gold/dwarf_reader.cc

   1 // dwarf_reader.cc -- parse dwarf2/3 debug information
   2
   3 // Copyright 2007, 2008, 2009, 2010, 2011, 2012 Free Software Foundation, Inc.
   4 // Written by Ian Lance Taylor <iant@google.com>.
   5
   6 // This file is part of gold.
   7
   8 // This program is free software; you can redistribute it and/or modify
   9 // it under the terms of the GNU General Public License as published by
  10 // the Free Software Foundation; either version 3 of the License, or
  11 // (at your option) any later version.
  12
  13 // This program is distributed in the hope that it will be useful,
  14 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  15 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16 // GNU General Public License for more details.
  17
  18 // You should have received a copy of the GNU General Public License
  19 // along with this program; if not, write to the Free Software
  20 // Foundation, Inc., 51 Franklin Street - Fifth Floor, Boston,
  21 // MA 02110-1301, USA.
  22
  23 #include "gold.h"
  24
  25 #include <algorithm>
  26 #include <vector>
  27
  28 #include "elfcpp_swap.h"
  29 #include "dwarf.h"
  30 #include "object.h"
  31 #include "parameters.h"
  32 #include "reloc.h"
  33 #include "dwarf_reader.h"
  34 #include "int_encoding.h"
  35 #include "compressed_output.h"
  36
  37 namespace gold {
  38
  39 // Class Sized_elf_reloc_mapper
  40
  41 // Initialize the relocation tracker for section RELOC_SHNDX.
  42
  43 template<int size, bool big_endian>
  44 bool
  45 Sized_elf_reloc_mapper<size, big_endian>::do_initialize(
  46     unsigned int reloc_shndx, unsigned int reloc_type)
  47 {
  48   this->reloc_type_ = reloc_type;
  49   return this->track_relocs_.initialize(this->object_, reloc_shndx,
  50                                         reloc_type);
  51 }
  52
  53 // Looks in the symtab to see what section a symbol is in.
  54
  55 template<int size, bool big_endian>
  56 unsigned int
  57 Sized_elf_reloc_mapper<size, big_endian>::symbol_section(
  58     unsigned int symndx, Address* value, bool* is_ordinary)
  59 {
  60   const int symsize = elfcpp::Elf_sizes<size>::sym_size;
  61   gold_assert((symndx + 1) * symsize <= this->symtab_size_);
  62   elfcpp::Sym<size, big_endian> elfsym(this->symtab_ + symndx * symsize);
  63   *value = elfsym.get_st_value();
  64   return this->object_->adjust_sym_shndx(symndx, elfsym.get_st_shndx(),
  65                                          is_ordinary);
  66 }
  67
  68 // Return the section index and offset within the section of
  69 // the target of the relocation for RELOC_OFFSET.
  70
  71 template<int size, bool big_endian>
  72 unsigned int
  73 Sized_elf_reloc_mapper<size, big_endian>::do_get_reloc_target(
  74     off_t reloc_offset, off_t* target_offset)
  75 {
  76   this->track_relocs_.advance(reloc_offset);
  77   if (reloc_offset != this->track_relocs_.next_offset())
  78     return 0;
  79   unsigned int symndx = this->track_relocs_.next_symndx();
  80   typename elfcpp::Elf_types<size>::Elf_Addr value;
  81   bool is_ordinary;
  82   unsigned int target_shndx = this->symbol_section(symndx, &value,
  83                                                    &is_ordinary);
  84   if (!is_ordinary)
  85     return 0;
  86   if (this->reloc_type_ == elfcpp::SHT_RELA)
  87     value += this->track_relocs_.next_addend();
  88   *target_offset = value;
  89   return target_shndx;
  90 }
  91
  92 static inline Elf_reloc_mapper*
  93 make_elf_reloc_mapper(Object* object, const unsigned char* symtab,
  94                       off_t symtab_size)
  95 {
  96   switch (parameters->size_and_endianness())
  97     {
  98 #ifdef HAVE_TARGET_32_LITTLE
  99       case Parameters::TARGET_32_LITTLE:
 100         return new Sized_elf_reloc_mapper<32, false>(object, symtab,
 101                                                      symtab_size);
 102 #endif
 103 #ifdef HAVE_TARGET_32_BIG
 104       case Parameters::TARGET_32_BIG:
 105         return new Sized_elf_reloc_mapper<32, true>(object, symtab,
 106                                                     symtab_size);
 107 #endif
 108 #ifdef HAVE_TARGET_64_LITTLE
 109       case Parameters::TARGET_64_LITTLE:
 110         return new Sized_elf_reloc_mapper<64, false>(object, symtab,
 111                                                      symtab_size);
 112 #endif
 113 #ifdef HAVE_TARGET_64_BIG
 114       case Parameters::TARGET_64_BIG:
 115         return new Sized_elf_reloc_mapper<64, true>(object, symtab,
 116                                                     symtab_size);
 117 #endif
 118       default:
 119         gold_unreachable();
 120     }
 121 }
 122
 123 // class Dwarf_abbrev_table
 124
 125 void
 126 Dwarf_abbrev_table::clear_abbrev_codes()
 127 {
 128   for (unsigned int code = 0; code < this->low_abbrev_code_max_; ++code)
 129     {
 130       if (this->low_abbrev_codes_[code] != NULL)
 131         {
 132           delete this->low_abbrev_codes_[code];
 133           this->low_abbrev_codes_[code] = NULL;
 134         }
 135     }
 136   for (Abbrev_code_table::iterator it = this->high_abbrev_codes_.begin();
 137        it != this->high_abbrev_codes_.end();
 138        ++it)
 139     {
 140       if (it->second != NULL)
 141         delete it->second;
 142     }
 143   this->high_abbrev_codes_.clear();
 144 }
 145
 146 // Read the abbrev table from an object file.
 147
 148 bool
 149 Dwarf_abbrev_table::do_read_abbrevs(
 150     Relobj* object,
 151     unsigned int abbrev_shndx,
 152     off_t abbrev_offset)
 153 {
 154   this->clear_abbrev_codes();
 155
 156   // If we don't have relocations, abbrev_shndx will be 0, and
 157   // we'll have to hunt for the .debug_abbrev section.
 158   if (abbrev_shndx == 0 && this->abbrev_shndx_ > 0)
 159     abbrev_shndx = this->abbrev_shndx_;
 160   else if (abbrev_shndx == 0)
 161     {
 162       for (unsigned int i = 1; i < object->shnum(); ++i)
 163         {
 164           std::string name = object->section_name(i);
 165           if (name == ".debug_abbrev")
 166             {
 167               abbrev_shndx = i;
 168               // Correct the offset.  For incremental update links, we have a
 169               // relocated offset that is relative to the output section, but
 170               // here we need an offset relative to the input section.
 171               abbrev_offset -= object->output_section_offset(i);
 172               break;
 173             }
 174         }
 175       if (abbrev_shndx == 0)
 176         return false;
 177     }
 178
 179   // Get the section contents and decompress if necessary.
 180   if (abbrev_shndx != this->abbrev_shndx_)
 181     {
 182       if (this->owns_buffer_ && this->buffer_ != NULL)
 183         {
 184           delete[] this->buffer_;
 185           this->owns_buffer_ = false;
 186         }
 187
 188       section_size_type buffer_size;
 189       this->buffer_ =
 190           object->decompressed_section_contents(abbrev_shndx,
 191                                                 &buffer_size,
 192                                                 &this->owns_buffer_);
 193       this->buffer_end_ = this->buffer_ + buffer_size;
 194       this->abbrev_shndx_ = abbrev_shndx;
 195     }
 196
 197   this->buffer_pos_ = this->buffer_ + abbrev_offset;
 198   return true;
 199 }
 200
 201 // Lookup the abbrev code entry for CODE.  This function is called
 202 // only when the abbrev code is not in the direct lookup table.
 203 // It may be in the hash table, it may not have been read yet,
 204 // or it may not exist in the abbrev table.
 205
 206 const Dwarf_abbrev_table::Abbrev_code*
 207 Dwarf_abbrev_table::do_get_abbrev(unsigned int code)
 208 {
 209   // See if the abbrev code is already in the hash table.
 210   Abbrev_code_table::const_iterator it = this->high_abbrev_codes_.find(code);
 211   if (it != this->high_abbrev_codes_.end())
 212     return it->second;
 213
 214   // Read and store abbrev code definitions until we find the
 215   // one we're looking for.
 216   for (;;)
 217     {
 218       // Read the abbrev code.  A zero here indicates the end of the
 219       // abbrev table.
 220       size_t len;
 221       if (this->buffer_pos_ >= this->buffer_end_)
 222         return NULL;
 223       uint64_t nextcode = read_unsigned_LEB_128(this->buffer_pos_, &len);
 224       if (nextcode == 0)
 225         {
 226           this->buffer_pos_ = this->buffer_end_;
 227           return NULL;
 228         }
 229       this->buffer_pos_ += len;
 230
 231       // Read the tag.
 232       if (this->buffer_pos_ >= this->buffer_end_)
 233         return NULL;
 234       uint64_t tag = read_unsigned_LEB_128(this->buffer_pos_, &len);
 235       this->buffer_pos_ += len;
 236
 237       // Read the has_children flag.
 238       if (this->buffer_pos_ >= this->buffer_end_)
 239         return NULL;
 240       bool has_children = *this->buffer_pos_ == elfcpp::DW_CHILDREN_yes;
 241       this->buffer_pos_ += 1;
 242
 243       // Read the list of (attribute, form) pairs.
 244       Abbrev_code* entry = new Abbrev_code(tag, has_children);
 245       for (;;)
 246         {
 247           // Read the attribute.
 248           if (this->buffer_pos_ >= this->buffer_end_)
 249             return NULL;
 250           uint64_t attr = read_unsigned_LEB_128(this->buffer_pos_, &len);
 251           this->buffer_pos_ += len;
 252
 253           // Read the form.
 254           if (this->buffer_pos_ >= this->buffer_end_)
 255             return NULL;
 256           uint64_t form = read_unsigned_LEB_128(this->buffer_pos_, &len);
 257           this->buffer_pos_ += len;
 258
 259           // A (0,0) pair terminates the list.
 260           if (attr == 0 && form == 0)
 261             break;
 262
 263           if (attr == elfcpp::DW_AT_sibling)
 264             entry->has_sibling_attribute = true;
 265
 266           entry->add_attribute(attr, form);
 267         }
 268
 269       this->store_abbrev(nextcode, entry);
 270       if (nextcode == code)
 271         return entry;
 272     }
 273
 274   return NULL;
 275 }
 276
 277 // class Dwarf_ranges_table
 278
 279 // Read the ranges table from an object file.
