = (*p)->write<size, big_endian>(oview, output_offset, offset,
					       address, addralign, cie_offset,
					       fde_encoding, eh_frame_hdr);
    }

  return offset;
}

// We track all the CIEs we see, and merge them when possible.  This
// works because each FDE holds an offset to the relevant CIE: we
// rewrite the FDEs to point to the merged CIE.  This is worthwhile
// because in a typical C++ program many FDEs in many different object
// files will use the same CIE.

// An equality operator for Cie.

bool
operator==(const Cie& cie1, const Cie& cie2)
{
  return (cie1.personality_name_ == cie2.personality_name_
	  && cie1.contents_ == cie2.contents_);
}

// A less-than operator for Cie.

bool
operator<(const Cie& cie1, const Cie& cie2)
{
  if (cie1.personality_name_ != cie2.personality_name_)
    return cie1.personality_name_ < cie2.personality_name_;
  return cie1.contents_ < cie2.contents_;
}

// Class Eh_frame.

Eh_frame::Eh_frame()
  : Output_section_data(Output_data::default_alignment()),
    eh_frame_hdr_(NULL),
    cie_offsets_(),
    unmergeable_cie_offsets_(),
    mappings_are_done_(false),
    final_data_size_(0)
{
}

// Skip an LEB128, updating *PP to point to the next character.
// Return false if we ran off the end of the string.

bool
Eh_frame::skip_leb128(const unsigned char** pp, const unsigned char* pend)
{
  const unsigned char* p;
  for (p = *pp; p < pend; ++p)
    {
      if ((*p & 0x80) == 0)
	{
	  *pp = p + 1;
	  return true;
	}
    }
  return false;
}

// Add input section SHNDX in OBJECT to an exception frame section.
// SYMBOLS is the contents of the symbol table section (size
// SYMBOLS_SIZE), SYMBOL_NAMES is the symbol names section (size
// SYMBOL_NAMES_SIZE).  RELOC_SHNDX is the index of a relocation
// section applying to SHNDX, or 0 if none, or -1U if more than one.
// RELOC_TYPE is the type of the reloc section if there is one, either
// SHT_REL or SHT_RELA.  We try to parse the input exception frame
// data into our data structures.  If we can't do it, we return false
// to mean that the section should be handled as a normal input
// section.

template<int size, bool big_endian>
Eh_frame::Eh_frame_section_disposition
Eh_frame::add_ehframe_input_section(
    Sized_relobj_file<size, big_endian>* object,
    const unsigned char* symbols,
    section_size_type symbols_size,
    const unsigned char* symbol_names,
    section_size_type symbol_names_size,
    unsigned int shndx,
    unsigned int reloc_shndx,
    unsigned int reloc_type)
{
  // Get the section contents.
  section_size_type contents_len;
  const unsigned char* pcontents = object->section_contents(shndx,
							    &contents_len,
							    false);
  if (contents_len == 0)
    return EH_EMPTY_SECTION;

  // If this is the marker section for the end of the data, then
  // return false to force it to be handled as an ordinary input
  // section.  If we don't do this, we won't correctly handle the case
  // of unrecognized .eh_frame sections.
  if (contents_len == 4
      && elfcpp::Swap<32, big_endian>::readval(pcontents) == 0)
    return EH_END_MARKER_SECTION;

  New_cies new_cies;
  if (!this->do_add_ehframe_input_section(object, symbols, symbols_size,
					  symbol_names, symbol_names_size,
					  shndx, reloc_shndx,
					  reloc_type, pcontents,
					  contents_len, &new_cies))
    {
      if (this->eh_frame_hdr_ != NULL)
	this->eh_frame_hdr_->found_unrecognized_eh_frame_section();

      for (New_cies::iterator p = new_cies.begin();
	   p != new_cies.end();
	   ++p)
	delete p->first;

      return EH_UNRECOGNIZED_SECTION;
    }

  // Now that we know we are using this section, record any new CIEs
  // that we found.
  for (New_cies::const_iterator p = new_cies.begin();
       p != new_cies.end();
       ++p)
    {
      if (p->second)
	this->cie_offsets_.insert(p->first);
      else
	this->unmergeable_cie_offsets_.push_back(p->first);
    }

  return EH_OPTIMIZABLE_SECTION;
}

// The bulk of the implementation of add_ehframe_input_section.

