gold/archive.cc

   1 // archive.cc -- archive support for gold
   2
   3 // Copyright 2006, 2007, 2008 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 <cerrno>
  26 #include <cstring>
  27 #include <climits>
  28 #include <vector>
  29 #include "libiberty.h"
  30 #include "filenames.h"
  31
  32 #include "elfcpp.h"
  33 #include "options.h"
  34 #include "mapfile.h"
  35 #include "fileread.h"
  36 #include "readsyms.h"
  37 #include "symtab.h"
  38 #include "object.h"
  39 #include "archive.h"
  40
  41 namespace gold
  42 {
  43
  44 // The header of an entry in the archive.  This is all readable text,
  45 // padded with spaces where necesary.  If the contents of an archive
  46 // are all text file, the entire archive is readable.
  47
  48 struct Archive::Archive_header
  49 {
  50   // The entry name.
  51   char ar_name[16];
  52   // The file modification time.
  53   char ar_date[12];
  54   // The user's UID in decimal.
  55   char ar_uid[6];
  56   // The user's GID in decimal.
  57   char ar_gid[6];
  58   // The file mode in octal.
  59   char ar_mode[8];
  60   // The file size in decimal.
  61   char ar_size[10];
  62   // The final magic code.
  63   char ar_fmag[2];
  64 };
  65
  66 // Archive methods.
  67
  68 const char Archive::armag[sarmag] =
  69 {
  70   '!', '<', 'a', 'r', 'c', 'h', '>', '\n'
  71 };
  72
  73 const char Archive::armagt[sarmag] =
  74 {
  75   '!', '<', 't', 'h', 'i', 'n', '>', '\n'
  76 };
  77
  78 const char Archive::arfmag[2] = { '`', '\n' };
  79
  80 // Set up the archive: read the symbol map and the extended name
  81 // table.
  82
  83 void
  84 Archive::setup()
  85 {
  86   // We need to ignore empty archives.
  87   if (this->input_file_->file().filesize() == sarmag)
  88     return;
  89
  90   // The first member of the archive should be the symbol table.
  91   std::string armap_name;
  92   section_size_type armap_size =
  93     convert_to_section_size_type(this->read_header(sarmag, false,
  94                                                    &armap_name, NULL));
  95   off_t off = sarmag;
  96   if (armap_name.empty())
  97     {
  98       this->read_armap(sarmag + sizeof(Archive_header), armap_size);
  99       off = sarmag + sizeof(Archive_header) + armap_size;
 100     }
 101   else if (!this->input_file_->options().whole_archive())
 102     gold_error(_("%s: no archive symbol table (run ranlib)"),
 103                this->name().c_str());
 104
 105   // See if there is an extended name table.  We cache these views
 106   // because it is likely that we will want to read the following
 107   // header in the add_symbols routine.
 108   if ((off & 1) != 0)
 109     ++off;
 110   std::string xname;
 111   section_size_type extended_size =
 112     convert_to_section_size_type(this->read_header(off, true, &xname, NULL));
 113   if (xname == "/")
 114     {
 115       const unsigned char* p = this->get_view(off + sizeof(Archive_header),
 116                                               extended_size, false, true);
 117       const char* px = reinterpret_cast<const char*>(p);
 118       this->extended_names_.assign(px, extended_size);
 119     }
 120 }
 121
 122 // Unlock any nested archives.
 123
 124 void
 125 Archive::unlock_nested_archives()
 126 {
 127   for (Nested_archive_table::iterator p = this->nested_archives_.begin();
 128        p != this->nested_archives_.end();
 129        ++p)
 130     {
 131       p->second->unlock(this->task_);
 132     }
 133 }
 134
 135 // Read the archive symbol map.
 136
 137 void
 138 Archive::read_armap(off_t start, section_size_type size)
 139 {
 140   // Read in the entire armap.
 141   const unsigned char* p = this->get_view(start, size, true, false);
 142
 143   // Numbers in the armap are always big-endian.