 280
 281 bool
 282 Dwarf_ranges_table::read_ranges_table(
 283     Relobj* object,
 284     const unsigned char* symtab,
 285     off_t symtab_size,
 286     unsigned int ranges_shndx)
 287 {
 288   // If we've already read this abbrev table, return immediately.
 289   if (this->ranges_shndx_ > 0
 290       && this->ranges_shndx_ == ranges_shndx)
 291     return true;
 292
 293   // If we don't have relocations, ranges_shndx will be 0, and
 294   // we'll have to hunt for the .debug_ranges section.
 295   if (ranges_shndx == 0 && this->ranges_shndx_ > 0)
 296     ranges_shndx = this->ranges_shndx_;
 297   else if (ranges_shndx == 0)
 298     {
 299       for (unsigned int i = 1; i < object->shnum(); ++i)
 300         {
 301           std::string name = object->section_name(i);
 302           if (name == ".debug_ranges")
 303             {
 304               ranges_shndx = i;
 305               this->output_section_offset_ = object->output_section_offset(i);
 306               break;
 307             }
 308         }
 309       if (ranges_shndx == 0)
 310         return false;
 311     }
 312
 313   // Get the section contents and decompress if necessary.
 314   if (ranges_shndx != this->ranges_shndx_)
 315     {
 316       if (this->owns_ranges_buffer_ && this->ranges_buffer_ != NULL)
 317         {
 318           delete[] this->ranges_buffer_;
 319           this->owns_ranges_buffer_ = false;
 320         }
 321
 322       section_size_type buffer_size;
 323       this->ranges_buffer_ =
 324           object->decompressed_section_contents(ranges_shndx,
 325                                                 &buffer_size,
 326                                                 &this->owns_ranges_buffer_);
 327       this->ranges_buffer_end_ = this->ranges_buffer_ + buffer_size;
 328       this->ranges_shndx_ = ranges_shndx;
 329     }
 330
 331   if (this->ranges_reloc_mapper_ != NULL)
 332     {
 333       delete this->ranges_reloc_mapper_;
 334       this->ranges_reloc_mapper_ = NULL;
 335     }
 336
 337   // For incremental objects, we have no relocations.
 338   if (object->is_incremental())
 339     return true;
 340
 341   // Find the relocation section for ".debug_ranges".
 342   unsigned int reloc_shndx = 0;
 343   unsigned int reloc_type = 0;
 344   for (unsigned int i = 0; i < object->shnum(); ++i)
 345     {
 346       reloc_type = object->section_type(i);
 347       if ((reloc_type == elfcpp::SHT_REL
 348            || reloc_type == elfcpp::SHT_RELA)
 349           && object->section_info(i) == ranges_shndx)
 350         {
 351           reloc_shndx = i;
 352           break;
 353         }
 354     }
 355
 356   this->ranges_reloc_mapper_ = make_elf_reloc_mapper(object, symtab,
 357                                                      symtab_size);
 358   this->ranges_reloc_mapper_->initialize(reloc_shndx, reloc_type);
 359
 360   return true;
 361 }
 362
 363 // Read a range list from section RANGES_SHNDX at offset RANGES_OFFSET.
 364
 365 Dwarf_range_list*
 366 Dwarf_ranges_table::read_range_list(
 367     Relobj* object,
 368     const unsigned char* symtab,
 369     off_t symtab_size,
 370     unsigned int addr_size,
 371     unsigned int ranges_shndx,
 372     off_t offset)
 373 {
 374   Dwarf_range_list* ranges;
 375
 376   if (!this->read_ranges_table(object, symtab, symtab_size, ranges_shndx))
 377     return NULL;
 378
 379   // Correct the offset.  For incremental update links, we have a
 380   // relocated offset that is relative to the output section, but
 381   // here we need an offset relative to the input section.
 382   offset -= this->output_section_offset_;
 383
 384   // Read the range list at OFFSET.
 385   ranges = new Dwarf_range_list();
 386   off_t base = 0;
 387   for (;
 388        this->ranges_buffer_ + offset < this->ranges_buffer_end_;
 389        offset += 2 * addr_size)
 390     {
 391       off_t start;
 392       off_t end;
 393
 394       // Read the raw contents of the section.
 395       if (addr_size == 4)
 396         {
 397           start = read_from_pointer<32>(this->ranges_buffer_ + offset);
 398           end = read_from_pointer<32>(this->ranges_buffer_ + offset + 4);
 399         }
 400       else
 401         {
 402           start = read_from_pointer<64>(this->ranges_buffer_ + offset);
 403           end = read_from_pointer<64>(this->ranges_buffer_ + offset + 8);
 404         }
 405
 406       // Check for relocations and adjust the values.
 407       unsigned int shndx1 = 0;
 408       unsigned int shndx2 = 0;
 409       if (this->ranges_reloc_mapper_ != NULL)
 410         {
 411           shndx1 =
 412               this->ranges_reloc_mapper_->get_reloc_target(offset, &start);
 413           shndx2 =
 414               this->ranges_reloc_mapper_->get_reloc_target(offset + addr_size,
 415                                                            &end);
 416         }
 417
 418       // End of list is marked by a pair of zeroes.
 419       if (shndx1 == 0 && start == 0 && end == 0)
 420         break;
 421
 422       // A "base address selection entry" is identified by
 423       // 0xffffffff for the first value of the pair.  The second
 424       // value is used as a base for subsequent range list entries.
 425       if (shndx1 == 0 && start == -1)
 426         base = end;
 427       else if (shndx1 == shndx2)
 428         {
 429           if (shndx1 == 0 || object->is_section_included(shndx1))
 430             ranges->add(shndx1, base + start, base + end);
 431         }
 432       else
 433         gold_warning(_("%s: DWARF info may be corrupt; offsets in a "
 434                        "range list entry are in different sections"),
 435                      object->name().c_str());
 436     }
 437
 438   return ranges;
 439 }
 440
 441 // class Dwarf_pubnames_table
 442
 443 // Read the pubnames section SHNDX from the object file.
 444
 445 bool
 446 Dwarf_pubnames_table::read_section(Relobj* object, unsigned int shndx)
 447 {
 448   section_size_type buffer_size;
 449
 450   // If we don't have relocations, shndx will be 0, and
 451   // we'll have to hunt for the .debug_pubnames/pubtypes section.
 452   if (shndx == 0)
 453     {
 454       const char* name = (this->is_pubtypes_
 455                           ? ".debug_pubtypes"
 456                           : ".debug_pubnames");
 457       for (unsigned int i = 1; i < object->shnum(); ++i)
 458         {
 459           if (object->section_name(i) == name)
 460             {
 461               shndx = i;
 462               this->output_section_offset_ = object->output_section_offset(i);
 463               break;
 464             }
 465         }
 466       if (shndx == 0)
 467         return false;
 468     }
 469
 470   this->buffer_ = object->decompressed_section_contents(shndx,
 471                                                         &buffer_size,
 472                                                         &this->owns_buffer_);
 473   if (this->buffer_ == NULL)
 474     return false;
 475   this->buffer_end_ = this->buffer_ + buffer_size;
 476   return true;
 477 }
 478
 479 // Read the header for the set at OFFSET.
 480
 481 bool
 482 Dwarf_pubnames_table::read_header(off_t offset)
 483 {
 484   // Correct the offset.  For incremental update links, we have a
 485   // relocated offset that is relative to the output section, but
 486   // here we need an offset relative to the input section.
 487   offset -= this->output_section_offset_;
 488
 489   if (offset < 0 || offset + 14 >= this->buffer_end_ - this->buffer_)
 490     return false;
 491
 492   const unsigned char* pinfo = this->buffer_ + offset;
 493
 494   // Read the unit_length field.
 495   uint32_t unit_length = read_from_pointer<32>(pinfo);
 496   pinfo += 4;
 497   if (unit_length == 0xffffffff)
 498     {
 499       unit_length = read_from_pointer<64>(pinfo);
 500       pinfo += 8;
 501       this->offset_size_ = 8;
 502     }
 503   else
 504     this->offset_size_ = 4;
 505
 506   // Check the version.
 507   unsigned int version = read_from_pointer<16>(pinfo);
 508   pinfo += 2;
 509   if (version != 2)
 510     return false;
 511
 512   // Skip the debug_info_offset and debug_info_size fields.
 513   pinfo += 2 * this->offset_size_;
 514
 515   if (pinfo >= this->buffer_end_)
 516     return false;
 517
 518   this->pinfo_ = pinfo;
 519   return true;
 520 }
 521
 522 // Read the next name from the set.
 523
 524 const char*
 525 Dwarf_pubnames_table::next_name()
 526 {
 527   const unsigned char* pinfo = this->pinfo_;
 528
 529   // Read the offset within the CU.  If this is zero, we have reached
 530   // the end of the list.
 531   uint32_t offset;
 532   if (this->offset_size_ == 4)
 533     offset = read_from_pointer<32>(&pinfo);
 534   else
 535     offset = read_from_pointer<64>(&pinfo);
 536   if (offset == 0)
 537     return NULL;
 538
 539   // Return a pointer to the string at the current location,
 540   // and advance the pointer to the next entry.
 541   const char* ret = reinterpret_cast<const char*>(pinfo);
 542   while (pinfo < this->buffer_end_ && *pinfo != '\0')
 543     ++pinfo;
 544   if (pinfo < this->buffer_end_)
 545     ++pinfo;
 546
 547   this->pinfo_ = pinfo;
 548   return ret;
 549 }
 550
 551 // class Dwarf_die
 552
 553 Dwarf_die::Dwarf_die(
 554     Dwarf_info_reader* dwinfo,
 555     off_t die_offset,
 556     Dwarf_die* parent)
 557   : dwinfo_(dwinfo), parent_(parent), die_offset_(die_offset),
 558     child_offset_(0), sibling_offset_(0), abbrev_code_(NULL), attributes_(),
 559     attributes_read_(false), name_(NULL), name_off_(-1), linkage_name_(NULL),
 560     linkage_name_off_(-1), string_shndx_(0), specification_(0),
 561     abstract_origin_(0)
 562 {
 563   size_t len;
 564   const unsigned char* pdie = dwinfo->buffer_at_offset(die_offset);
 565   if (pdie == NULL)
 566     return;
 567   unsigned int code = read_unsigned_LEB_128(pdie, &len);
 568   if (code == 0)
 569     {
 570       if (parent != NULL)
 571         parent->set_sibling_offset(die_offset + len);
 572       return;
 573     }
 574   this->attr_offset_ = len;
 575
 576   // Lookup the abbrev code in the abbrev table.
 577   this->abbrev_code_ = dwinfo->get_abbrev(code);
 578 }
 579
 580 // Read all the attributes of the DIE.
