db_iter.cc 24.7 KB
Newer Older
1 2 3 4 5
//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
//  This source code is licensed under the BSD-style license found in the
//  LICENSE file in the root directory of this source tree. An additional grant
//  of patent rights can be found in the PATENTS file in the same directory.
//
J
jorlow@chromium.org 已提交
6 7 8 9 10
// 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.

#include "db/db_iter.h"
11
#include <stdexcept>
12
#include <deque>
S
Stanislau Hlebik 已提交
13
#include <string>
S
Stanislau Hlebik 已提交
14
#include <limits>
J
jorlow@chromium.org 已提交
15 16 17

#include "db/filename.h"
#include "db/dbformat.h"
18 19 20 21
#include "rocksdb/env.h"
#include "rocksdb/options.h"
#include "rocksdb/iterator.h"
#include "rocksdb/merge_operator.h"
J
jorlow@chromium.org 已提交
22
#include "port/port.h"
23
#include "util/arena.h"
J
jorlow@chromium.org 已提交
24 25
#include "util/logging.h"
#include "util/mutexlock.h"
26
#include "util/perf_context_imp.h"
J
jorlow@chromium.org 已提交
27

28
namespace rocksdb {
J
jorlow@chromium.org 已提交
29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49

#if 0
static void DumpInternalIter(Iterator* iter) {
  for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
    ParsedInternalKey k;
    if (!ParseInternalKey(iter->key(), &k)) {
      fprintf(stderr, "Corrupt '%s'\n", EscapeString(iter->key()).c_str());
    } else {
      fprintf(stderr, "@ '%s'\n", k.DebugString().c_str());
    }
  }
}
#endif

// Memtables and sstables that make the DB representation contain
// (userkey,seq,type) => uservalue entries.  DBIter
// combines multiple entries for the same userkey found in the DB
// representation into a single entry while accounting for sequence
// numbers, deletion markers, overwrites, etc.
class DBIter: public Iterator {
 public:
50
  // The following is grossly complicated. TODO: clean it up
J
jorlow@chromium.org 已提交
51 52 53 54 55 56 57 58 59 60
  // Which direction is the iterator currently moving?
  // (1) When moving forward, the internal iterator is positioned at
  //     the exact entry that yields this->key(), this->value()
  // (2) When moving backwards, the internal iterator is positioned
  //     just before all entries whose user key == this->key().
  enum Direction {
    kForward,
    kReverse
  };

61 62 63
  DBIter(Env* env, const ImmutableCFOptions& ioptions,
         const Comparator* cmp, Iterator* iter, SequenceNumber s,
         bool arena_mode, uint64_t max_sequential_skip_in_iterations,
64
         const Slice* iterate_upper_bound = nullptr)
65 66
      : arena_mode_(arena_mode),
        env_(env),
67
        logger_(ioptions.info_log),
J
jorlow@chromium.org 已提交
68
        user_comparator_(cmp),
69
        user_merge_operator_(ioptions.merge_operator),
J
jorlow@chromium.org 已提交
70 71
        iter_(iter),
        sequence_(s),
J
jorlow@chromium.org 已提交
72
        direction_(kForward),
73
        valid_(false),
74
        current_entry_is_merged_(false),
75
        statistics_(ioptions.statistics),
76
        iterate_upper_bound_(iterate_upper_bound) {
L
Lei Jin 已提交
77
    RecordTick(statistics_, NO_ITERATORS);
78 79
    prefix_extractor_ = ioptions.prefix_extractor;
    max_skip_ = max_sequential_skip_in_iterations;
J
jorlow@chromium.org 已提交
80 81
  }
  virtual ~DBIter() {
82
    RecordTick(statistics_, NO_ITERATORS, -1);
83 84 85 86 87 88 89 90 91
    if (!arena_mode_) {
      delete iter_;
    } else {
      iter_->~Iterator();
    }
  }
  virtual void SetIter(Iterator* iter) {
    assert(iter_ == nullptr);
    iter_ = iter;
J
jorlow@chromium.org 已提交
92
  }
I
Igor Sugak 已提交
93 94
  virtual bool Valid() const override { return valid_; }
  virtual Slice key() const override {
J
jorlow@chromium.org 已提交
95
    assert(valid_);
96
    return saved_key_.GetKey();
J
jorlow@chromium.org 已提交
97
  }
I
Igor Sugak 已提交
98
  virtual Slice value() const override {
J
jorlow@chromium.org 已提交
99
    assert(valid_);
100 101
    return (direction_ == kForward && !current_entry_is_merged_) ?
      iter_->value() : saved_value_;
J
jorlow@chromium.org 已提交
102
  }
I
Igor Sugak 已提交
103
  virtual Status status() const override {
J
jorlow@chromium.org 已提交
104 105 106 107 108 109 110
    if (status_.ok()) {
      return iter_->status();
    } else {
      return status_;
    }
  }

