gold/merge.cc

   1 // merge.cc -- handle section merging for gold
   2
   3 // Copyright (C) 2006-2015 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 <cstdlib>
  26 #include <algorithm>
  27
  28 #include "merge.h"
  29 #include "compressed_output.h"
  30
  31 namespace gold
  32 {
  33
  34 // Class Object_merge_map.
  35
  36 // Destructor.
  37
  38 Object_merge_map::~Object_merge_map()
  39 {
  40   for (Section_merge_maps::iterator p = this->section_merge_maps_.begin();
  41        p != this->section_merge_maps_.end();
  42        ++p)
  43     delete p->second;
  44 }
  45
  46 // Get the Input_merge_map to use for an input section, or NULL.
  47
  48 const Object_merge_map::Input_merge_map*
  49 Object_merge_map::get_input_merge_map(unsigned int shndx) const
  50 {
  51   gold_assert(shndx != -1U);
  52   if (shndx == this->first_shnum_)
  53     return &this->first_map_;
  54   if (shndx == this->second_shnum_)
  55     return &this->second_map_;
  56   Section_merge_maps::const_iterator p = this->section_merge_maps_.find(shndx);
  57   if (p != this->section_merge_maps_.end())
  58     return p->second;
  59   return NULL;
  60 }
  61
  62 // Get or create the Input_merge_map to use for an input section.
  63
  64 Object_merge_map::Input_merge_map*
  65 Object_merge_map::get_or_make_input_merge_map(
  66     const Output_section_data* output_data, unsigned int shndx) {
  67   Input_merge_map* map = this->get_input_merge_map(shndx);
  68   if (map != NULL)
  69     {
  70       // For a given input section in a given object, every mapping
  71       // must be done with the same Merge_map.
  72       gold_assert(map->output_data == output_data);
  73       return map;
  74     }
  75
  76   // We need to create a new entry.
  77   if (this->first_shnum_ == -1U)
  78     {
  79       this->first_shnum_ = shndx;
  80       this->first_map_.output_data = output_data;
  81       return &this->first_map_;
  82     }
  83   if (this->second_shnum_ == -1U)
  84     {
  85       this->second_shnum_ = shndx;
  86       this->second_map_.output_data = output_data;
  87       return &this->second_map_;
  88     }
  89
  90   Input_merge_map* new_map = new Input_merge_map;
  91   new_map->output_data = output_data;
  92   this->section_merge_maps_[shndx] = new_map;
  93   return new_map;
  94 }
  95
  96 // Add a mapping.
  97
  98 void
  99 Object_merge_map::add_mapping(const Output_section_data* output_data,
 100                               unsigned int shndx,
 101                               section_offset_type input_offset,
 102                               section_size_type length,
 103                               section_offset_type output_offset)
 104 {
 105   Input_merge_map* map = this->get_or_make_input_merge_map(output_data, shndx);
 106
 107   // Try to merge the new entry in the last one we saw.
 108   if (!map->entries.empty())
 109     {
 110       Input_merge_entry& entry(map->entries.back());
 111
 112       // Use section_size_type to avoid signed/unsigned warnings.
 113       section_size_type input_offset_u = input_offset;
 114       section_size_type output_offset_u = output_offset;
 115
 116       // If this entry is not in order, we need to sort the vector
 117       // before looking anything up.
 118       if (input_offset_u < entry.input_offset + entry.length)
 119         {
 120           gold_assert(input_offset < entry.input_offset);
 121           gold_assert(input_offset_u + length
 122                       <= static_cast<section_size_type>(entry.input_offset));
 123           map->sorted = false;
 124         }
 125       else if (entry.input_offset + entry.length == input_offset_u
 126                && (output_offset == -1
 127                    ? entry.output_offset == -1
 128                    : entry.output_offset + entry.length == output_offset_u))
 129         {
 130           entry.length += length;
 131           return;
 132         }
 133     }
 134
 135   Input_merge_entry entry;
 136   entry.input_offset = input_offset;
 137   entry.length = length;
 138   entry.output_offset = output_offset;
 139   map->entries.push_back(entry);
 140 }
 141
 142 // Get the output offset for an input address.