 581
 582 bool
 583 Dwarf_die::read_attributes()
 584 {
 585   if (this->attributes_read_)
 586     return true;
 587
 588   gold_assert(this->abbrev_code_ != NULL);
 589
 590   const unsigned char* pdie =
 591       this->dwinfo_->buffer_at_offset(this->die_offset_);
 592   if (pdie == NULL)
 593     return false;
 594   const unsigned char* pattr = pdie + this->attr_offset_;
 595
 596   unsigned int nattr = this->abbrev_code_->attributes.size();
 597   this->attributes_.reserve(nattr);
 598   for (unsigned int i = 0; i < nattr; ++i)
 599     {
 600       size_t len;
 601       unsigned int attr = this->abbrev_code_->attributes[i].attr;
 602       unsigned int form = this->abbrev_code_->attributes[i].form;
 603       if (form == elfcpp::DW_FORM_indirect)
 604         {
 605           form = read_unsigned_LEB_128(pattr, &len);
 606           pattr += len;
 607         }
 608       off_t attr_off = this->die_offset_ + (pattr - pdie);
 609       bool ref_form = false;
 610       Attribute_value attr_value;
 611       attr_value.attr = attr;
 612       attr_value.form = form;
 613       attr_value.aux.shndx = 0;
 614       switch(form)
 615         {
 616           case elfcpp::DW_FORM_flag_present:
 617             attr_value.val.intval = 1;
 618             break;
 619           case elfcpp::DW_FORM_strp:
 620             {
 621               off_t str_off;
 622               if (this->dwinfo_->offset_size() == 4)
 623                 str_off = read_from_pointer<32>(&pattr);
 624               else
 625                 str_off = read_from_pointer<64>(&pattr);
 626               unsigned int shndx =
 627                   this->dwinfo_->lookup_reloc(attr_off, &str_off);
 628               attr_value.aux.shndx = shndx;
 629               attr_value.val.refval = str_off;
 630               break;
 631             }
 632           case elfcpp::DW_FORM_sec_offset:
 633             {
 634               off_t sec_off;
 635               if (this->dwinfo_->offset_size() == 4)
 636                 sec_off = read_from_pointer<32>(&pattr);
 637               else
 638                 sec_off = read_from_pointer<64>(&pattr);
 639               unsigned int shndx =
 640                   this->dwinfo_->lookup_reloc(attr_off, &sec_off);
 641               attr_value.aux.shndx = shndx;
 642               attr_value.val.refval = sec_off;
 643               ref_form = true;
 644               break;
 645             }
 646           case elfcpp::DW_FORM_addr:
 647           case elfcpp::DW_FORM_ref_addr:
 648             {
 649               off_t sec_off;
 650               if (this->dwinfo_->address_size() == 4)
 651                 sec_off = read_from_pointer<32>(&pattr);
 652               else
 653                 sec_off = read_from_pointer<64>(&pattr);
 654               unsigned int shndx =
 655                   this->dwinfo_->lookup_reloc(attr_off, &sec_off);
 656               attr_value.aux.shndx = shndx;
 657               attr_value.val.refval = sec_off;
 658               ref_form = true;
 659               break;
 660             }
 661           case elfcpp::DW_FORM_block1:
 662             attr_value.aux.blocklen = *pattr++;
 663             attr_value.val.blockval = pattr;
 664             pattr += attr_value.aux.blocklen;
 665             break;
 666           case elfcpp::DW_FORM_block2:
 667             attr_value.aux.blocklen = read_from_pointer<16>(&pattr);
 668             attr_value.val.blockval = pattr;
 669             pattr += attr_value.aux.blocklen;
 670             break;
 671           case elfcpp::DW_FORM_block4:
 672             attr_value.aux.blocklen = read_from_pointer<32>(&pattr);
 673             attr_value.val.blockval = pattr;
 674             pattr += attr_value.aux.blocklen;
 675             break;
 676           case elfcpp::DW_FORM_block:
 677           case elfcpp::DW_FORM_exprloc:
 678             attr_value.aux.blocklen = read_unsigned_LEB_128(pattr, &len);
 679             attr_value.val.blockval = pattr + len;
 680             pattr += len + attr_value.aux.blocklen;
 681             break;
 682           case elfcpp::DW_FORM_data1:
 683           case elfcpp::DW_FORM_flag:
 684             attr_value.val.intval = *pattr++;
 685             break;
 686           case elfcpp::DW_FORM_ref1:
 687             attr_value.val.refval = *pattr++;
 688             ref_form = true;
 689             break;
 690           case elfcpp::DW_FORM_data2:
 691             attr_value.val.intval = read_from_pointer<16>(&pattr);
 692             break;
 693           case elfcpp::DW_FORM_ref2:
 694             attr_value.val.refval = read_from_pointer<16>(&pattr);
 695             ref_form = true;
 696             break;
 697           case elfcpp::DW_FORM_data4:
 698             {
 699               off_t sec_off;
 700               sec_off = read_from_pointer<32>(&pattr);
 701               unsigned int shndx =
 702                   this->dwinfo_->lookup_reloc(attr_off, &sec_off);
 703               attr_value.aux.shndx = shndx;
 704               attr_value.val.intval = sec_off;
 705               break;
 706             }
 707           case elfcpp::DW_FORM_ref4:
 708             {
 709               off_t sec_off;
 710               sec_off = read_from_pointer<32>(&pattr);
 711               unsigned int shndx =
 712                   this->dwinfo_->lookup_reloc(attr_off, &sec_off);
 713               attr_value.aux.shndx = shndx;
 714               attr_value.val.refval = sec_off;
 715               ref_form = true;
 716               break;
 717             }
 718           case elfcpp::DW_FORM_data8:
 719             {
 720               off_t sec_off;
 721               sec_off = read_from_pointer<64>(&pattr);
 722               unsigned int shndx =
 723                   this->dwinfo_->lookup_reloc(attr_off, &sec_off);
 724               attr_value.aux.shndx = shndx;
 725               attr_value.val.intval = sec_off;
 726               break;
 727             }
 728           case elfcpp::DW_FORM_ref_sig8:
 729             attr_value.val.uintval = read_from_pointer<64>(&pattr);
 730             break;
 731           case elfcpp::DW_FORM_ref8:
 732             {
 733               off_t sec_off;
 734               sec_off = read_from_pointer<64>(&pattr);
 735               unsigned int shndx =
 736                   this->dwinfo_->lookup_reloc(attr_off, &sec_off);
 737               attr_value.aux.shndx = shndx;
 738               attr_value.val.refval = sec_off;
 739               ref_form = true;
 740               break;
 741             }
 742           case elfcpp::DW_FORM_ref_udata:
 743             attr_value.val.refval = read_unsigned_LEB_128(pattr, &len);
 744             ref_form = true;
 745             pattr += len;
 746             break;
 747           case elfcpp::DW_FORM_udata:
 748             attr_value.val.uintval = read_unsigned_LEB_128(pattr, &len);
 749             pattr += len;
 750             break;
 751           case elfcpp::DW_FORM_sdata:
 752             attr_value.val.intval = read_signed_LEB_128(pattr, &len);
 753             pattr += len;
 754             break;
 755           case elfcpp::DW_FORM_string:
 756             attr_value.val.stringval = reinterpret_cast<const char*>(pattr);
 757             len = strlen(attr_value.val.stringval);
 758             pattr += len + 1;
 759             break;
 760           default:
 761             return false;
 762         }
 763
 764       // Cache the most frequently-requested attributes.
 765       switch (attr)
 766         {
 767           case elfcpp::DW_AT_name:
 768             if (form == elfcpp::DW_FORM_string)
 769               this->name_ = attr_value.val.stringval;
 770             else if (form == elfcpp::DW_FORM_strp)
 771               {
 772                 // All indirect strings should refer to the same
 773                 // string section, so we just save the last one seen.
 774                 this->string_shndx_ = attr_value.aux.shndx;
 775                 this->name_off_ = attr_value.val.refval;
 776               }
 777             break;
 778           case elfcpp::DW_AT_linkage_name:
 779           case elfcpp::DW_AT_MIPS_linkage_name:
 780             if (form == elfcpp::DW_FORM_string)
 781               this->linkage_name_ = attr_value.val.stringval;
 782             else if (form == elfcpp::DW_FORM_strp)
 783               {
 784                 // All indirect strings should refer to the same
 785                 // string section, so we just save the last one seen.
 786                 this->string_shndx_ = attr_value.aux.shndx;
 787                 this->linkage_name_off_ = attr_value.val.refval;
 788               }
 789             break;
 790           case elfcpp::DW_AT_specification:
 791             if (ref_form)
 792               this->specification_ = attr_value.val.refval;
 793             break;
 794           case elfcpp::DW_AT_abstract_origin:
 795             if (ref_form)
 796               this->abstract_origin_ = attr_value.val.refval;
 797             break;
 798           case elfcpp::DW_AT_sibling:
 799             if (ref_form && attr_value.aux.shndx == 0)
 800               this->sibling_offset_ = attr_value.val.refval;
 801           default:
 802             break;
 803         }
 804
 805       this->attributes_.push_back(attr_value);
 806     }
 807
 808   // Now that we know where the next DIE begins, record the offset
 809   // to avoid later recalculation.
 810   if (this->has_children())
 811     this->child_offset_ = this->die_offset_ + (pattr - pdie);
 812   else
 813     this->sibling_offset_ = this->die_offset_ + (pattr - pdie);
 814
 815   this->attributes_read_ = true;
 816   return true;
 817 }
 818
 819 // Skip all the attributes of the DIE and return the offset of the next DIE.
