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   map->add_mapping(input_offset, length, output_offset);
 107 }
 108
 109 void
 110 Object_merge_map::Input_merge_map::add_mapping(
 111     section_offset_type input_offset, section_size_type length,
 112     section_offset_type output_offset) {
 113   // Try to merge the new entry in the last one we saw.
 114   if (!this->entries.empty())
 115     {
 116       Input_merge_entry& entry(this->entries.back());
 117
 118       // Use section_size_type to avoid signed/unsigned warnings.
 119       section_size_type input_offset_u = input_offset;
 120       section_size_type output_offset_u = output_offset;
 121
 122       // If this entry is not in order, we need to sort the vector
 123       // before looking anything up.
 124       if (input_offset_u < entry.input_offset + entry.length)
 125         {
 126           gold_assert(input_offset < entry.input_offset);
 127           gold_assert(input_offset_u + length
 128                       <= static_cast<section_size_type>(entry.input_offset));
 129           this->sorted = false;
 130         }
 131       else if (entry.input_offset + entry.length == input_offset_u
 132                && (output_offset == -1
 133                    ? entry.output_offset == -1
 134                    : entry.output_offset + entry.length == output_offset_u))
 135         {
 136           entry.length += length;
 137           return;
 138         }
 139     }
 140
 141   Input_merge_entry entry;
 142   entry.input_offset = input_offset;
 143   entry.length = length;
 144   entry.output_offset = output_offset;
 145   this->entries.push_back(entry);
 146 }
 147
 148 // Get the output offset for an input address.
 149
 150 bool
 151 Object_merge_map::get_output_offset(unsigned int shndx,
 152                                     section_offset_type input_offset,
 153                                     section_offset_type* output_offset)
 154 {
 155   Input_merge_map* map = this->get_input_merge_map(shndx);
 156   if (map == NULL)
 157     return false;
 158
 159   if (!map->sorted)
 160     {
 161       std::sort(map->entries.begin(), map->entries.end(),
 162                 Input_merge_compare());
 163       map->sorted = true;
 164     }
 165
 166   Input_merge_entry entry;
 167   entry.input_offset = input_offset;
 168   std::vector<Input_merge_entry>::const_iterator p =
 169     std::upper_bound(map->entries.begin(), map->entries.end(),
 170                      entry, Input_merge_compare());
 171   if (p == map->entries.begin())
 172     return false;
 173   --p;
 174   gold_assert(p->input_offset <= input_offset);
 175
 176   if (input_offset - p->input_offset
 177       >= static_cast<section_offset_type>(p->length))
 178     return false;
 179
 180   *output_offset = p->output_offset;
 181   if (*output_offset != -1)
 182     *output_offset += (input_offset - p->input_offset);
 183   return true;
 184 }
 185
 186 // Return whether this is the merge map for section SHNDX.
 187
 188 const Output_section_data*
 189 Object_merge_map::find_merge_section(unsigned int shndx) const {
 190   const Object_merge_map::Input_merge_map* map =
 191     this->get_input_merge_map(shndx);
 192   if (map == NULL)
 193     return NULL;
 194   return map->output_data;
 195 }
 196
 197 // Initialize a mapping from input offsets to output addresses.
 198
 199 template<int size>
 200 void
 201 Object_merge_map::initialize_input_to_output_map(
 202     unsigned int shndx,
 203     typename elfcpp::Elf_types<size>::Elf_Addr starting_address,
 204     Unordered_map<section_offset_type,
 205                   typename elfcpp::Elf_types<size>::Elf_Addr>* initialize_map)
 206 {
 207   Input_merge_map* map = this->get_input_merge_map(shndx);
 208   gold_assert(map != NULL);
 209
 210   gold_assert(initialize_map->empty());
 211   // We know how many entries we are going to add.
 212   // reserve_unordered_map takes an expected count of buckets, not a
 213   // count of elements, so double it to try to reduce collisions.
