1 // stringpool.cc -- a string pool for gold
10 #include "parameters.h"
11 #include "stringpool.h"
16 template<typename Stringpool_char>
17 Stringpool_template<Stringpool_char>::Stringpool_template()
18 : string_set_(), strings_(), strtab_size_(0), next_index_(1),
23 template<typename Stringpool_char>
24 Stringpool_template<Stringpool_char>::~Stringpool_template()
26 for (typename std::list<Stringdata*>::iterator p = this->strings_.begin();
27 p != this->strings_.end();
29 delete[] reinterpret_cast<char*>(*p);
32 // Return the length of a string of arbitrary character type.
34 template<typename Stringpool_char>
36 Stringpool_template<Stringpool_char>::string_length(const Stringpool_char* p)
44 // Specialize string_length for char. Maybe we could just use
45 // std::char_traits<>::length?
49 Stringpool_template<char>::string_length(const char* p)
54 // Equality comparison function.
56 template<typename Stringpool_char>
58 Stringpool_template<Stringpool_char>::Stringpool_eq::operator()(
59 const Stringpool_char* s1,
60 const Stringpool_char* s2) const
68 // Specialize equality comparison for char.
72 Stringpool_template<char>::Stringpool_eq::operator()(const char* s1,
75 return strcmp(s1, s2) == 0;
80 template<typename Stringpool_char>
82 Stringpool_template<Stringpool_char>::Stringpool_hash::operator()(
83 const Stringpool_char* s) const
85 // Fowler/Noll/Vo (FNV) hash (type FNV-1a).
86 if (sizeof(size_t) == 8)
88 size_t result = static_cast<size_t>(14695981039346656037ULL);
91 const char* p = reinterpret_cast<const char*>(s);
92 for (size_t i = 0; i < sizeof(Stringpool_char); ++i)
94 result &= (size_t) *p++;
95 result *= 1099511628211ULL;
103 size_t result = 2166136261UL;
106 const char* p = reinterpret_cast<const char*>(s);
107 for (size_t i = 0; i < sizeof(Stringpool_char); ++i)
109 result ^= (size_t) *p++;
110 result *= 16777619UL;
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.
121 template<typename Stringpool_char>
122 const Stringpool_char*
123 Stringpool_template<Stringpool_char>::add_string(const Stringpool_char* s,
126 // We are in trouble if we've already computed the string offsets.
127 gold_assert(this->strtab_size_ == 0);
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
133 const size_t key_mult = 1024;
134 gold_assert(key_mult >= buffer_size);
136 size_t len = (string_length(s) + 1) * sizeof(Stringpool_char);
140 if (len > buffer_size)
142 alc = sizeof(Stringdata) + len;
145 else if (this->strings_.empty())
146 alc = sizeof(Stringdata) + buffer_size;
149 Stringdata *psd = this->strings_.front();
150 if (len > psd->alc - psd->len)
151 alc = sizeof(Stringdata) + buffer_size;
154 char* ret = psd->data + psd->len;
158 *pkey = psd->index * key_mult + psd->len;
162 return reinterpret_cast<const Stringpool_char*>(ret);
166 Stringdata *psd = reinterpret_cast<Stringdata*>(new char[alc]);
167 psd->alc = alc - sizeof(Stringdata);
168 memcpy(psd->data, s, len);
170 psd->index = this->next_index_;
174 *pkey = psd->index * key_mult;
177 this->strings_.push_front(psd);
179 this->strings_.push_back(psd);
181 return reinterpret_cast<const Stringpool_char*>(psd->data);
184 // Add a string to a string pool.
186 template<typename Stringpool_char>
187 const Stringpool_char*
188 Stringpool_template<Stringpool_char>::add(const Stringpool_char* s, Key* pkey)
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.
196 typename String_set_type::const_iterator p = this->string_set_.find(s);
197 if (p != this->string_set_.end())
200 *pkey = p->second.first;
205 const Stringpool_char* ret = this->add_string(s, &k);
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);
220 // Add a prefix of a string to a string pool.
222 template<typename Stringpool_char>
223 const Stringpool_char*
224 Stringpool_template<Stringpool_char>::add(const Stringpool_char* s, size_t len,
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);
233 template<typename Stringpool_char>
234 const Stringpool_char*
235 Stringpool_template<Stringpool_char>::find(const Stringpool_char* s,
238 typename String_set_type::const_iterator p = this->string_set_.find(s);
239 if (p == this->string_set_.end())
243 *pkey = p->second.first;
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:
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.
259 template<typename Stringpool_char>
261 Stringpool_template<Stringpool_char>::Stringpool_sort_comparison::operator()(
262 const Stringpool_sort_info& sort_info1,
263 const Stringpool_sort_info& sort_info2) const
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)
280 // Return whether s1 is a suffix of s2.
282 template<typename Stringpool_char>
284 Stringpool_template<Stringpool_char>::is_suffix(const Stringpool_char* s1,
286 const Stringpool_char* s2,
291 return memcmp(s1, s2 + len2 - len1, len1 * sizeof(Stringpool_char)) == 0;
294 // Turn the stringpool into an ELF strtab: determine the offsets of
295 // each string in the table.
297 template<typename Stringpool_char>
299 Stringpool_template<Stringpool_char>::set_string_offsets()
301 if (this->strtab_size_ != 0)
303 // We've already computed the offsets.
307 const size_t charsize = sizeof(Stringpool_char);
309 // Offset 0 may be reserved for the empty string.
310 off_t offset = this->zero_null_ ? charsize : 0;
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)
319 for (typename String_set_type::iterator curr = this->string_set_.begin();
320 curr != this->string_set_.end();
323 if (this->zero_null_ && curr->first[0] == 0)
324 curr->second.second = 0;
327 curr->second.second = offset;
328 offset += (string_length(curr->first) + 1) * charsize;
334 size_t count = this->string_set_.size();
336 std::vector<Stringpool_sort_info> v;
339 for (typename String_set_type::iterator p = this->string_set_.begin();
340 p != this->string_set_.end();
342 v.push_back(Stringpool_sort_info(p, string_length(p->first)));
344 std::sort(v.begin(), v.end(), Stringpool_sort_comparison());
346 for (typename std::vector<Stringpool_sort_info>::iterator last = v.end(),
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)
362 curr->it->second.second = offset;
363 offset += (curr->string_length + 1) * charsize;
368 this->strtab_size_ = offset;
371 // Get the offset of a string in the ELF strtab. The string must
374 template<typename Stringpool_char>
376 Stringpool_template<Stringpool_char>::get_offset(const Stringpool_char* s)
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;
386 // Write the ELF strtab into the output file at the specified offset.
388 template<typename Stringpool_char>
390 Stringpool_template<Stringpool_char>::write(Output_file* of, off_t offset)
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_)
397 for (typename String_set_type::const_iterator p = this->string_set_.begin();
398 p != this->string_set_.end();
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);
405 // Instantiate the templates we need.
408 class Stringpool_template<char>;
411 class Stringpool_template<uint16_t>;
414 class Stringpool_template<uint32_t>;
416 } // End namespace gold.