summaryrefslogtreecommitdiffstats
path: root/src/kv/rocksdb_cache/BinnedLRUCache.cc
blob: 2e66c16eecdda321edd529f90f83a9687a5446f4 (plain)
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
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
// Copyright (c) 2018-Present Red Hat Inc.  All rights reserved.
//
// Copyright (c) 2011-2018, Facebook, Inc.  All rights reserved.
// This source code is licensed under both the GPLv2 and Apache 2.0 License
//
// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file. See the AUTHORS file for names of contributors.

#ifndef __STDC_FORMAT_MACROS
#define __STDC_FORMAT_MACROS
#endif

#include "BinnedLRUCache.h"

#include <stdio.h>
#include <stdlib.h>
#include <string>

#define dout_context cct
#define dout_subsys ceph_subsys_rocksdb
#undef dout_prefix
#define dout_prefix *_dout << "rocksdb: "

namespace rocksdb_cache {

BinnedLRUHandleTable::BinnedLRUHandleTable() : list_(nullptr), length_(0), elems_(0) {
  Resize();
}

BinnedLRUHandleTable::~BinnedLRUHandleTable() {
  ApplyToAllCacheEntries([](BinnedLRUHandle* h) {
    if (h->refs == 1) {
      h->Free();
    }
  });
  delete[] list_;
}

BinnedLRUHandle* BinnedLRUHandleTable::Lookup(const rocksdb::Slice& key, uint32_t hash) {
  return *FindPointer(key, hash);
}

BinnedLRUHandle* BinnedLRUHandleTable::Insert(BinnedLRUHandle* h) {
  BinnedLRUHandle** ptr = FindPointer(h->key(), h->hash);
  BinnedLRUHandle* old = *ptr;
  h->next_hash = (old == nullptr ? nullptr : old->next_hash);
  *ptr = h;
  if (old == nullptr) {
    ++elems_;
    if (elems_ > length_) {
      // Since each cache entry is fairly large, we aim for a small
      // average linked list length (<= 1).
      Resize();
    }
  }
  return old;
}

BinnedLRUHandle* BinnedLRUHandleTable::Remove(const rocksdb::Slice& key, uint32_t hash) {
  BinnedLRUHandle** ptr = FindPointer(key, hash);
  BinnedLRUHandle* result = *ptr;
  if (result != nullptr) {
    *ptr = result->next_hash;
    --elems_;
  }
  return result;
}

BinnedLRUHandle** BinnedLRUHandleTable::FindPointer(const rocksdb::Slice& key, uint32_t hash) {
  BinnedLRUHandle** ptr = &list_[hash & (length_ - 1)];
  while (*ptr != nullptr && ((*ptr)->hash != hash || key != (*ptr)->key())) {
    ptr = &(*ptr)->next_hash;
  }
  return ptr;
}

void BinnedLRUHandleTable::Resize() {
  uint32_t new_length = 16;
  while (new_length < elems_ * 1.5) {
    new_length *= 2;
  }
  BinnedLRUHandle** new_list = new BinnedLRUHandle*[new_length];
  memset(new_list, 0, sizeof(new_list[0]) * new_length);
  uint32_t count = 0;
  for (uint32_t i = 0; i < length_; i++) {
    BinnedLRUHandle* h = list_[i];
    while (h != nullptr) {
      BinnedLRUHandle* next = h->next_hash;
      uint32_t hash = h->hash;
      BinnedLRUHandle** ptr = &new_list[hash & (new_length - 1)];
      h->next_hash = *ptr;
      *ptr = h;
      h = next;
      count++;
    }
  }
  ceph_assert(elems_ == count);
  delete[] list_;
  list_ = new_list;
  length_ = new_length;
}

BinnedLRUCacheShard::BinnedLRUCacheShard(CephContext *c, size_t capacity, bool strict_capacity_limit,
                             double high_pri_pool_ratio)
    : cct(c),
      capacity_(0),
      high_pri_pool_usage_(0),
      strict_capacity_limit_(strict_capacity_limit),
      high_pri_pool_ratio_(high_pri_pool_ratio),
      high_pri_pool_capacity_(0),
      usage_(0),
      lru_usage_(0),
      age_bins(1) {
  shift_bins();
  // Make empty circular linked list
  lru_.next = &lru_;
  lru_.prev = &lru_;
  lru_low_pri_ = &lru_;
  SetCapacity(capacity);
}

