2016-02-10 00:12:00 +01:00
|
|
|
// Copyright (c) 2011-present, Facebook, Inc. All rights reserved.
|
2017-07-16 01:03:42 +02:00
|
|
|
// 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).
|
2013-10-16 23:59:46 +02:00
|
|
|
//
|
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.
|
|
|
|
|
|
|
|
#include "table/format.h"
|
|
|
|
|
2019-06-06 22:52:39 +02:00
|
|
|
#include <cinttypes>
|
2019-03-28 00:13:08 +01:00
|
|
|
#include <string>
|
2014-02-05 01:21:47 +01:00
|
|
|
|
2019-05-30 23:47:29 +02:00
|
|
|
#include "block_fetcher.h"
|
2019-06-01 02:19:43 +02:00
|
|
|
#include "logging/logging.h"
|
2019-05-31 02:39:43 +02:00
|
|
|
#include "memory/memory_allocator.h"
|
2017-04-06 04:02:00 +02:00
|
|
|
#include "monitoring/perf_context_imp.h"
|
|
|
|
#include "monitoring/statistics.h"
|
[RocksDB] Added nano second stopwatch and new perf counters to track block read cost
Summary: The pupose of this diff is to expose per user-call level precise timing of block read, so that we can answer questions like: a Get() costs me 100ms, is that somehow related to loading blocks from file system, or sth else? We will answer that with EXACTLY how many blocks have been read, how much time was spent on transfering the bytes from os, how much time was spent on checksum verification and how much time was spent on block decompression, just for that one Get. A nano second stopwatch was introduced to track time with higher precision. The cost/precision of the stopwatch is also measured in unit-test. On my dev box, retrieving one time instance costs about 30ns, on average. The deviation of timing results is good enough to track 100ns-1us level events. And the overhead could be safely ignored for 100us level events (10000 instances/s), for example, a viewstate thrift call.
Test Plan: perf_context_test, also testing with viewstate shadow traffic.
Reviewers: dhruba
Reviewed By: dhruba
CC: leveldb, xjin
Differential Revision: https://reviews.facebook.net/D12351
2013-06-04 08:09:15 +02:00
|
|
|
#include "rocksdb/env.h"
|
2019-05-30 23:47:29 +02:00
|
|
|
#include "table/block_based/block.h"
|
|
|
|
#include "table/block_based/block_based_table_reader.h"
|
2015-12-16 03:20:10 +01:00
|
|
|
#include "table/persistent_cache_helper.h"
|
2011-03-18 23:37:00 +01:00
|
|
|
#include "util/coding.h"
|
2015-01-09 21:57:11 +01:00
|
|
|
#include "util/compression.h"
|
2011-03-18 23:37:00 +01:00
|
|
|
#include "util/crc32c.h"
|
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
|
|
|
#include "util/file_reader_writer.h"
|
2016-07-19 18:44:03 +02:00
|
|
|
#include "util/stop_watch.h"
|
2017-03-16 03:22:52 +01:00
|
|
|
#include "util/string_util.h"
|
|
|
|
#include "util/xxhash.h"
|
2011-03-18 23:37:00 +01:00
|
|
|
|
2013-10-04 06:49:15 +02:00
|
|
|
namespace rocksdb {
|
2011-03-18 23:37:00 +01:00
|
|
|
|
2014-05-01 20:09:32 +02:00
|
|
|
extern const uint64_t kLegacyBlockBasedTableMagicNumber;
|
|
|
|
extern const uint64_t kBlockBasedTableMagicNumber;
|
2014-05-08 02:45:27 +02:00
|
|
|
|
|
|
|
#ifndef ROCKSDB_LITE
|
|
|
|
extern const uint64_t kLegacyPlainTableMagicNumber;
|
2014-05-01 20:09:32 +02:00
|
|
|
extern const uint64_t kPlainTableMagicNumber;
|
2014-05-08 02:45:27 +02:00
|
|
|
#else
|
|
|
|
// ROCKSDB_LITE doesn't have plain table
|
|
|
|
const uint64_t kLegacyPlainTableMagicNumber = 0;
|
|
|
|
const uint64_t kPlainTableMagicNumber = 0;
|
|
|
|
#endif
|
2014-05-01 20:09:32 +02:00
|
|
|
|
2016-07-19 18:44:03 +02:00
|
|
|
bool ShouldReportDetailedTime(Env* env, Statistics* stats) {
|
|
|
|
return env != nullptr && stats != nullptr &&
|
2019-03-01 19:39:00 +01:00
|
|
|
stats->get_stats_level() > kExceptDetailedTimers;
|
2016-07-19 18:44:03 +02:00
|
|
|
}
|
|
|
|
|
2011-03-18 23:37:00 +01:00
|
|
|
void BlockHandle::EncodeTo(std::string* dst) const {
|
|
|
|