 143
 144 bool
 145 Object_merge_map::get_output_offset(unsigned int shndx,
 146                                     section_offset_type input_offset,
 147                                     section_offset_type* output_offset)
 148 {
 149   Input_merge_map* map = this->get_input_merge_map(shndx);
 150   if (map == NULL)
 151     return false;
 152
 153   if (!map->sorted)
 154     {
 155       std::sort(map->entries.begin(), map->entries.end(),
 156                 Input_merge_compare());
 157       map->sorted = true;
 158     }
 159
 160   Input_merge_entry entry;
 161   entry.input_offset = input_offset;
 162   std::vector<Input_merge_entry>::const_iterator p =
 163     std::upper_bound(map->entries.begin(), map->entries.end(),
 164                      entry, Input_merge_compare());
 165   if (p == map->entries.begin())
 166     return false;
 167   --p;
 168   gold_assert(p->input_offset <= input_offset);
 169
 170   if (input_offset - p->input_offset
 171       >= static_cast<section_offset_type>(p->length))
 172     return false;
 173
 174   *output_offset = p->output_offset;
 175   if (*output_offset != -1)
 176     *output_offset += (input_offset - p->input_offset);
 177   return true;
 178 }
 179
 180 // Return whether this is the merge map for section SHNDX.
 181
 182 const Output_section_data*
 183 Object_merge_map::find_merge_section(unsigned int shndx) const {
 184   const Object_merge_map::Input_merge_map* map =
 185     this->get_input_merge_map(shndx);
 186   if (map == NULL)
 187     return NULL;
 188   return map->output_data;
 189 }
 190
 191 // Initialize a mapping from input offsets to output addresses.
 192
 193 template<int size>
 194 void
 195 Object_merge_map::initialize_input_to_output_map(
 196     unsigned int shndx,
 197     typename elfcpp::Elf_types<size>::Elf_Addr starting_address,
 198     Unordered_map<section_offset_type,
 199                   typename elfcpp::Elf_types<size>::Elf_Addr>* initialize_map)
 200 {
 201   Input_merge_map* map = this->get_input_merge_map(shndx);
 202   gold_assert(map != NULL);
 203
 204   gold_assert(initialize_map->empty());
 205   // We know how many entries we are going to add.
 206   // reserve_unordered_map takes an expected count of buckets, not a
 207   // count of elements, so double it to try to reduce collisions.
 208   reserve_unordered_map(initialize_map, map->entries.size() * 2);
 209
 210   for (Input_merge_map::Entries::const_iterator p = map->entries.begin();
 211        p != map->entries.end();
 212        ++p)
 213     {
 214       section_offset_type output_offset = p->output_offset;
 215       if (output_offset != -1)
 216         output_offset += starting_address;
 217       else
 218         {
 219           // If we see a relocation against an address we have chosen
 220           // to discard, we relocate to zero.  FIXME: We could also
 221           // issue a warning in this case; that would require
 222           // reporting this somehow and checking it in the routines in
 223           // reloc.h.
 224           output_offset = 0;
 225         }
 226       initialize_map->insert(std::make_pair(p->input_offset, output_offset));
 227     }
 228 }
 229
 230 // Class Output_merge_base.
 231
 232 // Return the output offset for an input offset.  The input address is
 233 // at offset OFFSET in section SHNDX in OBJECT.  If we know the
 234 // offset, set *POUTPUT and return true.  Otherwise return false.
 235
 236 bool
 237 Output_merge_base::do_output_offset(const Relobj* object,
 238                                     unsigned int shndx,
 239                                     section_offset_type offset,
 240                                     section_offset_type* poutput) const
 241 {
 242   return object->merge_output_offset(shndx, offset, poutput);
 243 }
 244
 245 // Record a merged input section for script processing.
 246
 247 void
 248 Output_merge_base::record_input_section(Relobj* relobj, unsigned int shndx)
 249 {
 250   gold_assert(this->keeps_input_sections_ && relobj != NULL);
 251   // If this is the first input section, record it.  We need do this because
 252   // this->input_sections_ is unordered.
 253   if (this->first_relobj_ == NULL)
 254     {
 255       this->first_relobj_ = relobj;
 256       this->first_shndx_ = shndx;
 257     }
 258
 259   std::pair<Input_sections::iterator, bool> result =
 260     this->input_sections_.insert(Section_id(relobj, shndx));
 261   // We should insert a merge section once only.
 262   gold_assert(result.second);
 263 }
 264
 265 // Class Output_merge_data.
 266
 267 // Compute the hash code for a fixed-size constant.
 268
 269 size_t
 270 Output_merge_data::Merge_data_hash::operator()(Merge_data_key k) const
 271 {
 272   const unsigned char* p = this->pomd_->constant(k);
 273   section_size_type entsize =
 274     convert_to_section_size_type(this->pomd_->entsize());
 275
 276   // Fowler/Noll/Vo (FNV) hash (type FNV-1a).
