rocksdb/util/coding.h

503 lines
16 KiB
C
Raw Normal View History

// 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.
//
// Encoding independent of machine byte order:
// * Fixed-length numbers are encoded with least-significant byte first
// (little endian, native order on Intel and others)
// * In addition we support variable length "varint" encoding
// * Strings are encoded prefixed by their length in varint format
#pragma once
#include <algorithm>
#include <stdint.h>
#include <string.h>
#include <string>
#include "rocksdb/write_batch.h"
#include "port/port.h"
// Some processors does not allow unaligned access to memory
#if defined(__sparc)
#define PLATFORM_UNALIGNED_ACCESS_NOT_ALLOWED
#endif
namespace ROCKSDB_NAMESPACE {
// The maximum length of a varint in bytes for 64-bit.
const unsigned int kMaxVarint64Length = 10;
// Standard Put... routines append to a string
extern void PutFixed16(std::string* dst, uint16_t value);
extern void PutFixed32(std::string* dst, uint32_t value);
extern void PutFixed64(std::string* dst, uint64_t value);
extern void PutVarint32(std::string* dst, uint32_t value);
Miscellaneous performance improvements Summary: I was investigating performance issues in the SstFileWriter and found all of the following: - The SstFileWriter::Add() function created a local InternalKey every time it was called generating a allocation and free each time. Changed to have an InternalKey member variable that can be reset with the new InternalKey::Set() function. - In SstFileWriter::Add() the smallest_key and largest_key values were assigned the result of a ToString() call, but it is simpler to just assign them directly from the user's key. - The Slice class had no move constructor so each time one was returned from a function a new one had to be allocated, the old data copied to the new, and the old one was freed. I added the move constructor which also required a copy constructor and assignment operator. - The BlockBuilder::CurrentSizeEstimate() function calculates the current estimate size, but was being called 2 or 3 times for each key added. I changed the class to maintain a running estimate (equal to the original calculation) so that the function can return an already calculated value. - The code in BlockBuilder::Add() that calculated the shared bytes between the last key and the new key duplicated what Slice::difference_offset does, so I replaced it with the standard function. - BlockBuilder::Add() had code to copy just the changed portion into the last key value (and asserted that it now matched the new key). It is more efficient just to copy the whole new key over. - Moved this same code up into the 'if (use_delta_encoding_)' since the last key value is only needed when delta encoding is on. - FlushBlockBySizePolicy::BlockAlmostFull calculated a standard deviation value each time it was called, but this information would only change if block_size of block_size_deviation changed, so I created a member variable to hold the value to avoid the calculation each time. - Each PutVarint??() function has a buffer and calls std::string::append(). Two or three calls in a row could share a buffer and a single call to std::string::append(). Some of these will be helpful outside of the SstFileWriter. I'm not 100% the addition of the move constructor is appropriate as I wonder why this wasn't done before - maybe because of compiler compatibility? I tried it on gcc 4.8 and 4.9. Test Plan: The changes should not affect the results so the existing tests should all still work and no new tests were added. The value of the changes was seen by manually testing the SstFileWriter class through MyRocks and adding timing code to identify problem areas. Reviewers: sdong, IslamAbdelRahman Reviewed By: IslamAbdelRahman Subscribers: andrewkr, dhruba Differential Revision: https://reviews.facebook.net/D59607
2016-06-13 16:57:43 +00:00
extern void PutVarint32Varint32(std::string* dst, uint32_t value1,
uint32_t value2);
extern void PutVarint32Varint32Varint32(std::string* dst, uint32_t value1,
uint32_t value2, uint32_t value3);
extern void PutVarint64(std::string* dst, uint64_t value);
Miscellaneous performance improvements Summary: I was investigating performance issues in the SstFileWriter and found all of the following: - The SstFileWriter::Add() function created a local InternalKey every time it was called generating a allocation and free each time. Changed to have an InternalKey member variable that can be reset with the new InternalKey::Set() function. - In SstFileWriter::Add() the smallest_key and largest_key values were assigned the result of a ToString() call, but it is simpler to just assign them directly from the user's key. - The Slice class had no move constructor so each time one was returned from a function a new one had to be allocated, the old data copied to the new, and the old one was freed. I added the move constructor which also required a copy constructor and assignment operator. - The