// Sanity check that all fields have been set
|
|
|
|
assert(offset_ != ~static_cast<uint64_t>(0));
|
|
|
|
assert(size_ != ~static_cast<uint64_t>(0));
|
2016-06-13 18:57:43 +02:00
|
|
|
PutVarint64Varint64(dst, offset_, size_);
|
2011-03-18 23:37:00 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
Status BlockHandle::DecodeFrom(Slice* input) {
|
2019-03-28 00:13:08 +01:00
|
|
|
if (GetVarint64(input, &offset_) && GetVarint64(input, &size_)) {
|
2011-03-18 23:37:00 +01:00
|
|
|
return Status::OK();
|
|
|
|
} else {
|
2016-11-05 17:10:51 +01:00
|
|
|
// reset in case failure after partially decoding
|
|
|
|
offset_ = 0;
|
|
|
|
size_ = 0;
|
2011-03-18 23:37:00 +01:00
|
|
|
return Status::Corruption("bad block handle");
|
|
|
|
}
|
|
|
|
}
|
2014-12-23 22:24:07 +01:00
|
|
|
|
2018-08-10 01:49:45 +02:00
|
|
|
Status BlockHandle::DecodeSizeFrom(uint64_t _offset, Slice* input) {
|
|
|
|
if (GetVarint64(input, &size_)) {
|
|
|
|
offset_ = _offset;
|
|
|
|
return Status::OK();
|
|
|
|
} else {
|
|
|
|
// reset in case failure after partially decoding
|
|
|
|
offset_ = 0;
|
|
|
|
size_ = 0;
|
|
|
|
return Status::Corruption("bad block handle");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2014-12-23 22:24:07 +01:00
|
|
|
// Return a string that contains the copy of handle.
|
|
|
|
std::string BlockHandle::ToString(bool hex) const {
|
|
|
|
std::string handle_str;
|
|
|
|
EncodeTo(&handle_str);
|
|
|
|
if (hex) {
|
2016-03-30 06:25:12 +02:00
|
|
|
return Slice(handle_str).ToString(true);
|
2014-12-23 22:24:07 +01:00
|
|
|
} else {
|
|
|
|
return handle_str;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2013-12-05 00:43:09 +01:00
|
|
|
const BlockHandle BlockHandle::kNullBlockHandle(0, 0);
|
2011-03-18 23:37:00 +01:00
|
|
|
|
Add an option to put first key of each sst block in the index (#5289)
Summary:
The first key is used to defer reading the data block until this file gets to the top of merging iterator's heap. For short range scans, most files never make it to the top of the heap, so this change can reduce read amplification by a lot sometimes.
Consider the following workload. There are a few data streams (we'll be calling them "logs"), each stream consisting of a sequence of blobs (we'll be calling them "records"). Each record is identified by log ID and a sequence number within the log. RocksDB key is concatenation of log ID and sequence number (big endian). Reads are mostly relatively short range scans, each within a single log. Writes are mostly sequential for each log, but writes to different logs are randomly interleaved. Compactions are disabled; instead, when we accumulate a few tens of sst files, we create a new column family and start writing to it.
So, a typical sst file consists of a few ranges of blocks, each range corresponding to one log ID (we use FlushBlockPolicy to cut blocks at log boundaries). A typical read would go like this. First, iterator Seek() reads one block from each sst file. Then a series of Next()s move through one sst file (since writes to each log are mostly sequential) until the subiterator reaches the end of this log in this sst file; then Next() switches to the next sst file and reads sequentially from that, and so on. Often a range scan will only return records from a small number of blocks in small number of sst files; in this case, the cost of initial Seek() reading one block from each file may be bigger than the cost of reading the actually useful blocks.
Neither iterate_upper_bound nor bloom filters can prevent reading one block from each file in Seek(). But this PR can: if the index contains first key from each block, we don't have to read the block until this block actually makes it to the top of merging iterator's heap, so for short range scans we won't read any blocks from most of the sst files.
