{{ message }}
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathshared_string.cpp
More file actions
269 lines (251 loc) · 9.58 KB
/
Copy pathshared_string.cpp
File metadata and controls
269 lines (251 loc) · 9.58 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
// Copyright (c) 2018 Dmitry Sokolov
//
// Distributed under the Boost Software License, Version 1.0. (See accompanying
// file LICENSE or copy at http://www.boost.org/LICENSE_1_0.txt)
//
#include "shared_string.hpp"
namespace utils {
namespace bits {
inline uint32_t reverse_bit_order(uint32_t x) {
x = ((x >> 1) & 0x55555555) | ((x & 0x55555555) << 1);
x = ((x >> 2) & 0x33333333) | ((x & 0x33333333) << 2);
x = ((x >> 4) & 0x0F0F0F0F) | ((x & 0x0F0F0F0F) << 4);
x = ((x >> 8) & 0x00FF00FF) | ((x & 0x00FF00FF) << 8);
return (x >> 16) | (x << 16);
}
} // namespace bits
literal_dictionary::literal_dictionary() : mem_{pmr::get_default_resource()} {
}
// Dictionary singleton accessor.
literal_dictionary& literal_dictionary::global() {
static literal_dictionary inst;
return inst;
}
literal_dictionary::string literal_dictionary::add(string_view str) {
return string(get_node(str));
}
// Add string to global dictionary.
const char* literal_dictionary::add_global_str(string_view str) {
return global().get_node(str)->data();
}
literal_dictionary::string literal_dictionary::add_global(string_view str) {
return string(global().get_node(str));
}
literal_dictionary::~literal_dictionary() {
// Free hashtable memory.
for(auto& segment : table_segments_) {
if(segment.data == nullptr)
break;
auto alloc_size = (segment.table_size - segment.prev_table_size)
* sizeof(literal_dictionary_node::ptr);
mem_->deallocate(
segment.data, alloc_size, std::alignment_of<node_ptr_t>::value);
}
// Free dictionary memory.
auto* dict_page = allocated_pages_;
while(dict_page) {
auto* next = dict_page->next;
mem_->deallocate(
dict_page, allocate_chunk_size,
std::alignment_of<node_ptr_t>::value);
dict_page = next;
}
}
// Dictionary node search/add method.
const literal_dictionary_node* literal_dictionary::get_node(string_view str) {
if(str.empty())
return &empty_node;
auto hash = static_cast<uint32_t>(std::hash<string_view>()(str));
const auto* segment = current_segment_.load();
if(segment != nullptr) {
// Fast lock-free search.
auto table_size = segment->table_size;
auto bucket_num = hash % table_size;
while(bucket_num < segment->prev_table_size)
--segment;
auto segment_index = bucket_num - segment->prev_table_size;
auto* node = segment->data[segment_index].load();
while(node != nullptr) {
if(node->hash == hash && node->str() == str)
return node;
if((node->hash % table_size) != bucket_num)
break;
node = node->next.load();
}
}
// If node has not been found, add new one.
return add_node(hash, str);
}
// Add new dictionary entry.
literal_dictionary_node* literal_dictionary::add_node(
uint32_t hash, string_view str) {
if(str.size() > max_string_size)
throw std::runtime_error("dictionary dict_string to big");
// Insertion should be mutually exclusive otherwise
// duplicate allocations of the same dict_string may occur.
std::lock_guard<std::mutex> lock(mtx_);
if(current_version_ == 0 && table_segments_[0].table_size == 0)
init_first_table_segment();
// Increase table when load factor = 1.
else if(
table_segments_[current_version_].table_size == size_
&& current_version_ < table_segment_count - 1)
init_next_table_segment();
// Calculate bucket segment and segment position.
auto* segment = &table_segments_[current_version_];
auto table_size = segment->table_size;
auto bucket_num = hash % table_size;
while(bucket_num < segment->prev_table_size)
--segment;
auto segment_index = bucket_num - segment->prev_table_size;
auto& bucket = segment->data[segment_index];
// Find bucket insertion point (using reverse bit order).
literal_dictionary_node* node = bucket.load();
literal_dictionary_node* prev = nullptr;
literal_dictionary_node* next = nullptr;
if(node != nullptr) {
auto shah = bits::reverse_bit_order(hash);
do {
// Check for bucket end.
if((node->hash % table_size) != bucket_num)
break;
// Check equal, may be already inserted by concurrent thread.
if(node->hash == hash && node->str() == str)
return node;
// Apply revers-bit ordering.
if(next == nullptr) {
auto node_shah = bits::reverse_bit_order(node->hash);
if(shah < node_shah)
next = node;
else
prev = node;
}
node = node->next.load();
} while(node != nullptr);
}
// Just allocate new node and add to list.