BlockBuilder::CurrentSizeEstimate() function calculates the current estimate size, but was being called 2 or 3 times for each key added. I changed the class to maintain a running estimate (equal to the original calculation) so that the function can return an already calculated value. - The code in BlockBuilder::Add() that calculated the shared bytes between the last key and the new key duplicated what Slice::difference_offset does, so I replaced it with the standard function. - BlockBuilder::Add() had code to copy just the changed portion into the last key value (and asserted that it now matched the new key). It is more efficient just to copy the whole new key over. - Moved this same code up into the 'if (use_delta_encoding_)' since the last key value is only needed when delta encoding is on. - FlushBlockBySizePolicy::BlockAlmostFull calculated a standard deviation value each time it was called, but this information would only change if block_size of block_size_deviation changed, so I created a member variable to hold the value to avoid the calculation each time. - Each PutVarint??() function has a buffer and calls std::string::append(). Two or three calls in a row could share a buffer and a single call to std::string::append(). Some of these will be helpful outside of the SstFileWriter. I'm not 100% the addition of the move constructor is appropriate as I wonder why this wasn't done before - maybe because of compiler compatibility? I tried it on gcc 4.8 and 4.9. Test Plan: The changes should not affect the results so the existing tests should all still work and no new tests were added. The value of the changes was seen by manually testing the SstFileWriter class through MyRocks and adding timing code to identify problem areas. Reviewers: sdong, IslamAbdelRahman Reviewed By: IslamAbdelRahman Subscribers: andrewkr, dhruba Differential Revision: https://reviews.facebook.net/D59607
2016-06-13 16:57:43 +00:00
extern void PutVarint64Varint64(std::string* dst, uint64_t value1,
uint64_t value2);
extern void PutVarint32Varint64(std::string* dst, uint32_t value1,
uint64_t value2);
extern void PutVarint32Varint32Varint64(std::string* dst, uint32_t value1,
uint32_t value2, uint64_t value3);
extern void PutLengthPrefixedSlice(std::string* dst, const Slice& value);
extern void PutLengthPrefixedSliceParts(std::string* dst,
const SliceParts& slice_parts);
Avoid user key copying for Get/Put/Write with user-timestamp (#5502) Summary: In previous https://github.com/facebook/rocksdb/issues/5079, we added user-specified timestamp to `DB::Get()` and `DB::Put()`. Limitation is that these two functions may cause extra memory allocation and key copy. The reason is that `WriteBatch` does not allocate extra memory for timestamps because it is not aware of timestamp size, and we did not provide an API to assign/update timestamp of each key within a `WriteBatch`. We address these issues in this PR by doing the following. 1. Add a `timestamp_size_` to `WriteBatch` so that `WriteBatch` can take timestamps into account when calling `WriteBatch::Put`, `WriteBatch::Delete`, etc. 2. Add APIs `WriteBatch::AssignTimestamp` and `WriteBatch::AssignTimestamps` so that application can assign/update timestamps for each key in a `WriteBatch`. 3. Avoid key copy in `GetImpl` by adding new constructor to `LookupKey`. Test plan (on devserver): ``` $make clean && COMPILE_WITH_ASAN=1 make -j32 all $./db_basic_test --gtest_filter=Timestamp/DBBasicTestWithTimestampWithParam.PutAndGet/* $make check ``` If the API extension looks good, I will add more unit tests. Some simple benchmark using db_bench. ``` $rm -rf /dev/shm/dbbench/* && TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=fillseq,readrandom -num=1000000 $rm -rf /dev/shm/dbbench/* && TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=fillrandom -num=1000000 -disable_wal=true ``` Master is at a78503bd6c80a3c4137df1962a972fe406b4d90b. ``` | | readrandom | fillrandom | | master | 15.53 MB/s | 25.97 MB/s | | PR5502 | 16.70 MB/s | 25.80 MB/s | ``` Pull Request resolved: https://github.com/facebook/rocksdb/pull/5502 Differential Revision: D16340894 Pulled By: riversand963 fbshipit-source-id: 51132cf792be07d1efc3ac33f5768c4ee2608bb8
2019-07-25 22:23:46 +00:00
extern void PutLengthPrefixedSlicePartsWithPadding(
std::string* dst, const SliceParts& slice_parts, size_t pad_sz);
// Standard Get... routines parse a value from the beginning of a Slice
// and advance the slice past the parsed value.
extern bool GetFixed64(Slice* input, uint64_t* value);
extern bool GetFixed32(Slice* input, uint32_t* value);
extern bool GetFixed16(Slice* input, uint16_t* value);
extern bool GetVarint32(Slice* input, uint32_t* value);
extern bool GetVarint64(Slice* input, uint64_t* value);
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 03:50:35 +00:00
extern bool GetVarsignedint64(Slice* input, int64_t* value);
extern bool GetLengthPrefixedSlice(Slice* input, Slice* result);
// This function assumes data is well-formed.