This PR does the deferred block loading inside value() call. This is not ideal: there's no good way to report an IO error from inside value(). As discussed with siying offline, it would probably be better to change InternalIterator's interface to explicitly fetch deferred value and get status. I'll do it in a separate PR.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/5289
Differential Revision: D15256423
Pulled By: al13n321
fbshipit-source-id: 750e4c39ce88e8d41662f701cf6275d9388ba46a
2019-06-25 05:50:35 +02:00
|
|
|
void IndexValue::EncodeTo(std::string* dst, bool have_first_key,
|
|
|
|
const BlockHandle* previous_handle) const {
|
|
|
|
if (previous_handle) {
|
|
|
|
assert(handle.offset() == previous_handle->offset() +
|
|
|
|
previous_handle->size() + kBlockTrailerSize);
|
|
|
|
PutVarsignedint64(dst, handle.size() - previous_handle->size());
|
|
|
|
} else {
|
|
|
|
handle.EncodeTo(dst);
|
|
|
|
}
|
|
|
|
assert(dst->size() != 0);
|
|
|
|
|
|
|
|
if (have_first_key) {
|
|
|
|
PutLengthPrefixedSlice(dst, first_internal_key);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
Status IndexValue::DecodeFrom(Slice* input, bool have_first_key,
|
|
|
|
const BlockHandle* previous_handle) {
|
|
|
|
if (previous_handle) {
|
|
|
|
int64_t delta;
|
|
|
|
if (!GetVarsignedint64(input, &delta)) {
|
|
|
|
return Status::Corruption("bad delta-encoded index value");
|
|
|
|
}
|
|
|
|
handle = BlockHandle(
|
|
|
|
previous_handle->offset() + previous_handle->size() + kBlockTrailerSize,
|
|
|
|
previous_handle->size() + delta);
|
|
|
|
} else {
|
|
|
|
Status s = handle.DecodeFrom(input);
|
|
|
|
if (!s.ok()) {
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!have_first_key) {
|
|
|
|
first_internal_key = Slice();
|
|
|
|
} else if (!GetLengthPrefixedSlice(input, &first_internal_key)) {
|
|
|
|
return Status::Corruption("bad first key in block info");
|
|
|
|
}
|
|
|
|
|
|
|
|
return Status::OK();
|
|
|
|
}
|
|
|
|
|
|
|
|
std::string IndexValue::ToString(bool hex, bool have_first_key) const {
|
|
|
|
std::string s;
|
|
|
|
EncodeTo(&s, have_first_key, nullptr);
|
|
|
|
if (hex) {
|
|
|
|
return Slice(s).ToString(true);
|
|
|
|
} else {
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2015-01-13 23:33:04 +01:00
|
|
|
namespace {
|
|
|
|
inline bool IsLegacyFooterFormat(uint64_t magic_number) {
|
|
|
|
return magic_number == kLegacyBlockBasedTableMagicNumber ||
|
|
|
|
magic_number == kLegacyPlainTableMagicNumber;
|
|
|
|
}
|
|
|
|
inline uint64_t UpconvertLegacyFooterFormat(uint64_t magic_number) {
|
|
|
|
if (magic_number == kLegacyBlockBasedTableMagicNumber) {
|
|
|
|
return kBlockBasedTableMagicNumber;
|
|
|
|
}
|
|
|
|
if (magic_number == kLegacyPlainTableMagicNumber) {
|
|
|
|
return kPlainTableMagicNumber;
|
|
|
|
}
|
|
|
|
assert(false);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
} // namespace
|
|
|
|
|
2014-05-01 20:09:32 +02:00
|
|
|
// legacy footer format:
|
|
|
|
// metaindex handle (varint64 offset, varint64 size)
|
|
|
|
// index handle (varint64 offset, varint64 size)
|
|
|
|
// <padding> to make the total size 2 * BlockHandle::kMaxEncodedLength
|
|
|
|
// table_magic_number (8 bytes)
|
|
|
|
// new footer format:
|
2017-08-24 04:31:40 +02:00
|
|
|
// checksum type (char, 1 byte)
|
2014-05-01 20:09:32 +02:00
|
|
|
// metaindex handle (varint64 offset, varint64 size)
|
|
|
|
// index handle (varint64 offset, varint64 size)
|
|
|
|
// <padding> to make the total size 2 * BlockHandle::kMaxEncodedLength + 1
|
|
|
|
// footer version (4 bytes)
|
|
|
|
// table_magic_number (8 bytes)
