2019-09-16 19:31:27 +02:00
|
|
|
// 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).
|
|
|
|
//
|
|
|
|
// 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.
|
|
|
|
|
|
|
|
#pragma once
|
2021-11-11 01:18:27 +01:00
|
|
|
#include <algorithm>
|
2019-09-16 19:31:27 +02:00
|
|
|
#include <atomic>
|
|
|
|
#include <sstream>
|
|
|
|
#include <string>
|
2020-06-29 23:51:57 +02:00
|
|
|
|
2021-11-11 01:18:27 +01:00
|
|
|
#include "file/readahead_file_info.h"
|
2019-09-16 19:31:27 +02:00
|
|
|
#include "port/port.h"
|
|
|
|
#include "rocksdb/env.h"
|
2020-06-29 23:51:57 +02:00
|
|
|
#include "rocksdb/options.h"
|
2019-09-16 19:31:27 +02:00
|
|
|
#include "util/aligned_buffer.h"
|
|
|
|
|
2020-02-20 21:07:53 +01:00
|
|
|
namespace ROCKSDB_NAMESPACE {
|
2019-09-16 19:31:27 +02:00
|
|
|
|
2021-11-11 01:18:27 +01:00
|
|
|
#define DEAFULT_DECREMENT 8 * 1024
|
|
|
|
|
2021-11-20 02:52:42 +01:00
|
|
|
struct IOOptions;
|
|
|
|
class RandomAccessFileReader;
|
|
|
|
|
2019-09-16 19:31:27 +02:00
|
|
|
// FilePrefetchBuffer is a smart buffer to store and read data from a file.
|
|
|
|
class FilePrefetchBuffer {
|
|
|
|
public:
|
2021-04-28 21:52:53 +02:00
|
|
|
static const int kMinNumFileReadsToStartAutoReadahead = 2;
|
2019-09-16 19:31:27 +02:00
|
|
|
// Constructor.
|
|
|
|
//
|
|
|
|
// All arguments are optional.
|
|
|
|
// readahead_size : the initial readahead size.
|
|
|
|
// max_readahead_size : the maximum readahead size.
|
|
|
|
// If max_readahead_size > readahead_size, the readahead size will be
|
|
|
|
// doubled on every IO until max_readahead_size is hit.
|
|
|
|
// Typically this is set as a multiple of readahead_size.
|
|
|
|
// max_readahead_size should be greater than equal to readahead_size.
|
|
|
|
// enable : controls whether reading from the buffer is enabled.
|
|
|
|
// If false, TryReadFromCache() always return false, and we only take stats
|
|
|
|
// for the minimum offset if track_min_offset = true.
|
|
|
|
// track_min_offset : Track the minimum offset ever read and collect stats on
|
|
|
|
// it. Used for adaptable readahead of the file footer/metadata.
|
2021-04-28 21:52:53 +02:00
|
|
|
// implicit_auto_readahead : Readahead is enabled implicitly by rocksdb after
|
|
|
|
// doing sequential scans for two times.
|
2019-09-16 19:31:27 +02:00
|
|
|
//
|
2021-11-20 02:52:42 +01:00
|
|
|
// Automatic readhead is enabled for a file if readahead_size
|
2019-09-16 19:31:27 +02:00
|
|
|
// and max_readahead_size are passed in.
|
|
|
|
// A user can construct a FilePrefetchBuffer without any arguments, but use
|
|
|
|
// `Prefetch` to load data into the buffer.