 144   const elfcpp::Elf_Word* pword = reinterpret_cast<const elfcpp::Elf_Word*>(p);
 145   unsigned int nsyms = elfcpp::Swap<32, true>::readval(pword);
 146   ++pword;
 147
 148   // Note that the addition is in units of sizeof(elfcpp::Elf_Word).
 149   const char* pnames = reinterpret_cast<const char*>(pword + nsyms);
 150   section_size_type names_size =
 151     reinterpret_cast<const char*>(p) + size - pnames;
 152   this->armap_names_.assign(pnames, names_size);
 153
 154   this->armap_.resize(nsyms);
 155
 156   section_offset_type name_offset = 0;
 157   for (unsigned int i = 0; i < nsyms; ++i)
 158     {
 159       this->armap_[i].name_offset = name_offset;
 160       this->armap_[i].file_offset = elfcpp::Swap<32, true>::readval(pword);
 161       name_offset += strlen(pnames + name_offset) + 1;
 162       ++pword;
 163     }
 164
 165   if (static_cast<section_size_type>(name_offset) > names_size)
 166     gold_error(_("%s: bad archive symbol table names"),
 167                this->name().c_str());
 168
 169   // This array keeps track of which symbols are for archive elements
 170   // which we have already included in the link.
 171   this->armap_checked_.resize(nsyms);
 172 }
 173
 174 // Read the header of an archive member at OFF.  Fail if something
 175 // goes wrong.  Return the size of the member.  Set *PNAME to the name
 176 // of the member.
 177
 178 off_t
 179 Archive::read_header(off_t off, bool cache, std::string* pname,
 180                      off_t* nested_off)
 181 {
 182   const unsigned char* p = this->get_view(off, sizeof(Archive_header), true,
 183                                           cache);
 184   const Archive_header* hdr = reinterpret_cast<const Archive_header*>(p);
 185   return this->interpret_header(hdr, off,  pname, nested_off);
 186 }
 187
 188 // Interpret the header of HDR, the header of the archive member at
 189 // file offset OFF.  Fail if something goes wrong.  Return the size of
 190 // the member.  Set *PNAME to the name of the member.
 191
 192 off_t
 193 Archive::interpret_header(const Archive_header* hdr, off_t off,
 194                           std::string* pname, off_t* nested_off) const
 195 {
 196   if (memcmp(hdr->ar_fmag, arfmag, sizeof arfmag) != 0)
 197     {
 198       gold_error(_("%s: malformed archive header at %zu"),
 199                  this->name().c_str(), static_cast<size_t>(off));
 200       return this->input_file_->file().filesize() - off;
 201     }
 202
 203   const int size_string_size = sizeof hdr->ar_size;
 204   char size_string[size_string_size + 1];
 205   memcpy(size_string, hdr->ar_size, size_string_size);
 206   char* ps = size_string + size_string_size;
 207   while (ps[-1] == ' ')
 208     --ps;
 209   *ps = '\0';
 210
 211   errno = 0;
 212   char* end;
 213   off_t member_size = strtol(size_string, &end, 10);
 214   if (*end != '\0'
 215       || member_size < 0
 216       || (member_size == LONG_MAX && errno == ERANGE))
 217     {
 218       gold_error(_("%s: malformed archive header size at %zu"),
 219                  this->name().c_str(), static_cast<size_t>(off));
 220       return this->input_file_->file().filesize() - off;
 221     }
 222
 223   if (hdr->ar_name[0] != '/')
 224     {
 225       const char* name_end = strchr(hdr->ar_name, '/');
 226       if (name_end == NULL
 227           || name_end - hdr->ar_name >= static_cast<int>(sizeof hdr->ar_name))
 228         {
 229           gold_error(_("%s: malformed archive header name at %zu"),
 230                      this->name().c_str(), static_cast<size_t>(off));
 231           return this->input_file_->file().filesize() - off;
 232         }
 233       pname->assign(hdr->ar_name, name_end - hdr->ar_name);
 234       if (nested_off != NULL)
 235         *nested_off = 0;
 236     }
 237   else if (hdr->ar_name[1] == ' ')
 238     {
 239       // This is the symbol table.