|
2011-03-18 23:37:00 +01:00
|
|
|
void Footer::EncodeTo(std::string* dst) const {
|
2015-01-13 23:33:04 +01:00
|
|
|
assert(HasInitializedTableMagicNumber());
|
|
|
|
if (IsLegacyFooterFormat(table_magic_number())) {
|
2014-05-01 20:09:32 +02:00
|
|
|
// has to be default checksum with legacy footer
|
|
|
|
assert(checksum_ == kCRC32c);
|
|
|
|
const size_t original_size = dst->size();
|
|
|
|
metaindex_handle_.EncodeTo(dst);
|
|
|
|
index_handle_.EncodeTo(dst);
|
|
|
|
dst->resize(original_size + 2 * BlockHandle::kMaxEncodedLength); // Padding
|
|
|
|
PutFixed32(dst, static_cast<uint32_t>(table_magic_number() & 0xffffffffu));
|
|
|
|
PutFixed32(dst, static_cast<uint32_t>(table_magic_number() >> 32));
|
|
|
|
assert(dst->size() == original_size + kVersion0EncodedLength);
|
|
|
|
} else {
|
|
|
|
const size_t original_size = dst->size();
|
|
|
|
dst->push_back(static_cast<char>(checksum_));
|
|
|
|
metaindex_handle_.EncodeTo(dst);
|
|
|
|
index_handle_.EncodeTo(dst);
|
2015-01-13 23:33:04 +01:00
|
|
|
dst->resize(original_size + kNewVersionsEncodedLength - 12); // Padding
|
|
|
|
PutFixed32(dst, version());
|
2014-05-01 20:09:32 +02:00
|
|
|
PutFixed32(dst, static_cast<uint32_t>(table_magic_number() & 0xffffffffu));
|
|
|
|
PutFixed32(dst, static_cast<uint32_t>(table_magic_number() >> 32));
|
2015-01-13 23:33:04 +01:00
|
|
|
assert(dst->size() == original_size + kNewVersionsEncodedLength);
|
2014-05-01 20:09:32 +02:00
|
|
|
}
|
2011-03-18 23:37:00 +01:00
|
|
|
}
|
2014-05-01 20:09:32 +02:00
|
|
|
|
2015-01-13 23:33:04 +01:00
|
|
|
Footer::Footer(uint64_t _table_magic_number, uint32_t _version)
|
|
|
|
: version_(_version),
|
2014-05-01 20:09:32 +02:00
|
|
|
checksum_(kCRC32c),
|
2015-01-13 23:33:04 +01:00
|
|
|
table_magic_number_(_table_magic_number) {
|
|
|
|
// This should be guaranteed by constructor callers
|
|
|
|
assert(!IsLegacyFooterFormat(_table_magic_number) || version_ == 0);
|
|
|
|
}
|
2011-03-18 23:37:00 +01:00
|
|
|
|
|
|
|
Status Footer::DecodeFrom(Slice* input) {
|
2015-01-13 23:33:04 +01:00
|
|
|
assert(!HasInitializedTableMagicNumber());
|
2013-03-01 03:04:58 +01:00
|
|
|
assert(input != nullptr);
|
2014-05-01 20:09:32 +02:00
|
|
|
assert(input->size() >= kMinEncodedLength);
|
2013-01-09 19:44:30 +01:00
|
|
|
|
2019-03-28 00:13:08 +01:00
|
|
|
const char* magic_ptr =
|
2014-05-01 20:09:32 +02:00
|
|
|
input->data() + input->size() - kMagicNumberLengthByte;
|
2011-03-18 23:37:00 +01:00
|
|
|
const uint32_t magic_lo = DecodeFixed32(magic_ptr);
|
|
|
|
const uint32_t magic_hi = DecodeFixed32(magic_ptr + 4);
|
2014-05-01 20:09:32 +02:00
|
|
|
uint64_t magic = ((static_cast<uint64_t>(magic_hi) << 32) |
|
|
|
|
(static_cast<uint64_t>(magic_lo)));
|
|
|
|
|
|
|
|
// We check for legacy formats here and silently upconvert them
|
|
|
|
bool legacy = IsLegacyFooterFormat(magic);
|
|
|
|
if (legacy) {
|
|
|
|
magic = UpconvertLegacyFooterFormat(magic);
|
|
|
|
}
|
2015-01-13 23:33:04 +01:00
|
|
|
set_table_magic_number(magic);
|
2011-03-18 23:37:00 +01:00
|
|
|
|
2014-05-01 20:09:32 +02:00
|
|
|
if (legacy) {
|
|
|
|
// The size is already asserted to be at least kMinEncodedLength
|
|
|
|
// at the beginning of the function
|
|
|
|
input->remove_prefix(input->size() - kVersion0EncodedLength);
|
2015-01-13 23:33:04 +01:00
|
|
|
version_ = 0 /* legacy */;
|
2014-05-01 20:09:32 +02:00
|
|
|
checksum_ = kCRC32c;
|
|
|
|
} else {
|
|
|
|
version_ = DecodeFixed32(magic_ptr - 4);
|
2015-01-13 23:33:04 +01:00
|
|
|
// Footer version 1 and higher will always occupy exactly this many bytes.
