forked from membase/ep-engine
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathstored-value.cc
More file actions
599 lines (524 loc) · 18.4 KB
/
stored-value.cc
File metadata and controls
599 lines (524 loc) · 18.4 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
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
/* -*- Mode: C++; tab-width: 4; c-basic-offset: 4; indent-tabs-mode: nil -*- */
#include "config.h"
#include <cassert>
#include <limits>
#include "stored-value.hh"
#ifndef DEFAULT_HT_SIZE
#define DEFAULT_HT_SIZE 1531
#endif
size_t HashTable::defaultNumBuckets = DEFAULT_HT_SIZE;
size_t HashTable::defaultNumLocks = 193;
enum stored_value_type HashTable::defaultStoredValueType = featured;
double StoredValue::mutation_mem_threshold = 0.9;
const int64_t StoredValue::state_id_cleared = -1;
const int64_t StoredValue::state_id_pending = -2;
const int64_t StoredValue::state_deleted_key = -3;
const int64_t StoredValue::state_non_existent_key = -4;
const int64_t StoredValue::state_temp_init = -5;
static ssize_t prime_size_table[] = {
3, 7, 13, 23, 47, 97, 193, 383, 769, 1531, 3067, 6143, 12289, 24571, 49157,
98299, 196613, 393209, 786433, 1572869, 3145721, 6291449, 12582917,
25165813, 50331653, 100663291, 201326611, 402653189, 805306357,
1610612741, -1
};
bool StoredValue::ejectValue(EPStats &stats, HashTable &ht) {
if (eligibleForEviction()) {
size_t oldsize = size();
size_t old_valsize = value->length();
blobval uval;
uval.len = valLength();
value_t sp(Blob::New(uval.chlen, sizeof(uval)));
extra.feature.resident = false;
timestampEviction();
value = sp;
size_t newsize = size();
size_t new_valsize = value->length();
// ejecting the value may increase the object size....
if (oldsize < newsize) {
increaseCacheSize(ht, newsize - oldsize);
} else if (newsize < oldsize) {
reduceCacheSize(ht, oldsize - newsize);
}
// Add or substract the key/meta data overhead differenece.
size_t old_keymeta_overhead = (oldsize - old_valsize);
size_t new_keymeta_overhead = (newsize - new_valsize);
if (old_keymeta_overhead < new_keymeta_overhead) {
increaseCurrentSize(stats, new_keymeta_overhead - old_keymeta_overhead);
} else if (new_keymeta_overhead < old_keymeta_overhead) {
reduceCurrentSize(stats, old_keymeta_overhead - new_keymeta_overhead);
}
++stats.numValueEjects;
++ht.numNonResidentItems;
++ht.numEjects;
if (isReferenced()) {
++ht.numReferencedEjects;
}
return true;
}
++stats.numFailedEjects;
return false;
}
void StoredValue::referenced(HashTable &ht) {
if (!_isSmall && extra.feature.nru == false) {
extra.feature.nru = true;
++ht.numReferenced;
}
}
bool StoredValue::isReferenced(bool reset, HashTable *ht) {
bool ret = false;
if (!_isSmall) {
ret = extra.feature.nru;
if (reset && extra.feature.nru) {
extra.feature.nru = false;
assert(ht);
--ht->numReferenced;
}
}
return ret;
}
bool StoredValue::unlocked_restoreValue(Item *itm, EPStats &stats,
HashTable &ht) {
// If cas == we loaded the object from our meta file, but
// we didn't know the size of the object.. Don't report
// this as an unexpected size change.
if (getCas() == 0 && !_isSmall) {
extra.feature.cas = itm->getCas();
flags = itm->getFlags();
extra.feature.exptime = itm->getExptime();
extra.feature.seqno = itm->getSeqno();
setValue(*itm, stats, ht, true);
if (!isResident()) {
--ht.numNonResidentItems;
}
markClean(NULL);
return true;
}
if (!isResident()) {
size_t oldsize = size();
size_t old_valsize = isDeleted() ? 0 : value->length();
if (itm->getValue()->length() != valLength()) {
if (valLength()) {
// generate warning msg only if valLength() > 0, otherwise,
// no warning is necessary since it is the very first
// background fetch after warmup.
int diff(static_cast<int>(valLength()) - // expected
static_cast<int>(itm->getValue()->length())); // got
getLogger()->log(EXTENSION_LOG_WARNING, NULL,
"Object unexpectedly changed size by %d bytes\n",
diff);
}
}
rel_time_t evicted_time(getEvictedTime());
stats.pagedOutTimeHisto.add(ep_current_time() - evicted_time);
extra.feature.resident = true;
value = itm->getValue();
size_t newsize = size();
size_t new_valsize = value->length();
if (oldsize < newsize) {
increaseCacheSize(ht, newsize - oldsize);
} else if (newsize < oldsize) {
reduceCacheSize(ht, oldsize - newsize);
}
// Add or substract the key/meta data overhead differenece.
