2016-02-10 00:12:00 +01:00
|
|
|
// Copyright (c) 2011-present, Facebook, Inc. All rights reserved.
|
2013-10-16 23:59:46 +02:00
|
|
|
// 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.
|
|
|
|
//
|
2011-03-18 23:37:00 +01:00
|
|
|
// 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.
|
|
|
|
|
2013-10-05 07:32:05 +02:00
|
|
|
#pragma once
|
2011-03-18 23:37:00 +01:00
|
|
|
#include <string>
|
|
|
|
#include <stdint.h>
|
2013-08-23 17:38:13 +02:00
|
|
|
#include "rocksdb/slice.h"
|
|
|
|
#include "rocksdb/status.h"
|
2013-09-02 08:23:40 +02:00
|
|
|
#include "rocksdb/options.h"
|
2013-10-29 01:54:09 +01:00
|
|
|
#include "rocksdb/table.h"
|
2011-03-18 23:37:00 +01:00
|
|
|
|
2017-03-04 03:09:43 +01:00
|
|
|
#include "port/port.h" // noexcept
|
|
|
|
#include "table/persistent_cache_options.h"
|
2016-09-02 23:16:31 +02:00
|
|
|
#include "util/cf_options.h"
|
2016-04-28 20:39:12 +02:00
|
|
|
|
2013-10-04 06:49:15 +02:00
|
|
|
namespace rocksdb {
|
2011-03-18 23:37:00 +01:00
|
|
|
|
|
|
|
class Block;
|
|
|
|
class RandomAccessFile;
|
|
|
|
struct ReadOptions;
|
|
|
|
|
2016-07-19 18:44:03 +02:00
|
|
|
extern bool ShouldReportDetailedTime(Env* env, Statistics* stats);
|
|
|
|
|
2014-02-05 01:21:47 +01:00
|
|
|
// the length of the magic number in bytes.
|
|
|
|
const int kMagicNumberLengthByte = 8;
|
|
|
|
|
2011-03-18 23:37:00 +01:00
|
|
|
// BlockHandle is a pointer to the extent of a file that stores a data
|
|
|
|
// block or a meta block.
|
|
|
|
class BlockHandle {
|
|
|
|
public:
|
|
|
|
BlockHandle();
|
2013-12-05 00:43:09 +01:00
|
|
|
BlockHandle(uint64_t offset, uint64_t size);
|
2011-03-18 23:37:00 +01:00
|
|
|
|
|
|
|
// The offset of the block in the file.
|
|
|
|
uint64_t offset() const { return offset_; }
|
2014-11-06 20:14:28 +01:00
|
|
|
void set_offset(uint64_t _offset) { offset_ = _offset; }
|
2011-03-18 23:37:00 +01:00
|
|
|
|
|
|
|
// The size of the stored block
|
|
|
|
uint64_t size() const { return size_; }
|
2014-11-06 20:14:28 +01:00
|
|
|
void set_size(uint64_t _size) { size_ = _size; }
|
2011-03-18 23:37:00 +01:00
|
|
|
|
|
|
|
void EncodeTo(std::string* dst) const;
|
|
|
|
Status DecodeFrom(Slice* input);
|
|
|
|
|
2014-12-23 22:24:07 +01:00
|
|
|
// Return a string that contains the copy of handle.
|
|
|
|
std::string ToString(bool hex = true) const;
|
|
|
|
|
2013-12-05 00:43:09 +01:00
|
|
|
// if the block handle's offset and size are both "0", we will view it
|
|
|
|
// as a null block handle that points to no where.