|
2014-05-01 20:09:32 +02:00
|
|
|
// It consists of the checksum type, two block handles, padding,
|
|
|
|
// a version number, and a magic number
|
2015-01-13 23:33:04 +01:00
|
|
|
if (input->size() < kNewVersionsEncodedLength) {
|
2014-09-11 02:00:00 +02:00
|
|
|
return Status::Corruption("input is too short to be an sstable");
|
2014-05-01 20:09:32 +02:00
|
|
|
} else {
|
2015-01-13 23:33:04 +01:00
|
|
|
input->remove_prefix(input->size() - kNewVersionsEncodedLength);
|
2014-05-01 20:09:32 +02:00
|
|
|
}
|
2014-11-06 20:14:28 +01:00
|
|
|
uint32_t chksum;
|
|
|
|
if (!GetVarint32(input, &chksum)) {
|
2014-05-01 20:09:32 +02:00
|
|
|
return Status::Corruption("bad checksum type");
|
|
|
|
}
|
2014-11-06 20:14:28 +01:00
|
|
|
checksum_ = static_cast<ChecksumType>(chksum);
|
2014-05-01 20:09:32 +02:00
|
|
|
}
|
|
|
|
|
2011-03-18 23:37:00 +01:00
|
|
|
Status result = metaindex_handle_.DecodeFrom(input);
|
|
|
|
if (result.ok()) {
|
|
|
|
result = index_handle_.DecodeFrom(input);
|
|
|
|
}
|
|
|
|
if (result.ok()) {
|
|
|
|
// We skip over any leftover data (just padding for now) in "input"
|
2014-05-01 20:09:32 +02:00
|
|
|
const char* end = magic_ptr + kMagicNumberLengthByte;
|
2011-03-18 23:37:00 +01:00
|
|
|
*input = Slice(end, input->data() + input->size() - end);
|
|
|
|
}
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
2014-12-23 22:24:07 +01:00
|
|
|
std::string Footer::ToString() const {
|
2018-02-16 01:43:23 +01:00
|
|
|
std::string result;
|
2014-12-23 22:24:07 +01:00
|
|
|
result.reserve(1024);
|
|
|
|
|
|
|
|
bool legacy = IsLegacyFooterFormat(table_magic_number_);
|
|
|
|
if (legacy) {
|
|
|
|
result.append("metaindex handle: " + metaindex_handle_.ToString() + "\n ");
|
|
|
|
result.append("index handle: " + index_handle_.ToString() + "\n ");
|
2015-04-24 04:17:57 +02:00
|
|
|
result.append("table_magic_number: " +
|
|
|
|
rocksdb::ToString(table_magic_number_) + "\n ");
|
2014-12-23 22:24:07 +01:00
|
|
|
} else {
|
2015-04-24 04:17:57 +02:00
|
|
|
result.append("checksum: " + rocksdb::ToString(checksum_) + "\n ");
|
2014-12-23 22:24:07 +01:00
|
|
|
result.append("metaindex handle: " + metaindex_handle_.ToString() + "\n ");
|
|
|
|
result.append("index handle: " + index_handle_.ToString() + "\n ");
|
2015-04-24 04:17:57 +02:00
|
|
|
result.append("footer version: " + rocksdb::ToString(version_) + "\n ");
|
|
|
|
result.append("table_magic_number: " +
|
|
|
|
rocksdb::ToString(table_magic_number_) + "\n ");
|
2014-12-23 22:24:07 +01:00
|
|
|
}
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
2017-08-11 20:59:13 +02:00
|
|
|
Status ReadFooterFromFile(RandomAccessFileReader* file,
|
|
|
|
FilePrefetchBuffer* prefetch_buffer,
|
|
|
|
uint64_t file_size, Footer* footer,
|
|
|
|
uint64_t enforce_table_magic_number) {
|
2014-05-01 20:09:32 +02:00
|
|
|
if (file_size < Footer::kMinEncodedLength) {
|
2019-03-28 00:13:08 +01:00
|
|
|
return Status::Corruption("file is too short (" + ToString(file_size) +
|
|
|
|
" bytes) to be an "
|
|
|
|
"sstable: " +
|
|
|
|
file->file_name());
|
2013-12-05 01:35:48 +01:00
|
|
|
}
|
|
|
|
|
2014-05-01 20:09:32 +02:00
|
|
|
char footer_space[Footer::kMaxEncodedLength];
|
2013-12-05 01:35:48 +01:00
|
|
|
Slice footer_input;
|
2014-11-13 20:39:30 +01:00
|
|
|
size_t read_offset =
|
|
|
|
(file_size > Footer::kMaxEncodedLength)
|
|
|
|
? static_cast<size_t>(file_size - Footer::kMaxEncodedLength)
|
|
|
|
: 0;
|
2017-08-11 20:59:13 +02:00
|
|
|
Status s;
|
|
|
|
if (prefetch_buffer == nullptr ||
|
|
|
|
!prefetch_buffer->TryReadFromCache(read_offset, Footer::kMaxEncodedLength,
|
|
|
|
&footer_input)) {
|
|
|
|
s = file->Read(read_offset, Footer::kMaxEncodedLength, &footer_input,
|
|
|
|
footer_space);
|
|
|
|
if (!s.ok()) return s;
|
|
|
|
}
|
2013-12-05 01:35:48 +01:00
|
|
|
|
|
|
|
// Check that we actually read the whole footer from the file. It may be
|
|
|
|
// that size isn't correct.