 214   reserve_unordered_map(initialize_map, map->entries.size() * 2);
 215
 216   for (Input_merge_map::Entries::const_iterator p = map->entries.begin();
 217        p != map->entries.end();
 218        ++p)
 219     {
 220       section_offset_type output_offset = p->output_offset;
 221       if (output_offset != -1)
 222         output_offset += starting_address;
 223       else
 224         {
 225           // If we see a relocation against an address we have chosen
 226           // to discard, we relocate to zero.  FIXME: We could also
 227           // issue a warning in this case; that would require
 228           // reporting this somehow and checking it in the routines in
 229           // reloc.h.
 230           output_offset = 0;
 231         }
 232       initialize_map->insert(std::make_pair(p->input_offset, output_offset));
 233     }
 234 }
 235
 236 // Class Output_merge_base.
 237
 238 // Return the output offset for an input offset.  The input address is
 239 // at offset OFFSET in section SHNDX in OBJECT.  If we know the
 240 // offset, set *POUTPUT and return true.  Otherwise return false.
 241
 242 bool
 243 Output_merge_base::do_output_offset(const Relobj* object,
 244                                     unsigned int shndx,
 245                                     section_offset_type offset,
 246                                     section_offset_type* poutput) const
 247 {
 248   return object->merge_output_offset(shndx, offset, poutput);
 249 }
 250
 251 // Record a merged input section for script processing.
 252
 253 void
 254 Output_merge_base::record_input_section(Relobj* relobj, unsigned int shndx)
 255 {
 256   gold_assert(this->keeps_input_sections_ && relobj != NULL);
 257   // If this is the first input section, record it.  We need do this because
 258   // this->input_sections_ is unordered.
 259   if (this->first_relobj_ == NULL)
 260     {
 261       this->first_relobj_ = relobj;
 262       this->first_shndx_ = shndx;
 263     }
 264
 265   std::pair<Input_sections::iterator, bool> result =
 266     this->input_sections_.insert(Section_id(relobj, shndx));
 267   // We should insert a merge section once only.
 268   gold_assert(result.second);
 269 }
 270
 271 // Class Output_merge_data.
 272
 273 // Compute the hash code for a fixed-size constant.
 274
 275 size_t
 276 Output_merge_data::Merge_data_hash::operator()(Merge_data_key k) const
 277 {
 278   const unsigned char* p = this->pomd_->constant(k);
 279   section_size_type entsize =
 280     convert_to_section_size_type(this->pomd_->entsize());
 281
 282   // Fowler/Noll/Vo (FNV) hash (type FNV-1a).
 283   if (sizeof(size_t) == 8)
 284     {
 285       size_t result = static_cast<size_t>(14695981039346656037ULL);
 286       for (section_size_type i = 0; i < entsize; ++i)
 287         {
 288           result &= (size_t) *p++;
 289           result *= 1099511628211ULL;
 290         }
 291       return result;
 292     }
 293   else
 294     {
 295       size_t result = 2166136261UL;
 296       for (section_size_type i = 0; i < entsize; ++i)
 297         {
 298           result ^= (size_t) *p++;
 299           result *= 16777619UL;
 300         }
 301       return result;
 302     }
 303 }
 304
 305 // Return whether one hash table key equals another.
 306
 307 bool
 308 Output_merge_data::Merge_data_eq::operator()(Merge_data_key k1,
 309                                              Merge_data_key k2) const
 310 {
 311   const unsigned char* p1 = this->pomd_->constant(k1);
 312   const unsigned char* p2 = this->pomd_->constant(k2);
 313   return memcmp(p1, p2, this->pomd_->entsize()) == 0;
 314 }
 315
 316 // Add a constant to the end of the section contents.
 317
 318 void
 319 Output_merge_data::add_constant(const unsigned char* p)
 320 {
 321   section_size_type entsize = convert_to_section_size_type(this->entsize());
 322   section_size_type addralign =
 323     convert_to_section_size_type(this->addralign());
 324   section_size_type addsize = std::max(entsize, addralign);
 325   if (this->len_ + addsize > this->alc_)
 326     {
 327       if (this->alc_ == 0)
 328         this->alc_ = 128 * addsize;
 329       else
 330         this->alc_ *= 2;
 331       this->p_ = static_cast<unsigned char*>(realloc(this->p_, this->alc_));
 332       if (this->p_ == NULL)
 333         gold_nomem();
 334     }
 335
 336   memcpy(this->p_ + this->len_, p, entsize);
 337   if (addsize > entsize)
 338     memset(this->p_ + this->len_ + entsize, 0, addsize - entsize);
 339   this->len_ += addsize;
 340 }
 341
 342 // Add the input section SHNDX in OBJECT to a merged output section
 343 // which holds fixed length constants.  Return whether we were able to
 344 // handle the section; if not, it will be linked as usual without
 345 // constant merging.