I
Igor Sugak 已提交
111 112 113 114 115
  virtual void Next() override;
  virtual void Prev() override;
  virtual void Seek(const Slice& target) override;
  virtual void SeekToFirst() override;
  virtual void SeekToLast() override;
J
jorlow@chromium.org 已提交
116

J
jorlow@chromium.org 已提交
117
 private:
118
  void ReverseToBackward();
S
Stanislau Hlebik 已提交
119 120 121 122 123 124
  void PrevInternal();
  void FindParseableKey(ParsedInternalKey* ikey, Direction direction);
  bool FindValueForCurrentKey();
  bool FindValueForCurrentKeyUsingSeek();
  void FindPrevUserKey();
  void FindNextUserKey();
125 126
  inline void FindNextUserEntry(bool skipping);
  void FindNextUserEntryInternal(bool skipping);
J
jorlow@chromium.org 已提交
127
  bool ParseKey(ParsedInternalKey* key);
128
  void MergeValuesNewToOld();
J
jorlow@chromium.org 已提交
129 130 131 132 133 134 135 136 137 138

  inline void ClearSavedValue() {
    if (saved_value_.capacity() > 1048576) {
      std::string empty;
      swap(empty, saved_value_);
    } else {
      saved_value_.clear();
    }
  }

139
  const SliceTransform* prefix_extractor_;
140
  bool arena_mode_;
J
jorlow@chromium.org 已提交
141
  Env* const env_;
I
Igor Canadi 已提交
142
  Logger* logger_;
J
jorlow@chromium.org 已提交
143
  const Comparator* const user_comparator_;
144
  const MergeOperator* const user_merge_operator_;
145
  Iterator* iter_;
J
jorlow@chromium.org 已提交
146
  SequenceNumber const sequence_;
J
jorlow@chromium.org 已提交
147

J
jorlow@chromium.org 已提交
148
  Status status_;
S
Stanislau Hlebik 已提交
149 150
  IterKey saved_key_;
  std::string saved_value_;
J
jorlow@chromium.org 已提交
151
  Direction direction_;
J
jorlow@chromium.org 已提交
152
  bool valid_;
153
  bool current_entry_is_merged_;
154
  Statistics* statistics_;
155
  uint64_t max_skip_;
156
  const Slice* iterate_upper_bound_;
J
jorlow@chromium.org 已提交
157 158 159 160 161 162 163 164 165

  // No copying allowed
  DBIter(const DBIter&);
  void operator=(const DBIter&);
};

inline bool DBIter::ParseKey(ParsedInternalKey* ikey) {
  if (!ParseInternalKey(iter_->key(), ikey)) {
    status_ = Status::Corruption("corrupted internal key in DBIter");
166 167
    Log(InfoLogLevel::ERROR_LEVEL,
        logger_, "corrupted internal key in DBIter: %s",
168
        iter_->key().ToString(true).c_str());
J
jorlow@chromium.org 已提交
169 170 171 172 173 174
    return false;
  } else {
    return true;
  }
}

J
jorlow@chromium.org 已提交
175 176 177
void DBIter::Next() {
  assert(valid_);

S
Stanislau Hlebik 已提交
178 179
  if (direction_ == kReverse) {
    FindNextUserKey();
J
jorlow@chromium.org 已提交
180 181 182
    direction_ = kForward;
    if (!iter_->Valid()) {
      iter_->SeekToFirst();
J
jorlow@chromium.org 已提交
183 184
    }
  }
J
jorlow@chromium.org 已提交
185

186 187 188 189 190 191
  // If the current value is merged, we might already hit end of iter_
  if (!iter_->Valid()) {
    valid_ = false;
    return;
  }
  FindNextUserEntry(true /* skipping the current user key */);
M
Manuel Ung 已提交
192 193 194 195 196 197 198
  if (statistics_ != nullptr) {
    RecordTick(statistics_, NUMBER_DB_NEXT);
    if (valid_) {
      RecordTick(statistics_, NUMBER_DB_NEXT_FOUND);
      RecordTick(statistics_, ITER_BYTES_READ, key().size() + value().size());
    }
  }
J
jorlow@chromium.org 已提交
199 200
}

