/bitcoin/src/leveldb/util/cache.cc
Line | Count | Source |
1 | | // Copyright (c) 2011 The LevelDB Authors. All rights reserved. |
2 | | // Use of this source code is governed by a BSD-style license that can be |
3 | | // found in the LICENSE file. See the AUTHORS file for names of contributors. |
4 | | |
5 | | #include <assert.h> |
6 | | #include <stdio.h> |
7 | | #include <stdlib.h> |
8 | | |
9 | | #include "leveldb/cache.h" |
10 | | #include "port/port.h" |
11 | | #include "port/thread_annotations.h" |
12 | | #include "util/hash.h" |
13 | | #include "util/mutexlock.h" |
14 | | |
15 | | namespace leveldb { |
16 | | |
17 | 66.5k | Cache::~Cache() {} |
18 | | |
19 | | namespace { |
20 | | |
21 | | // LRU cache implementation |
22 | | // |
23 | | // Cache entries have an "in_cache" boolean indicating whether the cache has a |
24 | | // reference on the entry. The only ways that this can become false without the |
25 | | // entry being passed to its "deleter" are via Erase(), via Insert() when |
26 | | // an element with a duplicate key is inserted, or on destruction of the cache. |
27 | | // |
28 | | // The cache keeps two linked lists of items in the cache. All items in the |
29 | | // cache are in one list or the other, and never both. Items still referenced |
30 | | // by clients but erased from the cache are in neither list. The lists are: |
31 | | // - in-use: contains the items currently referenced by clients, in no |
32 | | // particular order. (This list is used for invariant checking. If we |
33 | | // removed the check, elements that would otherwise be on this list could be |
34 | | // left as disconnected singleton lists.) |
35 | | // - LRU: contains the items not currently referenced by clients, in LRU order |
36 | | // Elements are moved between these lists by the Ref() and Unref() methods, |
37 | | // when they detect an element in the cache acquiring or losing its only |
38 | | // external reference. |
39 | | |
40 | | // An entry is a variable length heap-allocated structure. Entries |
41 | | // are kept in a circular doubly linked list ordered by access time. |
42 | | struct LRUHandle { |
43 | | void* value; |
44 | | void (*deleter)(const Slice&, void* value); |
45 | | LRUHandle* next_hash; |
46 | | LRUHandle* next; |
47 | | LRUHandle* prev; |
48 | | size_t charge; // TODO(opt): Only allow uint32_t? |
49 | | size_t key_length; |
50 | | bool in_cache; // Whether entry is in the cache. |
51 | | uint32_t refs; // References, including cache reference, if present. |
52 | | uint32_t hash; // Hash of key(); used for fast sharding and comparisons |
53 | | char key_data[1]; // Beginning of key |
54 | | |
55 | 60 | Slice key() const { |
56 | | // next_ is only equal to this if the LRU handle is the list head of an |
57 | | // empty list. List heads never have meaningful keys. |
58 | 60 | assert(next != this); Branch (58:5): [True: 60, False: 0]
|
59 | | |
60 | 60 | return Slice(key_data, key_length); |
61 | 60 | } |
62 | | }; |
63 | | |
64 | | // We provide our own simple hash table since it removes a whole bunch |
65 | | // of porting hacks and is also faster than some of the built-in hash |
66 | | // table implementations in some of the compiler/runtime combinations |
67 | | // we have tested. E.g., readrandom speeds up by ~5% over the g++ |
68 | | // 4.4.3's builtin hashtable. |
69 | | class HandleTable { |
70 | | public: |