 346
 347 bool
 348 Output_merge_data::do_add_input_section(Relobj* object, unsigned int shndx)
 349 {
 350   section_size_type len;
 351   bool is_new;
 352   const unsigned char* p = object->decompressed_section_contents(shndx, &len,
 353                                                                  &is_new);
 354
 355   section_size_type entsize = convert_to_section_size_type(this->entsize());
 356
 357   if (len % entsize != 0)
 358     {
 359       if (is_new)
 360         delete[] p;
 361       return false;
 362     }
 363
 364   this->input_count_ += len / entsize;
 365
 366   Object_merge_map* merge_map = object->get_or_create_merge_map();
 367   Object_merge_map::Input_merge_map* input_merge_map =
 368     merge_map->get_or_make_input_merge_map(this, shndx);
 369
 370   for (section_size_type i = 0; i < len; i += entsize, p += entsize)
 371     {
 372       // Add the constant to the section contents.  If we find that it
 373       // is already in the hash table, we will remove it again.
 374       Merge_data_key k = this->len_;
 375       this->add_constant(p);
 376
 377       std::pair<Merge_data_hashtable::iterator, bool> ins =
 378         this->hashtable_.insert(k);
 379
 380       if (!ins.second)
 381         {
 382           // Key was already present.  Remove the copy we just added.
 383           this->len_ -= entsize;
 384           k = *ins.first;
 385         }
 386
 387       // Record the offset of this constant in the output section.
 388       input_merge_map->add_mapping(i, entsize, k);
 389     }
 390
 391   // For script processing, we keep the input sections.
 392   if (this->keeps_input_sections())
 393     record_input_section(object, shndx);
 394
 395   if (is_new)
 396     delete[] p;
 397
 398   return true;
 399 }
 400
 401 // Set the final data size in a merged output section with fixed size
 402 // constants.
 403
 404 void
 405 Output_merge_data::set_final_data_size()
 406 {
 407   // Release the memory we don't need.
 408   this->p_ = static_cast<unsigned char*>(realloc(this->p_, this->len_));
 409   // An Output_merge_data object may be empty and realloc is allowed
 410   // to return a NULL pointer in this case.  An Output_merge_data is empty
 411   // if all its input sections have sizes that are not multiples of entsize.
 412   gold_assert(this->p_ != NULL || this->len_ == 0);
 413   this->set_data_size(this->len_);
 414 }
 415
 416 // Write the data of a merged output section with fixed size constants
 417 // to the file.
 418
 419 void
 420 Output_merge_data::do_write(Output_file* of)
 421 {
 422   of->write(this->offset(), this->p_, this->len_);
 423 }
 424
 425 // Write the data to a buffer.
 426
 427 void
 428 Output_merge_data::do_write_to_buffer(unsigned char* buffer)
 429 {
 430   memcpy(buffer, this->p_, this->len_);
 431 }
 432
 433 // Print merge stats to stderr.
 434
 435 void
 436 Output_merge_data::do_print_merge_stats(const char* section_name)
 437 {
 438   fprintf(stderr,
 439           _("%s: %s merged constants size: %lu; input: %zu; output: %zu\n"),
 440           program_name, section_name,
 441           static_cast<unsigned long>(this->entsize()),
 442           this->input_count_, this->hashtable_.size());
 443 }
 444
 445 // Class Output_merge_string.
 446
 447 // Add an input section to a merged string section.
