Commit | Line | Data |
---|---|---|
14bfc3f5 ILT |
1 | // stringpool.cc -- a string pool for gold |
2 | ||
3 | #include "gold.h" | |
4 | ||
5 | #include <cassert> | |
6 | #include <cstring> | |
61ba1cf9 ILT |
7 | #include <algorithm> |
8 | #include <vector> | |
14bfc3f5 | 9 | |
61ba1cf9 | 10 | #include "output.h" |
14bfc3f5 ILT |
11 | #include "stringpool.h" |
12 | ||
13 | namespace gold | |
14 | { | |
15 | ||
16 | Stringpool::Stringpool() | |
61ba1cf9 | 17 | : string_set_(), strings_(), strtab_size_(0) |
14bfc3f5 ILT |
18 | { |
19 | } | |
20 | ||
21 | Stringpool::~Stringpool() | |
22 | { | |
61ba1cf9 | 23 | for (std::list<Stringdata*>::iterator p = this->strings_.begin(); |
14bfc3f5 ILT |
24 | p != this->strings_.end(); |
25 | ++p) | |
26 | delete[] reinterpret_cast<char*>(*p); | |
27 | } | |
28 | ||
29 | // Hash function. | |
30 | ||
31 | size_t | |
32 | Stringpool::Stringpool_hash::operator()(const char* s) const | |
33 | { | |
34 | // Fowler/Noll/Vo (FNV) hash (type FNV-1a). | |
35 | if (sizeof(size_t) == 8) | |
36 | { | |
b3168e9d | 37 | size_t result = static_cast<size_t>(14695981039346656037ULL); |
14bfc3f5 ILT |
38 | while (*s != '\0') |
39 | { | |
40 | result &= (size_t) *s++; | |
41 | result *= 1099511628211ULL; | |
42 | } | |
43 | return result; | |
44 | } | |
45 | else | |
46 | { | |
47 | size_t result = 2166136261UL; | |
48 | while (*s != '\0') | |
49 | { | |
50 | result ^= (size_t) *s++; | |
51 | result *= 16777619UL; | |
52 | } | |
53 | return result; | |
54 | } | |
55 | } | |
56 | ||
57 | // Add a string to the list of canonical strings. Return a pointer to | |
58 | // the canonical string. | |
59 | ||
60 | const char* | |
61 | Stringpool::add_string(const char* s) | |
62 | { | |
63 | const size_t buffer_size = 1000; | |
64 | size_t len = strlen(s); | |
65 | ||
66 | size_t alc; | |
67 | bool front = true; | |
68 | if (len >= buffer_size) | |
69 | { | |
61ba1cf9 | 70 | alc = sizeof(Stringdata) + len; |
14bfc3f5 ILT |
71 | front = false; |
72 | } | |
73 | else if (this->strings_.empty()) | |
61ba1cf9 | 74 | alc = sizeof(Stringdata) + buffer_size; |
14bfc3f5 ILT |
75 | else |
76 | { | |
61ba1cf9 | 77 | Stringdata *psd = this->strings_.front(); |
14bfc3f5 | 78 | if (len >= psd->alc - psd->len) |
61ba1cf9 | 79 | alc = sizeof(Stringdata) + buffer_size; |
14bfc3f5 ILT |
80 | else |
81 | { | |
82 | char* ret = psd->data + psd->len; | |
83 | memcpy(ret, s, len + 1); | |
84 | psd->len += len + 1; | |
85 | return ret; | |
86 | } | |
87 | } | |
88 | ||
61ba1cf9 ILT |
89 | Stringdata *psd = reinterpret_cast<Stringdata*>(new char[alc]); |
90 | psd->alc = alc - sizeof(Stringdata); | |
14bfc3f5 ILT |
91 | memcpy(psd->data, s, len + 1); |
92 | psd->len = len + 1; | |
93 | if (front) | |
94 | this->strings_.push_front(psd); | |
95 | else | |
96 | this->strings_.push_back(psd); | |
97 | return psd->data; | |
98 | } | |
99 | ||
100 | // Add a string to a string pool. | |
101 | ||
102 | const char* | |
103 | Stringpool::add(const char* s) | |
104 | { | |
105 | // FIXME: This will look up the entry twice in the hash table. The | |
106 | // problem is that we can't insert S before we canonicalize it. I | |
a2fb1b05 | 107 | // don't think there is a way to handle this correctly with |
61ba1cf9 | 108 | // unordered_map, so this should be replaced with custom code to do |
14bfc3f5 ILT |
109 | // what we need, which is to return the empty slot. |
110 | ||
111 | String_set_type::const_iterator p = this->string_set_.find(s); | |
112 | if (p != this->string_set_.end()) | |
61ba1cf9 | 113 | return p->first; |
14bfc3f5 ILT |
114 | |
115 | const char* ret = this->add_string(s); | |
61ba1cf9 | 116 | std::pair<const char*, off_t> val(ret, 0); |
14bfc3f5 | 117 | std::pair<String_set_type::iterator, bool> ins = |
61ba1cf9 | 118 | this->string_set_.insert(val); |
14bfc3f5 ILT |
119 | assert(ins.second); |
120 | return ret; | |
121 | } | |
122 | ||
123 | // Add a prefix of a string to a string pool. | |
124 | ||
125 | const char* | |
126 | Stringpool::add(const char* s, size_t len) | |
127 | { | |
128 | // FIXME: This implementation should be rewritten when we rewrite | |
129 | // the hash table to avoid copying. | |