71 | 1.06M | HandleTable() : length_(0), elems_(0), list_(nullptr) { Resize(); } |
72 | 1.06M | ~HandleTable() { delete[] list_; } |
73 | | |
74 | 48 | LRUHandle* Lookup(const Slice& key, uint32_t hash) { |
75 | 48 | return *FindPointer(key, hash); |
76 | 48 | } |
77 | | |
78 | 24 | LRUHandle* Insert(LRUHandle* h) { |
79 | 24 | LRUHandle** ptr = FindPointer(h->key(), h->hash); |
80 | 24 | LRUHandle* old = *ptr; |
81 | 24 | h->next_hash = (old == nullptr ? nullptr : old->next_hash); Branch (81:21): [True: 24, False: 0]
|
82 | 24 | *ptr = h; |
83 | 24 | if (old == nullptr) { Branch (83:9): [True: 24, False: 0]
|
84 | 24 | ++elems_; |
85 | 24 | if (elems_ > length_) { Branch (85:11): [True: 0, False: 24]
|
86 | | // Since each cache entry is fairly large, we aim for a small |
87 | | // average linked list length (<= 1). |
88 | 0 | Resize(); |
89 | 0 | } |
90 | 24 | } |
91 | 24 | return old; |
92 | 24 | } |
93 | | |
94 | 0 | LRUHandle* Remove(const Slice& key, uint32_t hash) { |
95 | 0 | LRUHandle** ptr = FindPointer(key, hash); |
96 | 0 | LRUHandle* result = *ptr; |
97 | 0 | if (result != nullptr) { Branch (97:9): [True: 0, False: 0]
|
98 | 0 | *ptr = result->next_hash; |
99 | 0 | --elems_; |
100 | 0 | } |
101 | 0 | return result; |
102 | 0 | } |
103 | | |
104 | | private: |
105 | | // The table consists of an array of buckets where each bucket is |
106 | | // a linked list of cache entries that hash into the bucket. |
107 | | uint32_t length_; |
108 | | uint32_t elems_; |
109 | | LRUHandle** list_; |
110 | | |
111 | | // Return a pointer to slot that points to a cache entry that |
112 | | // matches key/hash. If there is no such cache entry, return a |
113 | | // pointer to the trailing slot in the corresponding linked list. |
114 | 72 | LRUHandle** FindPointer(const Slice& key, uint32_t hash) { |
115 | 72 | LRUHandle** ptr = &list_[hash & (length_ - 1)]; |
116 | 72 | while (*ptr != nullptr && ((*ptr)->hash != hash || key != (*ptr)->key())) { Branch (116:12): [True: 12, False: 60]
Branch (116:12): [True: 0, False: 72]
Branch (116:32): [True: 0, False: 12]
Branch (116:56): [True: 0, False: 12]
|
117 | 0 | ptr = &(*ptr)->next_hash; |
118 | 0 | } |
119 | 72 | return ptr; |
120 | 72 | } |
121 | | |
122 | 1.06M | void Resize() { |
123 | 1.06M | uint32_t new_length = 4; |
124 | 1.06M | while (new_length < elems_) { Branch (124:12): [True: 0, False: 1.06M]
|
125 | 0 | new_length *= 2; |
126 | 0 | } |
127 | 1.06M | LRUHandle** new_list = new LRUHandle*[new_length]; |
128 | 1.06M | memset(new_list, 0, sizeof(new_list[0]) * new_length); |
129 | 1.06M | uint32_t count = 0; |
130 | 1.06M | for (uint32_t i = 0; i < length_; i++) { Branch (130:26): [True: 0, False: 1.06M]
|
131 | 0 | LRUHandle* h = list_[i]; |
132 | 0 | while (h != nullptr) { Branch (132:14): [True: 0, False: 0]
|
133 | 0 | LRUHandle* next = h->next_hash; |
134 | 0 | uint32_t hash = h->hash; |
135 | 0 | LRUHandle** ptr = &new_list[hash & (new_length - 1)]; |
136 | 0 | h->next_hash = *ptr; |
137 | 0 | *ptr = h; |
138 | 0 | h = next; |
139 | 0 | count++; |
140 | 0 | } |
141 | 0 | } |
142 | 1.06M | assert(elems_ == count); Branch (142:5): [True: 1.06M, False: 0]
|
143 | 1.06M | delete[] list_; |
144 | 1.06M | list_ = new_list; |
145 | 1.06M | length_ = new_length; |
146 | 1.06M | } |
147 | | }; |
148 | | |
149 | | // A single shard of sharded cache. |
150 | | class LRUCache { |
151 | | public: |
152 | | LRUCache(); |
153 | | ~LRUCache(); |
154 | | |