|
|
|
|
bool IsNull() const {
|
|
|
|
return offset_ == 0 && size_ == 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static const BlockHandle& NullBlockHandle() {
|
|
|
|
return kNullBlockHandle;
|
|
|
|
}
|
|
|
|
|
2011-03-18 23:37:00 +01:00
|
|
|
// Maximum encoding length of a BlockHandle
|
|
|
|
enum { kMaxEncodedLength = 10 + 10 };
|
|
|
|
|
|
|
|
private:
|
2016-11-05 17:10:51 +01:00
|
|
|
uint64_t offset_;
|
|
|
|
uint64_t size_;
|
2013-12-05 00:43:09 +01:00
|
|
|
|
|
|
|
static const BlockHandle kNullBlockHandle;
|
2011-03-18 23:37:00 +01:00
|
|
|
};
|
|
|
|
|
2015-01-15 01:24:24 +01:00
|
|
|
inline uint32_t GetCompressFormatForVersion(CompressionType compression_type,
|
|
|
|
uint32_t version) {
|
|
|
|
// snappy is not versioned
|
|
|
|
assert(compression_type != kSnappyCompression &&
|
2016-04-20 07:54:24 +02:00
|
|
|
compression_type != kXpressCompression &&
|
2015-01-15 01:24:24 +01:00
|
|
|
compression_type != kNoCompression);
|
|
|
|
// As of version 2, we encode compressed block with
|
|
|
|
// compress_format_version == 2. Before that, the version is 1.
|
|
|
|
// DO NOT CHANGE THIS FUNCTION, it affects disk format
|
|
|
|
return version >= 2 ? 2 : 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
inline bool BlockBasedTableSupportedVersion(uint32_t version) {
|
|
|
|
return version <= 2;
|
|
|
|
}
|
|
|
|
|
2011-03-18 23:37:00 +01:00
|
|
|
// Footer encapsulates the fixed information stored at the tail
|
|
|
|
// end of every table file.
|
|
|
|
class Footer {
|
|
|
|
public:
|
2014-02-05 01:21:47 +01:00
|
|
|
// Constructs a footer without specifying its table magic number.
|
|
|
|
// In such case, the table magic number of such footer should be
|
|
|
|
// initialized via @ReadFooterFromFile().
|
2015-01-13 23:33:04 +01:00
|
|
|
// Use this when you plan to load Footer with DecodeFrom(). Never use this
|
|
|
|
// when you plan to EncodeTo.
|
|
|
|
Footer() : Footer(kInvalidTableMagicNumber, 0) {}
|
2014-02-05 01:21:47 +01:00
|
|
|
|
2015-01-13 23:33:04 +01:00
|
|
|
// Use this constructor when you plan to write out the footer using
|
|
|
|
// EncodeTo(). Never use this constructor with DecodeFrom().
|
|
|
|
Footer(uint64_t table_magic_number, uint32_t version);
|
2014-05-01 20:09:32 +02:00
|
|
|
|
|
|
|
// The version of the footer in this file
|
|
|
|
uint32_t version() const { return version_; }
|
|
|
|
|
|
|
|
// The checksum type used in this file
|
|
|
|
ChecksumType checksum() const { return checksum_; }
|
|
|
|
void set_checksum(const ChecksumType c) { checksum_ = c; }
|
2011-03-18 23:37:00 +01:00
|
|
|
|
|
|
|
// The block handle for the metaindex block of the table
|
|
|
|
const BlockHandle& metaindex_handle() const { return metaindex_handle_; }
|
|
|
|
void set_metaindex_handle(const BlockHandle& h) { metaindex_handle_ = h; }
|
|
|
|
|
|
|
|
// The block handle for the index block of the table
|
2014-05-01 20:09:32 +02:00
|
|
|
const BlockHandle& index_handle() const { return index_handle_; }
|
2014-02-05 01:21:47 +01:00
|
|
|
|
2014-05-01 20:09:32 +02:00
|
|
|
void set_index_handle(const BlockHandle& h) { index_handle_ = h; }
|
2011-03-18 23:37:00 +01:00
|
|
|
|
2014-02-05 01:21:47 +01:00
|
|
|
uint64_t table_magic_number() const { return table_magic_number_; }
|
|
|
|
|
2011-03-18 23:37:00 +01:00
|
|
|
void EncodeTo(std::string* dst) const;
|
2014-02-05 01:21:47 +01:00
|
|
|
|
2015-01-13 23:33:04 +01:00
|
|
|
// Set the current footer based on the input slice.
