| Directory: | cvmfs/ |
|---|---|
| File: | cvmfs/smallhash.h |
| Date: | 2026-03-15 02:35:27 |
| Exec | Total | Coverage | |
|---|---|---|---|
| Lines: | 259 | 263 | 98.5% |
| Branches: | 88 | 110 | 80.0% |
| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | /** | ||
| 2 | * This file is part of the CernVM File System. | ||
| 3 | */ | ||
| 4 | |||
| 5 | #ifndef CVMFS_SMALLHASH_H_ | ||
| 6 | #define CVMFS_SMALLHASH_H_ | ||
| 7 | |||
| 8 | #ifndef __STDC_FORMAT_MACROS | ||
| 9 | // NOLINTNEXTLINE | ||
| 10 | #define __STDC_FORMAT_MACROS | ||
| 11 | #endif | ||
| 12 | |||
| 13 | #include "duplex_testing.h" | ||
| 14 | #include <inttypes.h> | ||
| 15 | #include <pthread.h> | ||
| 16 | #include <stdint.h> | ||
| 17 | |||
| 18 | #include <algorithm> | ||
| 19 | #include <cassert> | ||
| 20 | #include <cstdlib> | ||
| 21 | #include <new> | ||
| 22 | |||
| 23 | #include "util/atomic.h" | ||
| 24 | #include "util/murmur.hxx" | ||
| 25 | #include "util/prng.h" | ||
| 26 | #include "util/smalloc.h" | ||
| 27 | |||
| 28 | /** | ||
| 29 | * Hash table with linear probing as collision resolution. Works only for | ||
| 30 | * a fixed (maximum) number of elements, i.e. no resizing. Load factor fixed | ||
| 31 | * to 0.7. | ||
| 32 | */ | ||
| 33 | template<class Key, class Value, class Derived> | ||
| 34 | class SmallHashBase { | ||
| 35 | FRIEND_TEST(T_Smallhash, InsertAndCopyMd5Slow); | ||
| 36 | |||
| 37 | public: | ||
| 38 | static const double kLoadFactor; // mainly useless for the dynamic version | ||
| 39 | static const double kThresholdGrow; // only used for resizable version | ||
| 40 | static const double kThresholdShrink; // only used for resizable version | ||
| 41 | |||
| 42 | 233136 | SmallHashBase() { | |
| 43 | 233136 | keys_ = NULL; | |
| 44 | 233136 | values_ = NULL; | |
| 45 | 233136 | hasher_ = NULL; | |
| 46 | 233136 | bytes_allocated_ = 0; | |
| 47 | 233136 | num_collisions_ = 0; | |
| 48 | 233136 | max_collisions_ = 0; | |
| 49 | |||
| 50 | // Properly initialized by Init() | ||
| 51 | 233136 | capacity_ = 0; | |
| 52 | 233136 | initial_capacity_ = 0; | |
| 53 | 233136 | size_ = 0; | |
| 54 | 233136 | } | |
| 55 | |||
| 56 | 232998 | ~SmallHashBase() { DeallocMemory(keys_, values_, capacity_); } | |
| 57 | |||
| 58 | 233064 | void Init(uint32_t expected_size, Key empty, | |
| 59 | uint32_t (*hasher)(const Key &key)) { | ||
| 60 | 233064 | hasher_ = hasher; | |
| 61 | 233064 | empty_key_ = empty; | |
| 62 | 233064 | capacity_ = static_cast<uint32_t>(static_cast<double>(expected_size) | |
| 63 | 233064 | / kLoadFactor); | |
| 64 | 233064 | initial_capacity_ = capacity_; | |
| 65 | 233064 | static_cast<Derived *>(this)->SetThresholds(); // No-op for fixed size | |
| 66 | 233064 | AllocMemory(); | |
| 67 | 233064 | this->DoClear(false); | |
| 68 | 233064 | } | |
| 69 | |||
| 70 | 119035202 | bool Lookup(const Key &key, Value *value) const { | |
| 71 | uint32_t bucket; | ||
| 72 | uint32_t collisions; | ||
| 73 |
1/2✓ Branch 1 taken 119096400 times.
✗ Branch 2 not taken.
|
119035202 | const bool found = DoLookup(key, &bucket, &collisions); |
| 74 |
2/2✓ Branch 0 taken 98593492 times.
