gold/stringpool.cc

   1 // stringpool.cc -- a string pool for gold
   2
   3 #include "gold.h"
   4
   5 #include <cstring>
   6 #include <algorithm>
   7 #include <vector>
   8
   9 #include "output.h"
  10 #include "parameters.h"
  11 #include "stringpool.h"
  12
  13 namespace gold
  14 {
  15
  16 template<typename Stringpool_char>
  17 Stringpool_template<Stringpool_char>::Stringpool_template()
  18   : string_set_(), strings_(), strtab_size_(0), next_index_(1),
  19     zero_null_(true)
  20 {
  21 }
  22
  23 template<typename Stringpool_char>
  24 Stringpool_template<Stringpool_char>::~Stringpool_template()
  25 {
  26   for (typename std::list<Stringdata*>::iterator p = this->strings_.begin();
  27        p != this->strings_.end();
  28        ++p)
  29     delete[] reinterpret_cast<char*>(*p);
  30 }
  31
  32 // Return the length of a string of arbitrary character type.
  33
  34 template<typename Stringpool_char>
  35 size_t
  36 Stringpool_template<Stringpool_char>::string_length(const Stringpool_char* p)
  37 {
  38   size_t len = 0;
  39   for (; *p != 0; ++p)
  40     ++len;
  41   return len;
  42 }
  43
  44 // Specialize string_length for char.  Maybe we could just use
  45 // std::char_traits<>::length?
  46
  47 template<>
  48 inline size_t
  49 Stringpool_template<char>::string_length(const char* p)
  50 {
  51   return strlen(p);
  52 }
  53
  54 // Equality comparison function.
  55
  56 template<typename Stringpool_char>
  57 bool
  58 Stringpool_template<Stringpool_char>::Stringpool_eq::operator()(
  59     const Stringpool_char* s1,
  60     const Stringpool_char* s2) const
  61 {
  62   while (*s1 != 0)
  63     if (*s1++ != *s2++)
  64       return false;
  65   return *s2 == 0;
  66 }
  67
  68 // Specialize equality comparison for char.
  69
  70 template<>
  71 bool
  72 Stringpool_template<char>::Stringpool_eq::operator()(const char* s1,
  73                                                      const char* s2) const
  74 {
  75   return strcmp(s1, s2) == 0;
  76 }
  77
  78 // Hash function.
  79
  80 template<typename Stringpool_char>
  81 size_t
  82 Stringpool_template<Stringpool_char>::Stringpool_hash::operator()(
  83     const Stringpool_char* s) const
  84 {
  85   // Fowler/Noll/Vo (FNV) hash (type FNV-1a).
  86   if (sizeof(size_t) == 8)
  87     {
  88       size_t result = static_cast<size_t>(14695981039346656037ULL);
  89       while (*s != 0)
  90         {
  91           const char* p = reinterpret_cast<const char*>(s);
  92           for (size_t i = 0; i < sizeof(Stringpool_char); ++i)
  93             {
  94               result &= (size_t) *p++;
  95               result *= 1099511628211ULL;
  96             }
  97           ++s;
  98         }
  99       return result;
 100     }
 101   else
 102     {
 103       size_t result = 2166136261UL;
 104       while (*s != 0)
 105         {
 106           const char* p = reinterpret_cast<const char*>(s);
 107           for (size_t i = 0; i < sizeof(Stringpool_char); ++i)
 108             {
 109               result ^= (size_t) *p++;
 110               result *= 16777619UL;
 111             }
 112           ++s;
 113         }
 114       return result;
 115     }
 116 }
 117
 118 // Add a string to the list of canonical strings.  Return a pointer to
 119 // the canonical string.  If PKEY is not NULL, set *PKEY to the key.
 120
 121 template<typename Stringpool_char>
 122 const Stringpool_char*
 123 Stringpool_template<Stringpool_char>::add_string(const Stringpool_char* s,
 124                                                  Key* pkey)
 125 {
 126   // We are in trouble if we've already computed the string offsets.
 127   gold_assert(this->strtab_size_ == 0);
 128
 129   // The size we allocate for a new Stringdata.
 130   const size_t buffer_size = 1000;
 131   // The amount we multiply the Stringdata index when calculating the
 132   // key.
