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