✓ Branch 1 taken 20502908 times.
|
119231920 | if (found) |
| 75 |
1/2✓ Branch 1 taken 15786 times.
✗ Branch 2 not taken.
|
98638644 | *value = values_[bucket]; |
| 76 | 119231920 | return found; | |
| 77 | } | ||
| 78 | |||
| 79 | /** | ||
| 80 | * Returns both the key and the value. That is useful if Key's equality | ||
| 81 | * operator implements an equivalence relation on Key. In this case, LookupEx | ||
| 82 | * returns the key representing the equivalence class that has been used | ||
| 83 | * during Insert(). | ||
| 84 | * Used to return a glue::InodeEx element when looking for an inode. | ||
| 85 | */ | ||
| 86 | 588 | bool LookupEx(Key *key, Value *value) const { | |
| 87 | 588 | uint32_t bucket = ScaleHash(*key); | |
| 88 |
2/2✓ Branch 1 taken 196 times.
✓ Branch 2 taken 98 times.
|
588 | while (!(keys_[bucket] == empty_key_)) { |
| 89 |
1/2✓ Branch 1 taken 196 times.
✗ Branch 2 not taken.
|
392 | if (keys_[bucket] == *key) { |
| 90 | 392 | *key = keys_[bucket]; | |
| 91 | 392 | *value = values_[bucket]; | |
| 92 | 392 | return true; | |
| 93 | } | ||
| 94 | ✗ | bucket = (bucket + 1) % capacity_; | |
| 95 | } | ||
| 96 | 196 | return false; | |
| 97 | } | ||
| 98 | |||
| 99 | 7043441 | bool Contains(const Key &key) const { | |
| 100 | uint32_t bucket; | ||
| 101 | uint32_t collisions; | ||
| 102 |
1/2✓ Branch 1 taken 7042762 times.
✗ Branch 2 not taken.
|
7043441 | const bool found = DoLookup(key, &bucket, &collisions); |
| 103 | 7043441 | return found; | |
| 104 | } | ||
| 105 | |||
| 106 | 2049913155 | void Insert(const Key &key, const Value &value) { | |
| 107 | 2049913155 | static_cast<Derived *>(this)->Grow(); // No-op if fixed-size | |
| 108 | 2041014103 | const bool overwritten = DoInsert(key, value, true); | |
| 109 | 2041403270 | size_ += !overwritten; // size + 1 if the key was not yet in the map | |
| 110 | 2041403270 | } | |
| 111 | |||
| 112 | 119771441 | bool Erase(const Key &key) { | |
| 113 | uint32_t bucket; | ||
| 114 | uint32_t collisions; | ||
| 115 |
1/2✓ Branch 1 taken 116146222 times.
✗ Branch 2 not taken.
|
119771441 | const bool found = DoLookup(key, &bucket, &collisions); |
| 116 |
1/2✓ Branch 0 taken 116157100 times.
✗ Branch 1 not taken.
|
116151019 | if (found) { |
| 117 | 116161560 | keys_[bucket] = empty_key_; | |
| 118 | 116161560 | size_--; | |
| 119 | 116161560 | bucket = (bucket + 1) % capacity_; | |
| 120 |
3/3✓ Branch 0 taken 177361306 times.
✓ Branch 1 taken 115522918 times.
✓ Branch 2 taken 370984 times.
|
293259668 | while (!(keys_[bucket] == empty_key_)) { |
| 121 | 177423044 | Key rehash = keys_[bucket]; | |
| 122 | 177423044 | keys_[bucket] = empty_key_; | |
| 123 |
1/2✓ Branch 1 taken 177098108 times.
✗ Branch 2 not taken.
|
177423044 | DoInsert(rehash, values_[bucket], false); |
| 124 | 177098108 | bucket = (bucket + 1) % capacity_; | |
| 125 | } | ||
| 126 |
1/2✓ Branch 1 taken 117446004 times.