201 202 203 204 205 206 207 208
// PRE: saved_key_ has the current user key if skipping
// POST: saved_key_ should have the next user key if valid_,
//       if the current entry is a result of merge
//           current_entry_is_merged_ => true
//           saved_value_             => the merged value
//
// NOTE: In between, saved_key_ can point to a user key that has
//       a delete marker
209
inline void DBIter::FindNextUserEntry(bool skipping) {
210
  PERF_TIMER_GUARD(find_next_user_entry_time);
211 212 213 214 215
  FindNextUserEntryInternal(skipping);
}

// Actual implementation of DBIter::FindNextUserEntry()
void DBIter::FindNextUserEntryInternal(bool skipping) {
J
jorlow@chromium.org 已提交
216 217 218
  // Loop until we hit an acceptable entry to yield
  assert(iter_->Valid());
  assert(direction_ == kForward);
219
  current_entry_is_merged_ = false;
220
  uint64_t num_skipped = 0;
J
jorlow@chromium.org 已提交
221
  do {
J
jorlow@chromium.org 已提交
222
    ParsedInternalKey ikey;
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

    if (ParseKey(&ikey)) {
      if (iterate_upper_bound_ != nullptr &&
          ikey.user_key.compare(*iterate_upper_bound_) >= 0) {
        break;
      }

      if (ikey.sequence <= sequence_) {
        if (skipping &&
           user_comparator_->Compare(ikey.user_key, saved_key_.GetKey()) <= 0) {
          num_skipped++;  // skip this entry
          PERF_COUNTER_ADD(internal_key_skipped_count, 1);
        } else {
          switch (ikey.type) {
            case kTypeDeletion:
              // Arrange to skip all upcoming entries for this key since
              // they are hidden by this deletion.
              saved_key_.SetKey(ikey.user_key);
              skipping = true;
              num_skipped = 0;
              PERF_COUNTER_ADD(internal_delete_skipped_count, 1);
              break;
            case kTypeValue:
              valid_ = true;
              saved_key_.SetKey(ikey.user_key);
              return;
            case kTypeMerge:
              // By now, we are sure the current ikey is going to yield a value
              saved_key_.SetKey(ikey.user_key);
              current_entry_is_merged_ = true;
              valid_ = true;
              MergeValuesNewToOld();  // Go to a different state machine
              return;
            default:
              assert(false);
              break;
          }
260
        }
J
jorlow@chromium.org 已提交
261
      }
J
jorlow@chromium.org 已提交
262
    }
263 264
    // If we have sequentially iterated via numerous keys and still not
    // found the next user-key, then it is better to seek so that we can
C
clark.kang 已提交
265
    // avoid too many key comparisons. We seek to the last occurrence of
266 267
    // our current key by looking for sequence number 0 and type deletion
    // (the smallest type).
268 269 270
    if (skipping && num_skipped > max_skip_) {
      num_skipped = 0;
      std::string last_key;
271
      AppendInternalKey(&last_key, ParsedInternalKey(saved_key_.GetKey(), 0,
272
                                                     kTypeDeletion));
273 274 275 276 277
      iter_->Seek(last_key);
      RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION);
    } else {
      iter_->Next();
    }
J
jorlow@chromium.org 已提交
278 279
  } while (iter_->Valid());
  valid_ = false;
J
jorlow@chromium.org 已提交
280 281
}

282 283 284 285 286 287 288
// Merge values of the same user key starting from the current iter_ position
// Scan from the newer entries to older entries.
// PRE: iter_->key() points to the first merge type entry
//      saved_key_ stores the user key
// POST: saved_value_ has the merged value for the user key
//       iter_ points to the next entry (or invalid)
void DBIter::MergeValuesNewToOld() {
D
Deon Nicholas 已提交
289
  if (!user_merge_operator_) {
290 291
    Log(InfoLogLevel::ERROR_LEVEL,
        logger_, "Options::merge_operator is null.");
292 293 294
    status_ = Status::InvalidArgument("user_merge_operator_ must be set.");
    valid_ = false;
    return;
D
Deon Nicholas 已提交
295
  }
296

297 298 299
  // Start the merge process by pushing the first operand
  std::deque<std::string> operands;
  operands.push_front(iter_->value().ToString());
300 301 302 303 304 305 306 307