 820
 821 off_t
 822 Dwarf_die::skip_attributes()
 823 {
 824   typedef Dwarf_abbrev_table::Attribute Attribute;
 825
 826   gold_assert(this->abbrev_code_ != NULL);
 827
 828   const unsigned char* pdie =
 829       this->dwinfo_->buffer_at_offset(this->die_offset_);
 830   if (pdie == NULL)
 831     return 0;
 832   const unsigned char* pattr = pdie + this->attr_offset_;
 833
 834   for (unsigned int i = 0; i < this->abbrev_code_->attributes.size(); ++i)
 835     {
 836       size_t len;
 837       unsigned int form = this->abbrev_code_->attributes[i].form;
 838       if (form == elfcpp::DW_FORM_indirect)
 839         {
 840           form = read_unsigned_LEB_128(pattr, &len);
 841           pattr += len;
 842         }
 843       switch(form)
 844         {
 845           case elfcpp::DW_FORM_flag_present:
 846             break;
 847           case elfcpp::DW_FORM_strp:
 848           case elfcpp::DW_FORM_sec_offset:
 849             pattr += this->dwinfo_->offset_size();
 850             break;
 851           case elfcpp::DW_FORM_addr:
 852           case elfcpp::DW_FORM_ref_addr:
 853             pattr += this->dwinfo_->address_size();
 854             break;
 855           case elfcpp::DW_FORM_block1:
 856             pattr += 1 + *pattr;
 857             break;
 858           case elfcpp::DW_FORM_block2:
 859             {
 860               uint16_t block_size;
 861               block_size = read_from_pointer<16>(&pattr);
 862               pattr += block_size;
 863               break;
 864             }
 865           case elfcpp::DW_FORM_block4:
 866             {
 867               uint32_t block_size;
 868               block_size = read_from_pointer<32>(&pattr);
 869               pattr += block_size;
 870               break;
 871             }
 872           case elfcpp::DW_FORM_block:
 873           case elfcpp::DW_FORM_exprloc:
 874             {
 875               uint64_t block_size;
 876               block_size = read_unsigned_LEB_128(pattr, &len);
 877               pattr += len + block_size;
 878               break;
 879             }
 880           case elfcpp::DW_FORM_data1:
 881           case elfcpp::DW_FORM_ref1:
 882           case elfcpp::DW_FORM_flag:
 883             pattr += 1;
 884             break;
 885           case elfcpp::DW_FORM_data2:
 886           case elfcpp::DW_FORM_ref2:
 887             pattr += 2;
 888             break;
 889           case elfcpp::DW_FORM_data4:
 890           case elfcpp::DW_FORM_ref4:
 891             pattr += 4;
 892             break;
 893           case elfcpp::DW_FORM_data8:
 894           case elfcpp::DW_FORM_ref8:
 895           case elfcpp::DW_FORM_ref_sig8:
 896             pattr += 8;
 897             break;
 898           case elfcpp::DW_FORM_ref_udata:
 899           case elfcpp::DW_FORM_udata:
 900             read_unsigned_LEB_128(pattr, &len);
 901             pattr += len;
 902             break;
 903           case elfcpp::DW_FORM_sdata:
 904             read_signed_LEB_128(pattr, &len);
 905             pattr += len;
 906             break;
 907           case elfcpp::DW_FORM_string:
 908             len = strlen(reinterpret_cast<const char*>(pattr));
 909             pattr += len + 1;
 910             break;
 911           default:
 912             return 0;
 913         }
 914     }
 915
 916   return this->die_offset_ + (pattr - pdie);
 917 }
 918
 919 // Get the name of the DIE and cache it.
 920
 921 void
 922 Dwarf_die::set_name()
 923 {
 924   if (this->name_ != NULL || !this->read_attributes())
 925     return;
 926   if (this->name_off_ != -1)
 927     this->name_ = this->dwinfo_->get_string(this->name_off_,
 928                                             this->string_shndx_);
 929 }
 930
 931 // Get the linkage name of the DIE and cache it.
 932
 933 void
 934 Dwarf_die::set_linkage_name()
 935 {
 936   if (this->linkage_name_ != NULL || !this->read_attributes())
 937     return;
 938   if (this->linkage_name_off_ != -1)
 939     this->linkage_name_ = this->dwinfo_->get_string(this->linkage_name_off_,
 940                                                     this->string_shndx_);
 941 }
 942
 943 // Return the value of attribute ATTR.
 944
 945 const Dwarf_die::Attribute_value*
 946 Dwarf_die::attribute(unsigned int attr)
 947 {
 948   if (!this->read_attributes())
 949     return NULL;
 950   for (unsigned int i = 0; i < this->attributes_.size(); ++i)
 951     {
 952       if (this->attributes_[i].attr == attr)
 953         return &this->attributes_[i];
 954     }
 955   return NULL;
 956 }
 957
 958 const char*
 959 Dwarf_die::string_attribute(unsigned int attr)
 960 {
 961   const Attribute_value* attr_val = this->attribute(attr);
 962   if (attr_val == NULL)
 963     return NULL;
 964   switch (attr_val->form)
 965     {
 966       case elfcpp::DW_FORM_string:
 967         return attr_val->val.stringval;
 968       case elfcpp::DW_FORM_strp:
 969         return this->dwinfo_->get_string(attr_val->val.refval,
 970                                          attr_val->aux.shndx);
 971       default:
 972         return NULL;
 973     }
 974 }
 975
 976 int64_t
 977 Dwarf_die::int_attribute(unsigned int attr)
 978 {
 979   const Attribute_value* attr_val = this->attribute(attr);
 980   if (attr_val == NULL)
 981     return 0;
 982   switch (attr_val->form)
 983     {
 984       case elfcpp::DW_FORM_flag_present:
 985       case elfcpp::DW_FORM_data1:
 986       case elfcpp::DW_FORM_flag:
 987       case elfcpp::DW_FORM_data2:
 988       case elfcpp::DW_FORM_data4:
 989       case elfcpp::DW_FORM_data8:
 990       case elfcpp::DW_FORM_sdata:
 991         return attr_val->val.intval;
 992       default:
 993         return 0;
 994     }
 995 }
 996
 997 uint64_t
 998 Dwarf_die::uint_attribute(unsigned int attr)
 999 {
1000   const Attribute_value* attr_val = this->attribute(attr);
1001   if (attr_val == NULL)
1002     return 0;
1003   switch (attr_val->form)
1004     {
1005       case elfcpp::DW_FORM_flag_present:
1006       case elfcpp::DW_FORM_data1:
1007       case elfcpp::DW_FORM_flag:
1008       case elfcpp::DW_FORM_data4:
1009       case elfcpp::DW_FORM_data8:
1010       case elfcpp::DW_FORM_ref_sig8:
1011       case elfcpp::DW_FORM_udata:
1012         return attr_val->val.uintval;
1013       default:
1014         return 0;
1015     }
1016 }
1017
1018 off_t
1019 Dwarf_die::ref_attribute(unsigned int attr, unsigned int* shndx)
1020 {
1021   const Attribute_value* attr_val = this->attribute(attr);
1022   if (attr_val == NULL)
1023     return -1;
1024   switch (attr_val->form)
1025     {
1026       case elfcpp::DW_FORM_sec_offset:
1027       case elfcpp::DW_FORM_addr:
1028       case elfcpp::DW_FORM_ref_addr:
1029       case elfcpp::DW_FORM_ref1:
1030       case elfcpp::DW_FORM_ref2:
1031       case elfcpp::DW_FORM_ref4:
1032       case elfcpp::DW_FORM_ref8:
1033       case elfcpp::DW_FORM_ref_udata:
1034         *shndx = attr_val->aux.shndx;
1035         return attr_val->val.refval;
1036       case elfcpp::DW_FORM_ref_sig8:
1037         *shndx = attr_val->aux.shndx;
1038         return attr_val->val.uintval;
1039       case elfcpp::DW_FORM_data4:
1040       case elfcpp::DW_FORM_data8:
1041         *shndx = attr_val->aux.shndx;
1042         return attr_val->val.intval;
1043       default:
1044         return -1;
1045     }
1046 }
1047
1048 off_t
1049 Dwarf_die::address_attribute(unsigned int attr, unsigned int* shndx)
1050 {
1051   const Attribute_value* attr_val = this->attribute(attr);
1052   if (attr_val == NULL || attr_val->form != elfcpp::DW_FORM_addr)
1053     return -1;
1054
1055   *shndx = attr_val->aux.shndx;
1056   return attr_val->val.refval;
1057 }
1058
1059 // Return the offset of this DIE's first child.
1060
1061 off_t
1062 Dwarf_die::child_offset()
1063 {
1064   gold_assert(this->abbrev_code_ != NULL);
1065   if (!this->has_children())
1066     return 0;
1067   if (this->child_offset_ == 0)
1068     this->child_offset_ = this->skip_attributes();
1069   return this->child_offset_;
1070 }
1071
1072 // Return the offset of this DIE's next sibling.
1073
1074 off_t
1075 Dwarf_die::sibling_offset()
1076 {
1077   gold_assert(this->abbrev_code_ != NULL);
1078
1079   if (this->sibling_offset_ != 0)
1080     return this->sibling_offset_;
1081
1082   if (!this->has_children())
1083     {
1084       this->sibling_offset_ = this->skip_attributes();
1085       return this->sibling_offset_;
1086     }
1087
1088   if (this->has_sibling_attribute())
1089     {
1090       if (!this->read_attributes())
1091         return 0;
1092       if (this->sibling_offset_ != 0)
1093         return this->sibling_offset_;
1094     }
1095
1096   // Skip over the children.
1097   off_t child_offset = this->child_offset();
1098   while (child_offset > 0)
1099     {
1100       Dwarf_die die(this->dwinfo_, child_offset, this);
1101       // The Dwarf_die ctor will set this DIE's sibling offset
1102       // when it reads a zero abbrev code.
1103       if (die.tag() == 0)
1104         break;
1105       child_offset = die.sibling_offset();
1106     }
1107
1108   // This should be set by now.  If not, there was a problem reading
1109   // the DWARF info, and we return 0.
1110   return this->sibling_offset_;
1111 }
1112
1113 // class Dwarf_info_reader
1114
1115 // Check that the pointer P is within the current compilation unit.
1116
1117 inline bool
1118 Dwarf_info_reader::check_buffer(const unsigned char* p) const
1119 {
1120   if (p > this->buffer_ + this->cu_offset_ + this->cu_length_)
1121     {
1122       gold_warning(_("%s: corrupt debug info in %s"),
1123                    this->object_->name().c_str(),
1124                    this->object_->section_name(this->shndx_).c_str());
1125       return false;
1126     }
1127   return true;
1128 }
1129
1130 // Begin parsing the debug info.  This calls visit_compilation_unit()
1131 // or visit_type_unit() for each compilation or type unit found in the
1132 // section, and visit_die() for each top-level DIE.
1133
1134 void
1135 Dwarf_info_reader::parse()
1136 {
1137   switch (parameters->size_and_endianness())
1138     {
1139 #ifdef HAVE_TARGET_32_LITTLE
1140       case Parameters::TARGET_32_LITTLE:
1141         this->do_parse<false>();
1142         break;
1143 #endif
1144 #ifdef HAVE_TARGET_32_BIG
1145       case Parameters::TARGET_32_BIG:
1146         this->do_parse<true>();
1147         break;
1148 #endif
1149 #ifdef HAVE_TARGET_64_LITTLE
1150       case Parameters::TARGET_64_LITTLE:
1151         this->do_parse<false>();
1152         break;
1153 #endif
1154 #ifdef HAVE_TARGET_64_BIG
1155       case Parameters::TARGET_64_BIG:
1156         this->do_parse<true>();
1157         break;
1158 #endif
1159       default:
1160         gold_unreachable();
1161     }
1162 }
1163
1164 template<bool big_endian>
1165 void
1166 Dwarf_info_reader::do_parse()
1167 {
1168   // Get the section contents and decompress if necessary.