 277   if (sizeof(size_t) == 8)
 278     {
 279       size_t result = static_cast<size_t>(14695981039346656037ULL);
 280       for (section_size_type i = 0; i < entsize; ++i)
 281         {
 282           result &= (size_t) *p++;
 283           result *= 1099511628211ULL;
 284         }
 285       return result;
 286     }
 287   else
 288     {
 289       size_t result = 2166136261UL;
 290       for (section_size_type i = 0; i < entsize; ++i)
 291         {
 292           result ^= (size_t) *p++;
 293           result *= 16777619UL;
 294         }
 295       return result;
 296     }
 297 }
 298
 299 // Return whether one hash table key equals another.
 300
 301 bool
 302 Output_merge_data::Merge_data_eq::operator()(Merge_data_key k1,
 303                                              Merge_data_key k2) const
 304 {
 305   const unsigned char* p1 = this->pomd_->constant(k1);
 306   const unsigned char* p2 = this->pomd_->constant(k2);
 307   return memcmp(p1, p2, this->pomd_->entsize()) == 0;
 308 }
 309
 310 // Add a constant to the end of the section contents.
 311
 312 void
 313 Output_merge_data::add_constant(const unsigned char* p)
 314 {
 315   section_size_type entsize = convert_to_section_size_type(this->entsize());
 316   section_size_type addralign =
 317     convert_to_section_size_type(this->addralign());
 318   section_size_type addsize = std::max(entsize, addralign);
 319   if (this->len_ + addsize > this->alc_)
 320     {
 321       if (this->alc_ == 0)
 322         this->alc_ = 128 * addsize;
 323       else
 324         this->alc_ *= 2;
 325       this->p_ = static_cast<unsigned char*>(realloc(this->p_, this->alc_));
 326       if (this->p_ == NULL)
 327         gold_nomem();
 328     }
 329
 330   memcpy(this->p_ + this->len_, p, entsize);
 331   if (addsize > entsize)
 332     memset(this->p_ + this->len_ + entsize, 0, addsize - entsize);
 333   this->len_ += addsize;
 334 }
 335
 336 // Add the input section SHNDX in OBJECT to a merged output section
 337 // which holds fixed length constants.  Return whether we were able to
 338 // handle the section; if not, it will be linked as usual without
 339 // constant merging.
 340
 341 bool
 342 Output_merge_data::do_add_input_section(Relobj* object, unsigned int shndx)
 343 {
 344   section_size_type len;
 345   bool is_new;
 346   const unsigned char* p = object->decompressed_section_contents(shndx, &len,
 347                                                                  &is_new);
 348
 349   section_size_type entsize = convert_to_section_size_type(this->entsize());
 350
 351   if (len % entsize != 0)
 352     {
 353       if (is_new)
 354         delete[] p;
 355       return false;
 356     }
 357
 358   this->input_count_ += len / entsize;
 359
 360   for (section_size_type i = 0; i < len; i += entsize, p += entsize)
 361     {
 362       // Add the constant to the section contents.  If we find that it
 363       // is already in the hash table, we will remove it again.
 364       Merge_data_key k = this->len_;
 365       this->add_constant(p);
 366
 367       std::pair<Merge_data_hashtable::iterator, bool> ins =
 368         this->hashtable_.insert(k);
 369
 370       if (!ins.second)
 371         {
 372           // Key was already present.  Remove the copy we just added.
 373           this->len_ -= entsize;
 374           k = *ins.first;
 375         }
 376
 377       // Record the offset of this constant in the output section.
 378       object->add_merge_mapping(this, shndx, i, entsize, k);
 379     }
 380
 381   // For script processing, we keep the input sections.
 382   if (this->keeps_input_sections())
 383     record_input_section(object, shndx);
 384
 385   if (is_new)
 386     delete[] p;
 387
 388   return true;
 389 }
 390
 391 // Set the final data size in a merged output section with fixed size
 392 // constants.
 393
 394 void
 395 Output_merge_data::set_final_data_size()
 396 {
 397   // Release the memory we don't need.
 398   this->p_ = static_cast<unsigned char*>(realloc(this->p_, this->len_));
 399   // An Output_merge_data object may be empty and realloc is allowed
 400   // to return a NULL pointer in this case.  An Output_merge_data is empty
 401   // if all its input sections have sizes that are not multiples of entsize.
 402   gold_assert(this->p_ != NULL || this->len_ == 0);
 403   this->set_data_size(this->len_);
 404 }
 405
 406 // Write the data of a merged output section with fixed size constants
 407 // to the file.
 408
 409 void
 410 Output_merge_data::do_write(Output_file* of)
 411 {
 412   of->write(this->offset(), this->p_, this->len_);
 413 }
 414
 415 // Write the data to a buffer.
