rocksdb/include/rocksdb/cleanable.h
Peter Dillinger 9d0cae7104 Eliminate unnecessary (slow) block cache Ref()ing in MultiGet (#9899)
Summary:
When MultiGet() determines that multiple query keys can be
served by examining the same data block in block cache (one Lookup()),
each PinnableSlice referring to data in that data block needs to hold
on to the block in cache so that they can be released at arbitrary
times by the API user. Historically this is accomplished with extra
calls to Ref() on the Handle from Lookup(), with each PinnableSlice
cleanup calling Release() on the Handle, but this creates extra
contention on the block cache for the extra Ref()s and Release()es,
especially because they hit the same cache shard repeatedly.

In the case of merge operands (possibly more cases?), the problem was
compounded by doing an extra Ref()+eventual Release() for each merge
operand for a key reusing a block (which could be the same key!), rather
than one Ref() per key. (Note: the non-shared case with `biter` was
already one per key.)

This change optimizes MultiGet not to rely on these extra, contentious
Ref()+Release() calls by instead, in the shared block case, wrapping
the cache Release() cleanup in a refcounted object referenced by the
PinnableSlices, such that after the last wrapped reference is released,
the cache entry is Release()ed. Relaxed atomic refcounts should be
much faster than mutex-guarded Ref() and Release(), and much less prone
to a performance cliff when MultiGet() does a lot of block sharing.

Note that I did not use std::shared_ptr, because that would require an
extra indirection object (shared_ptr itself new/delete) in order to
associate a ref increment/decrement with a Cleanable cleanup entry. (If
I assumed it was the size of two pointers, I could do some hackery to
make it work without the extra indirection, but that's too fragile.)

Some details:
* Fixed (removed) extra block cache tracing entries in cases of cache
entry reuse in MultiGet, but it's likely that in some other cases traces
are missing (XXX comment inserted)
* Moved existing implementations for cleanable.h from iterator.cc to
new cleanable.cc
* Improved API comments on Cleanable
* Added a public SharedCleanablePtr class to cleanable.h in case others
could benefit from the same pattern (potentially many Cleanables and/or
smart pointers referencing a shared Cleanable)
* Add a typedef for MultiGetContext::Mask
* Some variable renaming for clarity

Pull Request resolved: https://github.com/facebook/rocksdb/pull/9899

Test Plan:
Added unit tests for SharedCleanablePtr.

Greatly enhanced ability of existing tests to detect cache use-after-free.
* Release PinnableSlices from MultiGet as they are read rather than in
bulk (in db_test_util wrapper).
* In ASAN build, default to using a trivially small LRUCache for block_cache
so that entries are immediately erased when unreferenced. (Updated two
tests that depend on caching.) New ASAN testsuite running time seems
OK to me.

If I introduce a bug into my implementation where we skip the shared
cleanups on block reuse, ASAN detects the bug in
`db_basic_test *MultiGet*`. If I remove either of the above testing
enhancements, the bug is not detected.

Consider for follow-up work: manipulate or randomize ordering of
PinnableSlice use and release from MultiGet db_test_util wrapper. But in
typical cases, natural ordering gives pretty good functional coverage.

Performance test:
In the extreme (but possible) case of MultiGetting the same or adjacent keys
in a batch, throughput can improve by an order of magnitude.
`./db_bench -benchmarks=multireadrandom -db=/dev/shm/testdb -readonly -num=5 -duration=10 -threads=20 -multiread_batched -batch_size=200`
Before ops/sec, num=5: 1,384,394
Before ops/sec, num=500: 6,423,720
After ops/sec, num=500: 10,658,794
After ops/sec, num=5: 16,027,257

Also note that previously, with high parallelism, having query keys
concentrated in a single block was worse than spreading them out a bit. Now
concentrated in a single block is faster than spread out, which is hopefully
consistent with natural expectation.

Random query performance: with num=1000000, over 999 x 10s runs running before & after simultaneously (each -threads=12):
Before: multireadrandom [AVG    999 runs] : 1088699 (± 7344) ops/sec;  120.4 (± 0.8 ) MB/sec
After: multireadrandom [AVG    999 runs] : 1090402 (± 7230) ops/sec;  120.6 (± 0.8 ) MB/sec
Possibly better, possibly in the noise.