155 | | // Separate from constructor so caller can easily make an array of LRUCache |
156 | 1.06M | void SetCapacity(size_t capacity) { capacity_ = capacity; } |
157 | | |
158 | | // Like Cache methods, but with an extra "hash" parameter. |
159 | | Cache::Handle* Insert(const Slice& key, uint32_t hash, void* value, |
160 | | size_t charge, |
161 | | void (*deleter)(const Slice& key, void* value)); |
162 | | Cache::Handle* Lookup(const Slice& key, uint32_t hash); |
163 | | void Release(Cache::Handle* handle); |
164 | | void Erase(const Slice& key, uint32_t hash); |
165 | | void Prune(); |
166 | 0 | size_t TotalCharge() const { |
167 | 0 | MutexLock l(&mutex_); |
168 | 0 | return usage_; |
169 | 0 | } |
170 | | |
171 | | private: |
172 | | void LRU_Remove(LRUHandle* e); |
173 | | void LRU_Append(LRUHandle* list, LRUHandle* e); |
174 | | void Ref(LRUHandle* e); |
175 | | void Unref(LRUHandle* e); |
176 | | bool FinishErase(LRUHandle* e) EXCLUSIVE_LOCKS_REQUIRED(mutex_); |
177 | | |
178 | | // Initialized before use. |
179 | | size_t capacity_; |
180 | | |
181 | | // mutex_ protects the following state. |
182 | | mutable port::Mutex mutex_; |
183 | | size_t usage_ GUARDED_BY(mutex_); |
184 | | |
185 | | // Dummy head of LRU list. |
186 | | // lru.prev is newest entry, lru.next is oldest entry. |
187 | | // Entries have refs==1 and in_cache==true. |
188 | | LRUHandle lru_ GUARDED_BY(mutex_); |
189 | | |
190 | | // Dummy head of in-use list. |
191 | | // Entries are in use by clients, and have refs >= 2 and in_cache==true. |
192 | | LRUHandle in_use_ GUARDED_BY(mutex_); |
193 | | |
194 | | HandleTable table_ GUARDED_BY(mutex_); |
195 | | }; |
196 | | |
197 | 1.06M | LRUCache::LRUCache() : capacity_(0), usage_(0) { |
198 | | // Make empty circular linked lists. |
199 | 1.06M | lru_.next = &lru_; |
200 | 1.06M | lru_.prev = &lru_; |
201 | 1.06M | in_use_.next = &in_use_; |
202 | 1.06M | in_use_.prev = &in_use_; |
203 | 1.06M | } |
204 | | |
205 | 1.06M | LRUCache::~LRUCache() { |
206 | 1.06M | assert(in_use_.next == &in_use_); // Error if caller has an unreleased handle Branch (206:3): [True: 1.06M, False: 0]
|
207 | 1.06M | for (LRUHandle* e = lru_.next; e != &lru_;) { Branch (207:34): [True: 24, False: 1.06M]
|
208 | 24 | LRUHandle* next = e->next; |
209 | 24 | assert(e->in_cache); Branch (209:5): [True: 24, False: 0]
|
210 | 24 | e->in_cache = false; |
211 | 24 | assert(e->refs == 1); // Invariant of lru_ list. Branch (211:5): [True: 24, False: 0]
|
212 | 24 | Unref(e); |
213 | 24 | e = next; |
214 | 24 | } |
215 | 1.06M | } |
216 | | |
217 | 12 | void LRUCache::Ref(LRUHandle* e) { |
218 | 12 | if (e->refs == 1 && e->in_cache) { // If on lru_ list, move to in_use_ list. Branch (218:7): [True: 12, False: 0]
Branch (218:23): [True: 12, False: 0]
|
219 | 12 | LRU_Remove(e); |
220 | 12 | LRU_Append(&in_use_, e); |
221 | 12 | } |
222 | 12 | e->refs++; |
223 | 12 | } |
224 | | |
225 | 60 | void LRUCache::Unref(LRUHandle* e) { |
226 | 60 | assert(e->refs > 0); Branch (226:3): [True: 60, False: 0]
|
227 | 60 | e->refs--; |
228 | 60 | if (e->refs == 0) { // Deallocate. Branch (228:7): [True: 24, False: 36]
|
229 | 24 | assert(!e->in_cache); Branch (229:5): [True: 24, False: 0]
|
230 | 24 | (*e->deleter)(e->key(), e->value); |
231 | 24 | free(e); |
232 | 36 | } else if (e->in_cache && e->refs == 1) { Branch (232:14): [True: 36, False: 0]
Branch (232:29): [True: 36, False: 0]
|