BinnedLRUCacheShard::~BinnedLRUCacheShard() {}

bool BinnedLRUCacheShard::Unref(BinnedLRUHandle* e) {
  ceph_assert(e->refs > 0);
  e->refs--;
  return e->refs == 0;
}

// Call deleter and free

void BinnedLRUCacheShard::EraseUnRefEntries() {
  ceph::autovector<BinnedLRUHandle*> last_reference_list;
  {
    std::lock_guard<std::mutex> l(mutex_);
    while (lru_.next != &lru_) {
      BinnedLRUHandle* old = lru_.next;
      ceph_assert(old->InCache());
      ceph_assert(old->refs ==
             1);  // LRU list contains elements which may be evicted
      LRU_Remove(old);
      table_.Remove(old->key(), old->hash);
      old->SetInCache(false);
      Unref(old);
      usage_ -= old->charge;
      last_reference_list.push_back(old);
    }
  }

  for (auto entry : last_reference_list) {
    entry->Free();
  }
}

void BinnedLRUCacheShard::ApplyToAllCacheEntries(
  const std::function<void(const rocksdb::Slice& key,
                           void* value,
                           size_t charge,
                           DeleterFn)>& callback,
  bool thread_safe)
{
  if (thread_safe) {
    mutex_.lock();
  }
  table_.ApplyToAllCacheEntries(
    [callback](BinnedLRUHandle* h) {
      callback(h->key(), h->value, h->charge, h->deleter);
    });
  if (thread_safe) {
    mutex_.unlock();
  }
}

void BinnedLRUCacheShard::TEST_GetLRUList(BinnedLRUHandle** lru, BinnedLRUHandle** lru_low_pri) {
  *lru = &lru_;
  *lru_low_pri = lru_low_pri_;
}

size_t BinnedLRUCacheShard::TEST_GetLRUSize() {
  BinnedLRUHandle* lru_handle = lru_.next;
  size_t lru_size = 0;
  while (lru_handle != &lru_) {
    lru_size++;
    lru_handle = lru_handle->next;
  }
  return lru_size;
}

double BinnedLRUCacheShard::GetHighPriPoolRatio() const {
  std::lock_guard<std::mutex> l(mutex_);
  return high_pri_pool_ratio_;
}

size_t BinnedLRUCacheShard::GetHighPriPoolUsage() const {
  std::lock_guard<std::mutex> l(mutex_);
  return high_pri_pool_usage_;
}

void BinnedLRUCacheShard::LRU_Remove(BinnedLRUHandle* e) {
  ceph_assert(e->next != nullptr);
  ceph_assert(e->prev != nullptr);
  if (lru_low_pri_ == e) {
    lru_low_pri_ = e->prev;
  }
  e->next->prev = e->prev;
  e->prev->next = e->next;
  e->prev = e->next = nullptr;
  lru_usage_ -= e->charge;
  if (e->InHighPriPool()) {
    ceph_assert(high_pri_pool_usage_ >= e->charge);
    high_pri_pool_usage_ -= e->charge;
  } else {
    ceph_assert(*(e->age_bin) >= e->charge);
    *(e->age_bin) -= e->charge;
  }
}

void BinnedLRUCacheShard::LRU_Insert(BinnedLRUHandle* e) {
  ceph_assert(e->next == nullptr);
  ceph_assert(e->prev == nullptr);
  e->age_bin = age_bins.front();

  if (high_pri_pool_ratio_ > 0 && e->IsHighPri()) {
    // Inset "e" to head of LRU list.
    e->next = &lru_;
    e->prev = lru_.prev;
    e->prev->next = e;
    e->next->prev = e;
    e->SetInHighPriPool(true);
    high_pri_pool_usage_ += e->charge;
    MaintainPoolSize();
  } else {
    // Insert "e" to the head of low-pri pool. Note that when
    // high_pri_pool_ratio is 0, head of low-pri pool is also head of LRU list.
    e->next = lru_low_pri_->next;
    e->prev = lru_low_pri_;
    e->prev->next = e;
    e->next->prev = e;
    e->SetInHighPriPool(false);
    lru_low_pri_ = e;
    *(e->age_bin) += e->charge;
  }
  lru_usage_ += e->charge;
}