✗ Branch 2 not taken.
|
115836624 | static_cast<Derived *>(this)->Shrink(); // No-op if fixed-size |
| 127 | } | ||
| 128 | 117447373 | return found; | |
| 129 | } | ||
| 130 | |||
| 131 | 2383 | void Clear() { DoClear(true); } | |
| 132 | |||
| 133 | 5813 | uint64_t bytes_allocated() const { return bytes_allocated_; } | |
| 134 | 2148 | static double GetEntrySize() { | |
| 135 | 4296 | const double unit = sizeof(Key) + sizeof(Value); | |
| 136 | 4296 | return unit / kLoadFactor; | |
| 137 | } | ||
| 138 | |||
| 139 | void GetCollisionStats(uint64_t *num_collisions, | ||
| 140 | uint32_t *max_collisions) const { | ||
| 141 | *num_collisions = num_collisions_; | ||
| 142 | *max_collisions = max_collisions_; | ||
| 143 | } | ||
| 144 | |||
| 145 | // Careful with the direct access TODO: iterator | ||
| 146 | uint32_t capacity() const { return capacity_; } | ||
| 147 | 241124 | Key empty_key() const { return empty_key_; } | |
| 148 | 534166 | Key *keys() const { return keys_; } | |
| 149 | 637 | Value *values() const { return values_; } | |
| 150 | |||
| 151 | // Only needed by compat | ||
| 152 | ✗ | void SetHasher(uint32_t (*hasher)(const Key &key)) { hasher_ = hasher; } | |
| 153 | |||
| 154 | protected: | ||
| 155 | 2777979402 | uint32_t ScaleHash(const Key &key) const { | |
| 156 | 2777979402 | const double bucket = (static_cast<double>(hasher_(key)) | |
| 157 | 2769422058 | * static_cast<double>(capacity_) | |
| 158 | / static_cast<double>(static_cast<uint32_t>(-1))); | ||
| 159 | 2769422058 | return static_cast<uint32_t>(bucket) % capacity_; | |
| 160 | } | ||
| 161 | |||
| 162 | 395313 | void AllocMemory() { | |
| 163 | 395313 | keys_ = static_cast<Key *>(smmap(capacity_ * sizeof(Key))); | |
| 164 | 395385 | values_ = static_cast<Value *>(smmap(capacity_ * sizeof(Value))); | |
| 165 |
2/2✓ Branch 0 taken 1637299123 times.
✓ Branch 1 taken 199030 times.
|
3265274462 | for (uint32_t i = 0; i < capacity_; ++i) { |
| 166 | 3264879077 | /*keys_[i] =*/new (keys_ + i) Key(); | |
| 167 | } | ||
| 168 |
2/2✓ Branch 0 taken 1646762983 times.
✓ Branch 1 taken 179158 times.
|
3284162438 | for (uint32_t i = 0; i < capacity_; ++i) { |
| 169 | 3283806797 | /*values_[i] =*/new (values_ + i) Value(); | |
| 170 | } | ||
| 171 | 355641 | bytes_allocated_ = (sizeof(Key) + sizeof(Value)) * capacity_; | |
| 172 | 355641 | } | |
| 173 | |||
| 174 | 395319 | void DeallocMemory(Key *k, Value *v, uint32_t c) { | |
| 175 |
2/2✓ Branch 0 taken 1652730541 times.
✓ Branch 1 taken 198996 times.
|
3296139962 | for (uint32_t i = 0; i < c; ++i) { |
| 176 | 3295744643 | k[i].~Key(); | |
| 177 | } | ||
| 178 |
2/2✓ Branch 0 taken 1655301445 times.
✓ Branch 1 taken 198996 times.
|
3301281770 | for (uint32_t i = 0; i < c; ++i) { |
| 179 | 3300886451 | v[i].~Value(); | |
| 180 | } | ||
| 181 |
2/2✓ Branch 0 taken 198960 times.
✓ Branch 1 taken 36 times.
|
395319 | if (k) |
| 182 | 395247 | smunmap(k); | |
| 183 |
2/2✓ Branch 0 taken 198960 times.
✓ Branch 1 taken 36 times.
|
395319 | if (v) |
| 184 | 395247 | smunmap(v); | |
| 185 | 395319 | k = NULL; | |
| 186 | 395319 | v = NULL; | |
| 187 | 395319 | } | |
| 188 | |||
| 189 | // Returns true iff the key is overwritten | ||
| 190 | 2385817199 | bool DoInsert(const Key &key, const Value &value, | |
| 191 | const bool count_collisions) { | ||
| 192 | uint32_t bucket; | ||
| 193 | uint32_t collisions; | ||
| 194 |
1/2✓ Branch 1 taken 1214965774 times.