|
|
|
|
//
|
|
|
|
// REQUIRES: table_magic_number_ is not set (i.e.,
|
|
|
|
// HasInitializedTableMagicNumber() is true). The function will initialize the
|
|
|
|
// magic number
|
2011-03-18 23:37:00 +01:00
|
|
|
Status DecodeFrom(Slice* input);
|
|
|
|
|
2014-05-01 20:09:32 +02:00
|
|
|
// Encoded length of a Footer. Note that the serialization of a Footer will
|
|
|
|
// always occupy at least kMinEncodedLength bytes. If fields are changed
|
|
|
|
// the version number should be incremented and kMaxEncodedLength should be
|
|
|
|
// increased accordingly.
|
2011-03-18 23:37:00 +01:00
|
|
|
enum {
|
2014-05-01 20:09:32 +02:00
|
|
|
// Footer version 0 (legacy) will always occupy exactly this many bytes.
|
|
|
|
// It consists of two block handles, padding, and a magic number.
|
|
|
|
kVersion0EncodedLength = 2 * BlockHandle::kMaxEncodedLength + 8,
|
2015-01-13 23:33:04 +01:00
|
|
|
// Footer of versions 1 and higher will always occupy exactly this many
|
|
|
|
// bytes. It consists of the checksum type, two block handles, padding,
|
|
|
|
// a version number (bigger than 1), and a magic number
|
|
|
|
kNewVersionsEncodedLength = 1 + 2 * BlockHandle::kMaxEncodedLength + 4 + 8,
|
2014-05-01 20:09:32 +02:00
|
|
|
kMinEncodedLength = kVersion0EncodedLength,
|
2015-01-13 23:33:04 +01:00
|
|
|
kMaxEncodedLength = kNewVersionsEncodedLength,
|
2011-03-18 23:37:00 +01:00
|
|
|
};
|
|
|
|
|
2014-02-11 23:12:58 +01:00
|
|
|
static const uint64_t kInvalidTableMagicNumber = 0;
|
2014-02-05 01:21:47 +01:00
|
|
|
|
2014-12-23 22:24:07 +01:00
|
|
|
// convert this object to a human readable form
|
|
|
|
std::string ToString() const;
|
|
|
|
|
2011-03-18 23:37:00 +01:00
|
|
|
private:
|
2014-03-20 00:04:51 +01:00
|
|
|
// REQUIRES: magic number wasn't initialized.
|
|
|
|
void set_table_magic_number(uint64_t magic_number) {
|
|
|
|
assert(!HasInitializedTableMagicNumber());
|
|
|
|
table_magic_number_ = magic_number;
|
2014-02-05 01:21:47 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
// return true if @table_magic_number_ is set to a value different
|
|
|
|
// from @kInvalidTableMagicNumber.
|
|
|
|
bool HasInitializedTableMagicNumber() const {
|
|
|
|
return (table_magic_number_ != kInvalidTableMagicNumber);
|
|
|
|
}
|
|
|
|
|
2014-05-01 20:09:32 +02:00
|
|
|
uint32_t version_;
|
|
|
|
ChecksumType checksum_;
|
2011-03-18 23:37:00 +01:00
|
|
|
BlockHandle metaindex_handle_;
|
|
|
|
BlockHandle index_handle_;
|
2014-03-20 00:04:51 +01:00
|
|
|
uint64_t table_magic_number_ = 0;
|
2011-03-18 23:37:00 +01:00
|
|
|
};
|
|
|
|
|
2013-12-05 01:35:48 +01:00
|
|
|
// Read the footer from file
|
2015-01-13 23:33:04 +01:00
|
|
|
// If enforce_table_magic_number != 0, ReadFooterFromFile() will return
|
|
|
|
// corruption if table_magic number is not equal to enforce_table_magic_number
|
Move rate_limiter, write buffering, most perf context instrumentation and most random kill out of Env
Summary: We want to keep Env a think layer for better portability. Less platform dependent codes should be moved out of Env. In this patch, I create a wrapper of file readers and writers, and put rate limiting, write buffering, as well as most perf context instrumentation and random kill out of Env. It will make it easier to maintain multiple Env in the future.