Reviewed By: anand1976

Differential Revision: D35907003

Pulled By: pdillinger

fbshipit-source-id: bbd244d703649a8ca12d476f2d03853ed9d1a17e
2022-04-26 21:59:24 -07:00

129 lines
4.3 KiB
C++

// 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 "rocksdb/rocksdb_namespace.h"
namespace ROCKSDB_NAMESPACE {
class Cleanable {
public:
Cleanable();
// No copy constructor and copy assignment allowed.
Cleanable(Cleanable&) = delete;
Cleanable& operator=(Cleanable&) = delete;
// Executes all the registered cleanups
~Cleanable();
// Move constructor and move assignment is allowed.
Cleanable(Cleanable&&) noexcept;
Cleanable& operator=(Cleanable&&) noexcept;
// Clients are allowed to register function/arg1/arg2 triples that
// will be invoked when this iterator is destroyed.
//
// Note that unlike all of the preceding methods, this method is
// not abstract and therefore clients should not override it.
using CleanupFunction = void (*)(void* arg1, void* arg2);
// Add another Cleanup to the list
void RegisterCleanup(CleanupFunction function, void* arg1, void* arg2);
// Move the cleanups owned by this Cleanable to another Cleanable, adding to
// any existing cleanups it has
void DelegateCleanupsTo(Cleanable* other);
// DoCleanup and also resets the pointers for reuse
inline void Reset() {
DoCleanup();
cleanup_.function = nullptr;
cleanup_.next = nullptr;
}
inline bool HasCleanups() { return cleanup_.function != nullptr; }
protected:
struct Cleanup {
CleanupFunction function;
void* arg1;
void* arg2;
Cleanup* next;
};
Cleanup cleanup_;
// It also becomes the owner of c
void RegisterCleanup(Cleanup* c);
private:
// Performs all the cleanups. It does not reset the pointers. Making it
// private
// to prevent misuse
inline void DoCleanup() {
if (cleanup_.function != nullptr) {
(*cleanup_.function)(cleanup_.arg1, cleanup_.arg2);
for (Cleanup* c = cleanup_.next; c != nullptr;) {
(*c->function)(c->arg1, c->arg2);
Cleanup* next = c->next;
delete c;
c = next;
}
}
}
};
// A copyable, reference-counted pointer to a simple Cleanable that only
// performs registered cleanups after all copies are destroy. This is like
// shared_ptr<Cleanable> but works more efficiently with wrapping the pointer
// in an outer Cleanable (see RegisterCopyWith() and MoveAsCleanupTo()).
// WARNING: if you create a reference cycle, for example:
// SharedCleanablePtr scp;
// scp.Allocate();
// scp.RegisterCopyWith(&*scp);
// It will prevent cleanups from ever happening!
class SharedCleanablePtr {
public:
// Empy/null pointer
SharedCleanablePtr() {}
// Copy and move constructors and assignment
SharedCleanablePtr(const SharedCleanablePtr& from);
SharedCleanablePtr(SharedCleanablePtr&& from) noexcept;
SharedCleanablePtr& operator=(const SharedCleanablePtr& from);
SharedCleanablePtr& operator=(SharedCleanablePtr&& from) noexcept;
// Destructor (decrement refcount if non-null)
~SharedCleanablePtr();
// Create a new simple Cleanable and make this assign this pointer to it.
// (Reset()s first if necessary.)
void Allocate();
// Reset to empty/null (decrement refcount if previously non-null)
void Reset();
// Dereference to pointed-to Cleanable
Cleanable& operator*();
Cleanable* operator->();
// Get as raw pointer to Cleanable
Cleanable* get();
// Creates a (virtual) copy of this SharedCleanablePtr and registers its
// destruction with target, so that the cleanups registered with the
// Cleanable pointed to by this can only happen after the cleanups in the
// target Cleanable are run.
// No-op if this is empty (nullptr).
void RegisterCopyWith(Cleanable* target);
// Moves (virtually) this shared pointer to a new cleanup in the target.
// This is essentilly a move semantics version of RegisterCopyWith(), for
// performance optimization. No-op if this is empty (nullptr).
void MoveAsCleanupTo(Cleanable* target);
private:
struct Impl;
Impl* ptr_ = nullptr;
};
} // namespace ROCKSDB_NAMESPACE