✗ Branch 2 not taken.
|
2385817199 | const bool overwritten = DoLookup(key, &bucket, &collisions); |
| 195 |
2/2✓ Branch 0 taken 1039549434 times.
✓ Branch 1 taken 175416340 times.
|
2383711837 | if (count_collisions) { |
| 196 | 2040152633 | num_collisions_ += collisions; | |
| 197 | 2040152633 | max_collisions_ = std::max(collisions, max_collisions_); | |
| 198 | } | ||
| 199 | 2388126299 | keys_[bucket] = key; | |
| 200 |
1/2✓ Branch 1 taken 7296006 times.
✗ Branch 2 not taken.
|
2388126299 | values_[bucket] = value; |
| 201 | 2388126299 | return overwritten; | |
| 202 | } | ||
| 203 | |||
| 204 | 2779923905 | bool DoLookup(const Key &key, uint32_t *bucket, uint32_t *collisions) const { | |
| 205 | 2779923905 | *bucket = ScaleHash(key); | |
| 206 | 2768551839 | *collisions = 0; | |
| 207 |
3/3✓ Branch 0 taken 5187543481 times.
✓ Branch 1 taken 1088764384 times.
✓ Branch 2 taken 228340505 times.
|
9119782017 | while (!(keys_[*bucket] == empty_key_)) { |
| 208 |
3/3✓ Branch 0 taken 232801740 times.
✓ Branch 1 taken 4962356291 times.
✓ Branch 2 taken 108998642 times.
|
6766720055 | if (keys_[*bucket] == key) |
| 209 | 415489877 | return true; | |
| 210 | 6351230178 | *bucket = (*bucket + 1) % capacity_; | |
| 211 | 6351230178 | (*collisions)++; | |
| 212 | } | ||
| 213 | 2353061962 | return false; | |
| 214 | } | ||
| 215 | |||
| 216 | 395403 | void DoClear(const bool reset_capacity) { | |
| 217 |
2/2✓ Branch 0 taken 2383 times.
✓ Branch 1 taken 196665 times.
|
395403 | if (reset_capacity) |
| 218 | 3966 | static_cast<Derived *>(this)->ResetCapacity(); // No-op if fixed-size | |
| 219 |
2/2✓ Branch 0 taken 1648374325 times.
✓ Branch 1 taken 199048 times.
|
3287400314 | for (uint32_t i = 0; i < capacity_; ++i) |
| 220 | 3287004911 | keys_[i] = empty_key_; | |
| 221 | 395403 | size_ = 0; | |
| 222 | 395403 | } | |
| 223 | |||
| 224 | // Methods for resizable version | ||
| 225 | void SetThresholds() { } | ||
| 226 | void Grow() { } | ||
| 227 | void Shrink() { } | ||
| 228 | void ResetCapacity() { } | ||
| 229 | |||
| 230 | // Separate key and value arrays for better locality | ||
| 231 | Key *keys_; | ||
| 232 | Value *values_; | ||
| 233 | uint32_t capacity_; | ||
| 234 | uint32_t initial_capacity_; | ||
| 235 | uint32_t size_; | ||
| 236 | uint32_t (*hasher_)(const Key &key); | ||
| 237 | uint64_t bytes_allocated_; | ||
| 238 | uint64_t num_collisions_; | ||
| 239 | uint32_t max_collisions_; /**< maximum collisions for a single insert */ | ||
| 240 | Key empty_key_; | ||
| 241 | }; | ||
| 242 | |||
| 243 | |||
| 244 | template<class Key, class Value> | ||
| 245 | class SmallHashFixed | ||
| 246 | : public SmallHashBase<Key, Value, SmallHashFixed<Key, Value> > { | ||
| 247 | friend class SmallHashBase<Key, Value, SmallHashFixed<Key, Value> >; | ||
| 248 | |||
| 249 | protected: | ||
| 250 | // No-ops | ||
| 251 | 5795 | void SetThresholds() { } | |
| 252 | 76371 | void Grow() { } | |
| 253 | 7450 | void Shrink() { } | |
| 254 | 18 | void ResetCapacity() { } | |
| 255 | }; | ||
| 256 | |||
| 257 | |||
| 258 | template<class Key, class Value> | ||
| 259 | class SmallHashDynamic | ||