 448
 449 template<typename Char_type>
 450 bool
 451 Output_merge_string<Char_type>::do_add_input_section(Relobj* object,
 452                                                      unsigned int shndx)
 453 {
 454   section_size_type sec_len;
 455   bool is_new;
 456   const unsigned char* pdata = object->decompressed_section_contents(shndx,
 457                                                                      &sec_len,
 458                                                                      &is_new);
 459
 460   const Char_type* p = reinterpret_cast<const Char_type*>(pdata);
 461   const Char_type* pend = p + sec_len / sizeof(Char_type);
 462   const Char_type* pend0 = pend;
 463
 464   if (sec_len % sizeof(Char_type) != 0)
 465     {
 466       object->error(_("mergeable string section length not multiple of "
 467                       "character size"));
 468       if (is_new)
 469         delete[] pdata;
 470       return false;
 471     }
 472
 473   if (pend[-1] != 0)
 474     {
 475       gold_warning(_("%s: last entry in mergeable string section '%s' "
 476                      "not null terminated"),
 477                    object->name().c_str(),
 478                    object->section_name(shndx).c_str());
 479       // Find the end of the last NULL-terminated string in the buffer.
 480       while (pend0 > p && pend0[-1] != 0)
 481         --pend0;
 482     }
 483
 484   Merged_strings_list* merged_strings_list =
 485       new Merged_strings_list(object, shndx);
 486   this->merged_strings_lists_.push_back(merged_strings_list);
 487   Merged_strings& merged_strings = merged_strings_list->merged_strings;
 488
 489   // Count the number of non-null strings in the section and size the list.
 490   size_t count = 0;
 491   const Char_type* pt = p;
 492   while (pt < pend0)
 493     {
 494       size_t len = string_length(pt);
 495       if (len != 0)
 496         ++count;
 497       pt += len + 1;
 498     }
 499   if (pend0 < pend)
 500     ++count;
 501   merged_strings.reserve(count + 1);
 502
 503   // The index I is in bytes, not characters.
 504   section_size_type i = 0;
 505
 506   // We assume here that the beginning of the section is correctly
 507   // aligned, so each string within the section must retain the same
 508   // modulo.
 509   uintptr_t init_align_modulo = (reinterpret_cast<uintptr_t>(pdata)
 510                                  & (this->addralign() - 1));
 511   bool has_misaligned_strings = false;
 512
 513   while (p < pend)
 514     {
 515       size_t len = p < pend0 ? string_length(p) : pend - p;
 516
 517       // Within merge input section each string must be aligned.
 518       if (len != 0
 519           && ((reinterpret_cast<uintptr_t>(p) & (this->addralign() - 1))
 520               != init_align_modulo))
 521           has_misaligned_strings = true;
 522
 523       Stringpool::Key key;
 524       this->stringpool_.add_with_length(p, len, true, &key);
 525
 526       merged_strings.push_back(Merged_string(i, key));
 527       p += len + 1;
 528       i += (len + 1) * sizeof(Char_type);
 529     }
 530
 531   // Record the last offset in the input section so that we can
 532   // compute the length of the last string.
 533   merged_strings.push_back(Merged_string(i, 0));
 534
 535   this->input_count_ += count;
 536   this->input_size_ += i;
 537
 538   if (has_misaligned_strings)
 539     gold_warning(_("%s: section %s contains incorrectly aligned strings;"
 540                    " the alignment of those strings won't be preserved"),
 541                  object->name().c_str(),
 542                  object->section_name(shndx).c_str());
 543
 544   // For script processing, we keep the input sections.
 545   if (this->keeps_input_sections())
 546     record_input_section(object, shndx);
 547
 548   if (is_new)
 549     delete[] pdata;
 550
 551   return true;
 552 }
 553
 554 // Finalize the mappings from the input sections to the output
 555 // section, and return the final data size.
 556
 557 template<typename Char_type>
 558 section_size_type
 559 Output_merge_string<Char_type>::finalize_merged_data()
 560 {
 561   this->stringpool_.set_string_offsets();
 562
 563   for (typename Merged_strings_lists::const_iterator l =
 564          this->merged_strings_lists_.begin();
 565        l != this->merged_strings_lists_.end();
 566        ++l)
 567     {
 568       section_offset_type last_input_offset = 0;
 569       section_offset_type last_output_offset = 0;
 570       Relobj *object = (*l)->object;
 571       Object_merge_map* merge_map = object->get_or_create_merge_map();
 572       Object_merge_map::Input_merge_map* input_merge_map =
 573         merge_map->get_or_make_input_merge_map(this, (*l)->shndx);
 574
 575       for (typename Merged_strings::const_iterator p =
 576              (*l)->merged_strings.begin();
 577            p != (*l)->merged_strings.end();
 578            ++p)
 579         {
 580           section_size_type length = p->offset - last_input_offset;
 581           if (length > 0)
 582             input_merge_map->add_mapping(last_input_offset, length,
 583                                          last_output_offset);
 584           last_input_offset = p->offset;
 585           if (p->stringpool_key != 0)
 586             last_output_offset =
 587                 this->stringpool_.get_offset_from_key(p->stringpool_key);
 588         }
 589       delete *l;
 590     }
 591
 592   // Save some memory.  This also ensures that this function will work
 593   // if called twice, as may happen if Layout::set_segment_offsets
 594   // finds a better alignment.
