You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
 
rocksdb/table/format.h

289 lines
10 KiB

// 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
#include <stdint.h>
#include <string>
#ifdef ROCKSDB_MALLOC_USABLE_SIZE
#ifdef OS_FREEBSD
#include <malloc_np.h>
#else
#include <malloc.h>
#endif
#endif
#include "rocksdb/options.h"
#include "rocksdb/slice.h"
#include "rocksdb/status.h"
#include "rocksdb/table.h"
#include "options/cf_options.h"
#include "port/port.h" // noexcept
#include "table/persistent_cache_options.h"
#include "util/file_reader_writer.h"
#include "util/memory_allocator.h"
namespace rocksdb {
class RandomAccessFile;
struct ReadOptions;
extern bool ShouldReportDetailedTime(Env* env, Statistics* stats);
// the length of the magic number in bytes.
const int kMagicNumberLengthByte = 8;
// BlockHandle is a pointer to the extent of a file that stores a data
// block or a meta block.
class BlockHandle {
public:
BlockHandle();
BlockHandle(uint64_t offset, uint64_t size);
// The offset of the block in the file.
uint64_t offset() const { return offset_; }
void set_offset(uint64_t _offset) { offset_ = _offset; }
// The size of the stored block
uint64_t size() const { return size_; }
void set_size(uint64_t _size) { size_ = _size; }
void EncodeTo(std::string* dst) const;
Status DecodeFrom(Slice* input);
Status DecodeSizeFrom(uint64_t offset, Slice* input);
// Return a string that contains the copy of handle.
std::string ToString(bool hex = true) const;
// 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; }
// Maximum encoding length of a BlockHandle
enum { kMaxEncodedLength = 10 + 10 };
private:
uint64_t offset_;
uint64_t size_;
static const BlockHandle kNullBlockHandle;
};
inline uint32_t GetCompressFormatForVersion(
CompressionType compression_type, uint32_t version) {
#ifdef NDEBUG
(void)compression_type;
#endif
// snappy is not versioned
assert(compression_type != kSnappyCompression &&
compression_type != kXpressCompression &&
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 <= 4;
}
// Footer encapsulates the fixed information stored at the tail
// end of every table file.
class Footer {
public:
// Constructs a footer without specifying its table magic number.
// In such case, the table magic number of such footer should be
// initialized via @ReadFooterFromFile().
// Use this when you plan to load Footer with DecodeFrom(). Never use this
// when you plan to EncodeTo.
Footer() : Footer(kInvalidTableMagicNumber, 0) {}
// 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);
// 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; }
// 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
const BlockHandle& index_handle() const { return index_handle_; }
void set_index_handle(const BlockHandle& h) { index_handle_ = h; }
uint64_t table_magic_number() const { return table_magic_number_; }
void EncodeTo(std::string* dst) const;
// 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
Status DecodeFrom(Slice* input);
// 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.
enum {
// 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,
// 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,
kMinEncodedLength = kVersion0EncodedLength,
kMaxEncodedLength = kNewVersionsEncodedLength,
};
static const uint64_t kInvalidTableMagicNumber = 0;
// convert this object to a human readable form
std::string ToString() const;
private:
// REQUIRES: magic number wasn't initialized.
void set_table_magic_number(uint64_t magic_number) {
assert(!HasInitializedTableMagicNumber());
table_magic_number_ = magic_number;
}
// return true if @table_magic_number_ is set to a value different
// from @kInvalidTableMagicNumber.
bool HasInitializedTableMagicNumber() const {
return (table_magic_number_ != kInvalidTableMagicNumber);
}
uint32_t version_;
ChecksumType checksum_;
BlockHandle metaindex_handle_;
BlockHandle index_handle_;
uint64_t table_magic_number_ = 0;
};
// Read the footer from file
// If enforce_table_magic_number != 0, ReadFooterFromFile() will return
// corruption if table_magic number is not equal to enforce_table_magic_number
Status ReadFooterFromFile(RandomAccessFileReader* file,
FilePrefetchBuffer* prefetch_buffer,
uint64_t file_size, Footer* footer,
uint64_t enforce_table_magic_number = 0);
// 1-byte type + 32-bit crc
static const size_t kBlockTrailerSize = 5;
struct BlockContents {
Slice data; // Actual contents of data
bool cachable; // True iff data can be cached
CompressionType compression_type;
CacheAllocationPtr allocation;
BlockContents() : cachable(false), compression_type(kNoCompression) {}
BlockContents(const Slice& _data, bool _cachable,
CompressionType _compression_type)
: data(_data), cachable(_cachable), compression_type(_compression_type) {}
BlockContents(CacheAllocationPtr&& _data, size_t _size, bool _cachable,
CompressionType _compression_type)
: data(_data.get(), _size),
cachable(_cachable),
compression_type(_compression_type),
allocation(std::move(_data)) {}
BlockContents(std::unique_ptr<char[]>&& _data, size_t _size, bool _cachable,
CompressionType _compression_type)
: data(_data.get(), _size),
cachable(_cachable),
compression_type(_compression_type) {
allocation.reset(_data.release());
}
// The additional memory space taken by the block data.
size_t usable_size() const {
if (allocation.get() != nullptr) {
auto allocator = allocation.get_deleter().allocator;
if (allocator) {
return allocator->UsableSize(allocation.get(), data.size());
}
#ifdef ROCKSDB_MALLOC_USABLE_SIZE
return malloc_usable_size(allocation.get());
#else
return data.size();
#endif // ROCKSDB_MALLOC_USABLE_SIZE
} else {
return 0; // no extra memory is occupied by the data
}
}
BlockContents(BlockContents&& other) ROCKSDB_NOEXCEPT {
*this = std::move(other);
}
BlockContents& operator=(BlockContents&& other) {
data = std::move(other.data);
cachable = other.cachable;
compression_type = other.compression_type;
allocation = std::move(other.allocation);
return *this;
}
};
// Read the block identified by "handle" from "file". On failure
// return non-OK. On success fill *result and return OK.
extern Status ReadBlockContents(
RandomAccessFileReader* file, FilePrefetchBuffer* prefetch_buffer,
const Footer& footer, const ReadOptions& options, const BlockHandle& handle,
BlockContents* contents, const ImmutableCFOptions& ioptions,
bool do_uncompress = true, const Slice& compression_dict = Slice(),
const PersistentCacheOptions& cache_options = PersistentCacheOptions());
// 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.
// For description of compress_format_version and possible values, see
// util/compression.h
extern Status UncompressBlockContents(
const UncompressionContext& uncompression_ctx, const char* data, size_t n,
BlockContents* contents, uint32_t compress_format_version,
const ImmutableCFOptions& ioptions, MemoryAllocator* allocator = nullptr);
// 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 UncompressionContext& uncompression_ctx, const char* data, size_t n,
BlockContents* contents, uint32_t compress_format_version,
const ImmutableCFOptions& ioptions, MemoryAllocator* allocator = nullptr);
// Implementation details follow. Clients should ignore,
// 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.
inline BlockHandle::BlockHandle()
: BlockHandle(~static_cast<uint64_t>(0), ~static_cast<uint64_t>(0)) {}
inline BlockHandle::BlockHandle(uint64_t _offset, uint64_t _size)
: offset_(_offset), size_(_size) {}
} // namespace rocksdb