| 260 | : public SmallHashBase<Key, Value, SmallHashDynamic<Key, Value> > { | ||
| 261 | friend class SmallHashBase<Key, Value, SmallHashDynamic<Key, Value> >; | ||
| 262 | |||
| 263 | public: | ||
| 264 | typedef SmallHashBase<Key, Value, SmallHashDynamic<Key, Value> > Base; | ||
| 265 | static const double kThresholdGrow; | ||
| 266 | static const double kThresholdShrink; | ||
| 267 | |||
| 268 | 227341 | SmallHashDynamic() : Base() { | |
| 269 | 227341 | num_migrates_ = 0; | |
| 270 | |||
| 271 | // Properly set by Init | ||
| 272 | 227341 | threshold_grow_ = 0; | |
| 273 | 227341 | threshold_shrink_ = 0; | |
| 274 | 227341 | } | |
| 275 | |||
| 276 | SmallHashDynamic(const SmallHashDynamic<Key, Value> &other) : Base() { | ||
| 277 | num_migrates_ = 0; | ||
| 278 | CopyFrom(other); | ||
| 279 | } | ||
| 280 | |||
| 281 | 364 | SmallHashDynamic<Key, Value> &operator=( | |
| 282 | const SmallHashDynamic<Key, Value> &other) { | ||
| 283 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 200 times.
|
364 | if (&other == this) |
| 284 | ✗ | return *this; | |
| 285 | |||
| 286 | 364 | CopyFrom(other); | |
| 287 | 364 | return *this; | |
| 288 | } | ||
| 289 | |||
| 290 | 43702982 | uint32_t capacity() const { return Base::capacity_; } | |
| 291 | 1158854376 | uint32_t size() const { return Base::size_; } | |
| 292 | uint32_t num_migrates() const { return num_migrates_; } | ||
| 293 | |||
| 294 | protected: | ||
| 295 | 389446 | void SetThresholds() { | |
| 296 | 389446 | threshold_grow_ = static_cast<uint32_t>(static_cast<double>(capacity()) | |
| 297 | 389518 | * kThresholdGrow); | |
| 298 | 389518 | threshold_shrink_ = static_cast<uint32_t>(static_cast<double>(capacity()) | |
| 299 | 389518 | * kThresholdShrink); | |
| 300 | 389518 | } | |
| 301 | |||
| 302 | 2046842638 | void Grow() { | |
| 303 |
2/2✓ Branch 1 taken 54358 times.
✓ Branch 2 taken 1040668918 times.
|
2046842638 | if (size() > threshold_grow_) |
| 304 | 108716 | Migrate(capacity() * 2); | |
| 305 | 2042641000 | } | |
| 306 | |||
| 307 | 117139552 | void Shrink() { | |
| 308 |
2/2✓ Branch 1 taken 20968811 times.
✓ Branch 2 taken 96336765 times.
|
117139552 | if (size() < threshold_shrink_) { |
| 309 | 20973271 | const uint32_t target_capacity = capacity() / 2; | |
| 310 |
2/2✓ Branch 0 taken 24768 times.
✓ Branch 1 taken 20936663 times.
|
20965891 | if (target_capacity >= Base::initial_capacity_) |
| 311 | 24768 | Migrate(target_capacity); | |
| 312 | } | ||
| 313 | 117302656 | } | |
| 314 | |||
| 315 | 3948 | void ResetCapacity() { | |
| 316 | 3948 | Base::DeallocMemory(Base::keys_, Base::values_, Base::capacity_); | |
| 317 | 3948 | Base::capacity_ = Base::initial_capacity_; | |
| 318 | 3948 | Base::AllocMemory(); | |
| 319 | 3948 | SetThresholds(); | |
| 320 | 3948 | } | |
| 321 | |||
| 322 | private: | ||
| 323 | // Returns a random permutation of indices [0..N-1] that is allocated | ||
| 324 | // by smmap (Knuth's shuffle algorithm) | ||
| 325 | 49936 | uint32_t *ShuffleIndices(const uint32_t N) { | |
| 326 | 49936 | uint32_t *shuffled = static_cast<uint32_t *>(smmap(N * sizeof(uint32_t))); | |
| 327 | // Init with identity | ||
| 328 |
2/2✓ Branch 0 taken 403909512 times.