template<int size, bool big_endian>
bool
Eh_frame::do_add_ehframe_input_section(
    Sized_relobj_file<size, big_endian>* object,
    const unsigned char* symbols,
    section_size_type symbols_size,
    const unsigned char* symbol_names,
    section_size_type symbol_names_size,
    unsigned int shndx,
    unsigned int reloc_shndx,
    unsigned int reloc_type,
    const unsigned char* pcontents,
    section_size_type contents_len,
    New_cies* new_cies)
{
  Track_relocs<size, big_endian> relocs;

  const unsigned char* p = pcontents;
  const unsigned char* pend = p + contents_len;

  // Get the contents of the reloc section if any.
  if (!relocs.initialize(object, reloc_shndx, reloc_type))
    return false;

  // Keep track of which CIEs are at which offsets.
  Offsets_to_cie cies;

  while (p < pend)
    {
      if (pend - p < 4)
	return false;

      // There shouldn't be any relocations here.
      if (relocs.advance(p + 4 - pcontents) > 0)
	return false;

      unsigned int len = elfcpp::Swap<32, big_endian>::readval(p);
      p += 4;
      if (len == 0)
	{
	  // We should only find a zero-length entry at the end of the
	  // section.
	  if (p < pend)
	    return false;
	  break;
	}
      // We don't support a 64-bit .eh_frame.
      if (len == 0xffffffff)
	return false;
      if (static_cast<unsigned int>(pend - p) < len)
	return false;

      const unsigned char* const pentend = p + len;

      if (pend - p < 4)
	return false;
      if (relocs.advance(p + 4 - pcontents) > 0)
	return false;

      unsigned int id = elfcpp::Swap<32, big_endian>::readval(p);
      p += 4;

      if (id == 0)
	{
	  // CIE.
	  if (!this->read_cie(object, shndx, symbols, symbols_size,
			      symbol_names, symbol_names_size,
			      pcontents, p, pentend, &relocs, &cies,
			      new_cies))
	    return false;
	}
      else
	{
	  // FDE.
	  if (!this->read_fde(object, shndx, symbols, symbols_size,
			      pcontents, id, p, pentend, &relocs, &cies))
	    return false;
	}

      p = pentend;
    }

  return true;
}

// Read a CIE.  Return false if we can't parse the information.

template<int size, bool big_endian>
bool
Eh_frame::read_cie(Sized_relobj_file<size, big_endian>* object,
		   unsigned int shndx,
		   const unsigned char* symbols,
		   section_size_type symbols_size,
		   const unsigned char* symbol_names,
		   section_size_type symbol_names_size,
		   const unsigned char* pcontents,
		   const unsigned char* pcie,
		   const unsigned char* pcieend,
		   Track_relocs<size, big_endian>* relocs,
		   Offsets_to_cie* cies,
		   New_cies* new_cies)
{
  bool mergeable = true;

  // We need to find the personality routine if there is one, since we
  // can only merge CIEs which use the same routine.  We also need to
  // find the FDE encoding if there is one, so that we can read the PC
  // from the FDE.

  const unsigned char* p = pcie;

  if (pcieend - p < 1)
    return false;
  unsigned char version = *p++;
  if (version != 1 && version != 3)
    return false;

  const unsigned char* paug = p;
  const void* paugendv = memchr(p, '\0', pcieend - p);
  const unsigned char* paugend = static_cast<const unsigned char*>(paugendv);
  if (paugend == NULL)
    return false;
  p = paugend + 1;

  if (paug[0] == 'e' && paug[1] == 'h')
    {
      // This is a CIE from gcc before version 3.0.  We can't merge
      // these.  We can still read the FDEs.
      mergeable = false;
      paug += 2;
      if (*paug != '\0')
	return false;
      if (pcieend - p < size / 8)
	return false;
      p += size / 8;
    }

  // Skip the code alignment.
  if (!skip_leb128(&p, pcieend))
    return false;

  // Skip the data alignment.
  if (!skip_leb128(&p, pcieend))
    return false;