uint64_t BinnedLRUCacheShard::sum_bins(uint32_t start, uint32_t end) const {
  std::lock_guard<std::mutex> l(mutex_);
  auto size = age_bins.size();
  if (size < start) {
    return 0;
  }
  uint64_t bytes = 0;
  end = (size < end) ? size : end;
  for (auto i = start; i < end; i++) {
    bytes += *(age_bins[i]);
  }
  return bytes;
}

void BinnedLRUCacheShard::MaintainPoolSize() {
  while (high_pri_pool_usage_ > high_pri_pool_capacity_) {
    // Overflow last entry in high-pri pool to low-pri pool.
    lru_low_pri_ = lru_low_pri_->next;
    ceph_assert(lru_low_pri_ != &lru_);
    lru_low_pri_->SetInHighPriPool(false);
    high_pri_pool_usage_ -= lru_low_pri_->charge;
    *(lru_low_pri_->age_bin) += lru_low_pri_->charge;
  }
}

void BinnedLRUCacheShard::EvictFromLRU(size_t charge,
                                 ceph::autovector<BinnedLRUHandle*>* deleted) {
  while (usage_ + charge > capacity_ && lru_.next != &lru_) {
    BinnedLRUHandle* old = lru_.next;
    ceph_assert(old->InCache());
    ceph_assert(old->refs == 1);  // LRU list contains elements which may be evicted
    LRU_Remove(old);
    table_.Remove(old->key(), old->hash);
    old->SetInCache(false);
    Unref(old);
    usage_ -= old->charge;
    deleted->push_back(old);
  }
}

void BinnedLRUCacheShard::SetCapacity(size_t capacity) {
  ceph::autovector<BinnedLRUHandle*> last_reference_list;
  {
    std::lock_guard<std::mutex> l(mutex_);
    capacity_ = capacity;
    high_pri_pool_capacity_ = capacity_ * high_pri_pool_ratio_;
    EvictFromLRU(0, &last_reference_list);
  }
  // we free the entries here outside of mutex for
  // performance reasons
  for (auto entry : last_reference_list) {
    entry->Free();
  }
}

void BinnedLRUCacheShard::SetStrictCapacityLimit(bool strict_capacity_limit) {
  std::lock_guard<std::mutex> l(mutex_);
  strict_capacity_limit_ = strict_capacity_limit;
}

rocksdb::Cache::Handle* BinnedLRUCacheShard::Lookup(const rocksdb::Slice& key, uint32_t hash) {
  std::lock_guard<std::mutex> l(mutex_);
  BinnedLRUHandle* e = table_.Lookup(key, hash);
  if (e != nullptr) {
    ceph_assert(e->InCache());
    if (e->refs == 1) {
      LRU_Remove(e);
    }
    e->refs++;
    e->SetHit();
  }
  return reinterpret_cast<rocksdb::Cache::Handle*>(e);
}

bool BinnedLRUCacheShard::Ref(rocksdb::Cache::Handle* h) {
  BinnedLRUHandle* handle = reinterpret_cast<BinnedLRUHandle*>(h);
  std::lock_guard<std::mutex> l(mutex_);
  if (handle->InCache() && handle->refs == 1) {
    LRU_Remove(handle);
  }
  handle->refs++;
  return true;
}

void BinnedLRUCacheShard::SetHighPriPoolRatio(double high_pri_pool_ratio) {
  std::lock_guard<std::mutex> l(mutex_);
  high_pri_pool_ratio_ = high_pri_pool_ratio;
  high_pri_pool_capacity_ = capacity_ * high_pri_pool_ratio_;
  MaintainPoolSize();
}

bool BinnedLRUCacheShard::Release(rocksdb::Cache::Handle* handle, bool force_erase) {
  if (handle == nullptr) {
    return false;
  }
  BinnedLRUHandle* e = reinterpret_cast<BinnedLRUHandle*>(handle);
  bool last_reference = false;
  {
    std::lock_guard<std::mutex> l(mutex_);
    last_reference = Unref(e);
    if (last_reference) {
      usage_ -= e->charge;
    }
    if (e->refs == 1 && e->InCache()) {
      // The item is still in cache, and nobody else holds a reference to it
      if (usage_ > capacity_ || force_erase) {
        // the cache is full
        // The LRU list must be empty since the cache is full
        ceph_assert(!(usage_ > capacity_) || lru_.next == &lru_);
        // take this opportunity and remove the item
        table_.Remove(e->key(), e->hash);
        e->SetInCache(false);
        Unref(e);
        usage_ -= e->charge;
        last_reference = true;
      } else {
        // put the item on the list to be potentially freed
        LRU_Insert(e);
      }
    }
  }