1169   section_size_type buffer_size;
1170   bool buffer_is_new;
1171   this->buffer_ = this->object_->decompressed_section_contents(this->shndx_,
1172                                                                &buffer_size,
1173                                                                &buffer_is_new);
1174   if (this->buffer_ == NULL || buffer_size == 0)
1175     return;
1176   this->buffer_end_ = this->buffer_ + buffer_size;
1177
1178   // The offset of this input section in the output section.
1179   off_t section_offset = this->object_->output_section_offset(this->shndx_);
1180
1181   // Start tracking relocations for this section.
1182   this->reloc_mapper_ = make_elf_reloc_mapper(this->object_, this->symtab_,
1183                                               this->symtab_size_);
1184   this->reloc_mapper_->initialize(this->reloc_shndx_, this->reloc_type_);
1185
1186   // Loop over compilation units (or type units).
1187   unsigned int abbrev_shndx = 0;
1188   off_t abbrev_offset = 0;
1189   const unsigned char* pinfo = this->buffer_;
1190   while (pinfo < this->buffer_end_)
1191     {
1192       // Read the compilation (or type) unit header.
1193       const unsigned char* cu_start = pinfo;
1194       this->cu_offset_ = cu_start - this->buffer_;
1195       this->cu_length_ = this->buffer_end_ - cu_start;
1196
1197       // Read unit_length (4 or 12 bytes).
1198       if (!this->check_buffer(pinfo + 4))
1199         break;
1200       uint32_t unit_length =
1201           elfcpp::Swap_unaligned<32, big_endian>::readval(pinfo);
1202       pinfo += 4;
1203       if (unit_length == 0xffffffff)
1204         {
1205           if (!this->check_buffer(pinfo + 8))
1206             break;
1207           unit_length = elfcpp::Swap_unaligned<64, big_endian>::readval(pinfo);
1208           pinfo += 8;
1209           this->offset_size_ = 8;
1210         }
1211       else
1212         this->offset_size_ = 4;
1213       if (!this->check_buffer(pinfo + unit_length))
1214         break;
1215       const unsigned char* cu_end = pinfo + unit_length;
1216       this->cu_length_ = cu_end - cu_start;
1217       if (!this->check_buffer(pinfo + 2 + this->offset_size_ + 1))
1218         break;
1219
1220       // Read version (2 bytes).
1221       this->cu_version_ =
1222           elfcpp::Swap_unaligned<16, big_endian>::readval(pinfo);
1223       pinfo += 2;
1224
1225       // Read debug_abbrev_offset (4 or 8 bytes).
1226       if (this->offset_size_ == 4)
1227         abbrev_offset = elfcpp::Swap_unaligned<32, big_endian>::readval(pinfo);
1228       else
1229         abbrev_offset = elfcpp::Swap_unaligned<64, big_endian>::readval(pinfo);
1230       if (this->reloc_shndx_ > 0)
1231         {
1232           off_t reloc_offset = pinfo - this->buffer_;
1233           off_t value;
1234           abbrev_shndx =
1235               this->reloc_mapper_->get_reloc_target(reloc_offset, &value);
1236           if (abbrev_shndx == 0)
1237             return;
1238           if (this->reloc_type_ == elfcpp::SHT_REL)
1239             abbrev_offset += value;
1240           else
1241             abbrev_offset = value;
1242         }
1243       pinfo += this->offset_size_;
1244
1245       // Read address_size (1 byte).
1246       this->address_size_ = *pinfo++;
1247
1248       // For type units, read the two extra fields.
1249       uint64_t signature = 0;
1250       off_t type_offset = 0;
1251       if (this->is_type_unit_)
1252         {
1253           if (!this->check_buffer(pinfo + 8 + this->offset_size_))
1254             break;
1255
1256           // Read type_signature (8 bytes).
1257           signature = elfcpp::Swap_unaligned<64, big_endian>::readval(pinfo);
1258           pinfo += 8;
1259
1260           // Read type_offset (4 or 8 bytes).
1261           if (this->offset_size_ == 4)
1262             type_offset =
1263                 elfcpp::Swap_unaligned<32, big_endian>::readval(pinfo);
1264           else
1265             type_offset =
1266                 elfcpp::Swap_unaligned<64, big_endian>::readval(pinfo);
1267           pinfo += this->offset_size_;
1268         }
1269
1270       // Read the .debug_abbrev table.
1271       this->abbrev_table_.read_abbrevs(this->object_, abbrev_shndx,
1272                                        abbrev_offset);
1273
1274       // Visit the root DIE.
1275       Dwarf_die root_die(this,
1276                          pinfo - (this->buffer_ + this->cu_offset_),
1277                          NULL);
1278       if (root_die.tag() != 0)
1279         {
1280           // Visit the CU or TU.
1281           if (this->is_type_unit_)
1282             this->visit_type_unit(section_offset + this->cu_offset_,
1283                                   type_offset, signature, &root_die);
1284           else
1285             this->visit_compilation_unit(section_offset + this->cu_offset_,
1286                                          cu_end - cu_start, &root_die);
1287         }
1288
1289       // Advance to the next CU.
1290       pinfo = cu_end;
1291     }
1292
1293   if (buffer_is_new)
1294     {
1295       delete[] this->buffer_;
1296       this->buffer_ = NULL;
1297     }
1298 }
1299
1300 // Read the DWARF string table.
1301
1302 bool
1303 Dwarf_info_reader::do_read_string_table(unsigned int string_shndx)
1304 {
1305   Relobj* object = this->object_;
1306
1307   // If we don't have relocations, string_shndx will be 0, and
1308   // we'll have to hunt for the .debug_str section.
1309   if (string_shndx == 0)
1310     {
1311       for (unsigned int i = 1; i < this->object_->shnum(); ++i)
1312         {
1313           std::string name = object->section_name(i);
1314           if (name == ".debug_str")
1315             {
1316               string_shndx = i;
1317               this->string_output_section_offset_ =
1318                   object->output_section_offset(i);
1319               break;
1320             }
1321         }
1322       if (string_shndx == 0)
1323         return false;
1324     }
1325
1326   if (this->owns_string_buffer_ && this->string_buffer_ != NULL)
1327     {
1328       delete[] this->string_buffer_;
1329       this->owns_string_buffer_ = false;
1330     }
1331
1332   // Get the secton contents and decompress if necessary.
1333   section_size_type buffer_size;
1334   const unsigned char* buffer =
1335       object->decompressed_section_contents(string_shndx,
1336                                             &buffer_size,
1337                                             &this->owns_string_buffer_);
1338   this->string_buffer_ = reinterpret_cast<const char*>(buffer);
1339   this->string_buffer_end_ = this->string_buffer_ + buffer_size;
1340   this->string_shndx_ = string_shndx;
1341   return true;
1342 }
1343
1344 // Look for a relocation at offset ATTR_OFF in the dwarf info,
1345 // and return the section index and offset of the target.
1346
1347 unsigned int
1348 Dwarf_info_reader::lookup_reloc(off_t attr_off, off_t* target_off)
1349 {
1350   off_t value;
1351   attr_off += this->cu_offset_;
1352   unsigned int shndx = this->reloc_mapper_->get_reloc_target(attr_off, &value);
1353   if (shndx == 0)
1354     return 0;
1355   if (this->reloc_type_ == elfcpp::SHT_REL)
1356     *target_off += value;
1357   else
1358     *target_off = value;
1359   return shndx;
1360 }
1361
1362 // Return a string from the DWARF string table.
1363
1364 const char*
1365 Dwarf_info_reader::get_string(off_t str_off, unsigned int string_shndx)
1366 {
1367   if (!this->read_string_table(string_shndx))
1368     return NULL;
1369
1370   // Correct the offset.  For incremental update links, we have a
1371   // relocated offset that is relative to the output section, but
1372   // here we need an offset relative to the input section.
1373   str_off -= this->string_output_section_offset_;
1374
1375   const char* p = this->string_buffer_ + str_off;
1376
1377   if (p < this->string_buffer_ || p >= this->string_buffer_end_)
1378     return NULL;
1379
1380   return p;
1381 }
1382
1383 // The following are default, do-nothing, implementations of the
1384 // hook methods normally provided by a derived class.  We provide
1385 // default implementations rather than no implementation so that
1386 // a derived class needs to implement only the hooks that it needs
1387 // to use.
1388
1389 // Process a compilation unit and parse its child DIE.
1390
1391 void
1392 Dwarf_info_reader::visit_compilation_unit(off_t, off_t, Dwarf_die*)
1393 {
1394 }
1395
1396 // Process a type unit and parse its child DIE.
1397
1398 void
1399 Dwarf_info_reader::visit_type_unit(off_t, off_t, uint64_t, Dwarf_die*)
1400 {
1401 }
1402
1403 // class Sized_dwarf_line_info
1404
1405 struct LineStateMachine
1406 {
1407   int file_num;
1408   uint64_t address;
1409   int line_num;
1410   int column_num;
1411   unsigned int shndx;    // the section address refers to
1412   bool is_stmt;          // stmt means statement.
1413   bool basic_block;
1414   bool end_sequence;
1415 };
1416
1417 static void
1418 ResetLineStateMachine(struct LineStateMachine* lsm, bool default_is_stmt)
1419 {
1420   lsm->file_num = 1;
1421   lsm->address = 0;
1422   lsm->line_num = 1;
1423   lsm->column_num = 0;
1424   lsm->shndx = -1U;
1425   lsm->is_stmt = default_is_stmt;
1426   lsm->basic_block = false;
1427   lsm->end_sequence = false;
1428 }
1429
1430 template<int size, bool big_endian>
1431 Sized_dwarf_line_info<size, big_endian>::Sized_dwarf_line_info(
1432     Object* object,
1433     unsigned int read_shndx)
1434   : data_valid_(false), buffer_(NULL), buffer_start_(NULL),
1435     reloc_mapper_(NULL), symtab_buffer_(NULL), directories_(), files_(),
1436     current_header_index_(-1)
1437 {
1438   unsigned int debug_shndx;
1439
1440   for (debug_shndx = 1; debug_shndx < object->shnum(); ++debug_shndx)
1441     {
1442       // FIXME: do this more efficiently: section_name() isn't super-fast
1443       std::string name = object->section_name(debug_shndx);
1444       if (name == ".debug_line" || name == ".zdebug_line")
1445         {
1446           section_size_type buffer_size;
1447           bool is_new = false;
1448           this->buffer_ = object->decompressed_section_contents(debug_shndx,
1449                                                                 &buffer_size,
1450                                                                 &is_new);
1451           if (is_new)
1452             this->buffer_start_ = this->buffer_;
1453           this->buffer_end_ = this->buffer_ + buffer_size;
1454           break;
1455         }
1456     }
1457   if (this->buffer_ == NULL)
1458     return;
1459
1460   // Find the relocation section for ".debug_line".