  // Skip the return column.
  if (version == 1)
    {
      if (pcieend - p < 1)
	return false;
      ++p;
    }
  else
    {
      if (!skip_leb128(&p, pcieend))
	return false;
    }

  if (*paug == 'z')
    {
      ++paug;
      // Skip the augmentation size.
      if (!skip_leb128(&p, pcieend))
	return false;
    }

  unsigned char fde_encoding = elfcpp::DW_EH_PE_absptr;
  int per_offset = -1;
  while (*paug != '\0')
    {
      switch (*paug)
	{
	case 'L': // LSDA encoding.
	  if (pcieend - p < 1)
	    return false;
	  ++p;
	  break;

	case 'R': // FDE encoding.
	  if (pcieend - p < 1)
	    return false;
	  fde_encoding = *p;
	  switch (fde_encoding & 7)
	    {
	    case elfcpp::DW_EH_PE_absptr:
	    case elfcpp::DW_EH_PE_udata2:
	    case elfcpp::DW_EH_PE_udata4:
	    case elfcpp::DW_EH_PE_udata8:
	      break;
	    default:
	      // We don't expect to see any other cases here, and
	      // we're not prepared to handle them.
	      return false;
	    }
	  ++p;
	  break;

	case 'S':
	  break;

	case 'P':
	  // Personality encoding.
	  {
	    if (pcieend - p < 1)
	      return false;
	    unsigned char per_encoding = *p;
	    ++p;

	    if ((per_encoding & 0x60) == 0x60)
	      return false;
	    unsigned int per_width;
	    switch (per_encoding & 7)
	      {
	      case elfcpp::DW_EH_PE_udata2:
		per_width = 2;
		break;
	      case elfcpp::DW_EH_PE_udata4:
		per_width = 4;
		break;
	      case elfcpp::DW_EH_PE_udata8:
		per_width = 8;
		break;
	      case elfcpp::DW_EH_PE_absptr:
		per_width = size / 8;
		break;
	      default:
		return false;
	      }

	    if ((per_encoding & 0xf0) == elfcpp::DW_EH_PE_aligned)
	      {
		unsigned int len = p - pcie;
		len += per_width - 1;
		len &= ~ (per_width - 1);
		if (static_cast<unsigned int>(pcieend - p) < len)
		  return false;
		p += len;
	      }

	    per_offset = p - pcontents;

	    if (static_cast<unsigned int>(pcieend - p) < per_width)
	      return false;
	    p += per_width;
	  }
	  break;

	default:
	  return false;
	}

      ++paug;
    }

  const char* personality_name = "";
  if (per_offset != -1)
    {
      if (relocs->advance(per_offset) > 0)
	return false;
      if (relocs->next_offset() != per_offset)
	return false;

      unsigned int personality_symndx = relocs->next_symndx();
      if (personality_symndx == -1U)
	return false;

      if (personality_symndx < object->local_symbol_count())
	{
	  // We can only merge this CIE if the personality routine is
	  // a global symbol.  We can still read the FDEs.
	  mergeable = false;
	}
      else
	{
	  const int sym_size = elfcpp::Elf_sizes<size>::sym_size;
	  if (personality_symndx >= symbols_size / sym_size)
	    return false;
	  elfcpp::Sym<size, big_endian> sym(symbols
					    + (personality_symndx * sym_size));
	  unsigned int name_offset = sym.get_st_name();
	  if (name_offset >= symbol_names_size)
	    return false;
	  personality_name = (reinterpret_cast<const char*>(symbol_names)
			      + name_offset);
	}

      int r = relocs->advance(per_offset + 1);
      gold_assert(r == 1);
    }

  if (relocs->advance(pcieend - pcontents) > 0)
    return false;

  Cie cie(object, shndx, (pcie - 8) - pcontents, fde_encoding, 
	  personality_name, pcie, pcieend - pcie);
  Cie* cie_pointer = NULL;
  if (mergeable)
    {
      Cie_offsets::iterator find_cie = this->cie_offsets_.find(&cie);
      if (find_cie != this->cie_offsets_.end())
	cie_pointer = *find_cie;
      else
	{
	  // See if we already saw this CIE in this object file.
	  for (New_cies::const_iterator pc = new_cies->begin();
	       pc != new_cies->end();
	       ++pc)
	    {
	      if (*(pc->first) == cie)
		{
		  cie_pointer = pc->first;
		  break;
		}
	    }
	}
    }

  if (cie_pointer == NULL)
    {
      cie_pointer = new Cie(cie);
      new_cies->push_back(std::make_pair(cie_pointer, mergeable));
    }
  else
    {
      // We are deleting this CIE.  Record that in our mapping from
      // input sections to the output section.  At this point we don't
      // know for sure that we are doing a special mapping for this
      // input section, but that's OK--if we don't do a special
      // mapping, nobody will ever ask for the mapping we add here.
      object->add_merge_mapping(this, shndx, (pcie - 8) - pcontents,
                                pcieend - (pcie - 8), -1);
    }