  // free outside of mutex
  if (last_reference) {
    e->Free();
  }
  return last_reference;
}

rocksdb::Status BinnedLRUCacheShard::Insert(const rocksdb::Slice& key, uint32_t hash, void* value,
                             size_t charge,
                             DeleterFn deleter,
                             rocksdb::Cache::Handle** handle, rocksdb::Cache::Priority priority) {
  auto e = new BinnedLRUHandle();
  rocksdb::Status s;
  ceph::autovector<BinnedLRUHandle*> last_reference_list;

  e->value = value;
  e->deleter = deleter;
  e->charge = charge;
  e->key_length = key.size();
  e->key_data = new char[e->key_length];
  e->flags = 0;
  e->hash = hash;
  e->refs = (handle == nullptr
                 ? 1
                 : 2);  // One from BinnedLRUCache, one for the returned handle
  e->next = e->prev = nullptr;
  e->SetInCache(true);
  e->SetPriority(priority);
  std::copy_n(key.data(), e->key_length, e->key_data);

  {
    std::lock_guard<std::mutex> l(mutex_);
    // Free the space following strict LRU policy until enough space
    // is freed or the lru list is empty
    EvictFromLRU(charge, &last_reference_list);

    if (usage_ - lru_usage_ + charge > capacity_ &&
        (strict_capacity_limit_ || handle == nullptr)) {
      if (handle == nullptr) {
        // Don't insert the entry but still return ok, as if the entry inserted
        // into cache and get evicted immediately.
        last_reference_list.push_back(e);
      } else {
        delete e;
        *handle = nullptr;
        s = rocksdb::Status::Incomplete("Insert failed due to LRU cache being full.");
      }
    } else {
      // insert into the cache
      // note that the cache might get larger than its capacity if not enough
      // space was freed
      BinnedLRUHandle* old = table_.Insert(e);
      usage_ += e->charge;
      if (old != nullptr) {
        old->SetInCache(false);
        if (Unref(old)) {
          usage_ -= old->charge;
          // old is on LRU because it's in cache and its reference count
          // was just 1 (Unref returned 0)
          LRU_Remove(old);
          last_reference_list.push_back(old);
        }
      }
      if (handle == nullptr) {
        LRU_Insert(e);
      } else {
        *handle = reinterpret_cast<rocksdb::Cache::Handle*>(e);
      }
      s = rocksdb::Status::OK();
    }
  }

  // we free the entries here outside of mutex for
  // performance reasons
  for (auto entry : last_reference_list) {
    entry->Free();
  }

  return s;
}

void BinnedLRUCacheShard::Erase(const rocksdb::Slice& key, uint32_t hash) {
  BinnedLRUHandle* e;
  bool last_reference = false;
  {
    std::lock_guard<std::mutex> l(mutex_);
    e = table_.Remove(key, hash);
    if (e != nullptr) {
      last_reference = Unref(e);
      if (last_reference) {
        usage_ -= e->charge;
      }
      if (last_reference && e->InCache()) {
        LRU_Remove(e);
      }
      e->SetInCache(false);
    }
  }