|
2021-11-20 02:52:42 +01:00
|
|
|
FilePrefetchBuffer(size_t readahead_size = 0, size_t max_readahead_size = 0,
|
2021-04-28 21:52:53 +02:00
|
|
|
bool enable = true, bool track_min_offset = false,
|
|
|
|
bool implicit_auto_readahead = false)
|
2019-09-16 19:31:27 +02:00
|
|
|
: buffer_offset_(0),
|
2021-02-18 23:29:36 +01:00
|
|
|
readahead_size_(readahead_size),
|
2019-09-16 19:31:27 +02:00
|
|
|
max_readahead_size_(max_readahead_size),
|
2021-04-28 21:52:53 +02:00
|
|
|
initial_readahead_size_(readahead_size),
|
2019-09-16 19:31:27 +02:00
|
|
|
min_offset_read_(port::kMaxSizet),
|
|
|
|
enable_(enable),
|
2021-04-28 21:52:53 +02:00
|
|
|
track_min_offset_(track_min_offset),
|
|
|
|
implicit_auto_readahead_(implicit_auto_readahead),
|
|
|
|
prev_offset_(0),
|
|
|
|
prev_len_(0),
|
|
|
|
num_file_reads_(kMinNumFileReadsToStartAutoReadahead + 1) {}
|
2019-09-16 19:31:27 +02:00
|
|
|
|
|
|
|
// Load data into the buffer from a file.
|
|
|
|
// reader : the file reader.
|
|
|
|
// offset : the file offset to start reading from.
|
|
|
|
// n : the number of bytes to read.
|
|
|
|
// for_compaction : if prefetch is done for compaction read.
|
2020-06-29 23:51:57 +02:00
|
|
|
Status Prefetch(const IOOptions& opts, RandomAccessFileReader* reader,
|
|
|
|
uint64_t offset, size_t n, bool for_compaction = false);
|
2019-09-16 19:31:27 +02:00
|
|
|
|
2021-11-20 02:52:42 +01:00
|
|
|
// Tries returning the data for a file read from this buffer if that data is
|
2019-09-16 19:31:27 +02:00
|
|
|
// in the buffer.
|
|
|
|
// It handles tracking the minimum read offset if track_min_offset = true.
|
2021-02-18 23:29:36 +01:00
|
|
|
// It also does the exponential readahead when readahead_size is set as part
|
2019-09-16 19:31:27 +02:00
|
|
|
// of the constructor.
|
|
|
|
//
|
2021-11-20 02:52:42 +01:00
|
|
|
// opts : the IO options to use.
|
|
|
|
// reader : the file reader.
|
|
|
|
// offset : the file offset.
|
|
|
|
// n : the number of bytes.
|
|
|
|
// result : output buffer to put the data into.
|
|
|
|
// s : output status.
|
|
|
|
// for_compaction : true if cache read is done for compaction read.
|
|
|
|
bool TryReadFromCache(const IOOptions& opts, RandomAccessFileReader* reader,
|
|
|
|
uint64_t offset, size_t n, Slice* result, Status* s,
|
|
|
|
bool for_compaction = false);
|
2019-09-16 19:31:27 +02:00
|
|
|
|
|
|
|
// The minimum `offset` ever passed to TryReadFromCache(). This will nly be
|
|
|
|
// tracked if track_min_offset = true.
|
|
|
|
size_t min_offset_read() const { return min_offset_read_; }
|
|
|
|
|
2021-11-11 01:18:27 +01:00
|
|
|
void UpdateReadPattern(const uint64_t& offset, const size_t& len,
|
|
|
|
bool is_adaptive_readahead = false) {
|
|
|
|
if (is_adaptive_readahead) {
|
|
|
|
// Since this block was eligible for prefetch but it was found in
|
|
|
|
// cache, so check and decrease the readahead_size by 8KB (default)
|
|
|
|
// if eligible.