1461   // We expect these for relobjs (.o's) but not dynobjs (.so's).
1462   unsigned int reloc_shndx = 0;
1463   for (unsigned int i = 0; i < object->shnum(); ++i)
1464     {
1465       unsigned int reloc_sh_type = object->section_type(i);
1466       if ((reloc_sh_type == elfcpp::SHT_REL
1467            || reloc_sh_type == elfcpp::SHT_RELA)
1468           && object->section_info(i) == debug_shndx)
1469         {
1470           reloc_shndx = i;
1471           this->track_relocs_type_ = reloc_sh_type;
1472           break;
1473         }
1474     }
1475
1476   // Finally, we need the symtab section to interpret the relocs.
1477   if (reloc_shndx != 0)
1478     {
1479       unsigned int symtab_shndx;
1480       for (symtab_shndx = 0; symtab_shndx < object->shnum(); ++symtab_shndx)
1481         if (object->section_type(symtab_shndx) == elfcpp::SHT_SYMTAB)
1482           {
1483             this->symtab_buffer_ = object->section_contents(
1484                 symtab_shndx, &this->symtab_buffer_size_, false);
1485             break;
1486           }
1487       if (this->symtab_buffer_ == NULL)
1488         return;
1489     }
1490
1491   this->reloc_mapper_ =
1492       new Sized_elf_reloc_mapper<size, big_endian>(object,
1493                                                    this->symtab_buffer_,
1494                                                    this->symtab_buffer_size_);
1495   if (!this->reloc_mapper_->initialize(reloc_shndx, this->track_relocs_type_))
1496     return;
1497
1498   // Now that we have successfully read all the data, parse the debug
1499   // info.
1500   this->data_valid_ = true;
1501   this->read_line_mappings(read_shndx);
1502 }
1503
1504 // Read the DWARF header.
1505
1506 template<int size, bool big_endian>
1507 const unsigned char*
1508 Sized_dwarf_line_info<size, big_endian>::read_header_prolog(
1509     const unsigned char* lineptr)
1510 {
1511   uint32_t initial_length = elfcpp::Swap_unaligned<32, big_endian>::readval(lineptr);
1512   lineptr += 4;
1513
1514   // In DWARF2/3, if the initial length is all 1 bits, then the offset
1515   // size is 8 and we need to read the next 8 bytes for the real length.
1516   if (initial_length == 0xffffffff)
1517     {
1518       header_.offset_size = 8;
1519       initial_length = elfcpp::Swap_unaligned<64, big_endian>::readval(lineptr);
1520       lineptr += 8;
1521     }
1522   else
1523     header_.offset_size = 4;
1524
1525   header_.total_length = initial_length;
1526
1527   gold_assert(lineptr + header_.total_length <= buffer_end_);
1528
1529   header_.version = elfcpp::Swap_unaligned<16, big_endian>::readval(lineptr);
1530   lineptr += 2;
1531
1532   if (header_.offset_size == 4)
1533     header_.prologue_length = elfcpp::Swap_unaligned<32, big_endian>::readval(lineptr);
1534   else
1535     header_.prologue_length = elfcpp::Swap_unaligned<64, big_endian>::readval(lineptr);
1536   lineptr += header_.offset_size;
1537
1538   header_.min_insn_length = *lineptr;
1539   lineptr += 1;
1540
1541   header_.default_is_stmt = *lineptr;
1542   lineptr += 1;
1543
1544   header_.line_base = *reinterpret_cast<const signed char*>(lineptr);
1545   lineptr += 1;
1546
1547   header_.line_range = *lineptr;
1548   lineptr += 1;
1549
1550   header_.opcode_base = *lineptr;
1551   lineptr += 1;
1552
1553   header_.std_opcode_lengths.resize(header_.opcode_base + 1);
1554   header_.std_opcode_lengths[0] = 0;
1555   for (int i = 1; i < header_.opcode_base; i++)
1556     {
1557       header_.std_opcode_lengths[i] = *lineptr;
1558       lineptr += 1;
1559     }
1560
1561   return lineptr;
1562 }
1563
1564 // The header for a debug_line section is mildly complicated, because
1565 // the line info is very tightly encoded.
1566
1567 template<int size, bool big_endian>
1568 const unsigned char*
1569 Sized_dwarf_line_info<size, big_endian>::read_header_tables(
1570     const unsigned char* lineptr)
1571 {
1572   ++this->current_header_index_;
1573
1574   // Create a new directories_ entry and a new files_ entry for our new
1575   // header.  We initialize each with a single empty element, because
1576   // dwarf indexes directory and filenames starting at 1.
1577   gold_assert(static_cast<int>(this->directories_.size())
1578               == this->current_header_index_);
1579   gold_assert(static_cast<int>(this->files_.size())
1580               == this->current_header_index_);
1581   this->directories_.push_back(std::vector<std::string>(1));
1582   this->files_.push_back(std::vector<std::pair<int, std::string> >(1));
1583
1584   // It is legal for the directory entry table to be empty.
1585   if (*lineptr)
1586     {
1587       int dirindex = 1;
1588       while (*lineptr)
1589         {
1590           const char* dirname = reinterpret_cast<const char*>(lineptr);
1591           gold_assert(dirindex
1592                       == static_cast<int>(this->directories_.back().size()));
1593           this->directories_.back().push_back(dirname);
1594           lineptr += this->directories_.back().back().size() + 1;
1595           dirindex++;
1596         }
1597     }
1598   lineptr++;
1599
1600   // It is also legal for the file entry table to be empty.
1601   if (*lineptr)
1602     {
1603       int fileindex = 1;
1604       size_t len;
1605       while (*lineptr)
1606         {
1607           const char* filename = reinterpret_cast<const char*>(lineptr);
1608           lineptr += strlen(filename) + 1;
1609
1610           uint64_t dirindex = read_unsigned_LEB_128(lineptr, &len);
1611           lineptr += len;
1612
1613           if (dirindex >= this->directories_.back().size())
1614             dirindex = 0;
1615           int dirindexi = static_cast<int>(dirindex);
1616
1617           read_unsigned_LEB_128(lineptr, &len);   // mod_time
1618           lineptr += len;
1619
1620           read_unsigned_LEB_128(lineptr, &len);   // filelength
1621           lineptr += len;
1622
1623           gold_assert(fileindex
1624                       == static_cast<int>(this->files_.back().size()));
1625           this->files_.back().push_back(std::make_pair(dirindexi, filename));
1626           fileindex++;
1627         }
1628     }
1629   lineptr++;
1630
1631   return lineptr;
1632 }
1633
1634 // Process a single opcode in the .debug.line structure.
1635
1636 template<int size, bool big_endian>
1637 bool
1638 Sized_dwarf_line_info<size, big_endian>::process_one_opcode(
1639     const unsigned char* start, struct LineStateMachine* lsm, size_t* len)
1640 {
1641   size_t oplen = 0;
1642   size_t templen;
1643   unsigned char opcode = *start;
1644   oplen++;
1645   start++;
1646
1647   // If the opcode is great than the opcode_base, it is a special
1648   // opcode. Most line programs consist mainly of special opcodes.
1649   if (opcode >= header_.opcode_base)
1650     {
1651       opcode -= header_.opcode_base;
1652       const int advance_address = ((opcode / header_.line_range)
1653                                    * header_.min_insn_length);
1654       lsm->address += advance_address;
1655
1656       const int advance_line = ((opcode % header_.line_range)
1657                                 + header_.line_base);
1658       lsm->line_num += advance_line;
1659       lsm->basic_block = true;
1660       *len = oplen;
1661       return true;
1662     }
1663
1664   // Otherwise, we have the regular opcodes
1665   switch (opcode)
1666     {
1667     case elfcpp::DW_LNS_copy:
1668       lsm->basic_block = false;
1669       *len = oplen;
1670       return true;
1671
1672     case elfcpp::DW_LNS_advance_pc:
1673       {
1674         const uint64_t advance_address
1675             = read_unsigned_LEB_128(start, &templen);
1676         oplen += templen;
1677         lsm->address += header_.min_insn_length * advance_address;
1678       }
1679       break;
1680
1681     case elfcpp::DW_LNS_advance_line:
1682       {
1683         const uint64_t advance_line = read_signed_LEB_128(start, &templen);
1684         oplen += templen;
1685         lsm->line_num += advance_line;
1686       }
1687       break;
1688
1689     case elfcpp::DW_LNS_set_file:
1690       {
1691         const uint64_t fileno = read_unsigned_LEB_128(start, &templen);
1692         oplen += templen;
1693         lsm->file_num = fileno;
1694       }
1695       break;
1696
1697     case elfcpp::DW_LNS_set_column:
1698       {
1699         const uint64_t colno = read_unsigned_LEB_128(start, &templen);
1700         oplen += templen;
1701         lsm->column_num = colno;
1702       }
1703       break;
1704
1705     case elfcpp::DW_LNS_negate_stmt:
1706       lsm->is_stmt = !lsm->is_stmt;
1707       break;
1708
1709     case elfcpp::DW_LNS_set_basic_block:
1710       lsm->basic_block = true;
1711       break;
1712
1713     case elfcpp::DW_LNS_fixed_advance_pc:
1714       {
1715         int advance_address;
1716         advance_address = elfcpp::Swap_unaligned<16, big_endian>::readval(start);
1717         oplen += 2;
1718         lsm->address += advance_address;
1719       }
1720       break;
1721
1722     case elfcpp::DW_LNS_const_add_pc:
1723       {
1724         const int advance_address = (header_.min_insn_length
1725                                      * ((255 - header_.opcode_base)
1726                                         / header_.line_range));
1727         lsm->address += advance_address;
1728       }
1729       break;
1730
1731     case elfcpp::DW_LNS_extended_op:
1732       {
1733         const uint64_t extended_op_len
1734             = read_unsigned_LEB_128(start, &templen);
1735         start += templen;
1736         oplen += templen + extended_op_len;
1737
1738         const unsigned char extended_op = *start;
1739         start++;
1740
1741         switch (extended_op)
1742           {
1743           case elfcpp::DW_LNE_end_sequence:
1744             // This means that the current byte is the one immediately
1745             // after a set of instructions.  Record the current line
1746             // for up to one less than the current address.
1747             lsm->line_num = -1;
1748             lsm->end_sequence = true;
1749             *len = oplen;
1750             return true;
1751
1752           case elfcpp::DW_LNE_set_address:
1753             {
1754               lsm->address =
1755                 elfcpp::Swap_unaligned<size, big_endian>::readval(start);
1756               typename Reloc_map::const_iterator it
1757                   = this->reloc_map_.find(start - this->buffer_);
1758               if (it != reloc_map_.end())
1759                 {
1760                   // If this is a SHT_RELA section, then ignore the
1761                   // section contents.  This assumes that this is a
1762                   // straight reloc which just uses the reloc addend.