size_t old_keymeta_overhead = (oldsize - old_valsize);
size_t new_keymeta_overhead = (newsize - new_valsize);
if (old_keymeta_overhead < new_keymeta_overhead) {
increaseCurrentSize(stats, new_keymeta_overhead - old_keymeta_overhead);
} else if (new_keymeta_overhead < old_keymeta_overhead) {
reduceCurrentSize(stats, old_keymeta_overhead - new_keymeta_overhead);
}
--ht.numNonResidentItems;
return true;
}
return false;
}
mutation_type_t HashTable::insert(const Item &itm, bool eject, bool partial) {
assert(isActive());
if (!StoredValue::hasAvailableSpace(stats, itm)) {
return NOMEM;
}
assert(itm.getCas() != static_cast<uint64_t>(-1));
int bucket_num(0);
LockHolder lh = getLockedBucket(itm.getKey(), &bucket_num);
StoredValue *v = unlocked_find(itm.getKey(), bucket_num, true, false);
if (v == NULL) {
v = valFact(itm, values[bucket_num], *this);
v->markClean(NULL);
if (partial) {
v->extra.feature.resident = false;
++numNonResidentItems;
}
values[bucket_num] = v;
++numItems;
} else {
if (partial) {
// We don't have a better error code ;)
return INVALID_CAS;
}
// Verify that the CAS isn't changed
if (v->getCas() != itm.getCas()) {
if (v->getCas() == 0) {
v->extra.feature.cas = itm.getCas();
v->flags = itm.getFlags();
v->extra.feature.exptime = itm.getExptime();
v->extra.feature.seqno = itm.getSeqno();
} else {
return INVALID_CAS;
}
}
if (!v->isResident() && !v->isDeleted()) {
--numNonResidentItems;
}
v->setValue(const_cast<Item&>(itm), stats, *this, true);
}
v->markClean(NULL);
if (eject && !partial) {
v->ejectValue(stats, *this);
}
return NOT_FOUND;
}
bool StoredValue::unlocked_restoreMeta(Item *itm, ENGINE_ERROR_CODE status) {
assert(state_deleted_key != getId() && state_non_existent_key != getId());
switch(status) {
case ENGINE_SUCCESS:
assert(0 == itm->getValue()->length());
setSeqno(itm->getSeqno());
setCas(itm->getCas());
flags = itm->getFlags();
setExptime(itm->getExptime());
setStoredValueState(state_deleted_key);
return true;
case ENGINE_KEY_ENOENT:
setStoredValueState(state_non_existent_key);
return true;
default:
getLogger()->log(
EXTENSION_LOG_WARNING, NULL,
"The underlying storage returned error %d for get_meta\n", status);
return false;
}
}
static inline size_t getDefault(size_t x, size_t d) {
return x == 0 ? d : x;
}
size_t HashTable::getNumBuckets(size_t n) {
return getDefault(n, defaultNumBuckets);
}
size_t HashTable::getNumLocks(size_t n) {
return getDefault(n, defaultNumLocks);
}
/**
* Set the default number of hashtable buckets.
*/
void HashTable::setDefaultNumBuckets(size_t to) {
if (to != 0) {
defaultNumBuckets = to;
}
}
/**
* Set the default number of hashtable locks.
*/
void HashTable::setDefaultNumLocks(size_t to) {
if (to != 0) {
defaultNumLocks = to;
}
}
HashTableStatVisitor HashTable::clear(bool deactivate) {
HashTableStatVisitor rv;
if (!deactivate) {
// If not deactivating, assert we're already active.
assert(isActive());
}
MultiLockHolder mlh(mutexes, n_locks);
if (deactivate) {
setActiveState(false);
}
for (int i = 0; i < (int)size; i++) {
while (values[i]) {
StoredValue *v = values[i];
rv.visit(v);
values[i] = v->next;
delete v;
}
}
stats.currentSize.decr(rv.memSize - rv.valSize);
assert(stats.currentSize.get() < GIGANTOR);
numItems.set(0);
numTempItems.set(0);
numNonResidentItems.set(0);
memSize.set(0);
cacheSize.set(0);
numReferenced.set(0);
numReferencedEjects.set(0);
return rv;
}
void HashTable::resize(size_t newSize) {
assert(isActive());
// Due to the way hashing works, we can't fit anything larger than
// an int.
if (newSize > static_cast<size_t>(std::numeric_limits<int>::max())) {
return;
}
// Don't resize to the same size, either.
if (newSize == size) {
return;
}
MultiLockHolder mlh(mutexes, n_locks);
if (visitors.get() > 0) {
// Do not allow a resize while any visitors are actually
// processing. The next attempt will have to pick it up. New
// visitors cannot start doing meaningful work (we own all
// locks at this point).
return;
}
// Get a place for the new items.