extern Slice GetLengthPrefixedSlice(const char* data);
[RocksDB] BackupableDB Summary: In this diff I present you BackupableDB v1. You can easily use it to backup your DB and it will do incremental snapshots for you. Let's first describe how you would use BackupableDB. It's inheriting StackableDB interface so you can easily construct it with your DB object -- it will add a method RollTheSnapshot() to the DB object. When you call RollTheSnapshot(), current snapshot of the DB will be stored in the backup dir. To restore, you can just call RestoreDBFromBackup() on a BackupableDB (which is a static method) and it will restore all files from the backup dir. In the next version, it will even support automatic backuping every X minutes. There are multiple things you can configure: 1. backup_env and db_env can be different, which is awesome because then you can easily backup to HDFS or wherever you feel like. 2. sync - if true, it *guarantees* backup consistency on machine reboot 3. number of snapshots to keep - this will keep last N snapshots around if you want, for some reason, be able to restore from an earlier snapshot. All the backuping is done in incremental fashion - if we already have 00010.sst, we will not copy it again. *IMPORTANT* -- This is based on assumption that 00010.sst never changes - two files named 00010.sst from the same DB will always be exactly the same. Is this true? I always copy manifest, current and log files. 4. You can decide if you want to flush the memtables before you backup, or you're fine with backing up the log files -- either way, you get a complete and consistent view of the database at a time of backup. 5. More things you can find in BackupableDBOptions Here is the directory structure I use: backup_dir/CURRENT_SNAPSHOT - just 4 bytes holding the latest snapshot 0, 1, 2, ... - files containing serialized version of each snapshot - containing a list of files files/*.sst - sst files shared between snapshots - if one snapshot references 00010.sst and another one needs to backup it from the DB, it will just reference the same file files/ 0/, 1/, 2/, ... - snapshot directories containing private snapshot files - current, manifest and log files All the files are ref counted and deleted immediatelly when they get out of scope. Some other stuff in this diff: 1. Added GetEnv() method to the DB. Discussed with @haobo and we agreed that it seems right thing to do. 2. Fixed StackableDB interface. The way it was set up before, I was not able to implement BackupableDB. Test Plan: I have a unittest, but please don't look at this yet. I just hacked it up to help me with debugging. I will write a lot of good tests and update the diff. Also, `make asan_check` Reviewers: dhruba, haobo, emayanke Reviewed By: dhruba CC: leveldb, haobo Differential Revision: https://reviews.facebook.net/D14295
2013-12-09 22:06:52 +00:00
extern Slice GetSliceUntil(Slice* slice, char delimiter);
// Borrowed from
// https://github.com/facebook/fbthrift/blob/449a5f77f9f9bae72c9eb5e78093247eef185c04/thrift/lib/cpp/util/VarintUtils-inl.h#L202-L208
constexpr inline uint64_t i64ToZigzag(const int64_t l) {
return (static_cast<uint64_t>(l) << 1) ^ static_cast<uint64_t>(l >> 63);
}
inline int64_t zigzagToI64(uint64_t n) {
return (n >> 1) ^ -static_cast<int64_t>(n & 1);
}
// Pointer-based variants of GetVarint... These either store a value
// in *v and return a pointer just past the parsed value, or return
// nullptr on error. These routines only look at bytes in the range
// [p..limit-1]
extern const char* GetVarint32Ptr(const char* p,const char* limit, uint32_t* v);
extern const char* GetVarint64Ptr(const char* p,const char* limit, uint64_t* v);
inline const char* GetVarsignedint64Ptr(const char* p, const char* limit,
int64_t* value) {
uint64_t u = 0;
const char* ret = GetVarint64Ptr(p, limit, &u);
*value = zigzagToI64(u);
return ret;
}
// Returns the length of the varint32 or varint64 encoding of "v"
extern int VarintLength(uint64_t v);
// Lower-level versions of Put... that write directly into a character buffer
// REQUIRES: dst has enough space for the value being written
extern void EncodeFixed16(char* dst, uint16_t value);
extern void EncodeFixed32(char* dst, uint32_t value);
extern void EncodeFixed64(char* dst, uint64_t value);
// Lower-level versions of Put... that write directly into a character buffer
// and return a pointer just past the last byte written.
// REQUIRES: dst has enough space for the value being written
extern char* EncodeVarint32(char* dst, uint32_t value);
extern char* EncodeVarint64(char* dst, uint64_t value);
// Lower-level versions of Get... that read directly from a character buffer
// without any bounds checking.