 133   const size_t key_mult = 1024;
 134   gold_assert(key_mult >= buffer_size);
 135
 136   size_t len = (string_length(s) + 1) * sizeof(Stringpool_char);
 137
 138   size_t alc;
 139   bool front = true;
 140   if (len > buffer_size)
 141     {
 142       alc = sizeof(Stringdata) + len;
 143       front = false;
 144     }
 145   else if (this->strings_.empty())
 146     alc = sizeof(Stringdata) + buffer_size;
 147   else
 148     {
 149       Stringdata *psd = this->strings_.front();
 150       if (len > psd->alc - psd->len)
 151         alc = sizeof(Stringdata) + buffer_size;
 152       else
 153         {
 154           char* ret = psd->data + psd->len;
 155           memcpy(ret, s, len);
 156
 157           if (pkey != NULL)
 158             *pkey = psd->index * key_mult + psd->len;
 159
 160           psd->len += len;
 161
 162           return reinterpret_cast<const Stringpool_char*>(ret);
 163         }
 164     }
 165
 166   Stringdata *psd = reinterpret_cast<Stringdata*>(new char[alc]);
 167   psd->alc = alc - sizeof(Stringdata);
 168   memcpy(psd->data, s, len);
 169   psd->len = len;
 170   psd->index = this->next_index_;
 171   ++this->next_index_;
 172
 173   if (pkey != NULL)
 174     *pkey = psd->index * key_mult;
 175
 176   if (front)
 177     this->strings_.push_front(psd);
 178   else
 179     this->strings_.push_back(psd);
 180
 181   return reinterpret_cast<const Stringpool_char*>(psd->data);
 182 }
 183
 184 // Add a string to a string pool.
 185
 186 template<typename Stringpool_char>
 187 const Stringpool_char*
 188 Stringpool_template<Stringpool_char>::add(const Stringpool_char* s, Key* pkey)
 189 {
 190   // FIXME: This will look up the entry twice in the hash table.  The
 191   // problem is that we can't insert S before we canonicalize it.  I
 192   // don't think there is a way to handle this correctly with
 193   // unordered_map, so this should be replaced with custom code to do
 194   // what we need, which is to return the empty slot.
 195
 196   typename String_set_type::const_iterator p = this->string_set_.find(s);
 197   if (p != this->string_set_.end())
 198     {
 199       if (pkey != NULL)
 200         *pkey = p->second.first;
 201       return p->first;
 202     }
 203
 204   Key k;
 205   const Stringpool_char* ret = this->add_string(s, &k);
 206
 207   const off_t ozero = 0;
 208   std::pair<const Stringpool_char*, Val> element(ret,
 209                                                  std::make_pair(k, ozero));
 210   std::pair<typename String_set_type::iterator, bool> ins =
 211     this->string_set_.insert(element);
 212   gold_assert(ins.second);
 213
 214   if (pkey != NULL)
 215     *pkey = k;
 216
 217   return ret;
 218 }
 219
 220 // Add a prefix of a string to a string pool.
 221
 222 template<typename Stringpool_char>
 223 const Stringpool_char*
 224 Stringpool_template<Stringpool_char>::add(const Stringpool_char* s, size_t len,
 225                                  Key* pkey)
 226 {
 227   // FIXME: This implementation should be rewritten when we rewrite
 228   // the hash table to avoid copying.
 229   std::basic_string<Stringpool_char> st(s, len);
 230   return this->add(st, pkey);
 231 }
 232
 233 template<typename Stringpool_char>
 234 const Stringpool_char*
 235 Stringpool_template<Stringpool_char>::find(const Stringpool_char* s,
 236                                            Key* pkey) const
 237 {
 238   typename String_set_type::const_iterator p = this->string_set_.find(s);
 239   if (p == this->string_set_.end())
 240     return NULL;
 241
 242   if (pkey != NULL)
 243     *pkey = p->second.first;
 244
 245   return p->first;
 246 }
 247
 248 // Comparison routine used when sorting into an ELF strtab.  We want
 249 // to sort this so that when one string is a suffix of another, we
 250 // always see the shorter string immediately after the longer string.
 251 // For example, we want to see these strings in this order:
 252 //   abcd
 253 //   cd
 254 //   d
 255 // When strings are not suffixes, we don't care what order they are
 256 // in, but we need to ensure that suffixes wind up next to each other.
 257 // So we do a reversed lexicographic sort on the reversed string.
 258
 259 template<typename Stringpool_char>
 260 bool
 261 Stringpool_template<Stringpool_char>::Stringpool_sort_comparison::operator()(
 262   const Stringpool_sort_info& sort_info1,
 263   const Stringpool_sort_info& sort_info2) const
 264 {
 265   const Stringpool_char* s1 = sort_info1.it->first;
 266   const Stringpool_char* s2 = sort_info2.it->first;
 267   const size_t len1 = sort_info1.string_length;
 268   const size_t len2 = sort_info2.string_length;
 269   const size_t minlen = len1 < len2 ? len1 : len2;
 270   const Stringpool_char* p1 = s1 + len1 - 1;
 271   const Stringpool_char* p2 = s2 + len2 - 1;
 272   for (size_t i = minlen; i > 0; --i, --p1, --p2)
 273     {
 274       if (*p1 != *p2)
 275         return *p1 > *p2;
 276     }
 277   return len1 > len2;
 278 }
 279
 280 // Return whether s1 is a suffix of s2.