✓ Branch 1 taken 24968 times.
|
807868960 | for (unsigned i = 0; i < N; ++i) |
| 329 | 807819024 | shuffled[i] = i; | |
| 330 | // Shuffle (no shuffling for the last element) | ||
| 331 |
1/2✓ Branch 0 taken 383943172 times.
✗ Branch 1 not taken.
|
767661744 | for (unsigned i = 0; i < N - 1; ++i) { |
| 332 | 767886344 | const uint32_t swap_idx = i + g_prng.Next(N - i); | |
| 333 | 767611808 | const uint32_t tmp = shuffled[i]; | |
| 334 | 767611808 | shuffled[i] = shuffled[swap_idx]; | |
| 335 | 767611808 | shuffled[swap_idx] = tmp; | |
| 336 | } | ||
| 337 | 364 | return shuffled; | |
| 338 | } | ||
| 339 | |||
| 340 | 158180 | void Migrate(const uint32_t new_capacity) { | |
| 341 | 158180 | Key *old_keys = Base::keys_; | |
| 342 | 158180 | Value *old_values = Base::values_; | |
| 343 | 158180 | const uint32_t old_capacity = capacity(); | |
| 344 | 158108 | const uint32_t old_size = size(); | |
| 345 | |||
| 346 | 158180 | Base::capacity_ = new_capacity; | |
| 347 | 158180 | SetThresholds(); | |
| 348 | 158252 | Base::AllocMemory(); | |
| 349 | 158324 | Base::DoClear(false); | |
| 350 |
2/2✓ Branch 0 taken 24768 times.
✓ Branch 1 taken 54394 times.
|
158324 | if (new_capacity < old_capacity) { |
| 351 | 49536 | uint32_t *shuffled_indices = ShuffleIndices(old_capacity); | |
| 352 |
1/2✓ Branch 0 taken 348099048 times.
✗ Branch 1 not taken.
|
696191760 | for (uint32_t i = 0; i < old_capacity; ++i) { |
| 353 |
2/3✓ Branch 0 taken 87337332 times.
✓ Branch 1 taken 260761716 times.
✗ Branch 2 not taken.
|
696198096 | if (old_keys[shuffled_indices[i]] != Base::empty_key_) { |
| 354 | 174674664 | Base::Insert(old_keys[shuffled_indices[i]], | |
| 355 | 174674664 | old_values[shuffled_indices[i]]); | |
| 356 | } | ||
| 357 | } | ||
| 358 | ✗ | smunmap(shuffled_indices); | |
| 359 | } else { | ||
| 360 |
2/2✓ Branch 0 taken 652585401 times.
✓ Branch 1 taken 40318 times.
|
1305251438 | for (uint32_t i = 0; i < old_capacity; ++i) { |
| 361 |
3/3✓ Branch 0 taken 377419290 times.
✓ Branch 1 taken 237742024 times.
✓ Branch 2 taken 37424087 times.
|
1305170802 | if (old_keys[i] != Base::empty_key_) |
| 362 | 979403198 | Base::Insert(old_keys[i], old_values[i]); | |
| 363 | } | ||
| 364 | } | ||
| 365 |
1/2✗ Branch 1 not taken.
✓ Branch 2 taken 79162 times.
|
130172 | assert(size() == old_size); |
| 366 | |||
| 367 | 158324 | Base::DeallocMemory(old_keys, old_values, old_capacity); | |
| 368 | 158324 | num_migrates_++; | |
| 369 | 158324 | } | |
| 370 | |||
| 371 | 364 | void CopyFrom(const SmallHashDynamic<Key, Value> &other) { | |
| 372 | 364 | uint32_t *shuffled_indices = ShuffleIndices(other.capacity_); | |
| 373 |
2/2✓ Branch 0 taken 49548660 times.
✓ Branch 1 taken 200 times.
|
49552468 | for (uint32_t i = 0; i < other.capacity_; ++i) { |
| 374 |
2/3✗ Branch 0 not taken.
✓ Branch 1 taken 36001722 times.