  ParsedInternalKey ikey;
  for (iter_->Next(); iter_->Valid(); iter_->Next()) {
    if (!ParseKey(&ikey)) {
      // skip corrupted key
      continue;
    }

308
    if (!user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) {
309 310 311 312 313 314 315 316 317 318 319 320
      // hit the next user key, stop right here
      break;
    }

    if (kTypeDeletion == ikey.type) {
      // hit a delete with the same user key, stop right here
      // iter_ is positioned after delete
      iter_->Next();
      break;
    }

    if (kTypeValue == ikey.type) {
321 322 323
      // hit a put, merge the put value with operands and store the
      // final result in saved_value_. We are done!
      // ignore corruption if there is any.
I
Igor Canadi 已提交
324
      const Slice val = iter_->value();
325 326 327 328 329 330 331 332
      {
        StopWatchNano timer(env_, statistics_ != nullptr);
        PERF_TIMER_GUARD(merge_operator_time_nanos);
        user_merge_operator_->FullMerge(ikey.user_key, &val, operands,
                                        &saved_value_, logger_);
        RecordTick(statistics_, MERGE_OPERATION_TOTAL_TIME,
                   timer.ElapsedNanos());
      }
333 334 335 336 337 338
      // iter_ is positioned after put
      iter_->Next();
      return;
    }

    if (kTypeMerge == ikey.type) {
339 340
      // hit a merge, add the value as an operand and run associative merge.
      // when complete, add result to operands and continue.
I
Igor Canadi 已提交
341 342
      const Slice& val = iter_->value();
      operands.push_front(val.ToString());
343 344 345
    }
  }

346 347 348 349 350 351 352 353 354 355 356
  {
    StopWatchNano timer(env_, statistics_ != nullptr);
    PERF_TIMER_GUARD(merge_operator_time_nanos);
    // we either exhausted all internal keys under this user key, or hit
    // a deletion marker.
    // feed null as the existing value to the merge operator, such that
    // client can differentiate this scenario and do things accordingly.
    user_merge_operator_->FullMerge(saved_key_.GetKey(), nullptr, operands,
                                    &saved_value_, logger_);
    RecordTick(statistics_, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanos());
  }
357 358
}

J
jorlow@chromium.org 已提交
359 360
void DBIter::Prev() {
  assert(valid_);
S
Stanislau Hlebik 已提交
361
  if (direction_ == kForward) {
362
    ReverseToBackward();
S
Stanislau Hlebik 已提交
363 364
  }
  PrevInternal();
M
Manuel Ung 已提交
365 366 367 368 369 370 371
  if (statistics_ != nullptr) {
    RecordTick(statistics_, NUMBER_DB_PREV);
    if (valid_) {
      RecordTick(statistics_, NUMBER_DB_PREV_FOUND);
      RecordTick(statistics_, ITER_BYTES_READ, key().size() + value().size());
    }
  }
S
Stanislau Hlebik 已提交
372
}
J
jorlow@chromium.org 已提交
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
void DBIter::ReverseToBackward() {
  if (current_entry_is_merged_) {
    // Not placed in the same key. Need to call Prev() until finding the
    // previous key.
    if (!iter_->Valid()) {
      iter_->SeekToLast();
    }
    ParsedInternalKey ikey;
    FindParseableKey(&ikey, kReverse);
    while (iter_->Valid() &&
           user_comparator_->Compare(ikey.user_key, saved_key_.GetKey()) > 0) {
      iter_->Prev();
      FindParseableKey(&ikey, kReverse);
    }
  }
#ifndef NDEBUG
  if (iter_->Valid()) {
    ParsedInternalKey ikey;
    assert(ParseKey(&ikey));
    assert(user_comparator_->Compare(ikey.user_key, saved_key_.GetKey()) <= 0);
  }
#endif

  FindPrevUserKey();
  direction_ = kReverse;
}

S
Stanislau Hlebik 已提交
401 402 403 404
void DBIter::PrevInternal() {
  if (!iter_->Valid()) {
    valid_ = false;
    return;
405 406
  }