StoredValue **newValues = static_cast<StoredValue**>(calloc(newSize,
sizeof(StoredValue*)));
// If we can't allocate memory, don't move stuff around.
if (!newValues) {
return;
}
stats.memOverhead.decr(memorySize());
++numResizes;
// Set the new size so all the hashy stuff works.
size_t oldSize = size;
size = newSize;
ep_sync_synchronize();
// Move existing records into the new space.
for (size_t i = 0; i < oldSize; i++) {
while (values[i]) {
StoredValue *v = values[i];
values[i] = v->next;
int newBucket = getBucketForHash(hash(v->getKeyBytes(), v->getKeyLen()));
v->next = newValues[newBucket];
newValues[newBucket] = v;
}
}
// values still points to the old (now empty) table.
free(values);
values = newValues;
stats.memOverhead.incr(memorySize());
assert(stats.memOverhead.get() < GIGANTOR);
}
static size_t distance(size_t a, size_t b) {
return std::max(a, b) - std::min(a, b);
}
static size_t nearest(size_t n, size_t a, size_t b) {
return (distance(n, a) < distance(b, n)) ? a : b;
}
static bool isCurrently(size_t size, ssize_t a, ssize_t b) {
ssize_t current(static_cast<ssize_t>(size));
return (current == a || current == b);
}
void HashTable::resize() {
size_t ni = getNumItems();
int i(0);
size_t new_size(0);
// Figure out where in the prime table we are.
ssize_t target(static_cast<ssize_t>(ni));
for (i = 0; prime_size_table[i] > 0 && prime_size_table[i] < target; ++i) {
// Just looking...
}
if (prime_size_table[i] == -1) {
// We're at the end, take the biggest
new_size = prime_size_table[i-1];
} else if (prime_size_table[i] < static_cast<ssize_t>(defaultNumBuckets)) {
// Was going to be smaller than the configured ht_size.
new_size = defaultNumBuckets;
} else if (isCurrently(size, prime_size_table[i-1], prime_size_table[i])) {
// If one of the candidate sizes is the current size, maintain
// the current size in order to remain stable.
new_size = size;
} else {
// Somewhere in the middle, use the one we're closer to.
new_size = nearest(ni, prime_size_table[i-1], prime_size_table[i]);
}
resize(new_size);
}
void HashTable::visit(HashTableVisitor &visitor) {
if ((numItems.get() + numTempItems.get()) == 0 || !isActive()) {
return;
}
VisitorTracker vt(&visitors);
bool aborted = !visitor.shouldContinue();
size_t visited = 0;
for (int l = 0; isActive() && !aborted && l < static_cast<int>(n_locks); l++) {
LockHolder lh(mutexes[l]);
for (int i = l; i < static_cast<int>(size); i+= n_locks) {
assert(l == mutexForBucket(i));
StoredValue *v = values[i];
assert(v == NULL || i == getBucketForHash(hash(v->getKeyBytes(),
v->getKeyLen())));
while (v) {
visitor.visit(v);
v = v->next;
}
++visited;
}
lh.unlock();
aborted = !visitor.shouldContinue();
}
assert(aborted || visited == size);
}
void HashTable::visitDepth(HashTableDepthVisitor &visitor) {
if (numItems.get() == 0 || !isActive()) {
return;
}
size_t visited = 0;
VisitorTracker vt(&visitors);
for (int l = 0; l < static_cast<int>(n_locks); l++) {
LockHolder lh(mutexes[l]);
for (int i = l; i < static_cast<int>(size); i+= n_locks) {
size_t depth = 0;
StoredValue *p = values[i];
assert(p == NULL || i == getBucketForHash(hash(p->getKeyBytes(),
p->getKeyLen())));
size_t mem(0);
while (p) {
depth++;
mem += p->size();
p = p->next;
}
visitor.visit(i, depth, mem);
++visited;
}
}
assert(visited == size);
}
bool HashTable::setDefaultStorageValueType(const char *t) {
bool rv = false;
if (t && strcmp(t, "featured") == 0) {
setDefaultStorageValueType(featured);
rv = true;
} else if (t && strcmp(t, "small") == 0) {
setDefaultStorageValueType(small);
rv = true;
}
return rv;
}
void HashTable::setDefaultStorageValueType(enum stored_value_type t) {