✓ Branch 2 taken 13546938 times.
|
49552104 | if (other.keys_[shuffled_indices[i]] != other.empty_key_) { |
| 375 | 36000000 | this->Insert(other.keys_[shuffled_indices[i]], | |
| 376 | 36000000 | other.values_[shuffled_indices[i]]); | |
| 377 | } | ||
| 378 | } | ||
| 379 | 364 | smunmap(shuffled_indices); | |
| 380 | 364 | } | |
| 381 | |||
| 382 | uint32_t num_migrates_; | ||
| 383 | uint32_t threshold_grow_; | ||
| 384 | uint32_t threshold_shrink_; | ||
| 385 | static Prng g_prng; | ||
| 386 | }; | ||
| 387 | |||
| 388 | |||
| 389 | /** | ||
| 390 | * Distributes the key-value pairs over $n$ dynamic hash maps with individual | ||
| 391 | * mutexes. Hence low mutex contention, and benefits from multiple processors. | ||
| 392 | */ | ||
| 393 | template<class Key, class Value> | ||
| 394 | class MultiHash { | ||
| 395 | public: | ||
| 396 | 360 | MultiHash() { | |
| 397 | 360 | num_hashmaps_ = 0; | |
| 398 | 360 | hashmaps_ = NULL; | |
| 399 | 360 | locks_ = NULL; | |
| 400 | 360 | } | |
| 401 | |||
| 402 | 360 | void Init(const uint8_t num_hashmaps, const Key &empty_key, | |
| 403 | uint32_t (*hasher)(const Key &key)) { | ||
| 404 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 360 times.
|
360 | assert(num_hashmaps > 0); |
| 405 | 360 | const uint8_t N = num_hashmaps; | |
| 406 | 360 | num_hashmaps_ = N; | |
| 407 |
3/4✓ Branch 2 taken 15120 times.
✗ Branch 3 not taken.
✓ Branch 4 taken 15120 times.
✓ Branch 5 taken 360 times.
|
15480 | hashmaps_ = new SmallHashDynamic<Key, Value>[N](); |
| 408 | 360 | locks_ = static_cast<pthread_mutex_t *>( | |
| 409 | 360 | smalloc(N * sizeof(pthread_mutex_t))); | |
| 410 |
2/2✓ Branch 0 taken 15120 times.
✓ Branch 1 taken 360 times.
|
15480 | for (uint8_t i = 0; i < N; ++i) { |
| 411 | 15120 | int retval = pthread_mutex_init(&locks_[i], NULL); | |
| 412 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 15120 times.
|
15120 | assert(retval == 0); |
| 413 | 15120 | hashmaps_[i].Init(128, empty_key, hasher); | |
| 414 | } | ||
| 415 | 360 | } | |
| 416 | |||
| 417 | 360 | ~MultiHash() { | |
| 418 |
2/2✓ Branch 0 taken 15120 times.
✓ Branch 1 taken 360 times.
|
15480 | for (uint8_t i = 0; i < num_hashmaps_; ++i) { |
| 419 | 15120 | pthread_mutex_destroy(&locks_[i]); | |
| 420 | } | ||
| 421 | 360 | free(locks_); | |
| 422 |
3/4✓ Branch 0 taken 360 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 15120 times.
✓ Branch 3 taken 360 times.
|
15480 | delete[] hashmaps_; |
| 423 | 360 | } | |
| 424 | |||
| 425 | 36000000 | bool Lookup(const Key &key, Value *value) { | |
| 426 | 36000000 | uint8_t target = SelectHashmap(key); | |
| 427 | 36000000 | Lock(target); | |
| 428 | 36000000 | const bool result = hashmaps_[target].Lookup(key, value); | |
| 429 | 36000000 | Unlock(target); | |
| 430 | 36000000 | return result; | |
| 431 | } | ||
| 432 | |||
| 433 | 134473536 | void Insert(const Key &key, const Value &value) { | |
| 434 | 134473536 | uint8_t target = SelectHashmap(key); | |
| 435 | 131514156 | Lock(target); | |
| 436 | 135958212 | hashmaps_[target].Insert(key, value); | |
| 437 | 128274840 | Unlock(target); | |
| 438 | 140176008 | } | |
| 439 | |||
| 440 | 64799928 | void Erase(const Key &key) { | |
| 441 | 64799928 | uint8_t target = SelectHashmap(key); | |
| 442 | 62068608 | Lock(target); | |
| 443 | 66138156 | hashmaps_[target].Erase(key); | |
| 444 | 60534180 | Unlock(target); | |
| 445 | 67234032 | } | |
| 446 | |||
| 447 | 36 | void Clear() { | |
| 448 |
2/2✓ Branch 0 taken 1512 times.