 240       pname->clear();
 241     }
 242   else if (hdr->ar_name[1] == '/')
 243     {
 244       // This is the extended name table.
 245       pname->assign(1, '/');
 246     }
 247   else
 248     {
 249       errno = 0;
 250       long x = strtol(hdr->ar_name + 1, &end, 10);
 251       long y = 0;
 252       if (*end == ':')
 253         y = strtol(end + 1, &end, 10);
 254       if (*end != ' '
 255           || x < 0
 256           || (x == LONG_MAX && errno == ERANGE)
 257           || static_cast<size_t>(x) >= this->extended_names_.size())
 258         {
 259           gold_error(_("%s: bad extended name index at %zu"),
 260                      this->name().c_str(), static_cast<size_t>(off));
 261           return this->input_file_->file().filesize() - off;
 262         }
 263
 264       const char* name = this->extended_names_.data() + x;
 265       const char* name_end = strchr(name, '\n');
 266       if (static_cast<size_t>(name_end - name) > this->extended_names_.size()
 267           || name_end[-1] != '/')
 268         {
 269           gold_error(_("%s: bad extended name entry at header %zu"),
 270                      this->name().c_str(), static_cast<size_t>(off));
 271           return this->input_file_->file().filesize() - off;
 272         }
 273       pname->assign(name, name_end - 1 - name);
 274       if (nested_off != NULL)
 275         *nested_off = y;
 276     }
 277
 278   return member_size;
 279 }
 280
 281 // Select members from the archive and add them to the link.  We walk
 282 // through the elements in the archive map, and look each one up in
 283 // the symbol table.  If it exists as a strong undefined symbol, we
 284 // pull in the corresponding element.  We have to do this in a loop,
 285 // since pulling in one element may create new undefined symbols which
 286 // may be satisfied by other objects in the archive.
 287
 288 void
 289 Archive::add_symbols(Symbol_table* symtab, Layout* layout,
 290                      Input_objects* input_objects, Mapfile* mapfile)
 291 {
 292   if (this->input_file_->options().whole_archive())
 293     return this->include_all_members(symtab, layout, input_objects,
 294                                      mapfile);
 295
 296   input_objects->archive_start(this);
 297
 298   const size_t armap_size = this->armap_.size();
 299
 300   // This is a quick optimization, since we usually see many symbols
 301   // in a row with the same offset.  last_seen_offset holds the last
 302   // offset we saw that was present in the seen_offsets_ set.
 303   off_t last_seen_offset = -1;
 304
 305   // Track which symbols in the symbol table we've already found to be
 306   // defined.
 307
 308   bool added_new_object;
 309   do
 310     {
 311       added_new_object = false;
 312       for (size_t i = 0; i < armap_size; ++i)
 313         {
 314           if (this->armap_checked_[i])
 315             continue;
 316           if (this->armap_[i].file_offset == last_seen_offset)
 317             {
 318               this->armap_checked_[i] = true;
 319               continue;
 320             }
 321           if (this->seen_offsets_.find(this->armap_[i].file_offset)
 322               != this->seen_offsets_.end())
 323             {
 324               this->armap_checked_[i] = true;
 325               last_seen_offset = this->armap_[i].file_offset;
 326               continue;
 327             }
 328
 329           const char* sym_name = (this->armap_names_.data()
 330                                   + this->armap_[i].name_offset);
 331           Symbol* sym = symtab->lookup(sym_name);
 332           if (sym == NULL)
 333             {
 334               // Check whether the symbol was named in a -u option.