inline uint16_t DecodeFixed16(const char* ptr) {
if (port::kLittleEndian) {
// Load the raw bytes
uint16_t result;
memcpy(&result, ptr, sizeof(result)); // gcc optimizes this to a plain load
return result;
} else {
return ((static_cast<uint16_t>(static_cast<unsigned char>(ptr[0]))) |
(static_cast<uint16_t>(static_cast<unsigned char>(ptr[1])) << 8));
}
}
inline uint32_t DecodeFixed32(const char* ptr) {
if (port::kLittleEndian) {
// Load the raw bytes
uint32_t result;
memcpy(&result, ptr, sizeof(result)); // gcc optimizes this to a plain load
return result;
} else {
return ((static_cast<uint32_t>(static_cast<unsigned char>(ptr[0])))
| (static_cast<uint32_t>(static_cast<unsigned char>(ptr[1])) << 8)
| (static_cast<uint32_t>(static_cast<unsigned char>(ptr[2])) << 16)
| (static_cast<uint32_t>(static_cast<unsigned char>(ptr[3])) << 24));
}
}
inline uint64_t DecodeFixed64(const char* ptr) {
if (port::kLittleEndian) {
// Load the raw bytes
uint64_t result;
memcpy(&result, ptr, sizeof(result)); // gcc optimizes this to a plain load
return result;
} else {
uint64_t lo = DecodeFixed32(ptr);
uint64_t hi = DecodeFixed32(ptr + 4);
return (hi << 32) | lo;
}
}
// Internal routine for use by fallback path of GetVarint32Ptr
extern const char* GetVarint32PtrFallback(const char* p,
const char* limit,
uint32_t* value);
inline const char* GetVarint32Ptr(const char* p,
const char* limit,
uint32_t* value) {
if (p < limit) {
uint32_t result = *(reinterpret_cast<const unsigned char*>(p));
if ((result & 128) == 0) {
*value = result;
return p + 1;
}
}
return GetVarint32PtrFallback(p, limit, value);
}
// -- Implementation of the functions declared above
inline void EncodeFixed16(char* buf, uint16_t value) {
if (port::kLittleEndian) {
memcpy(buf, &value, sizeof(value));
} else {
buf[0] = value & 0xff;
buf[1] = (value >> 8) & 0xff;
}
}
inline void EncodeFixed32(char* buf, uint32_t value) {
if (port::kLittleEndian) {
memcpy(buf, &value, sizeof(value));
} else {
buf[0] = value & 0xff;
buf[1] = (value >> 8) & 0xff;
buf[2] = (value >> 16) & 0xff;
buf[3] = (value >> 24) & 0xff;
}
}
inline void EncodeFixed64(char* buf, uint64_t value) {
if (port::kLittleEndian) {
memcpy(buf, &value, sizeof(value));
} else {
buf[0] = value & 0xff;
buf[1] = (value >> 8) & 0xff;
buf[2] = (value >> 16) & 0xff;
buf[3] = (value >> 24) & 0xff;
buf[4] = (value >> 32) & 0xff;
buf[5] = (value >> 40) & 0xff;
buf[6] = (value >> 48) & 0xff;
buf[7] = (value >> 56) & 0xff;
}
}
Miscellaneous performance improvements Summary: I was investigating performance issues in the SstFileWriter and found all of the following: - The SstFileWriter::Add() function created a local InternalKey every time it was called generating a allocation and free each time. Changed to have an InternalKey member variable that can be reset with the new InternalKey::Set() function. - In SstFileWriter::Add() the smallest_key and largest_key values were assigned the result of a ToString() call, but it is simpler to just assign them directly from the user's key. - The Slice class had no move constructor so each time one was returned from a function a new one had to be allocated, the old data copied to the new, and the old one was freed. I added the move constructor which also required a copy constructor and assignment operator. - The BlockBuilder::CurrentSizeEstimate() function calculates the current estimate size, but was being called 2 or 3 times for each key added. I changed the class to maintain a running estimate (equal to the original calculation) so that the function can return an already calculated value. - The code in BlockBuilder::Add() that calculated the shared bytes between the last key and the new key duplicated what Slice::difference_offset does, so I replaced it with the standard function. - BlockBuilder::Add() had code to copy just the changed portion into the last key value (and asserted that it now matched the new key). It is more efficient just to copy the whole new key over. - Moved this same code up into the 'if (use_delta_encoding_)' since the last key value is only needed when delta encoding is on. - FlushBlockBySizePolicy::BlockAlmostFull calculated a standard deviation value each time it was called, but this information would only change if block_size of block_size_deviation changed, so I created a member variable to hold the value to avoid the calculation each time. - Each PutVarint??() function has a buffer and calls std::string::append(). Two or three calls in a row could share a buffer and a single call to std::string::append(). Some of these will be helpful outside of the SstFileWriter. I'm not 100% the addition of the move constructor is appropriate as I wonder why this wasn't done before - maybe because of compiler compatibility? I tried it on gcc 4.8 and 4.9. Test Plan: The changes should not affect the results so the existing tests should all still work and no new tests were added. The value of the changes was seen by manually testing the SstFileWriter class through MyRocks and adding timing code to identify problem areas. Reviewers: sdong, IslamAbdelRahman Reviewed By: IslamAbdelRahman Subscribers: andrewkr, dhruba Differential Revision: https://reviews.facebook.net/D59607