S
Stanislau Hlebik 已提交
407 408 409
  ParsedInternalKey ikey;

  while (iter_->Valid()) {
L
Lei Jin 已提交
410
    saved_key_.SetKey(ExtractUserKey(iter_->key()));
S
Stanislau Hlebik 已提交
411 412
    if (FindValueForCurrentKey()) {
      valid_ = true;
J
jorlow@chromium.org 已提交
413 414 415
      if (!iter_->Valid()) {
        return;
      }
S
Stanislau Hlebik 已提交
416
      FindParseableKey(&ikey, kReverse);
417
      if (user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) {
S
Stanislau Hlebik 已提交
418
        FindPrevUserKey();
J
jorlow@chromium.org 已提交
419
      }
S
Stanislau Hlebik 已提交
420
      return;
J
jorlow@chromium.org 已提交
421
    }
S
Stanislau Hlebik 已提交
422 423 424 425
    if (!iter_->Valid()) {
      break;
    }
    FindParseableKey(&ikey, kReverse);
426
    if (user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) {
S
Stanislau Hlebik 已提交
427 428 429 430 431 432
      FindPrevUserKey();
    }
  }
  // We haven't found any key - iterator is not valid
  assert(!iter_->Valid());
  valid_ = false;
J
jorlow@chromium.org 已提交
433 434
}

S
Stanislau Hlebik 已提交
435 436 437 438 439 440 441 442 443
// This function checks, if the entry with biggest sequence_number <= sequence_
// is non kTypeDeletion. If it's not, we save value in saved_value_
bool DBIter::FindValueForCurrentKey() {
  assert(iter_->Valid());
  // Contains operands for merge operator.
  std::deque<std::string> operands;
  // last entry before merge (could be kTypeDeletion or kTypeValue)
  ValueType last_not_merge_type = kTypeDeletion;
  ValueType last_key_entry_type = kTypeDeletion;
J
jorlow@chromium.org 已提交
444

S
Stanislau Hlebik 已提交
445 446 447 448 449
  ParsedInternalKey ikey;
  FindParseableKey(&ikey, kReverse);

  size_t num_skipped = 0;
  while (iter_->Valid() && ikey.sequence <= sequence_ &&
450
         user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) {
S
Stanislau Hlebik 已提交
451 452 453 454 455 456 457 458 459 460 461 462 463 464 465
    // We iterate too much: let's use Seek() to avoid too much key comparisons
    if (num_skipped >= max_skip_) {
      return FindValueForCurrentKeyUsingSeek();
    }

    last_key_entry_type = ikey.type;
    switch (last_key_entry_type) {
      case kTypeValue:
        operands.clear();
        saved_value_ = iter_->value().ToString();
        last_not_merge_type = kTypeValue;
        break;
      case kTypeDeletion:
        operands.clear();
        last_not_merge_type = kTypeDeletion;
466
        PERF_COUNTER_ADD(internal_delete_skipped_count, 1);
S
Stanislau Hlebik 已提交
467 468 469 470 471 472 473 474 475
        break;
      case kTypeMerge:
        assert(user_merge_operator_ != nullptr);
        operands.push_back(iter_->value().ToString());
        break;
      default:
        assert(false);
    }

476
    PERF_COUNTER_ADD(internal_key_skipped_count, 1);
477
    assert(user_comparator_->Equal(ikey.user_key, saved_key_.GetKey()));
S
Stanislau Hlebik 已提交
478 479 480 481 482 483 484 485 486 487 488
    iter_->Prev();
    ++num_skipped;
    FindParseableKey(&ikey, kReverse);
  }

  switch (last_key_entry_type) {
    case kTypeDeletion:
      valid_ = false;
      return false;
    case kTypeMerge:
      if (last_not_merge_type == kTypeDeletion) {
489 490
        StopWatchNano timer(env_, statistics_ != nullptr);
        PERF_TIMER_GUARD(merge_operator_time_nanos);
S
Stanislau Hlebik 已提交
491 492
        user_merge_operator_->FullMerge(saved_key_.GetKey(), nullptr, operands,
                                        &saved_value_, logger_);
493 494
        RecordTick(statistics_, MERGE_OPERATION_TOTAL_TIME,
                   timer.ElapsedNanos());
495
      } else {
S
Stanislau Hlebik 已提交
496 497 498
        assert(last_not_merge_type == kTypeValue);
        std::string last_put_value = saved_value_;
        Slice temp_slice(last_put_value);
499 500 501 502 503 504 505 506
        {
          StopWatchNano timer(env_, statistics_ != nullptr);
          PERF_TIMER_GUARD(merge_operator_time_nanos);
          user_merge_operator_->FullMerge(saved_key_.GetKey(), &temp_slice,
                                          operands, &saved_value_, logger_);
          RecordTick(statistics_, MERGE_OPERATION_TOTAL_TIME,
                     timer.ElapsedNanos());
        }
507
      }
S
Stanislau Hlebik 已提交
508 509 510 511 512 513 514
      break;
    case kTypeValue:
      // do nothing - we've already has value in saved_value_
      break;
    default:
      assert(false);
      break;
J
jorlow@chromium.org 已提交
515
  }
S
Stanislau Hlebik 已提交
516 517 518
  valid_ = true;
  return true;
}
J
jorlow@chromium.org 已提交
519