|
2014-05-01 20:09:32 +02:00
|
|
|
if (footer_input.size() < Footer::kMinEncodedLength) {
|
2019-03-28 00:13:08 +01:00
|
|
|
return Status::Corruption("file is too short (" + ToString(file_size) +
|
|
|
|
" bytes) to be an "
|
|
|
|
"sstable" +
|
|
|
|
file->file_name());
|
2013-12-05 01:35:48 +01:00
|
|
|
}
|
|
|
|
|
2015-01-13 23:33:04 +01:00
|
|
|
s = footer->DecodeFrom(&footer_input);
|
|
|
|
if (!s.ok()) {
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
if (enforce_table_magic_number != 0 &&
|
|
|
|
enforce_table_magic_number != footer->table_magic_number()) {
|
2017-06-29 06:26:03 +02:00
|
|
|
return Status::Corruption(
|
2019-03-28 00:13:08 +01:00
|
|
|
"Bad table magic number: expected " +
|
|
|
|
ToString(enforce_table_magic_number) + ", found " +
|
|
|
|
ToString(footer->table_magic_number()) + " in " + file->file_name());
|
2015-01-13 23:33:04 +01:00
|
|
|
}
|
|
|
|
return Status::OK();
|
2013-12-05 01:35:48 +01:00
|
|
|
}
|
|
|
|
|
2018-06-05 21:51:05 +02:00
|
|
|
Status UncompressBlockContentsForCompressionType(
|
2019-01-19 04:10:17 +01:00
|
|
|
const UncompressionInfo& uncompression_info, const char* data, size_t n,
|
2018-06-05 21:51:05 +02:00
|
|
|
BlockContents* contents, uint32_t format_version,
|
2018-10-26 23:27:09 +02:00
|
|
|
const ImmutableCFOptions& ioptions, MemoryAllocator* allocator) {
|
2018-10-03 02:21:54 +02:00
|
|
|
CacheAllocationPtr ubuf;
|
2016-06-11 03:20:54 +02:00
|
|
|
|
2019-01-19 04:10:17 +01:00
|
|
|
assert(uncompression_info.type() != kNoCompression &&
|
2018-06-05 21:51:05 +02:00
|
|
|
"Invalid compression type");
|
2016-06-11 03:20:54 +02:00
|
|
|
|
2019-03-28 00:13:08 +01:00
|
|
|
StopWatchNano timer(ioptions.env, ShouldReportDetailedTime(
|
|
|
|
ioptions.env, ioptions.statistics));
|
2012-06-29 04:26:43 +02:00
|
|
|
int decompress_size = 0;
|
2019-01-19 04:10:17 +01:00
|
|
|
switch (uncompression_info.type()) {
|
2011-03-23 00:24:02 +01:00
|
|
|
case kSnappyCompression: {
|
2011-07-21 04:40:18 +02:00
|
|
|
size_t ulength = 0;
|
2012-12-20 23:25:06 +01:00
|
|
|
static char snappy_corrupt_msg[] =
|
2019-03-28 00:13:08 +01:00
|
|
|
"Snappy not supported or corrupted Snappy compressed block contents";
|
2015-01-09 21:57:11 +01:00
|
|
|
if (!Snappy_GetUncompressedLength(data, n, &ulength)) {
|
2012-12-20 23:25:06 +01:00
|
|
|
return Status::Corruption(snappy_corrupt_msg);
|
2011-03-18 23:37:00 +01:00
|
|
|
}
|
2018-10-03 02:21:54 +02:00
|
|
|
ubuf = AllocateBlock(ulength, allocator);
|
2015-01-09 21:57:11 +01:00
|
|
|
if (!Snappy_Uncompress(data, n, ubuf.get())) {
|
2012-12-20 23:25:06 +01:00
|
|
|
return Status::Corruption(snappy_corrupt_msg);
|
2011-07-21 04:40:18 +02:00
|
|
|
}
|
2018-11-14 02:00:49 +01:00
|
|
|
*contents = BlockContents(std::move(ubuf), ulength);
|
2011-03-18 23:37:00 +01:00
|
|
|
break;
|
|
|
|
}
|
2012-06-28 08:41:33 +02:00
|
|
|
case kZlibCompression:
|
2018-10-03 02:21:54 +02:00
|
|
|
ubuf = Zlib_Uncompress(
|
2019-01-19 04:10:17 +01:00
|
|
|
uncompression_info, data, n, &decompress_size,
|