Test Plan: Run all existing unit tests.
Reviewers: anthony, kradhakrishnan, IslamAbdelRahman, yhchiang, igor
Reviewed By: igor
Subscribers: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D42321
2015-07-18 01:16:11 +02:00
|
|
|
Status ReadFooterFromFile(RandomAccessFileReader* file, uint64_t file_size,
|
2015-01-13 23:33:04 +01:00
|
|
|
Footer* footer,
|
|
|
|
uint64_t enforce_table_magic_number = 0);
|
2013-12-05 01:35:48 +01:00
|
|
|
|
2011-03-18 23:37:00 +01:00
|
|
|
// 1-byte type + 32-bit crc
|
|
|
|
static const size_t kBlockTrailerSize = 5;
|
|
|
|
|
2012-04-17 17:36:46 +02:00
|
|
|
struct BlockContents {
|
|
|
|
Slice data; // Actual contents of data
|
|
|
|
bool cachable; // True iff data can be cached
|
2013-09-02 08:23:40 +02:00
|
|
|
CompressionType compression_type;
|
2014-08-16 00:05:09 +02:00
|
|
|
std::unique_ptr<char[]> allocation;
|
|
|
|
|
2014-09-18 00:08:19 +02:00
|
|
|
BlockContents() : cachable(false), compression_type(kNoCompression) {}
|
|
|
|
|
|
|
|
BlockContents(const Slice& _data, bool _cachable,
|
2015-07-13 21:11:05 +02:00
|
|
|
CompressionType _compression_type)
|
|
|
|
: data(_data), cachable(_cachable), compression_type(_compression_type) {}
|
2014-09-18 00:08:19 +02:00
|
|
|
|
|
|
|
BlockContents(std::unique_ptr<char[]>&& _data, size_t _size, bool _cachable,
|
2015-07-13 21:11:05 +02:00
|
|
|
CompressionType _compression_type)
|
|
|
|
: data(_data.get(), _size),
|
|
|
|
cachable(_cachable),
|
|
|
|
compression_type(_compression_type),
|
|
|
|
allocation(std::move(_data)) {}
|
|
|
|
|
2016-04-28 20:39:12 +02:00
|
|
|
BlockContents(BlockContents&& other) ROCKSDB_NOEXCEPT { *this = std::move(other); }
|
2015-07-02 01:13:49 +02:00
|
|
|
|
|
|
|
BlockContents& operator=(BlockContents&& other) {
|
2015-07-08 01:58:20 +02:00
|
|
|
data = std::move(other.data);
|
|
|
|
cachable = other.cachable;
|
|
|
|
compression_type = other.compression_type;
|
|
|
|
allocation = std::move(other.allocation);
|
|
|
|
return *this;
|
2015-07-02 01:13:49 +02:00
|
|
|
}
|
2012-04-17 17:36:46 +02:00
|
|
|
};
|
|
|
|
|
|
|
|
// Read the block identified by "handle" from "file". On failure
|
|
|
|
// return non-OK. On success fill *result and return OK.
|
2015-12-16 03:20:10 +01:00
|
|
|
extern Status ReadBlockContents(
|
|
|
|
RandomAccessFileReader* file, const Footer& footer,
|
|
|
|
const ReadOptions& options, const BlockHandle& handle,
|
2016-07-19 18:44:03 +02:00
|
|
|
BlockContents* contents, const ImmutableCFOptions &ioptions,
|
|
|
|
bool do_uncompress = true, const Slice& compression_dict = Slice(),
|
|
|
|
const PersistentCacheOptions& cache_options = PersistentCacheOptions());
|
2013-09-02 08:23:40 +02:00
|
|
|
|
|
|
|
// The 'data' points to the raw block contents read in from file.