 335               if (!parameters->options().is_undefined(sym_name))
 336                 continue;
 337             }
 338           else if (!sym->is_undefined())
 339             {
 340               this->armap_checked_[i] = true;
 341               continue;
 342             }
 343           else if (sym->binding() == elfcpp::STB_WEAK)
 344             continue;
 345
 346           // We want to include this object in the link.
 347           last_seen_offset = this->armap_[i].file_offset;
 348           this->seen_offsets_.insert(last_seen_offset);
 349           this->armap_checked_[i] = true;
 350
 351           std::string why;
 352           if (sym == NULL)
 353             {
 354               why = "-u ";
 355               why += sym_name;
 356             }
 357           this->include_member(symtab, layout, input_objects,
 358                                last_seen_offset, mapfile, sym, why.c_str());
 359
 360           added_new_object = true;
 361         }
 362     }
 363   while (added_new_object);
 364
 365   input_objects->archive_stop(this);
 366 }
 367
 368 // An archive member iterator.
 369
 370 class Archive::const_iterator
 371 {
 372  public:
 373   // The header of an archive member.  This is what this iterator
 374   // points to.
 375   struct Header
 376   {
 377     // The name of the member.
 378     std::string name;
 379     // The file offset of the member.
 380     off_t off;
 381     // The file offset of a nested archive member.
 382     off_t nested_off;
 383     // The size of the member.
 384     off_t size;
 385   };
 386
 387   const_iterator(Archive* archive, off_t off)
 388     : archive_(archive), off_(off)
 389   { this->read_next_header(); }
 390
 391   const Header&
 392   operator*() const
 393   { return this->header_; }
 394
 395   const Header*
 396   operator->() const
 397   { return &this->header_; }
 398
 399   const_iterator&
 400   operator++()
 401   {
 402     if (this->off_ == this->archive_->file().filesize())
 403       return *this;
 404     this->off_ += sizeof(Archive_header);
 405     if (!this->archive_->is_thin_archive())
 406       this->off_ += this->header_.size;
 407     if ((this->off_ & 1) != 0)
 408       ++this->off_;
 409     this->read_next_header();
 410     return *this;
 411   }
 412
 413   const_iterator
 414   operator++(int)
 415   {
 416     const_iterator ret = *this;
 417     ++*this;
 418     return ret;
 419   }
 420
 421   bool
 422   operator==(const const_iterator p) const
 423   { return this->off_ == p->off; }
 424
 425   bool
 426   operator!=(const const_iterator p) const
 427   { return this->off_ != p->off; }
 428
 429  private:
 430   void
 431   read_next_header();
 432
 433   // The underlying archive.
 434   Archive* archive_;
 435   // The current offset in the file.
 436   off_t off_;
 437   // The current archive header.
 438   Header header_;
 439 };
 440
 441 // Read the next archive header.
 442
 443 void
 444 Archive::const_iterator::read_next_header()
 445 {
 446   off_t filesize = this->archive_->file().filesize();
 447   while (true)
 448     {
 449       if (filesize - this->off_ < static_cast<off_t>(sizeof(Archive_header)))
 450         {
 451           if (filesize != this->off_)
 452             {
 453               gold_error(_("%s: short archive header at %zu"),
 454                          this->archive_->filename().c_str(),
 455                          static_cast<size_t>(this->off_));
 456               this->off_ = filesize;
 457             }
 458           this->header_.off = filesize;
 459           return;
 460         }
 461
 462       unsigned char buf[sizeof(Archive_header)];
 463       this->archive_->file().read(this->off_, sizeof(Archive_header), buf);
 464
 465       const Archive_header* hdr = reinterpret_cast<const Archive_header*>(buf);
 466       this->header_.size =
 467         this->archive_->interpret_header(hdr, this->off_, &this->header_.name,
 468                                          &this->header_.nested_off);
 469       this->header_.off = this->off_;
 470
 471       // Skip special members.