  // Record this CIE plus the offset in the input section.
  cies->insert(std::make_pair(pcie - pcontents, cie_pointer));

  return true;
}

// Read an FDE.  Return false if we can't parse the information.

template<int size, bool big_endian>
bool
Eh_frame::read_fde(Sized_relobj_file<size, big_endian>* object,
		   unsigned int shndx,
		   const unsigned char* symbols,
		   section_size_type symbols_size,
		   const unsigned char* pcontents,
		   unsigned int offset,
		   const unsigned char* pfde,
		   const unsigned char* pfdeend,
		   Track_relocs<size, big_endian>* relocs,
		   Offsets_to_cie* cies)
{
  // OFFSET is the distance between the 4 bytes before PFDE to the
  // start of the CIE.  The offset we recorded for the CIE is 8 bytes
  // after the start of the CIE--after the length and the zero tag.
  unsigned int cie_offset = (pfde - 4 - pcontents) - offset + 8;
  Offsets_to_cie::const_iterator pcie = cies->find(cie_offset);
  if (pcie == cies->end())
    return false;
  Cie* cie = pcie->second;

  int pc_size = 0;
  switch (cie->fde_encoding() & 7)
    {
    case elfcpp::DW_EH_PE_udata2:
      pc_size = 2;
      break;
    case elfcpp::DW_EH_PE_udata4:
      pc_size = 4;
      break;
    case elfcpp::DW_EH_PE_udata8:
      gold_assert(size == 64);
      pc_size = 8;
      break;
    case elfcpp::DW_EH_PE_absptr:
      pc_size = size == 32 ? 4 : 8;
      break;
    default:
      // All other cases were rejected in Eh_frame::read_cie.
      gold_unreachable();
    }

  // The FDE should start with a reloc to the start of the code which
  // it describes.
  if (relocs->advance(pfde - pcontents) > 0)
    return false;
  if (relocs->next_offset() != pfde - pcontents)
    {
      // In an object produced by a relocatable link, gold may have
      // discarded a COMDAT group in the previous link, but not the
      // corresponding FDEs. In that case, gold will have discarded
      // the relocations, so the FDE will have a non-relocatable zero
      // (regardless of whether the PC encoding is absolute, pc-relative,
      // or data-relative) instead of a pointer to the start of the code.

      uint64_t pc_value = 0;
      switch (pc_size)
	{
	case 2:
	  pc_value = elfcpp::Swap<16, big_endian>::readval(pfde);
	  break;
	case 4:
	  pc_value = elfcpp::Swap<32, big_endian>::readval(pfde);
	  break;
	case 8:
	  pc_value = elfcpp::Swap_unaligned<64, big_endian>::readval(pfde);
	  break;
	default:
	  gold_unreachable();
	}

      if (pc_value == 0)
	{
	  // This FDE applies to a discarded function.  We
	  // can discard this FDE.
	  object->add_merge_mapping(this, shndx, (pfde - 8) - pcontents,
				    pfdeend - (pfde - 8), -1);
	  return true;
	}

      // Otherwise, reject the FDE.
      return false;
    }

  unsigned int symndx = relocs->next_symndx();
  if (symndx == -1U)
    return false;

  // There can be another reloc in the FDE, if the CIE specifies an
  // LSDA (language specific data area).  We currently don't care.  We
  // will care later if we want to optimize the LSDA from an absolute
  // pointer to a PC relative offset when generating a shared library.
  relocs->advance(pfdeend - pcontents);

  // Find the section index for code that this FDE describes.
  // If we have discarded the section, we can also discard the FDE.
  unsigned int fde_shndx;
  const int sym_size = elfcpp::Elf_sizes<size>::sym_size;
  if (symndx >= symbols_size / sym_size)
    return false;
  elfcpp::Sym<size, big_endian> sym(symbols + symndx * sym_size);
  bool is_ordinary;
  fde_shndx = object->adjust_sym_shndx(symndx, sym.get_st_shndx(),
				       &is_ordinary);
  bool is_discarded = (is_ordinary
		       && fde_shndx != elfcpp::SHN_UNDEF
		       && fde_shndx < object->shnum()
		       && !object->is_section_included(fde_shndx));