S
Stanislau Hlebik 已提交
520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538
// This function is used in FindValueForCurrentKey.
// We use Seek() function instead of Prev() to find necessary value
bool DBIter::FindValueForCurrentKeyUsingSeek() {
  std::string last_key;
  AppendInternalKey(&last_key, ParsedInternalKey(saved_key_.GetKey(), sequence_,
                                                 kValueTypeForSeek));
  iter_->Seek(last_key);
  RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION);

  // assume there is at least one parseable key for this user key
  ParsedInternalKey ikey;
  FindParseableKey(&ikey, kForward);

  if (ikey.type == kTypeValue || ikey.type == kTypeDeletion) {
    if (ikey.type == kTypeValue) {
      saved_value_ = iter_->value().ToString();
      valid_ = true;
      return true;
    }
J
jorlow@chromium.org 已提交
539
    valid_ = false;
S
Stanislau Hlebik 已提交
540 541 542 543 544 545 546
    return false;
  }

  // kTypeMerge. We need to collect all kTypeMerge values and save them
  // in operands
  std::deque<std::string> operands;
  while (iter_->Valid() &&
547
         user_comparator_->Equal(ikey.user_key, saved_key_.GetKey()) &&
S
Stanislau Hlebik 已提交
548 549 550 551 552 553 554
         ikey.type == kTypeMerge) {
    operands.push_front(iter_->value().ToString());
    iter_->Next();
    FindParseableKey(&ikey, kForward);
  }

  if (!iter_->Valid() ||
555
      !user_comparator_->Equal(ikey.user_key, saved_key_.GetKey()) ||
S
Stanislau Hlebik 已提交
556
      ikey.type == kTypeDeletion) {
557 558 559 560 561 562 563
    {
      StopWatchNano timer(env_, statistics_ != nullptr);
      PERF_TIMER_GUARD(merge_operator_time_nanos);
      user_merge_operator_->FullMerge(saved_key_.GetKey(), nullptr, operands,
                                      &saved_value_, logger_);
      RecordTick(statistics_, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanos());
    }
S
Stanislau Hlebik 已提交
564 565
    // Make iter_ valid and point to saved_key_
    if (!iter_->Valid() ||
566
        !user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) {
S
Stanislau Hlebik 已提交
567 568 569
      iter_->Seek(last_key);
      RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION);
    }
J
jorlow@chromium.org 已提交
570
    valid_ = true;
S
Stanislau Hlebik 已提交
571 572 573
    return true;
  }

I
Igor Canadi 已提交
574
  const Slice& val = iter_->value();
575 576 577 578 579 580 581
  {
    StopWatchNano timer(env_, statistics_ != nullptr);
    PERF_TIMER_GUARD(merge_operator_time_nanos);
    user_merge_operator_->FullMerge(saved_key_.GetKey(), &val, operands,
                                    &saved_value_, logger_);
    RecordTick(statistics_, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanos());
  }
S
Stanislau Hlebik 已提交
582 583 584 585 586 587 588 589 590 591 592 593 594 595 596
  valid_ = true;
  return true;
}

// Used in Next to change directions
// Go to next user key
// Don't use Seek(),
// because next user key will be very close
void DBIter::FindNextUserKey() {
  if (!iter_->Valid()) {
    return;
  }
  ParsedInternalKey ikey;
  FindParseableKey(&ikey, kForward);
  while (iter_->Valid() &&
597
         !user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) {
S
Stanislau Hlebik 已提交
598 599 600 601 602 603 604 605 606 607 608 609 610
    iter_->Next();
    FindParseableKey(&ikey, kForward);
  }
}