defaultStoredValueType = t;
}
enum stored_value_type HashTable::getDefaultStorageValueType() {
return defaultStoredValueType;
}
const char* HashTable::getDefaultStorageValueTypeStr() {
const char *rv = "unknown";
switch(getDefaultStorageValueType()) {
case small: rv = "small"; break;
case featured: rv = "featured"; break;
default: abort();
}
return rv;
}
add_type_t HashTable::unlocked_add(int &bucket_num,
const Item &val,
bool isDirty,
bool storeVal) {
StoredValue *v = unlocked_find(val.getKey(), bucket_num, true);
add_type_t rv = ADD_SUCCESS;
if (v && !v->isDeleted() && !v->isExpired(ep_real_time())) {
rv = ADD_EXISTS;
} else {
Item &itm = const_cast<Item&>(val);
itm.setCas();
if (!StoredValue::hasAvailableSpace(stats, itm)) {
return ADD_NOMEM;
}
if (v) {
rv = (v->isDeleted() || v->isExpired(ep_real_time())) ? ADD_UNDEL : ADD_SUCCESS;
v->setValue(itm, stats, *this, false);
if (isDirty) {
v->markDirty();
} else {
v->markClean(NULL);
}
} else {
v = valFact(itm, values[bucket_num], *this, isDirty);
values[bucket_num] = v;
if (v->isTempItem()) {
++numTempItems;
} else {
++numItems;
}
/**
* Possibly, this item is being recreated. Conservatively assign
* it a seqno that is greater than the greatest seqno of all
* deleted items seen so far.
*/
uint64_t seqno = getMaxDeletedSeqno() + 1;
v->setSeqno(seqno);
itm.setSeqno(seqno);
}
if (!storeVal) {
v->ejectValue(stats, *this);
}
if (v->isTempItem()) {
v->resetValue();
} else {
v->referenced(*this);
}
}
return rv;
}
add_type_t HashTable::unlocked_addTempDeletedItem(int &bucket_num,
const std::string &key) {
assert(isActive());
Item itm(key.c_str(), key.length(), (size_t)0, (uint32_t)0, (time_t)0,
0, StoredValue::state_temp_init);
// if a temp item for a possibly deleted, set it non-resident by resetting
// the value cuz normally a new item added is considered resident which does
// not apply for temp item.
return unlocked_add(bucket_num, itm, false, true);
}
void StoredValue::setMutationMemoryThreshold(double memThreshold) {
if (memThreshold > 0.0 && memThreshold <= 1.0) {
mutation_mem_threshold = memThreshold;
}
}
void StoredValue::increaseCacheSize(HashTable &ht, size_t by) {
ht.cacheSize.incr(by);
assert(ht.cacheSize.get() < GIGANTOR);
ht.memSize.incr(by);
assert(ht.memSize.get() < GIGANTOR);
}
void StoredValue::reduceCacheSize(HashTable &ht, size_t by) {
ht.cacheSize.decr(by);
assert(ht.cacheSize.get() < GIGANTOR);
ht.memSize.decr(by);
assert(ht.memSize.get() < GIGANTOR);
}
void StoredValue::increaseCurrentSize(EPStats &st, size_t by) {
st.currentSize.incr(by);
assert(st.currentSize.get() < GIGANTOR);
}
void StoredValue::reduceCurrentSize(EPStats &st, size_t by) {
size_t val;
do {
val = st.currentSize.get();
assert(val >= by);
} while (!st.currentSize.cas(val, val - by));;
}
void StoredValue::increaseMetaDataSize(HashTable &ht, size_t by) {
ht.metaDataMemory.incr(by);
assert(ht.metaDataMemory.get() < GIGANTOR);
}
void StoredValue::reduceMetaDataSize(HashTable &ht, size_t by) {
ht.metaDataMemory.decr(by);
assert(ht.metaDataMemory.get() < GIGANTOR);
}
/**
* Is there enough space for this thing?
*/
bool StoredValue::hasAvailableSpace(EPStats &st, const Item &itm) {
double newSize = static_cast<double>(st.getTotalMemoryUsed() +
sizeof(StoredValue) + itm.getNKey());
double maxSize= static_cast<double>(st.getMaxDataSize()) * mutation_mem_threshold;
return newSize <= maxSize;
}
Item* StoredValue::toItem(bool lck, uint16_t vbucket) const {
return new Item(getKey(), getFlags(), getExptime(),
value,
lck ? static_cast<uint64_t>(-1) : getCas(),
id, vbucket, getSeqno());
}