  // Fetch the address range field from the FDE. The offset and size
  // of the field depends on the PC encoding given in the CIE, but
  // it is always an absolute value. If the address range is 0, this
  // FDE corresponds to a function that was discarded during optimization
  // (too late to discard the corresponding FDE).
  uint64_t address_range = 0;
  switch (pc_size)
    {
    case 2:
      address_range = elfcpp::Swap<16, big_endian>::readval(pfde + 2);
      break;
    case 4:
      address_range = elfcpp::Swap<32, big_endian>::readval(pfde + 4);
      break;
    case 8:
      address_range = elfcpp::Swap_unaligned<64, big_endian>::readval(pfde + 8);
      break;
    default:
      gold_unreachable();
    }

  if (is_discarded || address_range == 0)
    {
      // This FDE applies to a discarded function.  We
      // can discard this FDE.
      object->add_merge_mapping(this, shndx, (pfde - 8) - pcontents,
                                pfdeend - (pfde - 8), -1);
      return true;
    }

  cie->add_fde(new Fde(object, shndx, (pfde - 8) - pcontents,
		       pfde, pfdeend - pfde));

  return true;
}

// Add unwind information for a PLT.

void
Eh_frame::add_ehframe_for_plt(Output_data* plt, const unsigned char* cie_data,
			      size_t cie_length, const unsigned char* fde_data,
			      size_t fde_length)
{
  Cie cie(NULL, 0, 0, elfcpp::DW_EH_PE_pcrel | elfcpp::DW_EH_PE_sdata4, "",
	  cie_data, cie_length);
  Cie_offsets::iterator find_cie = this->cie_offsets_.find(&cie);
  Cie* pcie;
  if (find_cie != this->cie_offsets_.end())
    pcie = *find_cie;
  else
    {
      gold_assert(!this->mappings_are_done_);
      pcie = new Cie(cie);
      this->cie_offsets_.insert(pcie);
    }

  Fde* fde = new Fde(plt, fde_data, fde_length, this->mappings_are_done_);
  pcie->add_fde(fde);

  if (this->mappings_are_done_)
    this->final_data_size_ += align_address(fde_length + 8, this->addralign());
}

// Remove all post-map unwind information for a PLT.

void
Eh_frame::remove_ehframe_for_plt(Output_data* plt,
				 const unsigned char* cie_data,
				 size_t cie_length)
{
  if (!this->mappings_are_done_)
    return;

  Cie cie(NULL, 0, 0, elfcpp::DW_EH_PE_pcrel | elfcpp::DW_EH_PE_sdata4, "",
	  cie_data, cie_length);
  Cie_offsets::iterator find_cie = this->cie_offsets_.find(&cie);
  gold_assert (find_cie != this->cie_offsets_.end());
  Cie* pcie = *find_cie;

  while (pcie->fde_count() != 0)
    {
      const Fde* fde = pcie->last_fde();
      if (!fde->post_map(plt))
	break;
      size_t length = fde->length();
      this->final_data_size_ -= align_address(length + 8, this->addralign());
      pcie->remove_fde();
    }
}

// Return the number of FDEs.

unsigned int
Eh_frame::fde_count() const
{
  unsigned int ret = 0;
  for (Unmergeable_cie_offsets::const_iterator p =
	 this->unmergeable_cie_offsets_.begin();
       p != this->unmergeable_cie_offsets_.end();
       ++p)
    ret += (*p)->fde_count();
  for (Cie_offsets::const_iterator p = this->cie_offsets_.begin();
       p != this->cie_offsets_.end();
       ++p)
    ret += (*p)->fde_count();
  return ret;
}

// Set the final data size.

void
Eh_frame::set_final_data_size()
{
  // We can be called more than once if Layout::set_segment_offsets
  // finds a better mapping.  We don't want to add all the mappings
  // again.
  if (this->mappings_are_done_)
    {
      this->set_data_size(this->final_data_size_);
      return;
    }

  section_offset_type output_start = 0;
  if (this->is_offset_valid())
    output_start = this->offset() - this->output_section()->offset();
  section_offset_type output_offset = output_start;

  for (Unmergeable_cie_offsets::iterator p =
	 this->unmergeable_cie_offsets_.begin();
       p != this->unmergeable_cie_offsets_.end();
       ++p)
    output_offset = (*p)->set_output_offset(output_offset,
					    this->addralign(),
					    this);

  for (Cie_offsets::iterator p = this->cie_offsets_.begin();
       p != this->cie_offsets_.end();
       ++p)
    output_offset = (*p)->set_output_offset(output_offset,
					    this->addralign(),
					    this);

  this->mappings_are_done_ = true;
  this->final_data_size_ = output_offset - output_start;

  gold_assert((output_offset & (this->addralign() - 1)) == 0);
  this->set_data_size(this->final_data_size_);
}

// Return an output offset for an input offset.