2016-06-13 16:57:43 +00:00
// Pull the last 8 bits and cast it to a character
inline void PutFixed16(std::string* dst, uint16_t value) {
if (port::kLittleEndian) {
dst->append(const_cast<const char*>(reinterpret_cast<char*>(&value)),
sizeof(value));
} else {
char buf[sizeof(value)];
EncodeFixed16(buf, value);
dst->append(buf, sizeof(buf));
}
}
inline void PutFixed32(std::string* dst, uint32_t value) {
if (port::kLittleEndian) {
dst->append(const_cast<const char*>(reinterpret_cast<char*>(&value)),
sizeof(value));
} else {
char buf[sizeof(value)];
EncodeFixed32(buf, value);
dst->append(buf, sizeof(buf));
}
}
inline void PutFixed64(std::string* dst, uint64_t value) {
if (port::kLittleEndian) {
dst->append(const_cast<const char*>(reinterpret_cast<char*>(&value)),
sizeof(value));
} else {
char buf[sizeof(value)];
EncodeFixed64(buf, value);
dst->append(buf, sizeof(buf));
}
}
inline void PutVarint32(std::string* dst, uint32_t v) {
char buf[5];
char* ptr = EncodeVarint32(buf, v);
dst->append(buf, static_cast<size_t>(ptr - buf));
}
Miscellaneous performance improvements Summary: I was investigating performance issues in the SstFileWriter and found all of the following: - The SstFileWriter::Add() function created a local InternalKey every time it was called generating a allocation and free each time. Changed to have an InternalKey member variable that can be reset with the new InternalKey::Set() function. - In SstFileWriter::Add() the smallest_key and largest_key values were assigned the result of a ToString() call, but it is simpler to just assign them directly from the user's key. - The Slice class had no move constructor so each time one was returned from a function a new one had to be allocated, the old data copied to the new, and the old one was freed. I added the move constructor which also required a copy constructor and assignment operator. - The BlockBuilder::CurrentSizeEstimate() function calculates the current estimate size, but was being called 2 or 3 times for each key added. I changed the class to maintain a running estimate (equal to the original calculation) so that the function can return an already calculated value. - The code in BlockBuilder::Add() that calculated the shared bytes between the last key and the new key duplicated what Slice::difference_offset does, so I replaced it with the standard function. - BlockBuilder::Add() had code to copy just the changed portion into the last key value (and asserted that it now matched the new key). It is more efficient just to copy the whole new key over. - Moved this same code up into the 'if (use_delta_encoding_)' since the last key value is only needed when delta encoding is on. - FlushBlockBySizePolicy::BlockAlmostFull calculated a standard deviation value each time it was called, but this information would only change if block_size of block_size_deviation changed, so I created a member variable to hold the value to avoid the calculation each time. - Each PutVarint??() function has a buffer and calls std::string::append(). Two or three calls in a row could share a buffer and a single call to std::string::append(). Some of these will be helpful outside of the SstFileWriter. I'm not 100% the addition of the move constructor is appropriate as I wonder why this wasn't done before - maybe because of compiler compatibility? I tried it on gcc 4.8 and 4.9. Test Plan: The changes should not affect the results so the existing tests should all still work and no new tests were added. The value of the changes was seen by manually testing the SstFileWriter class through MyRocks and adding timing code to identify problem areas. Reviewers: sdong, IslamAbdelRahman Reviewed By: IslamAbdelRahman Subscribers: andrewkr, dhruba Differential Revision: https://reviews.facebook.net/D59607
2016-06-13 16:57:43 +00:00
inline void PutVarint32Varint32(std::string* dst, uint32_t v1, uint32_t v2) {
char buf[10];
char* ptr = EncodeVarint32(buf, v1);
ptr = EncodeVarint32(ptr, v2);
dst->append(buf, static_cast<size_t>(ptr - buf));
}
inline void PutVarint32Varint32Varint32(std::string* dst, uint32_t v1,
uint32_t v2, uint32_t v3) {
char buf[15];
char* ptr = EncodeVarint32(buf, v1);
ptr = EncodeVarint32(ptr, v2);
ptr = EncodeVarint32(ptr, v3);
dst->append(buf, static_cast<size_t>(ptr - buf));
}
inline char* EncodeVarint64(char* dst, uint64_t v) {
static const unsigned int B = 128;
unsigned char* ptr = reinterpret_cast<unsigned char*>(dst);
while (v >= B) {
*(ptr++) = (v & (B - 1)) | B;
v >>= 7;
}
*(ptr++) = static_cast<unsigned char>(v);
return reinterpret_cast<char*>(ptr);
}
inline void PutVarint64(std::string* dst, uint64_t v) {
char buf[kMaxVarint64Length];
char* ptr = EncodeVarint64(buf, v);
dst->append(buf, static_cast<size_t>(ptr - buf));
}
inline void PutVarsignedint64(std::string* dst, int64_t v) {
char buf[kMaxVarint64Length];
// Using Zigzag format to convert signed to unsigned