 472       if (!this->header_.name.empty() && this->header_.name != "/")
 473         return;
 474
 475       this->off_ += sizeof(Archive_header) + this->header_.size;
 476       if ((this->off_ & 1) != 0)
 477         ++this->off_;
 478     }
 479 }
 480
 481 // Initial iterator.
 482
 483 Archive::const_iterator
 484 Archive::begin()
 485 {
 486   return Archive::const_iterator(this, sarmag);
 487 }
 488
 489 // Final iterator.
 490
 491 Archive::const_iterator
 492 Archive::end()
 493 {
 494   return Archive::const_iterator(this, this->input_file_->file().filesize());
 495 }
 496
 497 // Include all the archive members in the link.  This is for --whole-archive.
 498
 499 void
 500 Archive::include_all_members(Symbol_table* symtab, Layout* layout,
 501                              Input_objects* input_objects, Mapfile* mapfile)
 502 {
 503   input_objects->archive_start(this);
 504
 505   for (Archive::const_iterator p = this->begin();
 506        p != this->end();
 507        ++p)
 508     this->include_member(symtab, layout, input_objects, p->off,
 509                          mapfile, NULL, "--whole-archive");
 510
 511   input_objects->archive_stop(this);
 512 }
 513
 514 // Return the number of members in the archive.  This is only used for
 515 // reports.
 516
 517 size_t
 518 Archive::count_members()
 519 {
 520   size_t ret = 0;
 521   for (Archive::const_iterator p = this->begin();
 522        p != this->end();
 523        ++p)
 524     ++ret;
 525   return ret;
 526 }
 527
 528 // Include an archive member in the link.  OFF is the file offset of
 529 // the member header.  WHY is the reason we are including this member.
 530
 531 void
 532 Archive::include_member(Symbol_table* symtab, Layout* layout,
 533                         Input_objects* input_objects, off_t off,
 534                         Mapfile* mapfile, Symbol* sym, const char* why)
 535 {
 536   std::string n;
 537   off_t nested_off;
 538   this->read_header(off, false, &n, &nested_off);
 539
 540   if (mapfile != NULL)
 541     mapfile->report_include_archive_member(this, n, sym, why);
 542
 543   Input_file* input_file;
 544   off_t memoff;
 545
 546   if (!this->is_thin_archive_)
 547     {
 548       input_file = this->input_file_;
 549       memoff = off + static_cast<off_t>(sizeof(Archive_header));
 550     }
 551   else
 552     {
 553       // Adjust a relative pathname so that it is relative
 554       // to the directory containing the archive.
 555       if (!IS_ABSOLUTE_PATH(n.c_str()))
 556         {
 557           const char *arch_path = this->name().c_str();
 558           const char *basename = lbasename(arch_path);
 559           if (basename > arch_path)
 560             n.replace(0, 0, this->name().substr(0, basename - arch_path));
 561         }
 562       if (nested_off > 0)
 563         {
 564           // This is a member of a nested archive.  Open the containing
 565           // archive if we don't already have it open, then do a recursive
 566           // call to include the member from that archive.
 567           Archive* arch;
 568           Nested_archive_table::const_iterator p =
 569             this->nested_archives_.find(n);
 570           if (p != this->nested_archives_.end())
 571             arch = p->second;
 572           else
 573             {
 574               Input_file_argument* input_file_arg =
 575                 new Input_file_argument(n.c_str(), false, "", false,
 576                                         parameters->options());
 577               input_file = new Input_file(input_file_arg);
 578               if (!input_file->open(parameters->options(), *this->dirpath_,
 579                                     this->task_))
 580                 return;
 581               arch = new Archive(n, input_file, false, this->dirpath_,
 582                                  this->task_);
 583               arch->setup();
 584               std::pair<Nested_archive_table::iterator, bool> ins =
 585                 this->nested_archives_.insert(std::make_pair(n, arch));
 586               gold_assert(ins.second);
 587             }
 588           arch->include_member(symtab, layout, input_objects, nested_off,
 589                                NULL, NULL, NULL);
 590           return;
 591         }
 592       // This is an external member of a thin archive.  Open the
 593       // file as a regular relocatable object file.