auto* new_node = allocate_node(hash, str);
new_node->next = next;
++size_;
if(prev != nullptr)
prev->next.store(new_node);
else
bucket.store(new_node);
return new_node;
}
// Allocate new dictionary node.
literal_dictionary_node* literal_dictionary::allocate_node(
uint32_t hash, string_view str) {
constexpr size_t node_align =
std::alignment_of<literal_dictionary_node>::value;
size_t node_size = sizeof(literal_dictionary_node) + str.size() + 1;
if(current_page_ != nullptr)
current_page_ =
std::align(node_align, node_size, current_page_, remain_page_size_);
// Allocate new page if no more space left.
if(current_page_ == nullptr) {
auto* page = static_cast<dict_page_t*>(mem_->allocate(
allocate_chunk_size, std::alignment_of<dict_page_t>::value));
total_allocated_size_ += allocate_chunk_size;
page->next = allocated_pages_;
allocated_pages_ = page;
current_page_ = page + 1;
remain_page_size_ = allocate_chunk_size - sizeof(dict_page_t);
current_page_ =
std::align(node_align, node_size, current_page_, remain_page_size_);
}
// Construct node and copy dict_string.
auto* node = reinterpret_cast<literal_dictionary_node*>(current_page_);
current_page_ = static_cast<char*>(current_page_) + node_size;
remain_page_size_ -= node_size;
new(node) literal_dictionary_node();
node->hash = hash;
node->size = static_cast<uint32_t>(str.size());
char* data = reinterpret_cast<char*>(node + 1);
std::char_traits<char>::copy(data, str.data(), str.size());
data[str.size()] = '\0';
return node;
}
// Allocate data for new table segment (uninitialized).
void literal_dictionary::allocate_table_segment(size_t segment_num) {
auto& segment = table_segments_[segment_num];
if(segment_num == 0)
segment.table_size = table_initial_size;
else {
segment.prev_table_size = table_initial_size << (segment_num - 1);
segment.table_size = segment.prev_table_size << 1;
}
auto alloc_size = (segment.table_size - segment.prev_table_size)
* sizeof(literal_dictionary_node::ptr);
segment.data = static_cast<node_ptr_t*>(
mem_->allocate(alloc_size, std::alignment_of<node_ptr_t>::value));
total_allocated_size_ += alloc_size;
}
// Allocate and fill first table segment.
void literal_dictionary::init_first_table_segment() {
allocate_table_segment(0);
std::uninitialized_fill_n(
table_segments_[0].data, table_initial_size, nullptr);
current_segment_.store(&table_segments_[0]);
}
// Allocate and fill new table segment.
void literal_dictionary::init_next_table_segment() {
allocate_table_segment(current_version_ + 1);
auto& new_segment = table_segments_[current_version_ + 1];
auto old_table_size = table_segments_[current_version_].table_size;
auto new_table_size = new_segment.table_size;
// Iterate over all segments and split their buckets
// saving right part in new segment.
auto* segment = &table_segments_[0];
for(size_t i = 0; i < old_table_size; ++i) {
if(i >= segment->table_size)
++segment;
auto segment_index = i - segment->prev_table_size;
literal_dictionary_node* right_bucket =
segment->data[segment_index].load();
// Find list split position.
while(right_bucket != nullptr
&& (right_bucket->hash % old_table_size) == i
&& (right_bucket->hash % new_table_size) == i)
right_bucket = right_bucket->next.load();
new(&new_segment.data[i]) node_ptr_t(right_bucket);
}
++current_version_;
current_segment_.store(&new_segment);
}
literal_dictionary::iterator& literal_dictionary::iterator::operator++() {
if(dict_ == nullptr)
return *this;
if(node_ == nullptr) {
bucket_position_ = 0;
position_ = 0;
segment_ = &dict_->table_segments_[0];
last_segment_ = dict_->current_segment_.load();
if(last_segment_ == nullptr)
return *this;
}
else {
node_ = node_->next.load();
if(node_ != nullptr
&& node_->hash % segment_->table_size == position_) {
++bucket_position_;
return *this;
}
node_ = nullptr;
bucket_position_ = 0;
++position_;
}
while(node_ == nullptr) {
if(position_ >= segment_->table_size) {
if(segment_ == last_segment_)
return *this;
++segment_;
}
node_ = segment_->data[position_ - segment_->prev_table_size].load();
if(node_ != nullptr && node_->hash % segment_->table_size == position_)
return *this;
++position_;
}
return *this;
}
} // namespace utils
You can’t perform that action at this time.