✓ Branch 1 taken 36 times.
|
1548 | for (uint8_t i = 0; i < num_hashmaps_; ++i) |
| 449 | 1512 | Lock(i); | |
| 450 |
2/2✓ Branch 0 taken 1512 times.
✓ Branch 1 taken 36 times.
|
1548 | for (uint8_t i = 0; i < num_hashmaps_; ++i) |
| 451 | 1512 | hashmaps_[i].Clear(); | |
| 452 |
2/2✓ Branch 0 taken 1512 times.
✓ Branch 1 taken 36 times.
|
1548 | for (uint8_t i = 0; i < num_hashmaps_; ++i) |
| 453 | 1512 | Unlock(i); | |
| 454 | 36 | } | |
| 455 | |||
| 456 | 360 | uint8_t num_hashmaps() const { return num_hashmaps_; } | |
| 457 | |||
| 458 | 288 | void GetSizes(uint32_t *sizes) { | |
| 459 |
2/2✓ Branch 0 taken 12096 times.
✓ Branch 1 taken 288 times.
|
12384 | for (uint8_t i = 0; i < num_hashmaps_; ++i) { |
| 460 | 12096 | Lock(i); | |
| 461 | 12096 | sizes[i] = hashmaps_[i].size(); | |
| 462 | 12096 | Unlock(i); | |
| 463 | } | ||
| 464 | 288 | } | |
| 465 | |||
| 466 | void GetCollisionStats(uint64_t *num_collisions, uint32_t *max_collisions) { | ||
| 467 | for (uint8_t i = 0; i < num_hashmaps_; ++i) { | ||
| 468 | Lock(i); | ||
| 469 | hashmaps_[i].GetCollisionStats(&num_collisions[i], &max_collisions[i]); | ||
| 470 | Unlock(i); | ||
| 471 | } | ||
| 472 | } | ||
| 473 | |||
| 474 | private: | ||
| 475 | 233647992 | inline uint8_t SelectHashmap(const Key &key) { | |
| 476 | 233647992 | uint32_t hash = MurmurHash2(&key, sizeof(key), 0x37); | |
| 477 | 230882076 | double bucket = static_cast<double>(hash) | |
| 478 | 230882076 | * static_cast<double>(num_hashmaps_) | |
| 479 | / static_cast<double>(static_cast<uint32_t>(-1)); | ||
| 480 | 230882076 | return static_cast<uint32_t>(bucket) % num_hashmaps_; | |
| 481 | } | ||
| 482 | |||
| 483 | 228616380 | inline void Lock(const uint8_t target) { | |
| 484 | 228616380 | int retval = pthread_mutex_lock(&locks_[target]); | |
| 485 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 246496752 times.
|
246496752 | assert(retval == 0); |
| 486 | 246496752 | } | |
| 487 | |||
| 488 | 225008172 | inline void Unlock(const uint8_t target) { | |
| 489 | 225008172 | int retval = pthread_mutex_unlock(&locks_[target]); | |
| 490 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 247897296 times.
|
247897296 | assert(retval == 0); |
| 491 | 247897296 | } | |
| 492 | |||
| 493 | uint8_t num_hashmaps_; | ||
| 494 | SmallHashDynamic<Key, Value> *hashmaps_; | ||
| 495 | pthread_mutex_t *locks_; | ||
| 496 | }; | ||
| 497 | |||
| 498 | |||
| 499 | // initialize the static fields | ||
| 500 | template<class Key, class Value> | ||
| 501 | Prng SmallHashDynamic<Key, Value>::g_prng; | ||
| 502 | |||
| 503 | template<class Key, class Value, class Derived> | ||
| 504 | const double SmallHashBase<Key, Value, Derived>::kLoadFactor = 0.75; | ||
| 505 | |||
| 506 | template<class Key, class Value> | ||
| 507 | const double SmallHashDynamic<Key, Value>::kThresholdGrow = 0.75; | ||
| 508 | |||
| 509 | template<class Key, class Value> | ||
| 510 | const double SmallHashDynamic<Key, Value>::kThresholdShrink = 0.25; | ||
| 511 | |||
| 512 | #endif // CVMFS_SMALLHASH_H_ | ||
| 513 |