2018-10-03 02:21:54 +02:00
|
|
|
GetCompressFormatForVersion(kZlibCompression, format_version),
|
|
|
|
allocator);
|
2012-06-28 08:41:33 +02:00
|
|
|
if (!ubuf) {
|
2014-10-01 07:27:39 +02:00
|
|
|
static char zlib_corrupt_msg[] =
|
2019-03-28 00:13:08 +01:00
|
|
|
"Zlib not supported or corrupted Zlib compressed block contents";
|
2012-12-20 23:25:06 +01:00
|
|
|
return Status::Corruption(zlib_corrupt_msg);
|
2012-06-28 08:41:33 +02:00
|
|
|
}
|
2018-11-14 02:00:49 +01:00
|
|
|
*contents = BlockContents(std::move(ubuf), decompress_size);
|
2012-06-28 08:41:33 +02:00
|
|
|
break;
|
2012-06-29 04:26:43 +02:00
|
|
|
case kBZip2Compression:
|
2018-10-03 02:21:54 +02:00
|
|
|
ubuf = BZip2_Uncompress(
|
2015-01-15 01:24:24 +01:00
|
|
|
data, n, &decompress_size,
|
2018-10-03 02:21:54 +02:00
|
|
|
GetCompressFormatForVersion(kBZip2Compression, format_version),
|
|
|
|
allocator);
|
2012-06-29 04:26:43 +02:00
|
|
|
if (!ubuf) {
|
2014-10-01 07:27:39 +02:00
|
|
|
static char bzip2_corrupt_msg[] =
|
2019-03-28 00:13:08 +01:00
|
|
|
"Bzip2 not supported or corrupted Bzip2 compressed block contents";
|
2012-12-20 23:25:06 +01:00
|
|
|
return Status::Corruption(bzip2_corrupt_msg);
|
2012-06-29 04:26:43 +02:00
|
|
|
}
|
2018-11-14 02:00:49 +01:00
|
|
|
*contents = BlockContents(std::move(ubuf), decompress_size);
|
2012-06-29 04:26:43 +02:00
|
|
|
break;
|
2014-02-08 03:12:30 +01:00
|
|
|
case kLZ4Compression:
|
2018-10-03 02:21:54 +02:00
|
|
|
ubuf = LZ4_Uncompress(
|
2019-01-19 04:10:17 +01:00
|
|
|
uncompression_info, data, n, &decompress_size,
|
2018-10-03 02:21:54 +02:00
|
|
|
GetCompressFormatForVersion(kLZ4Compression, format_version),
|
|
|
|
allocator);
|
2014-02-08 03:12:30 +01:00
|
|
|
if (!ubuf) {
|
2014-10-01 07:27:39 +02:00
|
|
|
static char lz4_corrupt_msg[] =
|
2019-03-28 00:13:08 +01:00
|
|
|
"LZ4 not supported or corrupted LZ4 compressed block contents";
|
2014-02-08 03:12:30 +01:00
|
|
|
return Status::Corruption(lz4_corrupt_msg);
|
|
|
|
}
|
2018-11-14 02:00:49 +01:00
|
|
|
*contents = BlockContents(std::move(ubuf), decompress_size);
|
2014-02-08 03:12:30 +01:00
|
|
|
break;
|
|
|
|
case kLZ4HCCompression:
|
2018-10-03 02:21:54 +02:00
|
|
|
ubuf = LZ4_Uncompress(
|
2019-01-19 04:10:17 +01:00
|
|
|
uncompression_info, data, n, &decompress_size,
|
2018-10-03 02:21:54 +02:00
|
|
|
GetCompressFormatForVersion(kLZ4HCCompression, format_version),
|
|
|
|
allocator);
|
2014-02-08 03:12:30 +01:00
|
|
|
if (!ubuf) {
|
2014-10-01 07:27:39 +02:00
|
|
|
static char lz4hc_corrupt_msg[] =
|
2019-03-28 00:13:08 +01:00
|
|
|
"LZ4HC not supported or corrupted LZ4HC compressed block contents";
|
2014-02-08 03:12:30 +01:00
|
|
|
return Status::Corruption(lz4hc_corrupt_msg);
|
|
|
|
}
|
2018-11-14 02:00:49 +01:00
|
|
|
*contents = BlockContents(std::move(ubuf), decompress_size);
|
2014-02-08 03:12:30 +01:00
|
|
|
break;
|
2016-04-20 07:54:24 +02:00
|
|
|
case kXpressCompression:
|
2018-10-03 02:21:54 +02:00
|
|
|
// XPRESS allocates memory internally, thus no support for custom
|
|
|
|
// allocator.