1763                   // The reloc addend has already been included in the
1764                   // symbol value.
1765                   if (this->track_relocs_type_ == elfcpp::SHT_RELA)
1766                     lsm->address = 0;
1767                   // Add in the symbol value.
1768                   lsm->address += it->second.second;
1769                   lsm->shndx = it->second.first;
1770                 }
1771               else
1772                 {
1773                   // If we're a normal .o file, with relocs, every
1774                   // set_address should have an associated relocation.
1775                   if (this->input_is_relobj())
1776                     this->data_valid_ = false;
1777                 }
1778               break;
1779             }
1780           case elfcpp::DW_LNE_define_file:
1781             {
1782               const char* filename  = reinterpret_cast<const char*>(start);
1783               templen = strlen(filename) + 1;
1784               start += templen;
1785
1786               uint64_t dirindex = read_unsigned_LEB_128(start, &templen);
1787
1788               if (dirindex >= this->directories_.back().size())
1789                 dirindex = 0;
1790               int dirindexi = static_cast<int>(dirindex);
1791
1792               // This opcode takes two additional ULEB128 parameters
1793               // (mod_time and filelength), but we don't use those
1794               // values.  Because OPLEN already tells us how far to
1795               // skip to the next opcode, we don't need to read
1796               // them at all.
1797
1798               this->files_.back().push_back(std::make_pair(dirindexi,
1799                                                            filename));
1800             }
1801             break;
1802           }
1803       }
1804       break;
1805
1806     default:
1807       {
1808         // Ignore unknown opcode  silently
1809         for (int i = 0; i < header_.std_opcode_lengths[opcode]; i++)
1810           {
1811             size_t templen;
1812             read_unsigned_LEB_128(start, &templen);
1813             start += templen;
1814             oplen += templen;
1815           }
1816       }
1817       break;
1818   }
1819   *len = oplen;
1820   return false;
1821 }
1822
1823 // Read the debug information at LINEPTR and store it in the line
1824 // number map.
1825
1826 template<int size, bool big_endian>
1827 unsigned const char*
1828 Sized_dwarf_line_info<size, big_endian>::read_lines(unsigned const char* lineptr,
1829                                                     unsigned int shndx)
1830 {
1831   struct LineStateMachine lsm;
1832
1833   // LENGTHSTART is the place the length field is based on.  It is the
1834   // point in the header after the initial length field.
1835   const unsigned char* lengthstart = buffer_;
1836
1837   // In 64 bit dwarf, the initial length is 12 bytes, because of the
1838   // 0xffffffff at the start.
1839   if (header_.offset_size == 8)
1840     lengthstart += 12;
1841   else
1842     lengthstart += 4;
1843
1844   while (lineptr < lengthstart + header_.total_length)
1845     {
1846       ResetLineStateMachine(&lsm, header_.default_is_stmt);
1847       while (!lsm.end_sequence)
1848         {
1849           size_t oplength;
1850           bool add_line = this->process_one_opcode(lineptr, &lsm, &oplength);
1851           if (add_line
1852               && (shndx == -1U || lsm.shndx == -1U || shndx == lsm.shndx))
1853             {
1854               Offset_to_lineno_entry entry
1855                   = { static_cast<off_t>(lsm.address),
1856                       this->current_header_index_,
1857                       static_cast<unsigned int>(lsm.file_num),
1858                       true, lsm.line_num };
1859               std::vector<Offset_to_lineno_entry>&
1860                 map(this->line_number_map_[lsm.shndx]);
1861               // If we see two consecutive entries with the same
1862               // offset and a real line number, then mark the first
1863               // one as non-canonical.
1864               if (!map.empty()
1865                   && (map.back().offset == static_cast<off_t>(lsm.address))
1866                   && lsm.line_num != -1
1867                   && map.back().line_num != -1)
1868                 map.back().last_line_for_offset = false;
1869               map.push_back(entry);
1870             }
1871           lineptr += oplength;
1872         }
1873     }
1874
1875   return lengthstart + header_.total_length;
1876 }
1877
1878 // Read the relocations into a Reloc_map.
1879
1880 template<int size, bool big_endian>
1881 void
1882 Sized_dwarf_line_info<size, big_endian>::read_relocs()
1883 {
1884   if (this->symtab_buffer_ == NULL)
1885     return;
1886
1887   off_t value;
1888   off_t reloc_offset;
1889   while ((reloc_offset = this->reloc_mapper_->next_offset()) != -1)
1890     {
1891       const unsigned int shndx =
1892           this->reloc_mapper_->get_reloc_target(reloc_offset, &value);
1893
1894       // There is no reason to record non-ordinary section indexes, or
1895       // SHN_UNDEF, because they will never match the real section.
1896       if (shndx != 0)
1897         this->reloc_map_[reloc_offset] = std::make_pair(shndx, value);
1898
1899       this->reloc_mapper_->advance(reloc_offset + 1);
1900     }
1901 }
1902
1903 // Read the line number info.
1904
1905 template<int size, bool big_endian>
1906 void
1907 Sized_dwarf_line_info<size, big_endian>::read_line_mappings(unsigned int shndx)
1908 {
1909   gold_assert(this->data_valid_ == true);
1910
1911   this->read_relocs();
1912   while (this->buffer_ < this->buffer_end_)
1913     {
1914       const unsigned char* lineptr = this->buffer_;
1915       lineptr = this->read_header_prolog(lineptr);
1916       lineptr = this->read_header_tables(lineptr);
1917       lineptr = this->read_lines(lineptr, shndx);
1918       this->buffer_ = lineptr;
1919     }
1920
1921   // Sort the lines numbers, so addr2line can use binary search.
1922   for (typename Lineno_map::iterator it = line_number_map_.begin();
1923        it != line_number_map_.end();
1924        ++it)
1925     // Each vector needs to be sorted by offset.
1926     std::sort(it->second.begin(), it->second.end());
1927 }
1928
1929 // Some processing depends on whether the input is a .o file or not.
1930 // For instance, .o files have relocs, and have .debug_lines
1931 // information on a per section basis.  .so files, on the other hand,
1932 // lack relocs, and offsets are unique, so we can ignore the section
1933 // information.
1934
1935 template<int size, bool big_endian>
1936 bool
1937 Sized_dwarf_line_info<size, big_endian>::input_is_relobj()
1938 {
1939   // Only .o files have relocs and the symtab buffer that goes with them.
1940   return this->symtab_buffer_ != NULL;
1941 }
1942
1943 // Given an Offset_to_lineno_entry vector, and an offset, figure out
1944 // if the offset points into a function according to the vector (see
1945 // comments below for the algorithm).  If it does, return an iterator
1946 // into the vector that points to the line-number that contains that
1947 // offset.  If not, it returns vector::end().
1948
1949 static std::vector<Offset_to_lineno_entry>::const_iterator
1950 offset_to_iterator(const std::vector<Offset_to_lineno_entry>* offsets,
1951                    off_t offset)
1952 {
1953   const Offset_to_lineno_entry lookup_key = { offset, 0, 0, true, 0 };
1954
1955   // lower_bound() returns the smallest offset which is >= lookup_key.
1956   // If no offset in offsets is >= lookup_key, returns end().
1957   std::vector<Offset_to_lineno_entry>::const_iterator it
1958       = std::lower_bound(offsets->begin(), offsets->end(), lookup_key);
1959
1960   // This code is easiest to understand with a concrete example.
1961   // Here's a possible offsets array:
1962   // {{offset = 3211, header_num = 0, file_num = 1, last, line_num = 16},  // 0
1963   //  {offset = 3224, header_num = 0, file_num = 1, last, line_num = 20},  // 1
1964   //  {offset = 3226, header_num = 0, file_num = 1, last, line_num = 22},  // 2
1965   //  {offset = 3231, header_num = 0, file_num = 1, last, line_num = 25},  // 3
1966   //  {offset = 3232, header_num = 0, file_num = 1, last, line_num = -1},  // 4
1967   //  {offset = 3232, header_num = 0, file_num = 1, last, line_num = 65},  // 5
1968   //  {offset = 3235, header_num = 0, file_num = 1, last, line_num = 66},  // 6
1969   //  {offset = 3236, header_num = 0, file_num = 1, last, line_num = -1},  // 7
1970   //  {offset = 5764, header_num = 0, file_num = 1, last, line_num = 48},  // 8
1971   //  {offset = 5764, header_num = 0, file_num = 1,!last, line_num = 47},  // 9
1972   //  {offset = 5765, header_num = 0, file_num = 1, last, line_num = 49},  // 10
1973   //  {offset = 5767, header_num = 0, file_num = 1, last, line_num = 50},  // 11
1974   //  {offset = 5768, header_num = 0, file_num = 1, last, line_num = 51},  // 12
1975   //  {offset = 5773, header_num = 0, file_num = 1, last, line_num = -1},  // 13
1976   //  {offset = 5787, header_num = 1, file_num = 1, last, line_num = 19},  // 14
1977   //  {offset = 5790, header_num = 1, file_num = 1, last, line_num = 20},  // 15
1978   //  {offset = 5793, header_num = 1, file_num = 1, last, line_num = 67},  // 16
1979   //  {offset = 5793, header_num = 1, file_num = 1, last, line_num = -1},  // 17
1980   //  {offset = 5793, header_num = 1, file_num = 1,!last, line_num = 66},  // 18
1981   //  {offset = 5795, header_num = 1, file_num = 1, last, line_num = 68},  // 19
1982   //  {offset = 5798, header_num = 1, file_num = 1, last, line_num = -1},  // 20
1983   // The entries with line_num == -1 mark the end of a function: the
1984   // associated offset is one past the last instruction in the
1985   // function.  This can correspond to the beginning of the next
1986   // function (as is true for offset 3232); alternately, there can be
1987   // a gap between the end of one function and the start of the next
1988   // (as is true for some others, most obviously from 3236->5764).
1989   //
1990   // Case 1: lookup_key has offset == 10.  lower_bound returns
1991   //         offsets[0].  Since it's not an exact match and we're
1992   //         at the beginning of offsets, we return end() (invalid).
1993   // Case 2: lookup_key has offset 10000.  lower_bound returns
1994   //         offset[21] (end()).  We return end() (invalid).
1995   // Case 3: lookup_key has offset == 3211.  lower_bound matches
1996   //         offsets[0] exactly, and that's the entry we return.
1997   // Case 4: lookup_key has offset == 3232.  lower_bound returns
1998   //         offsets[4].  That's an exact match, but indicates
1999   //         end-of-function.  We check if offsets[5] is also an
2000   //         exact match but not end-of-function.  It is, so we
2001   //         return offsets[5].