// Go to previous user_key
void DBIter::FindPrevUserKey() {
  if (!iter_->Valid()) {
    return;
  }
  size_t num_skipped = 0;
  ParsedInternalKey ikey;
  FindParseableKey(&ikey, kReverse);
611 612 613 614 615 616 617 618 619 620 621 622 623 624 625
  int cmp;
  while (iter_->Valid() && ((cmp = user_comparator_->Compare(
                                 ikey.user_key, saved_key_.GetKey())) == 0 ||
                            (cmp > 0 && ikey.sequence > sequence_))) {
    if (cmp == 0) {
      if (num_skipped >= max_skip_) {
        num_skipped = 0;
        IterKey last_key;
        last_key.SetInternalKey(ParsedInternalKey(
            saved_key_.GetKey(), kMaxSequenceNumber, kValueTypeForSeek));
        iter_->Seek(last_key.GetKey());
        RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION);
      } else {
        ++num_skipped;
      }
S
Stanislau Hlebik 已提交
626 627 628 629 630 631 632 633 634 635 636 637 638 639
    }
    iter_->Prev();
    FindParseableKey(&ikey, kReverse);
  }
}

// Skip all unparseable keys
void DBIter::FindParseableKey(ParsedInternalKey* ikey, Direction direction) {
  while (iter_->Valid() && !ParseKey(ikey)) {
    if (direction == kReverse) {
      iter_->Prev();
    } else {
      iter_->Next();
    }
J
jorlow@chromium.org 已提交
640 641
  }
}
J
jorlow@chromium.org 已提交
642

J
jorlow@chromium.org 已提交
643
void DBIter::Seek(const Slice& target) {
L
Lei Jin 已提交
644
  StopWatch sw(env_, statistics_, DB_SEEK);
645 646 647
  saved_key_.Clear();
  // now savved_key is used to store internal key.
  saved_key_.SetInternalKey(target, sequence_);
648 649 650 651 652 653

  {
    PERF_TIMER_GUARD(seek_internal_seek_time);
    iter_->Seek(saved_key_.GetKey());
  }

M
Manuel Ung 已提交
654
  RecordTick(statistics_, NUMBER_DB_SEEK);
J
jorlow@chromium.org 已提交
655
  if (iter_->Valid()) {
656 657
    direction_ = kForward;
    ClearSavedValue();
658
    FindNextUserEntry(false /* not skipping */);
M
Manuel Ung 已提交
659 660 661 662 663 664
    if (statistics_ != nullptr) {
      if (valid_) {
        RecordTick(statistics_, NUMBER_DB_SEEK_FOUND);
        RecordTick(statistics_, ITER_BYTES_READ, key().size() + value().size());
      }
    }
J
jorlow@chromium.org 已提交
665 666 667 668
  } else {
    valid_ = false;
  }
}
J
jorlow@chromium.org 已提交
669

J
jorlow@chromium.org 已提交
670
void DBIter::SeekToFirst() {
S
Stanislau Hlebik 已提交
671
  // Don't use iter_::Seek() if we set a prefix extractor
672
  // because prefix seek will be used.
673
  if (prefix_extractor_ != nullptr) {
S
Stanislau Hlebik 已提交
674 675
    max_skip_ = std::numeric_limits<uint64_t>::max();
  }
J
jorlow@chromium.org 已提交
676 677
  direction_ = kForward;
  ClearSavedValue();
678 679 680 681 682 683

  {
    PERF_TIMER_GUARD(seek_internal_seek_time);
    iter_->SeekToFirst();
  }

M
Manuel Ung 已提交
684
  RecordTick(statistics_, NUMBER_DB_SEEK);
J
jorlow@chromium.org 已提交
685
  if (iter_->Valid()) {
686
    FindNextUserEntry(false /* not skipping */);
M
Manuel Ung 已提交
687 688 689 690 691 692
    if (statistics_ != nullptr) {
      if (valid_) {
        RecordTick(statistics_, NUMBER_DB_SEEK_FOUND);
        RecordTick(statistics_, ITER_BYTES_READ, key().size() + value().size());
      }
    }
J
jorlow@chromium.org 已提交
693 694
  } else {
    valid_ = false;
J
jorlow@chromium.org 已提交
695 696 697
  }
}

J
jorlow@chromium.org 已提交
698
void DBIter::SeekToLast() {
S
Stanislau Hlebik 已提交
699
  // Don't use iter_::Seek() if we set a prefix extractor
700
  // because prefix seek will be used.
701
  if (prefix_extractor_ != nullptr) {
S
Stanislau Hlebik 已提交
702 703
    max_skip_ = std::numeric_limits<uint64_t>::max();
  }
J
jorlow@chromium.org 已提交
704 705
  direction_ = kReverse;
  ClearSavedValue();
706 707 708 709 710