char* ptr = EncodeVarint64(buf, i64ToZigzag(v));
dst->append(buf, static_cast<size_t>(ptr - buf));
}
Miscellaneous performance improvements Summary: I was investigating performance issues in the SstFileWriter and found all of the following: - The SstFileWriter::Add() function created a local InternalKey every time it was called generating a allocation and free each time. Changed to have an InternalKey member variable that can be reset with the new InternalKey::Set() function. - In SstFileWriter::Add() the smallest_key and largest_key values were assigned the result of a ToString() call, but it is simpler to just assign them directly from the user's key. - The Slice class had no move constructor so each time one was returned from a function a new one had to be allocated, the old data copied to the new, and the old one was freed. I added the move constructor which also required a copy constructor and assignment operator. - The BlockBuilder::CurrentSizeEstimate() function calculates the current estimate size, but was being called 2 or 3 times for each key added. I changed the class to maintain a running estimate (equal to the original calculation) so that the function can return an already calculated value. - The code in BlockBuilder::Add() that calculated the shared bytes between the last key and the new key duplicated what Slice::difference_offset does, so I replaced it with the standard function. - BlockBuilder::Add() had code to copy just the changed portion into the last key value (and asserted that it now matched the new key). It is more efficient just to copy the whole new key over. - Moved this same code up into the 'if (use_delta_encoding_)' since the last key value is only needed when delta encoding is on. - FlushBlockBySizePolicy::BlockAlmostFull calculated a standard deviation value each time it was called, but this information would only change if block_size of block_size_deviation changed, so I created a member variable to hold the value to avoid the calculation each time. - Each PutVarint??() function has a buffer and calls std::string::append(). Two or three calls in a row could share a buffer and a single call to std::string::append(). Some of these will be helpful outside of the SstFileWriter. I'm not 100% the addition of the move constructor is appropriate as I wonder why this wasn't done before - maybe because of compiler compatibility? I tried it on gcc 4.8 and 4.9. Test Plan: The changes should not affect the results so the existing tests should all still work and no new tests were added. The value of the changes was seen by manually testing the SstFileWriter class through MyRocks and adding timing code to identify problem areas. Reviewers: sdong, IslamAbdelRahman Reviewed By: IslamAbdelRahman Subscribers: andrewkr, dhruba Differential Revision: https://reviews.facebook.net/D59607
2016-06-13 16:57:43 +00:00
inline void PutVarint64Varint64(std::string* dst, uint64_t v1, uint64_t v2) {
char buf[20];
char* ptr = EncodeVarint64(buf, v1);
ptr = EncodeVarint64(ptr, v2);
dst->append(buf, static_cast<size_t>(ptr - buf));
}
inline void PutVarint32Varint64(std::string* dst, uint32_t v1, uint64_t v2) {
char buf[15];
char* ptr = EncodeVarint32(buf, v1);
ptr = EncodeVarint64(ptr, v2);
dst->append(buf, static_cast<size_t>(ptr - buf));
}
inline void PutVarint32Varint32Varint64(std::string* dst, uint32_t v1,
uint32_t v2, uint64_t v3) {
char buf[20];
char* ptr = EncodeVarint32(buf, v1);
ptr = EncodeVarint32(ptr, v2);
ptr = EncodeVarint64(ptr, v3);
dst->append(buf, static_cast<size_t>(ptr - buf));
}
inline void PutLengthPrefixedSlice(std::string* dst, const Slice& value) {
PutVarint32(dst, static_cast<uint32_t>(value.size()));
dst->append(value.data(), value.size());
}
Avoid user key copying for Get/Put/Write with user-timestamp (#5502) Summary: In previous https://github.com/facebook/rocksdb/issues/5079, we added user-specified timestamp to `DB::Get()` and `DB::Put()`. Limitation is that these two functions may cause extra memory allocation and key copy. The reason is that `WriteBatch` does not allocate extra memory for timestamps because it is not aware of timestamp size, and we did not provide an API to assign/update timestamp of each key within a `WriteBatch`. We address these issues in this PR by doing the following. 1. Add a `timestamp_size_` to `WriteBatch` so that `WriteBatch` can take timestamps into account when calling `WriteBatch::Put`, `WriteBatch::Delete`, etc. 2. Add APIs `WriteBatch::AssignTimestamp` and `WriteBatch::AssignTimestamps` so that application can assign/update timestamps for each key in a `WriteBatch`. 3. Avoid key copy in `GetImpl` by adding new constructor to `LookupKey`. Test plan (on devserver): ``` $make clean && COMPILE_WITH_ASAN=1 make -j32 all $./db_basic_test --gtest_filter=Timestamp/DBBasicTestWithTimestampWithParam.PutAndGet/* $make check ``` If the API extension looks good, I will add more unit tests. Some simple benchmark using db_bench. ``` $rm -rf /dev/shm/dbbench/* && TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=fillseq,readrandom -num=1000000 $rm -rf /dev/shm/dbbench/* && TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=fillrandom -num=1000000 -disable_wal=true ``` Master is at a78503bd6c80a3c4137df1962a972fe406b4d90b. ``` | | readrandom | fillrandom | | master | 15.53 MB/s | 25.97 MB/s | | PR5502 | 16.70 MB/s | 25.80 MB/s | ``` Pull Request resolved: https://github.com/facebook/rocksdb/pull/5502 Differential Revision: D16340894 Pulled By: riversand963 fbshipit-source-id: 51132cf792be07d1efc3ac33f5768c4ee2608bb8