  // mutex not held here
  // last_reference will only be true if e != nullptr
  if (last_reference) {
    e->Free();
  }
}

size_t BinnedLRUCacheShard::GetUsage() const {
  std::lock_guard<std::mutex> l(mutex_);
  return usage_;
}

size_t BinnedLRUCacheShard::GetPinnedUsage() const {
  std::lock_guard<std::mutex> l(mutex_);
  ceph_assert(usage_ >= lru_usage_);
  return usage_ - lru_usage_;
}

void BinnedLRUCacheShard::shift_bins() {
  std::lock_guard<std::mutex> l(mutex_);
  age_bins.push_front(std::make_shared<uint64_t>(0));
}

uint32_t BinnedLRUCacheShard::get_bin_count() const {
  std::lock_guard<std::mutex> l(mutex_);
  return age_bins.capacity();
}

void BinnedLRUCacheShard::set_bin_count(uint32_t count) {
  std::lock_guard<std::mutex> l(mutex_);
  age_bins.set_capacity(count);
}

std::string BinnedLRUCacheShard::GetPrintableOptions() const {
  const int kBufferSize = 200;
  char buffer[kBufferSize];
  {
    std::lock_guard<std::mutex> l(mutex_);
    snprintf(buffer, kBufferSize, "    high_pri_pool_ratio: %.3lf\n",
             high_pri_pool_ratio_);
  }
  return std::string(buffer);
}

DeleterFn BinnedLRUCacheShard::GetDeleter(rocksdb::Cache::Handle* h) const
{
  auto* handle = reinterpret_cast<BinnedLRUHandle*>(h);
  return handle->deleter;
}

BinnedLRUCache::BinnedLRUCache(CephContext *c, 
                               size_t capacity, 
                               int num_shard_bits,
                               bool strict_capacity_limit, 
                               double high_pri_pool_ratio)
    : ShardedCache(capacity, num_shard_bits, strict_capacity_limit), cct(c) {
  num_shards_ = 1 << num_shard_bits;
  // TODO: Switch over to use mempool
  int rc = posix_memalign((void**) &shards_, 
                          CACHE_LINE_SIZE, 
                          sizeof(BinnedLRUCacheShard) * num_shards_);
  if (rc != 0) {
    throw std::bad_alloc();
  } 
  size_t per_shard = (capacity + (num_shards_ - 1)) / num_shards_;
  for (int i = 0; i < num_shards_; i++) {
    new (&shards_[i])
        BinnedLRUCacheShard(c, per_shard, strict_capacity_limit, high_pri_pool_ratio);
  }
}

BinnedLRUCache::~BinnedLRUCache() {
  for (int i = 0; i < num_shards_; i++) {
    shards_[i].~BinnedLRUCacheShard();
  }
  aligned_free(shards_);
}

CacheShard* BinnedLRUCache::GetShard(int shard) {
  return reinterpret_cast<CacheShard*>(&shards_[shard]);
}

const CacheShard* BinnedLRUCache::GetShard(int shard) const {
  return reinterpret_cast<CacheShard*>(&shards_[shard]);
}

void* BinnedLRUCache::Value(Handle* handle) {
  return reinterpret_cast<const BinnedLRUHandle*>(handle)->value;
}

size_t BinnedLRUCache::GetCharge(Handle* handle) const {
  return reinterpret_cast<const BinnedLRUHandle*>(handle)->charge;
}

uint32_t BinnedLRUCache::GetHash(Handle* handle) const {
  return reinterpret_cast<const BinnedLRUHandle*>(handle)->hash;
}

void BinnedLRUCache::DisownData() {
// Do not drop data if compile with ASAN to suppress leak warning.
#ifndef __SANITIZE_ADDRESS__
  shards_ = nullptr;
#endif  // !__SANITIZE_ADDRESS__
}

#if (ROCKSDB_MAJOR >= 7 || (ROCKSDB_MAJOR == 6 && ROCKSDB_MINOR >= 22))
DeleterFn BinnedLRUCache::GetDeleter(Handle* handle) const
{
  return reinterpret_cast<const BinnedLRUHandle*>(handle)->deleter;
}
#endif

size_t BinnedLRUCache::TEST_GetLRUSize() {
  size_t lru_size_of_all_shards = 0;
  for (int i = 0; i < num_shards_; i++) {
    lru_size_of_all_shards += shards_[i].TEST_GetLRUSize();
  }
  return lru_size_of_all_shards;
}

void BinnedLRUCache::SetHighPriPoolRatio(double high_pri_pool_ratio) {
  for (int i = 0; i < num_shards_; i++) {
    shards_[i].SetHighPriPoolRatio(high_pri_pool_ratio);
  }
}

double BinnedLRUCache::GetHighPriPoolRatio() const {
  double result = 0.0;
  if (num_shards_ > 0) {
    result = shards_[0].GetHighPriPoolRatio();
  }
  return result;
}

size_t BinnedLRUCache::GetHighPriPoolUsage() const {
  // We will not lock the cache when getting the usage from shards.
  size_t usage = 0;
  for (int s = 0; s < num_shards_; s++) {
    usage += shards_[s].GetHighPriPoolUsage();
  }
  return usage;
}