  {
    PERF_TIMER_GUARD(seek_internal_seek_time);
    iter_->SeekToLast();
  }
711 712 713 714 715 716 717 718 719 720 721
  // When the iterate_upper_bound is set to a value,
  // it will seek to the last key before the
  // ReadOptions.iterate_upper_bound
  if (iter_->Valid() && iterate_upper_bound_ != nullptr) {
    saved_key_.SetKey(*iterate_upper_bound_);
    std::string last_key;
    AppendInternalKey(&last_key,
                      ParsedInternalKey(saved_key_.GetKey(), kMaxSequenceNumber,
                                        kValueTypeForSeek));

    iter_->Seek(last_key);
S
Stanislau Hlebik 已提交
722

723 724 725 726 727 728 729 730 731 732
    if (!iter_->Valid()) {
      iter_->SeekToLast();
    } else {
      iter_->Prev();
      if (!iter_->Valid()) {
        valid_ = false;
        return;
      }
    }
  }
S
Stanislau Hlebik 已提交
733
  PrevInternal();
M
Manuel Ung 已提交
734 735 736 737 738 739 740
  if (statistics_ != nullptr) {
    RecordTick(statistics_, NUMBER_DB_SEEK);
    if (valid_) {
      RecordTick(statistics_, NUMBER_DB_SEEK_FOUND);
      RecordTick(statistics_, ITER_BYTES_READ, key().size() + value().size());
    }
  }
J
jorlow@chromium.org 已提交
741 742
}

743
Iterator* NewDBIterator(Env* env, const ImmutableCFOptions& ioptions,
744 745
                        const Comparator* user_key_comparator,
                        Iterator* internal_iter,
746
                        const SequenceNumber& sequence,
747
                        uint64_t max_sequential_skip_in_iterations,
748
                        const Slice* iterate_upper_bound) {
749 750 751
  return new DBIter(env, ioptions, user_key_comparator, internal_iter, sequence,
                    false, max_sequential_skip_in_iterations,
                    iterate_upper_bound);
752 753
}

I
Igor Canadi 已提交
754
ArenaWrappedDBIter::~ArenaWrappedDBIter() { db_iter_->~DBIter(); }
755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776

void ArenaWrappedDBIter::SetDBIter(DBIter* iter) { db_iter_ = iter; }

void ArenaWrappedDBIter::SetIterUnderDBIter(Iterator* iter) {
  static_cast<DBIter*>(db_iter_)->SetIter(iter);
}

inline bool ArenaWrappedDBIter::Valid() const { return db_iter_->Valid(); }
inline void ArenaWrappedDBIter::SeekToFirst() { db_iter_->SeekToFirst(); }
inline void ArenaWrappedDBIter::SeekToLast() { db_iter_->SeekToLast(); }
inline void ArenaWrappedDBIter::Seek(const Slice& target) {
  db_iter_->Seek(target);
}
inline void ArenaWrappedDBIter::Next() { db_iter_->Next(); }
inline void ArenaWrappedDBIter::Prev() { db_iter_->Prev(); }
inline Slice ArenaWrappedDBIter::key() const { return db_iter_->key(); }
inline Slice ArenaWrappedDBIter::value() const { return db_iter_->value(); }
inline Status ArenaWrappedDBIter::status() const { return db_iter_->status(); }
void ArenaWrappedDBIter::RegisterCleanup(CleanupFunction function, void* arg1,
                                         void* arg2) {
  db_iter_->RegisterCleanup(function, arg1, arg2);
}
J
jorlow@chromium.org 已提交
777

778
ArenaWrappedDBIter* NewArenaWrappedDbIterator(
779 780
    Env* env, const ImmutableCFOptions& ioptions,
    const Comparator* user_key_comparator,
781
    const SequenceNumber& sequence,
782
    uint64_t max_sequential_skip_in_iterations,
783
    const Slice* iterate_upper_bound) {
784 785 786
  ArenaWrappedDBIter* iter = new ArenaWrappedDBIter();
  Arena* arena = iter->GetArena();
  auto mem = arena->AllocateAligned(sizeof(DBIter));
787 788 789
  DBIter* db_iter = new (mem) DBIter(env, ioptions, user_key_comparator,
      nullptr, sequence, true, max_sequential_skip_in_iterations,
      iterate_upper_bound);
790

791
  iter->SetDBIter(db_iter);
792

793
  return iter;
J
jorlow@chromium.org 已提交
794 795
}

796
}  // namespace rocksdb