233 | | // No longer in use; move to lru_ list. |
234 | 36 | LRU_Remove(e); |
235 | 36 | LRU_Append(&lru_, e); |
236 | 36 | } |
237 | 60 | } |
238 | | |
239 | 48 | void LRUCache::LRU_Remove(LRUHandle* e) { |
240 | 48 | e->next->prev = e->prev; |
241 | 48 | e->prev->next = e->next; |
242 | 48 | } |
243 | | |
244 | 72 | void LRUCache::LRU_Append(LRUHandle* list, LRUHandle* e) { |
245 | | // Make "e" newest entry by inserting just before *list |
246 | 72 | e->next = list; |
247 | 72 | e->prev = list->prev; |
248 | 72 | e->prev->next = e; |
249 | 72 | e->next->prev = e; |
250 | 72 | } |
251 | | |
252 | 48 | Cache::Handle* LRUCache::Lookup(const Slice& key, uint32_t hash) { |
253 | 48 | MutexLock l(&mutex_); |
254 | 48 | LRUHandle* e = table_.Lookup(key, hash); |
255 | 48 | if (e != nullptr) { Branch (255:7): [True: 12, False: 36]
|
256 | 12 | Ref(e); |
257 | 12 | } |
258 | 48 | return reinterpret_cast<Cache::Handle*>(e); |
259 | 48 | } |
260 | | |
261 | 36 | void LRUCache::Release(Cache::Handle* handle) { |
262 | 36 | MutexLock l(&mutex_); |
263 | 36 | Unref(reinterpret_cast<LRUHandle*>(handle)); |
264 | 36 | } |
265 | | |
266 | | Cache::Handle* LRUCache::Insert(const Slice& key, uint32_t hash, void* value, |
267 | | size_t charge, |
268 | | void (*deleter)(const Slice& key, |
269 | 24 | void* value)) { |
270 | 24 | MutexLock l(&mutex_); |
271 | | |
272 | 24 | LRUHandle* e = |
273 | 24 | reinterpret_cast<LRUHandle*>(malloc(sizeof(LRUHandle) - 1 + key.size())); |
274 | 24 | e->value = value; |
275 | 24 | e->deleter = deleter; |
276 | 24 | e->charge = charge; |
277 | 24 | e->key_length = key.size(); |
278 | 24 | e->hash = hash; |
279 | 24 | e->in_cache = false; |
280 | 24 | e->refs = 1; // for the returned handle. |
281 | 24 | memcpy(e->key_data, key.data(), key.size()); |
282 | | |
283 | 24 | if (capacity_ > 0) { Branch (283:7): [True: 24, False: 0]
|
284 | 24 | e->refs++; // for the cache's reference. |
285 | 24 | e->in_cache = true; |
286 | 24 | LRU_Append(&in_use_, e); |
287 | 24 | usage_ += charge; |
288 | 24 | FinishErase(table_.Insert(e)); |
289 | 24 | } else { // don't cache. (capacity_==0 is supported and turns off caching.) |
290 | | // next is read by key() in an assert, so it must be initialized |
291 | 0 | e->next = nullptr; |
292 | 0 | } |
293 | 24 | while (usage_ > capacity_ && lru_.next != &lru_) { Branch (293:10): [True: 0, False: 24]
Branch (293:32): [True: 0, False: 0]
|
294 | 0 | LRUHandle* old = lru_.next; |
295 | 0 | assert(old->refs == 1); Branch (295:5): [True: 0, False: 0]
|
296 | 0 | bool erased = FinishErase(table_.Remove(old->key(), old->hash)); |
297 | 0 | if (!erased) { // to avoid unused variable when compiled NDEBUG Branch (297:9): [True: 0, False: 0]
|
298 | 0 | assert(erased); Branch (298:7): [True: 0, False: 0]
|
299 | 0 | } |
300 | 0 | } |
301 | | |
302 | 24 | return reinterpret_cast<Cache::Handle*>(e); |
303 | 24 | } |
304 | | |
305 | | // If e != nullptr, finish removing *e from the cache; it has already been |
306 | | // removed from the hash table. Return whether e != nullptr. |
307 | 24 | bool LRUCache::FinishErase(LRUHandle* e) { |
308 | 24 | if (e != nullptr) { Branch (308:7): [True: 0, False: 24]
|
309 | 0 | assert(e->in_cache); Branch (309:5): [True: 0, False: 0]
|
310 | 0 | LRU_Remove(e); |
311 | 0 | e->in_cache = false; |
312 | 0 | usage_ -= e->charge; |
313 | 0 | Unref(e); |
314 | 0 | } |
315 | 24 | return e != nullptr; |
316 | 24 | } |