bool
Eh_frame::do_output_offset(const Relobj* object, unsigned int shndx,
			   section_offset_type offset,
			   section_offset_type* poutput) const
{
  return object->merge_output_offset(shndx, offset, poutput);
}

// Write the data to the output file.

void
Eh_frame::do_write(Output_file* of)
{
  const off_t offset = this->offset();
  const off_t oview_size = this->data_size();
  unsigned char* const oview = of->get_output_view(offset, oview_size);

  switch (parameters->size_and_endianness())
    {
#ifdef HAVE_TARGET_32_LITTLE
    case Parameters::TARGET_32_LITTLE:
      this->do_sized_write<32, false>(oview);
      break;
#endif
#ifdef HAVE_TARGET_32_BIG
    case Parameters::TARGET_32_BIG:
      this->do_sized_write<32, true>(oview);
      break;
#endif
#ifdef HAVE_TARGET_64_LITTLE
    case Parameters::TARGET_64_LITTLE:
      this->do_sized_write<64, false>(oview);
      break;
#endif
#ifdef HAVE_TARGET_64_BIG
    case Parameters::TARGET_64_BIG:
      this->do_sized_write<64, true>(oview);
      break;
#endif
    default:
      gold_unreachable();
    }

  of->write_output_view(offset, oview_size, oview);
}

// Write the data to the output file--template version.

template<int size, bool big_endian>
void
Eh_frame::do_sized_write(unsigned char* oview)
{
  uint64_t address = this->address();
  unsigned int addralign = this->addralign();
  section_offset_type o = 0;
  const off_t output_offset = this->offset() - this->output_section()->offset();
  Post_fdes post_fdes;
  for (Unmergeable_cie_offsets::iterator p =
	 this->unmergeable_cie_offsets_.begin();
       p != this->unmergeable_cie_offsets_.end();
       ++p)
    o = (*p)->write<size, big_endian>(oview, output_offset, o, address,
				      addralign, this->eh_frame_hdr_,
				      &post_fdes);
  for (Cie_offsets::iterator p = this->cie_offsets_.begin();
       p != this->cie_offsets_.end();
       ++p)
    o = (*p)->write<size, big_endian>(oview, output_offset, o, address,
				      addralign, this->eh_frame_hdr_,
				      &post_fdes);
  for (Post_fdes::iterator p = post_fdes.begin();
       p != post_fdes.end();
       ++p)
    o = (*p).fde->write<size, big_endian>(oview, output_offset, o, address,
					  addralign, (*p).cie_offset,
					  (*p).fde_encoding,
					  this->eh_frame_hdr_);
}

#ifdef HAVE_TARGET_32_LITTLE
template
Eh_frame::Eh_frame_section_disposition
Eh_frame::add_ehframe_input_section<32, false>(
    Sized_relobj_file<32, false>* object,
    const unsigned char* symbols,
    section_size_type symbols_size,
    const unsigned char* symbol_names,
    section_size_type symbol_names_size,
    unsigned int shndx,
    unsigned int reloc_shndx,
    unsigned int reloc_type);
#endif

#ifdef HAVE_TARGET_32_BIG
template
Eh_frame::Eh_frame_section_disposition
Eh_frame::add_ehframe_input_section<32, true>(
    Sized_relobj_file<32, true>* object,
    const unsigned char* symbols,
    section_size_type symbols_size,
    const unsigned char* symbol_names,
    section_size_type symbol_names_size,
    unsigned int shndx,
    unsigned int reloc_shndx,
    unsigned int reloc_type);
#endif

#ifdef HAVE_TARGET_64_LITTLE
template
Eh_frame::Eh_frame_section_disposition
Eh_frame::add_ehframe_input_section<64, false>(
    Sized_relobj_file<64, false>* object,
    const unsigned char* symbols,
    section_size_type symbols_size,
    const unsigned char* symbol_names,
    section_size_type symbol_names_size,
    unsigned int shndx,
    unsigned int reloc_shndx,
    unsigned int reloc_type);
#endif

#ifdef HAVE_TARGET_64_BIG
template
Eh_frame::Eh_frame_section_disposition
Eh_frame::add_ehframe_input_section<64, true>(
    Sized_relobj_file<64, true>* object,
    const unsigned char* symbols,
    section_size_type symbols_size,
    const unsigned char* symbol_names,
    section_size_type symbol_names_size,
    unsigned int shndx,
    unsigned int reloc_shndx,
    unsigned int reloc_type);
#endif

} // End namespace gold.