|
|
|
|
DecreaseReadAheadIfEligible(offset, len);
|
|
|
|
}
|
2021-04-28 21:52:53 +02:00
|
|
|
prev_offset_ = offset;
|
|
|
|
prev_len_ = len;
|
|
|
|
}
|
|
|
|
|
|
|
|
bool IsBlockSequential(const size_t& offset) {
|
|
|
|
return (prev_len_ == 0 || (prev_offset_ + prev_len_ == offset));
|
|
|
|
}
|
|
|
|
|
|
|
|
void ResetValues() {
|
|
|
|
num_file_reads_ = 1;
|
|
|
|
readahead_size_ = initial_readahead_size_;
|
|
|
|
}
|
|
|
|
|
2021-11-11 01:18:27 +01:00
|
|
|
void GetReadaheadState(ReadaheadFileInfo::ReadaheadInfo* readahead_info) {
|
|
|
|
readahead_info->readahead_size = readahead_size_;
|
|
|
|
readahead_info->num_file_reads = num_file_reads_;
|
|
|
|
}
|
|
|
|
|
|
|
|
void DecreaseReadAheadIfEligible(uint64_t offset, size_t size,
|
|
|
|
size_t value = DEAFULT_DECREMENT) {
|
|
|
|
// Decrease the readahead_size if
|
|
|
|
// - its enabled internally by RocksDB (implicit_auto_readahead_) and,
|
|
|
|
// - readahead_size is greater than 0 and,
|
|
|
|
// - this block would have called prefetch API if not found in cache for
|
|
|
|
// which conditions are:
|
|
|
|
// - few/no bytes are in buffer and,
|
|
|
|
// - block is sequential with the previous read and,
|
|
|
|
// - num_file_reads_ + 1 (including this read) >
|
|
|
|
// kMinNumFileReadsToStartAutoReadahead
|
|
|
|
if (implicit_auto_readahead_ && readahead_size_ > 0) {
|
|
|
|
if ((offset + size > buffer_offset_ + buffer_.CurrentSize()) &&
|
|
|
|
IsBlockSequential(offset) &&
|
|
|
|
(num_file_reads_ + 1 > kMinNumFileReadsToStartAutoReadahead)) {
|
|
|
|
readahead_size_ =
|
|
|
|
std::max(initial_readahead_size_,
|
|
|
|
(readahead_size_ >= value ? readahead_size_ - value : 0));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-09-16 19:31:27 +02:00
|
|
|
private:
|
|
|
|
AlignedBuffer buffer_;
|
|
|
|
uint64_t buffer_offset_;
|
|
|
|
size_t readahead_size_;
|
2021-11-11 01:18:27 +01:00
|
|
|
// FilePrefetchBuffer object won't be created from Iterator flow if
|
|
|
|
// max_readahead_size_ = 0.
|
2019-09-16 19:31:27 +02:00
|
|
|
size_t max_readahead_size_;
|
2021-04-28 21:52:53 +02:00
|
|
|
size_t initial_readahead_size_;
|
2019-09-16 19:31:27 +02:00
|
|
|
// The minimum `offset` ever passed to TryReadFromCache().
|
|
|
|
size_t min_offset_read_;
|
|
|
|
// if false, TryReadFromCache() always return false, and we only take stats
|
|
|
|
// for track_min_offset_ if track_min_offset_ = true
|
|
|
|
bool enable_;
|
|
|
|
// If true, track minimum `offset` ever passed to TryReadFromCache(), which
|
|
|
|
// can be fetched from min_offset_read().
|
|
|
|
bool track_min_offset_;
|
2021-04-28 21:52:53 +02:00
|
|
|
|
2021-11-11 01:18:27 +01:00
|
|
|
// implicit_auto_readahead is enabled by rocksdb internally after 2
|
|
|
|
// sequential IOs.
|
2021-04-28 21:52:53 +02:00
|
|
|
bool implicit_auto_readahead_;
|
2021-11-11 01:18:27 +01:00
|
|
|
uint64_t prev_offset_;
|
2021-04-28 21:52:53 +02:00
|
|
|
size_t prev_len_;
|
2021-11-11 01:18:27 +01:00
|
|
|
int64_t num_file_reads_;
|
2019-09-16 19:31:27 +02:00
|
|
|
};
|
2020-02-20 21:07:53 +01:00
|
|
|
} // namespace ROCKSDB_NAMESPACE
|