|
2016-04-20 07:54:24 +02:00
|
|
|
ubuf.reset(XPRESS_Uncompress(data, n, &decompress_size));
|
|
|
|
if (!ubuf) {
|
|
|
|
static char xpress_corrupt_msg[] =
|
2019-03-28 00:13:08 +01:00
|
|
|
"XPRESS not supported or corrupted XPRESS compressed block "
|
|
|
|
"contents";
|
2016-04-20 07:54:24 +02:00
|
|
|
return Status::Corruption(xpress_corrupt_msg);
|
|
|
|
}
|
2018-11-14 02:00:49 +01:00
|
|
|
*contents = BlockContents(std::move(ubuf), decompress_size);
|
2016-04-20 07:54:24 +02:00
|
|
|
break;
|
2016-09-02 00:28:40 +02:00
|
|
|
case kZSTD:
|
2015-08-28 00:40:42 +02:00
|
|
|
case kZSTDNotFinalCompression:
|
2019-01-19 04:10:17 +01:00
|
|
|
ubuf = ZSTD_Uncompress(uncompression_info, data, n, &decompress_size,
|
2018-10-03 02:21:54 +02:00
|
|
|
allocator);
|
2015-08-28 00:40:42 +02:00
|
|
|
if (!ubuf) {
|
|
|
|
static char zstd_corrupt_msg[] =
|
|
|
|
"ZSTD not supported or corrupted ZSTD compressed block contents";
|
|
|
|
return Status::Corruption(zstd_corrupt_msg);
|
|
|
|
}
|
2018-11-14 02:00:49 +01:00
|
|
|
*contents = BlockContents(std::move(ubuf), decompress_size);
|
2015-08-28 00:40:42 +02:00
|
|
|
break;
|
2011-03-18 23:37:00 +01:00
|
|
|
default:
|
|
|
|
return Status::Corruption("bad block type");
|
|
|
|
}
|
2015-12-16 03:20:10 +01:00
|
|
|
|
2019-03-28 00:13:08 +01:00
|
|
|
if (ShouldReportDetailedTime(ioptions.env, ioptions.statistics)) {
|
2019-02-28 19:14:19 +01:00
|
|
|
RecordTimeToHistogram(ioptions.statistics, DECOMPRESSION_TIMES_NANOS,
|
|
|
|
timer.ElapsedNanos());
|
2016-07-19 18:44:03 +02:00
|
|
|
}
|
2019-02-28 19:14:19 +01:00
|
|
|
RecordTimeToHistogram(ioptions.statistics, BYTES_DECOMPRESSED,
|
|
|
|
contents->data.size());
|
2017-12-14 19:17:22 +01:00
|
|
|
RecordTick(ioptions.statistics, NUMBER_BLOCK_DECOMPRESSED);
|
2016-07-19 18:44:03 +02:00
|
|
|
|
2011-03-18 23:37:00 +01:00
|
|
|
return Status::OK();
|
|
|
|
}
|
|
|
|
|
2016-06-11 03:20:54 +02:00
|
|
|
//
|
|
|
|
// The 'data' points to the raw block contents that was 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.
|
|
|
|
// format_version is the block format as defined in include/rocksdb/table.h
|
2019-01-19 04:10:17 +01:00
|
|
|
Status UncompressBlockContents(const UncompressionInfo& uncompression_info,
|
2018-06-04 21:04:52 +02:00
|
|
|
const char* data, size_t n,
|
2016-06-11 03:20:54 +02:00
|
|
|
BlockContents* contents, uint32_t format_version,
|
2018-10-03 02:21:54 +02:00
|
|
|
const ImmutableCFOptions& ioptions,
|
2018-10-26 23:27:09 +02:00
|
|
|
MemoryAllocator* allocator) {
|
2016-06-11 03:20:54 +02:00
|
|
|
assert(data[n] != kNoCompression);
|
2019-01-19 04:10:17 +01:00
|
|
|
assert(data[n] == uncompression_info.type());
|
|
|
|
return UncompressBlockContentsForCompressionType(uncompression_info, data, n,
|
2018-10-03 02:21:54 +02:00
|
|
|
contents, format_version,
|
|
|
|
ioptions, allocator);
|
2016-06-11 03:20:54 +02:00
|
|
|
}
|
|
|
|
|
2013-10-04 06:49:15 +02:00
|
|
|
} // namespace rocksdb
|