 281
 282 template<typename Stringpool_char>
 283 bool
 284 Stringpool_template<Stringpool_char>::is_suffix(const Stringpool_char* s1,
 285                                                 size_t len1,
 286                                                 const Stringpool_char* s2,
 287                                                 size_t len2)
 288 {
 289   if (len1 > len2)
 290     return false;
 291   return memcmp(s1, s2 + len2 - len1, len1 * sizeof(Stringpool_char)) == 0;
 292 }
 293
 294 // Turn the stringpool into an ELF strtab: determine the offsets of
 295 // each string in the table.
 296
 297 template<typename Stringpool_char>
 298 void
 299 Stringpool_template<Stringpool_char>::set_string_offsets()
 300 {
 301   if (this->strtab_size_ != 0)
 302     {
 303       // We've already computed the offsets.
 304       return;
 305     }
 306
 307   const size_t charsize = sizeof(Stringpool_char);
 308
 309   // Offset 0 may be reserved for the empty string.
 310   off_t offset = this->zero_null_ ? charsize : 0;
 311
 312   // Sorting to find suffixes can take over 25% of the total CPU time
 313   // used by the linker.  Since it's merely an optimization to reduce
 314   // the strtab size, and gives a relatively small benefit (it's
 315   // typically rare for a symbol to be a suffix of another), we only
 316   // take the time to sort when the user asks for heavy optimization.
 317   if (parameters->optimization_level() < 2)
 318     {
 319       for (typename String_set_type::iterator curr = this->string_set_.begin();
 320            curr != this->string_set_.end();
 321            curr++)
 322         {
 323           if (this->zero_null_ && curr->first[0] == 0)
 324             curr->second.second = 0;
 325           else
 326             {
 327               curr->second.second = offset;
 328               offset += (string_length(curr->first) + 1) * charsize;
 329             }
 330         }
 331     }
 332   else
 333     {
 334       size_t count = this->string_set_.size();
 335
 336       std::vector<Stringpool_sort_info> v;
 337       v.reserve(count);
 338
 339       for (typename String_set_type::iterator p = this->string_set_.begin();
 340            p != this->string_set_.end();
 341            ++p)
 342         v.push_back(Stringpool_sort_info(p, string_length(p->first)));
 343
 344       std::sort(v.begin(), v.end(), Stringpool_sort_comparison());
 345
 346       for (typename std::vector<Stringpool_sort_info>::iterator last = v.end(),
 347              curr = v.begin();
 348            curr != v.end();
 349            last = curr++)
 350         {
 351           if (this->zero_null_ && curr->it->first[0] == 0)
 352             curr->it->second.second = 0;
 353           else if (last != v.end()
 354                    && is_suffix(curr->it->first, curr->string_length,
 355                                 last->it->first, last->string_length))
 356             curr->it->second.second = (last->it->second.second
 357                                        + ((last->string_length
 358                                            - curr->string_length)
 359                                           * charsize));
 360           else
 361             {
 362               curr->it->second.second = offset;
 363               offset += (curr->string_length + 1) * charsize;
 364             }
 365         }
 366     }
 367
 368   this->strtab_size_ = offset;
 369 }
 370
 371 // Get the offset of a string in the ELF strtab.  The string must
 372 // exist.
 373
 374 template<typename Stringpool_char>
 375 off_t
 376 Stringpool_template<Stringpool_char>::get_offset(const Stringpool_char* s)
 377   const
 378 {
 379   gold_assert(this->strtab_size_ != 0);
 380   typename String_set_type::const_iterator p = this->string_set_.find(s);
 381   if (p != this->string_set_.end())
 382     return p->second.second;
 383   gold_unreachable();
 384 }
 385
 386 // Write the ELF strtab into the output file at the specified offset.
 387
 388 template<typename Stringpool_char>
 389 void
 390 Stringpool_template<Stringpool_char>::write(Output_file* of, off_t offset)
 391 {
 392   gold_assert(this->strtab_size_ != 0);
 393   unsigned char* viewu = of->get_output_view(offset, this->strtab_size_);
 394   char* view = reinterpret_cast<char*>(viewu);
 395   if (this->zero_null_)
 396     view[0] = '\0';
 397   for (typename String_set_type::const_iterator p = this->string_set_.begin();
 398        p != this->string_set_.end();
 399        ++p)
 400     memcpy(view + p->second.second, p->first,
 401            (string_length(p->first) + 1) * sizeof(Stringpool_char));
 402   of->write_output_view(offset, this->strtab_size_, viewu);
 403 }
 404
 405 // Instantiate the templates we need.
 406
 407 template
 408 class Stringpool_template<char>;
 409
 410 template
 411 class Stringpool_template<uint16_t>;
 412
 413 template
 414 class Stringpool_template<uint32_t>;
 415
 416 } // End namespace gold.