 595   this->merged_strings_lists_.clear();
 596
 597   return this->stringpool_.get_strtab_size();
 598 }
 599
 600 template<typename Char_type>
 601 void
 602 Output_merge_string<Char_type>::set_final_data_size()
 603 {
 604   const off_t final_data_size = this->finalize_merged_data();
 605   this->set_data_size(final_data_size);
 606 }
 607
 608 // Write out a merged string section.
 609
 610 template<typename Char_type>
 611 void
 612 Output_merge_string<Char_type>::do_write(Output_file* of)
 613 {
 614   this->stringpool_.write(of, this->offset());
 615 }
 616
 617 // Write a merged string section to a buffer.
 618
 619 template<typename Char_type>
 620 void
 621 Output_merge_string<Char_type>::do_write_to_buffer(unsigned char* buffer)
 622 {
 623   this->stringpool_.write_to_buffer(buffer, this->data_size());
 624 }
 625
 626 // Return the name of the types of string to use with
 627 // do_print_merge_stats.
 628
 629 template<typename Char_type>
 630 const char*
 631 Output_merge_string<Char_type>::string_name()
 632 {
 633   gold_unreachable();
 634   return NULL;
 635 }
 636
 637 template<>
 638 const char*
 639 Output_merge_string<char>::string_name()
 640 {
 641   return "strings";
 642 }
 643
 644 template<>
 645 const char*
 646 Output_merge_string<uint16_t>::string_name()
 647 {
 648   return "16-bit strings";
 649 }
 650
 651 template<>
 652 const char*
 653 Output_merge_string<uint32_t>::string_name()
 654 {
 655   return "32-bit strings";
 656 }
 657
 658 // Print merge stats to stderr.
 659
 660 template<typename Char_type>
 661 void
 662 Output_merge_string<Char_type>::do_print_merge_stats(const char* section_name)
 663 {
 664   char buf[200];
 665   snprintf(buf, sizeof buf, "%s merged %s", section_name, this->string_name());
 666   fprintf(stderr, _("%s: %s input bytes: %zu\n"),
 667           program_name, buf, this->input_size_);
 668   fprintf(stderr, _("%s: %s input strings: %zu\n"),
 669           program_name, buf, this->input_count_);
 670   this->stringpool_.print_stats(buf);
 671 }
 672
 673 // Instantiate the templates we need.
 674
 675 template
 676 class Output_merge_string<char>;
 677
 678 template
 679 class Output_merge_string<uint16_t>;
 680
 681 template
 682 class Output_merge_string<uint32_t>;
 683
 684 #if defined(HAVE_TARGET_32_LITTLE) || defined(HAVE_TARGET_32_BIG)
 685 template
 686 void
 687 Object_merge_map::initialize_input_to_output_map<32>(
 688     unsigned int shndx,
 689     elfcpp::Elf_types<32>::Elf_Addr starting_address,
 690     Unordered_map<section_offset_type, elfcpp::Elf_types<32>::Elf_Addr>*);
 691 #endif
 692
 693 #if defined(HAVE_TARGET_64_LITTLE) || defined(HAVE_TARGET_64_BIG)
 694 template
 695 void
 696 Object_merge_map::initialize_input_to_output_map<64>(
 697     unsigned int shndx,
 698     elfcpp::Elf_types<64>::Elf_Addr starting_address,
 699     Unordered_map<section_offset_type, elfcpp::Elf_types<64>::Elf_Addr>*);
 700 #endif
 701
 702 } // End namespace gold.