|
|
|
|
// This method allocates a new heap buffer and the raw block
|
|
|
|
// contents are uncompresed into this buffer. This buffer is
|
|
|
|
// returned via 'result' and it is upto the caller to
|
|
|
|
// free this buffer.
|
2015-01-15 01:24:24 +01:00
|
|
|
// For description of compress_format_version and possible values, see
|
|
|
|
// util/compression.h
|
2014-08-16 00:05:09 +02:00
|
|
|
extern Status UncompressBlockContents(const char* data, size_t n,
|
2015-01-15 01:24:24 +01:00
|
|
|
BlockContents* contents,
|
Shared dictionary compression using reference block
Summary:
This adds a new metablock containing a shared dictionary that is used
to compress all data blocks in the SST file. The size of the shared dictionary
is configurable in CompressionOptions and defaults to 0. It's currently only
used for zlib/lz4/lz4hc, but the block will be stored in the SST regardless of
the compression type if the user chooses a nonzero dictionary size.
During compaction, computes the dictionary by randomly sampling the first
output file in each subcompaction. It pre-computes the intervals to sample
by assuming the output file will have the maximum allowable length. In case
the file is smaller, some of the pre-computed sampling intervals can be beyond
end-of-file, in which case we skip over those samples and the dictionary will
be a bit smaller. After the dictionary is generated using the first file in a
subcompaction, it is loaded into the compression library before writing each
block in each subsequent file of that subcompaction.
On the read path, gets the dictionary from the metablock, if it exists. Then,
loads that dictionary into the compression library before reading each block.
Test Plan: new unit test
Reviewers: yhchiang, IslamAbdelRahman, cyan, sdong
Reviewed By: sdong
Subscribers: andrewkr, yoshinorim, kradhakrishnan, dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D52287
2016-04-28 02:36:03 +02:00
|
|
|
uint32_t compress_format_version,
|
2016-07-19 18:44:03 +02:00
|
|
|
const Slice& compression_dict,
|
|
|
|
const ImmutableCFOptions &ioptions);
|
2011-03-18 23:37:00 +01:00
|
|
|
|
2016-06-11 03:20:54 +02:00
|
|
|
// This is an extension to UncompressBlockContents that accepts
|
|
|
|
// a specific compression type. This is used by un-wrapped blocks
|
|
|
|
// with no compression header.
|
|
|
|
extern Status UncompressBlockContentsForCompressionType(
|
|
|
|
const char* data, size_t n, BlockContents* contents,
|
|
|
|
uint32_t compress_format_version, const Slice& compression_dict,
|
2016-07-19 18:44:03 +02:00
|
|
|
CompressionType compression_type, const ImmutableCFOptions &ioptions);
|
2016-06-11 03:20:54 +02:00
|
|
|
|
2011-03-18 23:37:00 +01:00
|
|
|
// Implementation details follow. Clients should ignore,
|
|
|
|
|
2016-11-05 17:10:51 +01:00
|
|
|
// TODO(andrewkr): we should prefer one way of representing a null/uninitialized
|
|
|
|
// BlockHandle. Currently we use zeros for null and use negation-of-zeros for
|
|
|
|
// uninitialized.
|
2011-03-18 23:37:00 +01:00
|
|
|
inline BlockHandle::BlockHandle()
|
2013-12-05 00:43:09 +01:00
|
|
|
: BlockHandle(~static_cast<uint64_t>(0),
|
|
|
|
~static_cast<uint64_t>(0)) {
|
|
|
|
}
|
|
|
|
|
2014-11-06 20:14:28 +01:00
|
|
|
inline BlockHandle::BlockHandle(uint64_t _offset, uint64_t _size)
|
|
|
|
: offset_(_offset), size_(_size) {}
|
2011-03-18 23:37:00 +01:00
|
|
|
|
2013-10-04 06:49:15 +02:00
|
|
|
} // namespace rocksdb
|