130 | std::string st(s, len); | |
131 | return this->add(st); | |
132 | } | |
133 | ||
61ba1cf9 ILT |
134 | const char* |
135 | Stringpool::find(const char* s) const | |
136 | { | |
137 | String_set_type::const_iterator p = this->string_set_.find(s); | |
138 | if (p == this->string_set_.end()) | |
139 | return NULL; | |
140 | return p->first; | |
141 | } | |
142 | ||
143 | // Comparison routine used when sorting into an ELF strtab. We want | |
144 | // to sort this so that when one string is a suffix of another, we | |
145 | // always see the shorter string immediately after the longer string. | |
146 | // For example, we want to see these strings in this order: | |
147 | // abcd | |
148 | // cd | |
149 | // d | |
150 | // When strings are not suffixes, we don't care what order they are | |
151 | // in, but we need to ensure that suffixes wind up next to each other. | |
152 | // So we do a reversed lexicographic sort on the reversed string. | |
153 | ||
154 | bool | |
155 | Stringpool::Stringpool_sort_comparison::operator()( | |
156 | String_set_type::iterator it1, | |
157 | String_set_type::iterator it2) const | |
158 | { | |
159 | const char* s1 = it1->first; | |
160 | const char* s2 = it2->first; | |
161 | int len1 = strlen(s1); | |
162 | int len2 = strlen(s2); | |
163 | int minlen = len1 < len2 ? len1 : len2; | |
164 | const char* p1 = s1 + len1 - 1; | |
165 | const char* p2 = s2 + len2 - 1; | |
166 | for (int i = minlen - 1; i >= 0; --i, --p1, --p2) | |
167 | { | |
168 | if (*p1 != *p2) | |
169 | return *p1 > *p2; | |
170 | } | |
171 | return len1 > len2; | |
172 | } | |
173 | ||
174 | // Return whether s1 is a suffix of s2. | |
175 | ||
176 | bool | |
177 | Stringpool::is_suffix(const char* s1, const char* s2) | |
178 | { | |
179 | size_t len1 = strlen(s1); | |
180 | size_t len2 = strlen(s2); | |
181 | if (len1 > len2) | |
182 | return false; | |
183 | return strcmp(s1, s2 + len2 - len1) == 0; | |
184 | } | |
185 | ||
186 | // Turn the stringpool into an ELF strtab: determine the offsets of | |
187 | // each string in the table. | |
188 | ||
189 | void | |
190 | Stringpool::set_string_offsets() | |
191 | { | |
192 | size_t count = this->string_set_.size(); | |
193 | ||
194 | std::vector<String_set_type::iterator> v; | |
195 | v.reserve(count); | |
196 | ||
197 | for (String_set_type::iterator p = this->string_set_.begin(); | |
198 | p != this->string_set_.end(); | |
199 | ++p) | |
200 | v.push_back(p); | |
201 | ||
202 | std::sort(v.begin(), v.end(), Stringpool_sort_comparison()); | |
203 | ||
204 | // Offset 0 is reserved for the empty string. | |
205 | off_t offset = 1; | |
206 | for (size_t i = 0; i < count; ++i) | |
207 | { | |
208 | if (v[i]->first[0] == '\0') | |
209 | v[i]->second = 0; | |
210 | else if (i > 0 && Stringpool::is_suffix(v[i]->first, v[i - 1]->first)) | |
211 | v[i]->second = (v[i - 1]->second | |
212 | + strlen(v[i - 1]->first) | |
213 | - strlen(v[i]->first)); | |
214 | else | |
215 | { | |
216 | v[i]->second = offset; | |
217 | offset += strlen(v[i]->first) + 1; | |
218 | } | |
219 | } | |
220 | ||
221 | this->strtab_size_ = offset; | |
222 | } | |
223 | ||
224 | // Get the offset of a string in the ELF strtab. The string must | |
225 | // exist. | |
226 | ||
227 | off_t | |
228 | Stringpool::get_offset(const char* s) const | |
229 | { | |
230 | String_set_type::const_iterator p = this->string_set_.find(s); | |
231 | if (p != this->string_set_.end()) | |
232 | return p->second; | |
233 | abort(); | |
234 | } | |
235 | ||
236 | // Write the ELF strtab into the output file at the specified offset. | |
237 | ||
238 | void | |
239 | Stringpool::write(Output_file* of, off_t offset) | |
240 | { | |
241 | unsigned char* viewu = of->get_output_view(offset, this->strtab_size_); | |
242 | char* view = reinterpret_cast<char*>(viewu); | |
243 | view[0] = '\0'; | |
244 | for (String_set_type::const_iterator p = this->string_set_.begin(); | |
245 | p != this->string_set_.end(); | |
246 | ++p) | |
247 | strcpy(view + p->second, p->first); | |
248 | of->write_output_view(offset, this->strtab_size_, viewu); | |
249 | } | |
250 | ||
14bfc3f5 | 251 | } // End namespace gold. |