 594       Input_file_argument* input_file_arg =
 595           new Input_file_argument(n.c_str(), false, "", false,
 596                                   this->input_file_->options());
 597       input_file = new Input_file(input_file_arg);
 598       if (!input_file->open(parameters->options(), *this->dirpath_,
 599                             this->task_))
 600         {
 601           return;
 602         }
 603       memoff = 0;
 604     }
 605
 606   off_t filesize = input_file->file().filesize();
 607   int read_size = elfcpp::Elf_sizes<64>::ehdr_size;
 608   if (filesize - memoff < read_size)
 609     read_size = filesize - memoff;
 610
 611   if (read_size < 4)
 612     {
 613       gold_error(_("%s: member at %zu is not an ELF object"),
 614                  this->name().c_str(), static_cast<size_t>(off));
 615       return;
 616     }
 617
 618   const unsigned char* ehdr = input_file->file().get_view(memoff, 0, read_size,
 619                                                           true, false);
 620
 621   static unsigned char elfmagic[4] =
 622     {
 623       elfcpp::ELFMAG0, elfcpp::ELFMAG1,
 624       elfcpp::ELFMAG2, elfcpp::ELFMAG3
 625     };
 626   if (memcmp(ehdr, elfmagic, 4) != 0)
 627     {
 628       gold_error(_("%s: member at %zu is not an ELF object"),
 629                  this->name().c_str(), static_cast<size_t>(off));
 630       return;
 631     }
 632
 633   Object* obj = make_elf_object((std::string(this->input_file_->filename())
 634                                  + "(" + n + ")"),
 635                                 input_file, memoff, ehdr, read_size);
 636
 637   if (input_objects->add_object(obj))
 638     {
 639       Read_symbols_data sd;
 640       obj->read_symbols(&sd);
 641       obj->layout(symtab, layout, &sd);
 642       obj->add_symbols(symtab, &sd);
 643     }
 644   else
 645     {
 646       // FIXME: We need to close the descriptor here.
 647       delete obj;
 648     }
 649
 650   if (this->is_thin_archive_)
 651     {
 652       // Opening the file locked it.  Unlock it now.
 653       input_file->file().unlock(this->task_);
 654     }
 655 }
 656
 657 // Add_archive_symbols methods.
 658
 659 Add_archive_symbols::~Add_archive_symbols()
 660 {
 661   if (this->this_blocker_ != NULL)
 662     delete this->this_blocker_;
 663   // next_blocker_ is deleted by the task associated with the next
 664   // input file.
 665 }
 666
 667 // Return whether we can add the archive symbols.  We are blocked by
 668 // this_blocker_.  We block next_blocker_.  We also lock the file.
 669
 670 Task_token*
 671 Add_archive_symbols::is_runnable()
 672 {
 673   if (this->this_blocker_ != NULL && this->this_blocker_->is_blocked())
 674     return this->this_blocker_;
 675   return NULL;
 676 }
 677
 678 void
 679 Add_archive_symbols::locks(Task_locker* tl)
 680 {
 681   tl->add(this, this->next_blocker_);
 682   tl->add(this, this->archive_->token());
 683 }
 684
 685 void
 686 Add_archive_symbols::run(Workqueue*)
 687 {
 688   this->archive_->add_symbols(this->symtab_, this->layout_,
 689                               this->input_objects_, this->mapfile_);
 690
 691   this->archive_->unlock_nested_archives();
 692
 693   this->archive_->release();
 694   this->archive_->clear_uncached_views();
 695
 696   if (this->input_group_ != NULL)
 697     this->input_group_->add_archive(this->archive_);
 698   else
 699     {
 700       // We no longer need to know about this archive.
 701       delete this->archive_;
 702       this->archive_ = NULL;
 703     }
 704 }
 705
 706 } // End namespace gold.