317 | | |
318 | 0 | void LRUCache::Erase(const Slice& key, uint32_t hash) { |
319 | 0 | MutexLock l(&mutex_); |
320 | 0 | FinishErase(table_.Remove(key, hash)); |
321 | 0 | } |
322 | | |
323 | 0 | void LRUCache::Prune() { |
324 | 0 | MutexLock l(&mutex_); |
325 | 0 | while (lru_.next != &lru_) { Branch (325:10): [True: 0, False: 0]
|
326 | 0 | LRUHandle* e = lru_.next; |
327 | 0 | assert(e->refs == 1); Branch (327:5): [True: 0, False: 0]
|
328 | 0 | bool erased = FinishErase(table_.Remove(e->key(), e->hash)); |
329 | 0 | if (!erased) { // to avoid unused variable when compiled NDEBUG Branch (329:9): [True: 0, False: 0]
|
330 | 0 | assert(erased); Branch (330:7): [True: 0, False: 0]
|
331 | 0 | } |
332 | 0 | } |
333 | 0 | } |
334 | | |
335 | | static const int kNumShardBits = 4; |
336 | | static const int kNumShards = 1 << kNumShardBits; |
337 | | |
338 | | class ShardedLRUCache : public Cache { |
339 | | private: |
340 | | LRUCache shard_[kNumShards]; |
341 | | port::Mutex id_mutex_; |
342 | | uint64_t last_id_; |
343 | | |
344 | 72 | static inline uint32_t HashSlice(const Slice& s) { |
345 | 72 | return Hash(s.data(), s.size(), 0); |
346 | 72 | } |
347 | | |
348 | 108 | static uint32_t Shard(uint32_t hash) { return hash >> (32 - kNumShardBits); } |
349 | | |
350 | | public: |
351 | 66.5k | explicit ShardedLRUCache(size_t capacity) : last_id_(0) { |
352 | 66.5k | const size_t per_shard = (capacity + (kNumShards - 1)) / kNumShards; |
353 | 1.13M | for (int s = 0; s < kNumShards; s++) { Branch (353:21): [True: 1.06M, False: 66.5k]
|
354 | 1.06M | shard_[s].SetCapacity(per_shard); |
355 | 1.06M | } |
356 | 66.5k | } |
357 | 66.5k | ~ShardedLRUCache() override {} |
358 | | Handle* Insert(const Slice& key, void* value, size_t charge, |
359 | 24 | void (*deleter)(const Slice& key, void* value)) override { |
360 | 24 | const uint32_t hash = HashSlice(key); |
361 | 24 | return shard_[Shard(hash)].Insert(key, hash, value, charge, deleter); |
362 | 24 | } |
363 | 48 | Handle* Lookup(const Slice& key) override { |
364 | 48 | const uint32_t hash = HashSlice(key); |
365 | 48 | return shard_[Shard(hash)].Lookup(key, hash); |
366 | 48 | } |
367 | 36 | void Release(Handle* handle) override { |
368 | 36 | LRUHandle* h = reinterpret_cast<LRUHandle*>(handle); |
369 | 36 | shard_[Shard(h->hash)].Release(handle); |
370 | 36 | } |
371 | 0 | void Erase(const Slice& key) override { |
372 | 0 | const uint32_t hash = HashSlice(key); |
373 | 0 | shard_[Shard(hash)].Erase(key, hash); |
374 | 0 | } |
375 | 36 | void* Value(Handle* handle) override { |
376 | 36 | return reinterpret_cast<LRUHandle*>(handle)->value; |
377 | 36 | } |
378 | 24 | uint64_t NewId() override { |
379 | 24 | MutexLock l(&id_mutex_); |
380 | 24 | return ++(last_id_); |
381 | 24 | } |
382 | 0 | void Prune() override { |
383 | 0 | for (int s = 0; s < kNumShards; s++) { Branch (383:21): [True: 0, False: 0]
|
384 | 0 | shard_[s].Prune(); |
385 | 0 | } |
386 | 0 | } |
387 | 0 | size_t TotalCharge() const override { |
388 | 0 | size_t total = 0; |
389 | 0 | for (int s = 0; s < kNumShards; s++) { Branch (389:21): [True: 0, False: 0]
|
390 | 0 | total += shard_[s].TotalCharge(); |
391 | 0 | } |
392 | 0 | return total; |
393 | 0 | } |
394 | | }; |
395 | | |
396 | | } // end anonymous namespace |
397 | | |
398 | 66.5k | Cache* NewLRUCache(size_t capacity) { return new ShardedLRUCache(capacity); } |
399 | | |
400 | | } // namespace leveldb |