 416
 417 void
 418 Output_merge_data::do_write_to_buffer(unsigned char* buffer)
 419 {
 420   memcpy(buffer, this->p_, this->len_);
 421 }
 422
 423 // Print merge stats to stderr.
 424
 425 void
 426 Output_merge_data::do_print_merge_stats(const char* section_name)
 427 {
 428   fprintf(stderr,
 429           _("%s: %s merged constants size: %lu; input: %zu; output: %zu\n"),
 430           program_name, section_name,
 431           static_cast<unsigned long>(this->entsize()),
 432           this->input_count_, this->hashtable_.size());
 433 }
 434
 435 // Class Output_merge_string.
 436
 437 // Add an input section to a merged string section.
 438
 439 template<typename Char_type>
 440 bool
 441 Output_merge_string<Char_type>::do_add_input_section(Relobj* object,
 442                                                      unsigned int shndx)
 443 {
 444   section_size_type sec_len;
 445   bool is_new;
 446   const unsigned char* pdata = object->decompressed_section_contents(shndx,
 447                                                                      &sec_len,
 448                                                                      &is_new);
 449
 450   const Char_type* p = reinterpret_cast<const Char_type*>(pdata);
 451   const Char_type* pend = p + sec_len / sizeof(Char_type);
 452   const Char_type* pend0 = pend;
 453
 454   if (sec_len % sizeof(Char_type) != 0)
 455     {
 456       object->error(_("mergeable string section length not multiple of "
 457                       "character size"));
 458       if (is_new)
 459         delete[] pdata;
 460       return false;
 461     }
 462
 463   if (pend[-1] != 0)
 464     {
 465       gold_warning(_("%s: last entry in mergeable string section '%s' "
 466                      "not null terminated"),
 467                    object->name().c_str(),
 468                    object->section_name(shndx).c_str());
 469       // Find the end of the last NULL-terminated string in the buffer.
 470       while (pend0 > p && pend0[-1] != 0)
 471         --pend0;
 472     }
 473
 474   Merged_strings_list* merged_strings_list =
 475       new Merged_strings_list(object, shndx);
 476   this->merged_strings_lists_.push_back(merged_strings_list);
 477   Merged_strings& merged_strings = merged_strings_list->merged_strings;
 478
 479   // Count the number of non-null strings in the section and size the list.
 480   size_t count = 0;
 481   const Char_type* pt = p;
 482   while (pt < pend0)
 483     {
 484       size_t len = string_length(pt);
 485       if (len != 0)
 486         ++count;
 487       pt += len + 1;
 488     }
 489   if (pend0 < pend)
 490     ++count;
 491   merged_strings.reserve(count + 1);
 492
 493   // The index I is in bytes, not characters.
 494   section_size_type i = 0;
 495
 496   // We assume here that the beginning of the section is correctly
 497   // aligned, so each string within the section must retain the same
 498   // modulo.
 499   uintptr_t init_align_modulo = (reinterpret_cast<uintptr_t>(pdata)
 500                                  & (this->addralign() - 1));
 501   bool has_misaligned_strings = false;
 502
 503   while (p < pend)
 504     {
 505       size_t len = p < pend0 ? string_length(p) : pend - p;
 506
 507       // Within merge input section each string must be aligned.
 508       if (len != 0
 509           && ((reinterpret_cast<uintptr_t>(p) & (this->addralign() - 1))
 510               != init_align_modulo))
 511           has_misaligned_strings = true;
 512
 513       Stringpool::Key key;
 514       this->stringpool_.add_with_length(p, len, true, &key);
 515
 516       merged_strings.push_back(Merged_string(i, key));
 517       p += len + 1;
 518       i += (len + 1) * sizeof(Char_type);
 519     }
 520
 521   // Record the last offset in the input section so that we can
 522   // compute the length of the last string.
 523   merged_strings.push_back(Merged_string(i, 0));
 524
 525   this->input_count_ += count;
 526   this->input_size_ += i;
 527
 528   if (has_misaligned_strings)
 529     gold_warning(_("%s: section %s contains incorrectly aligned strings;"
 530                    " the alignment of those strings won't be preserved"),
 531                  object->name().c_str(),
 532                  object->section_name(shndx).c_str());
 533
 534   // For script processing, we keep the input sections.
 535   if (this->keeps_input_sections())
 536     record_input_section(object, shndx);
 537
 538   if (is_new)
 539     delete[] pdata;
 540
 541   return true;
 542 }
 543
 544 // Finalize the mappings from the input sections to the output
 545 // section, and return the final data size.
