Skip to main content

Browse the archive

Revision 4fc216649ddb40763963140eac3eb95890b32486 authored by Cheng Chang on 20 March 2020, 23:15:40 UTC, committed by Facebook GitHub Bot on 20 March 2020, 23:33:26 UTC 
Support direct IO in RandomAccessFileReader::MultiRead (#6446)
 
Summary:
By supporting direct IO in RandomAccessFileReader::MultiRead, the benefits of parallel IO (IO uring) and direct IO can be combined.

In direct IO mode, read requests are aligned and merged together before being issued to RandomAccessFile::MultiRead, so blocks in the original requests might share the same underlying buffer, the shared buffers are returned in `aligned_bufs`, which is a new parameter of the `MultiRead` API.

For example, suppose alignment requirement for direct IO is 4KB, one request is (offset: 1KB, len: 1KB), another request is (offset: 3KB, len: 1KB), then since they all belong to page (offset: 0, len: 4KB), `MultiRead` only reads the page with direct IO into a buffer on heap, and returns 2 Slices referencing regions in that same buffer. See `random_access_file_reader_test.cc` for more examples.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6446

Test Plan: Added a new test `random_access_file_reader_test.cc`.

Reviewed By: anand1976

Differential Revision: D20097518

Pulled By: cheng-chang

fbshipit-source-id: ca48a8faf9c3af146465c102ef6b266a363e78d1
1 parent 5fd152b
Raw File
snapshot_checker.h 
// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
//  This source code is licensed under both the GPLv2 (found in the
//  COPYING file in the root directory) and Apache 2.0 License
//  (found in the LICENSE.Apache file in the root directory).

#pragma once
#include "rocksdb/types.h"

namespace ROCKSDB_NAMESPACE {

enum class SnapshotCheckerResult : int {
  kInSnapshot = 0,
  kNotInSnapshot = 1,
  // In case snapshot is released and the checker has no clue whether
  // the given sequence is visible to the snapshot.
  kSnapshotReleased = 2,
};

// Callback class that control GC of duplicate keys in flush/compaction.
class SnapshotChecker {
 public:
  virtual ~SnapshotChecker() {}
  virtual SnapshotCheckerResult CheckInSnapshot(
      SequenceNumber sequence, SequenceNumber snapshot_sequence) const = 0;
};

class DisableGCSnapshotChecker : public SnapshotChecker {
 public:
  virtual ~DisableGCSnapshotChecker() {}
  virtual SnapshotCheckerResult CheckInSnapshot(
      SequenceNumber /*sequence*/,
      SequenceNumber /*snapshot_sequence*/) const override {
    // By returning kNotInSnapshot, we prevent all the values from being GCed
    return SnapshotCheckerResult::kNotInSnapshot;
  }
  static DisableGCSnapshotChecker* Instance() { return &instance_; }

 protected:
  static DisableGCSnapshotChecker instance_;
  explicit DisableGCSnapshotChecker() {}
};

class WritePreparedTxnDB;

// Callback class created by WritePreparedTxnDB to check if a key
// is visible by a snapshot.
class WritePreparedSnapshotChecker : public SnapshotChecker {
 public:
  explicit WritePreparedSnapshotChecker(WritePreparedTxnDB* txn_db);
  virtual ~WritePreparedSnapshotChecker() {}

  virtual SnapshotCheckerResult CheckInSnapshot(
      SequenceNumber sequence, SequenceNumber snapshot_sequence) const override;

 private:
#ifndef ROCKSDB_LITE
  const WritePreparedTxnDB* const txn_db_;
#endif  // !ROCKSDB_LITE
};

}  // namespace ROCKSDB_NAMESPACE
back to top