2019-07-25 22:23:46 +00:00
inline void PutLengthPrefixedSliceParts(std::string* dst, size_t total_bytes,
const SliceParts& slice_parts) {
for (int i = 0; i < slice_parts.num_parts; ++i) {
total_bytes += slice_parts.parts[i].size();
}
PutVarint32(dst, static_cast<uint32_t>(total_bytes));
for (int i = 0; i < slice_parts.num_parts; ++i) {
dst->append(slice_parts.parts[i].data(), slice_parts.parts[i].size());
}
}
Avoid user key copying for Get/Put/Write with user-timestamp (#5502) Summary: In previous https://github.com/facebook/rocksdb/issues/5079, we added user-specified timestamp to `DB::Get()` and `DB::Put()`. Limitation is that these two functions may cause extra memory allocation and key copy. The reason is that `WriteBatch` does not allocate extra memory for timestamps because it is not aware of timestamp size, and we did not provide an API to assign/update timestamp of each key within a `WriteBatch`. We address these issues in this PR by doing the following. 1. Add a `timestamp_size_` to `WriteBatch` so that `WriteBatch` can take timestamps into account when calling `WriteBatch::Put`, `WriteBatch::Delete`, etc. 2. Add APIs `WriteBatch::AssignTimestamp` and `WriteBatch::AssignTimestamps` so that application can assign/update timestamps for each key in a `WriteBatch`. 3. Avoid key copy in `GetImpl` by adding new constructor to `LookupKey`. Test plan (on devserver): ``` $make clean && COMPILE_WITH_ASAN=1 make -j32 all $./db_basic_test --gtest_filter=Timestamp/DBBasicTestWithTimestampWithParam.PutAndGet/* $make check ``` If the API extension looks good, I will add more unit tests. Some simple benchmark using db_bench. ``` $rm -rf /dev/shm/dbbench/* && TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=fillseq,readrandom -num=1000000 $rm -rf /dev/shm/dbbench/* && TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=fillrandom -num=1000000 -disable_wal=true ``` Master is at a78503bd6c80a3c4137df1962a972fe406b4d90b. ``` | | readrandom | fillrandom | | master | 15.53 MB/s | 25.97 MB/s | | PR5502 | 16.70 MB/s | 25.80 MB/s | ``` Pull Request resolved: https://github.com/facebook/rocksdb/pull/5502 Differential Revision: D16340894 Pulled By: riversand963 fbshipit-source-id: 51132cf792be07d1efc3ac33f5768c4ee2608bb8
2019-07-25 22:23:46 +00:00
inline void PutLengthPrefixedSliceParts(std::string* dst,
const SliceParts& slice_parts) {
PutLengthPrefixedSliceParts(dst, /*total_bytes=*/0, slice_parts);
}
inline void PutLengthPrefixedSlicePartsWithPadding(
std::string* dst, const SliceParts& slice_parts, size_t pad_sz) {
PutLengthPrefixedSliceParts(dst, /*total_bytes=*/pad_sz, slice_parts);
dst->append(pad_sz, '\0');
}
inline int VarintLength(uint64_t v) {
int len = 1;
while (v >= 128) {
v >>= 7;
len++;
}
return len;
}
inline bool GetFixed64(Slice* input, uint64_t* value) {
if (input->size() < sizeof(uint64_t)) {
return false;
}
*value = DecodeFixed64(input->data());
input->remove_prefix(sizeof(uint64_t));
return true;
}
inline bool GetFixed32(Slice* input, uint32_t* value) {
if (input->size() < sizeof(uint32_t)) {
return false;
}
*value = DecodeFixed32(input->data());
input->remove_prefix(sizeof(uint32_t));
return true;
}
inline bool GetFixed16(Slice* input, uint16_t* value) {
if (input->size() < sizeof(uint16_t)) {
return false;
}
*value = DecodeFixed16(input->data());
input->remove_prefix(sizeof(uint16_t));
return true;
}
inline bool GetVarint32(Slice* input, uint32_t* value) {
const char* p = input->data();
const char* limit = p + input->size();
const char* q = GetVarint32Ptr(p, limit, value);
if (q == nullptr) {
return false;
} else {
*input = Slice(q, static_cast<size_t>(limit - q));
return true;
}
}
inline bool GetVarint64(Slice* input, uint64_t* value) {
const char* p = input->data();
const char* limit = p + input->size();
const char* q = GetVarint64Ptr(p, limit, value);
if (q == nullptr) {
return false;
} else {
*input = Slice(q, static_cast<size_t>(limit - q));
return true;
}
}
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 03:50:35 +00:00
inline bool GetVarsignedint64(Slice* input, int64_t* value) {
const char* p = input->data();
const char* limit = p + input->size();
const char* q = GetVarsignedint64Ptr(p, limit, value);
if (q == nullptr) {
return false;
} else {
*input = Slice(q, static_cast<size_t>(limit - q));
return true;
}
}
// Swaps between big and little endian. Can be used to in combination
// with the little-endian encoding/decoding functions to encode/decode
// big endian.