 546
 547 template<typename Char_type>
 548 section_size_type
 549 Output_merge_string<Char_type>::finalize_merged_data()
 550 {
 551   this->stringpool_.set_string_offsets();
 552
 553   for (typename Merged_strings_lists::const_iterator l =
 554          this->merged_strings_lists_.begin();
 555        l != this->merged_strings_lists_.end();
 556        ++l)
 557     {
 558       section_offset_type last_input_offset = 0;
 559       section_offset_type last_output_offset = 0;
 560       for (typename Merged_strings::const_iterator p =
 561              (*l)->merged_strings.begin();
 562            p != (*l)->merged_strings.end();
 563            ++p)
 564         {
 565           section_size_type length = p->offset - last_input_offset;
 566           if (length > 0)
 567             (*l)->object->add_merge_mapping(this, (*l)->shndx,
 568                               last_input_offset, length, last_output_offset);
 569           last_input_offset = p->offset;
 570           if (p->stringpool_key != 0)
 571             last_output_offset =
 572                 this->stringpool_.get_offset_from_key(p->stringpool_key);
 573         }
 574       delete *l;
 575     }
 576
 577   // Save some memory.  This also ensures that this function will work
 578   // if called twice, as may happen if Layout::set_segment_offsets
 579   // finds a better alignment.
 580   this->merged_strings_lists_.clear();
 581
 582   return this->stringpool_.get_strtab_size();
 583 }
 584
 585 template<typename Char_type>
 586 void
 587 Output_merge_string<Char_type>::set_final_data_size()
 588 {
 589   const off_t final_data_size = this->finalize_merged_data();
 590   this->set_data_size(final_data_size);
 591 }
 592
 593 // Write out a merged string section.
 594
 595 template<typename Char_type>
 596 void
 597 Output_merge_string<Char_type>::do_write(Output_file* of)
 598 {
 599   this->stringpool_.write(of, this->offset());
 600 }
 601
 602 // Write a merged string section to a buffer.
 603
 604 template<typename Char_type>
 605 void
 606 Output_merge_string<Char_type>::do_write_to_buffer(unsigned char* buffer)
 607 {
 608   this->stringpool_.write_to_buffer(buffer, this->data_size());
 609 }
 610
 611 // Return the name of the types of string to use with
 612 // do_print_merge_stats.
 613
 614 template<typename Char_type>
 615 const char*
 616 Output_merge_string<Char_type>::string_name()
 617 {
 618   gold_unreachable();
 619   return NULL;
 620 }
 621
 622 template<>
 623 const char*
 624 Output_merge_string<char>::string_name()
 625 {
 626   return "strings";
 627 }
 628
 629 template<>
 630 const char*
 631 Output_merge_string<uint16_t>::string_name()
 632 {
 633   return "16-bit strings";
 634 }
 635
 636 template<>
 637 const char*
 638 Output_merge_string<uint32_t>::string_name()
 639 {
 640   return "32-bit strings";
 641 }
 642
 643 // Print merge stats to stderr.
 644
 645 template<typename Char_type>
 646 void
 647 Output_merge_string<Char_type>::do_print_merge_stats(const char* section_name)
 648 {
 649   char buf[200];
 650   snprintf(buf, sizeof buf, "%s merged %s", section_name, this->string_name());
 651   fprintf(stderr, _("%s: %s input bytes: %zu\n"),
 652           program_name, buf, this->input_size_);
 653   fprintf(stderr, _("%s: %s input strings: %zu\n"),
 654           program_name, buf, this->input_count_);
 655   this->stringpool_.print_stats(buf);
 656 }
 657
 658 // Instantiate the templates we need.
 659
 660 template
 661 class Output_merge_string<char>;
 662
 663 template
 664 class Output_merge_string<uint16_t>;
 665
 666 template
 667 class Output_merge_string<uint32_t>;
 668
 669 #if defined(HAVE_TARGET_32_LITTLE) || defined(HAVE_TARGET_32_BIG)
 670 template
 671 void
 672 Object_merge_map::initialize_input_to_output_map<32>(
 673     unsigned int shndx,
 674     elfcpp::Elf_types<32>::Elf_Addr starting_address,
 675     Unordered_map<section_offset_type, elfcpp::Elf_types<32>::Elf_Addr>*);
 676 #endif
 677
 678 #if defined(HAVE_TARGET_64_LITTLE) || defined(HAVE_TARGET_64_BIG)
 679 template
 680 void
 681 Object_merge_map::initialize_input_to_output_map<64>(
 682     unsigned int shndx,
 683     elfcpp::Elf_types<64>::Elf_Addr starting_address,
 684     Unordered_map<section_offset_type, elfcpp::Elf_types<64>::Elf_Addr>*);
 685 #endif
 686
 687 } // End namespace gold.