2002   // Case 5: lookup_key has offset == 3214.  lower_bound returns
2003   //         offsets[1].  Since it's not an exact match, we back
2004   //         up to the offset that's < lookup_key, offsets[0].
2005   //         We note offsets[0] is a valid entry (not end-of-function),
2006   //         so that's the entry we return.
2007   // Case 6: lookup_key has offset == 4000.  lower_bound returns
2008   //         offsets[8].  Since it's not an exact match, we back
2009   //         up to offsets[7].  Since offsets[7] indicates
2010   //         end-of-function, we know lookup_key is between
2011   //         functions, so we return end() (not a valid offset).
2012   // Case 7: lookup_key has offset == 5794.  lower_bound returns
2013   //         offsets[19].  Since it's not an exact match, we back
2014   //         up to offsets[16].  Note we back up to the *first*
2015   //         entry with offset 5793, not just offsets[19-1].
2016   //         We note offsets[16] is a valid entry, so we return it.
2017   //         If offsets[16] had had line_num == -1, we would have
2018   //         checked offsets[17].  The reason for this is that
2019   //         16 and 17 can be in an arbitrary order, since we sort
2020   //         only by offset and last_line_for_offset.  (Note it
2021   //         doesn't help to use line_number as a tertiary sort key,
2022   //         since sometimes we want the -1 to be first and sometimes
2023   //         we want it to be last.)
2024
2025   // This deals with cases (1) and (2).
2026   if ((it == offsets->begin() && offset < it->offset)
2027       || it == offsets->end())
2028     return offsets->end();
2029
2030   // This deals with cases (3) and (4).
2031   if (offset == it->offset)
2032     {
2033       while (it != offsets->end()
2034              && it->offset == offset
2035              && it->line_num == -1)
2036         ++it;
2037       if (it == offsets->end() || it->offset != offset)
2038         return offsets->end();
2039       else
2040         return it;
2041     }
2042
2043   // This handles the first part of case (7) -- we back up to the
2044   // *first* entry that has the offset that's behind us.
2045   gold_assert(it != offsets->begin());
2046   std::vector<Offset_to_lineno_entry>::const_iterator range_end = it;
2047   --it;
2048   const off_t range_value = it->offset;
2049   while (it != offsets->begin() && (it-1)->offset == range_value)
2050     --it;
2051
2052   // This handles cases (5), (6), and (7): if any entry in the
2053   // equal_range [it, range_end) has a line_num != -1, it's a valid
2054   // match.  If not, we're not in a function.  The line number we saw
2055   // last for an offset will be sorted first, so it'll get returned if
2056   // it's present.
2057   for (; it != range_end; ++it)
2058     if (it->line_num != -1)
2059       return it;
2060   return offsets->end();
2061 }
2062
2063 // Returns the canonical filename:lineno for the address passed in.
2064 // If other_lines is not NULL, appends the non-canonical lines
2065 // assigned to the same address.
2066
2067 template<int size, bool big_endian>
2068 std::string
2069 Sized_dwarf_line_info<size, big_endian>::do_addr2line(
2070     unsigned int shndx,
2071     off_t offset,
2072     std::vector<std::string>* other_lines)
2073 {
2074   if (this->data_valid_ == false)
2075     return "";
2076
2077   const std::vector<Offset_to_lineno_entry>* offsets;
2078   // If we do not have reloc information, then our input is a .so or
2079   // some similar data structure where all the information is held in
2080   // the offset.  In that case, we ignore the input shndx.
2081   if (this->input_is_relobj())
2082     offsets = &this->line_number_map_[shndx];
2083   else
2084     offsets = &this->line_number_map_[-1U];
2085   if (offsets->empty())
2086     return "";
2087
2088   typename std::vector<Offset_to_lineno_entry>::const_iterator it
2089       = offset_to_iterator(offsets, offset);
2090   if (it == offsets->end())
2091     return "";
2092
2093   std::string result = this->format_file_lineno(*it);
2094   if (other_lines != NULL)
2095     for (++it; it != offsets->end() && it->offset == offset; ++it)
2096       {
2097         if (it->line_num == -1)
2098           continue;  // The end of a previous function.
2099         other_lines->push_back(this->format_file_lineno(*it));
2100       }
2101   return result;
2102 }
2103
2104 // Convert the file_num + line_num into a string.
2105
2106 template<int size, bool big_endian>
2107 std::string
2108 Sized_dwarf_line_info<size, big_endian>::format_file_lineno(
2109     const Offset_to_lineno_entry& loc) const
2110 {
2111   std::string ret;
2112
2113   gold_assert(loc.header_num < static_cast<int>(this->files_.size()));
2114   gold_assert(loc.file_num
2115               < static_cast<unsigned int>(this->files_[loc.header_num].size()));
2116   const std::pair<int, std::string>& filename_pair
2117       = this->files_[loc.header_num][loc.file_num];
2118   const std::string& filename = filename_pair.second;
2119
2120   gold_assert(loc.header_num < static_cast<int>(this->directories_.size()));
2121   gold_assert(filename_pair.first
2122               < static_cast<int>(this->directories_[loc.header_num].size()));
2123   const std::string& dirname
2124       = this->directories_[loc.header_num][filename_pair.first];
2125
2126   if (!dirname.empty())
2127     {
2128       ret += dirname;
2129       ret += "/";
2130     }
2131   ret += filename;
2132   if (ret.empty())
2133     ret = "(unknown)";
2134
2135   char buffer[64];   // enough to hold a line number
2136   snprintf(buffer, sizeof(buffer), "%d", loc.line_num);
2137   ret += ":";
2138   ret += buffer;
2139
2140   return ret;
2141 }
2142
2143 // Dwarf_line_info routines.
2144
2145 static unsigned int next_generation_count = 0;
2146
2147 struct Addr2line_cache_entry
2148 {
2149   Object* object;
2150   unsigned int shndx;
2151   Dwarf_line_info* dwarf_line_info;
2152   unsigned int generation_count;
2153   unsigned int access_count;
2154
2155   Addr2line_cache_entry(Object* o, unsigned int s, Dwarf_line_info* d)
2156       : object(o), shndx(s), dwarf_line_info(d),
2157         generation_count(next_generation_count), access_count(0)
2158   {
2159     if (next_generation_count < (1U << 31))
2160       ++next_generation_count;
2161   }
2162 };
2163 // We expect this cache to be small, so don't bother with a hashtable
2164 // or priority queue or anything: just use a simple vector.
2165 static std::vector<Addr2line_cache_entry> addr2line_cache;
2166
2167 std::string
2168 Dwarf_line_info::one_addr2line(Object* object,
2169                                unsigned int shndx, off_t offset,
2170                                size_t cache_size,
2171                                std::vector<std::string>* other_lines)
2172 {
2173   Dwarf_line_info* lineinfo = NULL;
2174   std::vector<Addr2line_cache_entry>::iterator it;
2175
2176   // First, check the cache.  If we hit, update the counts.
2177   for (it = addr2line_cache.begin(); it != addr2line_cache.end(); ++it)
2178     {
2179       if (it->object == object && it->shndx == shndx)
2180         {
2181           lineinfo = it->dwarf_line_info;
2182           it->generation_count = next_generation_count;
2183           // We cap generation_count at 2^31 -1 to avoid overflow.
2184           if (next_generation_count < (1U << 31))
2185             ++next_generation_count;
2186           // We cap access_count at 31 so 2^access_count doesn't overflow
2187           if (it->access_count < 31)
2188             ++it->access_count;
2189           break;
2190         }
2191     }
2192
2193   // If we don't hit the cache, create a new object and insert into the
2194   // cache.
2195   if (lineinfo == NULL)
2196   {
2197     switch (parameters->size_and_endianness())
2198       {
2199 #ifdef HAVE_TARGET_32_LITTLE
2200         case Parameters::TARGET_32_LITTLE:
2201           lineinfo = new Sized_dwarf_line_info<32, false>(object, shndx); break;
2202 #endif
2203 #ifdef HAVE_TARGET_32_BIG
2204         case Parameters::TARGET_32_BIG:
2205           lineinfo = new Sized_dwarf_line_info<32, true>(object, shndx); break;
2206 #endif
2207 #ifdef HAVE_TARGET_64_LITTLE
2208         case Parameters::TARGET_64_LITTLE:
2209           lineinfo = new Sized_dwarf_line_info<64, false>(object, shndx); break;
2210 #endif
2211 #ifdef HAVE_TARGET_64_BIG
2212         case Parameters::TARGET_64_BIG:
2213           lineinfo = new Sized_dwarf_line_info<64, true>(object, shndx); break;
2214 #endif
2215         default:
2216           gold_unreachable();
2217       }
2218     addr2line_cache.push_back(Addr2line_cache_entry(object, shndx, lineinfo));
2219   }
2220
2221   // Now that we have our object, figure out the answer
2222   std::string retval = lineinfo->addr2line(shndx, offset, other_lines);
2223
2224   // Finally, if our cache has grown too big, delete old objects.  We
2225   // assume the common (probably only) case is deleting only one object.
2226   // We use a pretty simple scheme to evict: function of LRU and MFU.
2227   while (addr2line_cache.size() > cache_size)
2228     {
2229       unsigned int lowest_score = ~0U;
2230       std::vector<Addr2line_cache_entry>::iterator lowest
2231           = addr2line_cache.end();
2232       for (it = addr2line_cache.begin(); it != addr2line_cache.end(); ++it)
2233         {
2234           const unsigned int score = (it->generation_count
2235                                       + (1U << it->access_count));
2236           if (score < lowest_score)
2237             {
2238               lowest_score = score;
2239               lowest = it;
2240             }
2241         }
2242       if (lowest != addr2line_cache.end())
2243         {
2244           delete lowest->dwarf_line_info;
2245           addr2line_cache.erase(lowest);
2246         }
2247     }
2248
2249   return retval;
2250 }
2251
2252 void
2253 Dwarf_line_info::clear_addr2line_cache()
2254 {
2255   for (std::vector<Addr2line_cache_entry>::iterator it = addr2line_cache.begin();
2256        it != addr2line_cache.end();
2257        ++it)
2258     delete it->dwarf_line_info;
2259   addr2line_cache.clear();
2260 }
2261
2262 #ifdef HAVE_TARGET_32_LITTLE
2263 template
2264 class Sized_dwarf_line_info<32, false>;
2265 #endif
2266
2267 #ifdef HAVE_TARGET_32_BIG
2268 template
2269 class Sized_dwarf_line_info<32, true>;
2270 #endif
2271
2272 #ifdef HAVE_TARGET_64_LITTLE
2273 template
2274 class Sized_dwarf_line_info<64, false>;
2275 #endif
2276
2277 #ifdef HAVE_TARGET_64_BIG
2278 template
2279 class Sized_dwarf_line_info<64, true>;
2280 #endif
2281
2282 } // End namespace gold.