template <typename T>
inline T EndianSwapValue(T v) {
static_assert(std::is_integral<T>::value, "non-integral type");
#ifdef _MSC_VER
if (sizeof(T) == 2) {
return static_cast<T>(_byteswap_ushort(static_cast<uint16_t>(v)));
} else if (sizeof(T) == 4) {
return static_cast<T>(_byteswap_ulong(static_cast<uint32_t>(v)));
} else if (sizeof(T) == 8) {
return static_cast<T>(_byteswap_uint64(static_cast<uint64_t>(v)));
}
#else
if (sizeof(T) == 2) {
return static_cast<T>(__builtin_bswap16(static_cast<uint16_t>(v)));
} else if (sizeof(T) == 4) {
return static_cast<T>(__builtin_bswap32(static_cast<uint32_t>(v)));
} else if (sizeof(T) == 8) {
return static_cast<T>(__builtin_bswap64(static_cast<uint64_t>(v)));
}
#endif
// Recognized by clang as bswap, but not by gcc :(
T ret_val = 0;
for (size_t i = 0; i < sizeof(T); ++i) {
ret_val |= ((v >> (8 * i)) & 0xff) << (8 * (sizeof(T) - 1 - i));
}
return ret_val;
}
inline bool GetLengthPrefixedSlice(Slice* input, Slice* result) {
uint32_t len = 0;
if (GetVarint32(input, &len) && input->size() >= len) {
*result = Slice(input->data(), len);
input->remove_prefix(len);
return true;
} else {
return false;
}
}
inline Slice GetLengthPrefixedSlice(const char* data) {
uint32_t len = 0;
// +5: we assume "data" is not corrupted
// unsigned char is 7 bits, uint32_t is 32 bits, need 5 unsigned char
auto p = GetVarint32Ptr(data, data + 5 /* limit */, &len);
return Slice(p, len);
}
inline Slice GetSliceUntil(Slice* slice, char delimiter) {
uint32_t len = 0;
for (len = 0; len < slice->size() && slice->data()[len] != delimiter; ++len) {
// nothing
}
Slice ret(slice->data(), len);
slice->remove_prefix(len + ((len < slice->size()) ? 1 : 0));
return ret;
}
template<class T>
#ifdef ROCKSDB_UBSAN_RUN
#if defined(__clang__)
__attribute__((__no_sanitize__("alignment")))
#elif defined(__GNUC__)
__attribute__((__no_sanitize_undefined__))
#endif
#endif
inline void PutUnaligned(T *memory, const T &value) {
#if defined(PLATFORM_UNALIGNED_ACCESS_NOT_ALLOWED)
char *nonAlignedMemory = reinterpret_cast<char*>(memory);
memcpy(nonAlignedMemory, reinterpret_cast<const char*>(&value), sizeof(T));
#else
*memory = value;
#endif
}
template<class T>
#ifdef ROCKSDB_UBSAN_RUN
#if defined(__clang__)
__attribute__((__no_sanitize__("alignment")))
#elif defined(__GNUC__)
__attribute__((__no_sanitize_undefined__))
#endif
#endif
inline void GetUnaligned(const T *memory, T *value) {
#if defined(PLATFORM_UNALIGNED_ACCESS_NOT_ALLOWED)
char *nonAlignedMemory = reinterpret_cast<char*>(value);
memcpy(nonAlignedMemory, reinterpret_cast<const char*>(memory), sizeof(T));
#else
*value = *memory;
#endif
}
} // namespace ROCKSDB_NAMESPACE