// PriCache

int64_t BinnedLRUCache::request_cache_bytes(PriorityCache::Priority pri, uint64_t total_cache) const
{
  int64_t assigned = get_cache_bytes(pri);
  int64_t request = 0;

  switch(pri) {
  // PRI0 is for rocksdb's high priority items (indexes/filters)
  case PriorityCache::Priority::PRI0:
    {
      // Because we want the high pri cache to grow independently of the low
      // pri cache, request a chunky allocation independent of the other
      // priorities.
      request = PriorityCache::get_chunk(GetHighPriPoolUsage(), total_cache);
      break;
    }
  case PriorityCache::Priority::LAST:
    {
      auto max = get_bin_count();
      request = GetUsage();
      request -= GetHighPriPoolUsage();
      request -= sum_bins(0, max);
      break;
    }
  default:
    {
      ceph_assert(pri > 0 && pri < PriorityCache::Priority::LAST);
      auto prev_pri = static_cast<PriorityCache::Priority>(pri - 1);
      uint64_t start = get_bins(prev_pri);
      uint64_t end = get_bins(pri);
      request = sum_bins(start, end);
      break;
    }
  }
  request = (request > assigned) ? request - assigned : 0;
  ldout(cct, 10) << __func__ << " Priority: " << static_cast<uint32_t>(pri)
                 << " Request: " << request << dendl;
  return request;
}

int64_t BinnedLRUCache::commit_cache_size(uint64_t total_bytes)
{
  size_t old_bytes = GetCapacity();
  int64_t new_bytes = PriorityCache::get_chunk(
      get_cache_bytes(), total_bytes);
  ldout(cct, 10) << __func__ << " old: " << old_bytes
                 << " new: " << new_bytes << dendl;
  SetCapacity((size_t) new_bytes);

  double ratio = 0;
  if (new_bytes > 0) {
    int64_t pri0_bytes = get_cache_bytes(PriorityCache::Priority::PRI0);
    ratio = (double) pri0_bytes / new_bytes;
  }
  ldout(cct, 5) << __func__ << " High Pri Pool Ratio set to " << ratio << dendl;
  SetHighPriPoolRatio(ratio);
  return new_bytes;
}

void BinnedLRUCache::shift_bins() {
  for (int s = 0; s < num_shards_; s++) {
    shards_[s].shift_bins();
  }
}

uint64_t BinnedLRUCache::sum_bins(uint32_t start, uint32_t end) const {
  uint64_t bytes = 0;
  for (int s = 0; s < num_shards_; s++) {
    bytes += shards_[s].sum_bins(start, end);
  }
  return bytes;
}

uint32_t BinnedLRUCache::get_bin_count() const {
  uint32_t result = 0;
  if (num_shards_ > 0) {
    result = shards_[0].get_bin_count();
  }
  return result;
}

void BinnedLRUCache::set_bin_count(uint32_t count) {
  for (int s = 0; s < num_shards_; s++) {
    shards_[s].set_bin_count(count);
  }
}

std::shared_ptr<rocksdb::Cache> NewBinnedLRUCache(
    CephContext *c, 
    size_t capacity,
    int num_shard_bits,
    bool strict_capacity_limit,
    double high_pri_pool_ratio) {
  if (num_shard_bits >= 20) {
    return nullptr;  // the cache cannot be sharded into too many fine pieces
  }
  if (high_pri_pool_ratio < 0.0 || high_pri_pool_ratio > 1.0) {
    // invalid high_pri_pool_ratio
    return nullptr;
  }
  if (num_shard_bits < 0) {
    num_shard_bits = GetDefaultCacheShardBits(capacity);
  }
  return std::make_shared<BinnedLRUCache>(
      c, capacity, num_shard_bits, strict_capacity_limit, high_pri_pool_ratio);
}

}  // namespace rocksdb_cache