1
0
mirror of https://github.com/facebook/rocksdb.git synced 2026-10-23 08:09:02 +03:00
Files
rocksdb/monitoring/thread_status_updater.h
Josh Kang 4b983da055 Spell out special member functions (rule of five) (#15387)
Summary:
Pull Request resolved: https://github.com/facebook/rocksdb/pull/15387

## Summary

RocksDB's `.clang-tidy` enables `cppcoreguidelines-special-member-functions`,
`cppcoreguidelines-virtual-class-destructor` and the noexcept-move checks,
and clang warns `-Wdeprecated-copy-with-user-provided-dtor`. A full scan
reports 476 findings. In most of them, a class declares a destructor (or
copy operations) but leaves the rest of copy/move implicit. That has two
costs. Readers can't tell whether a class may be copied or moved. And
resource-owning classes keep an implicit copy that would double-free if
anyone ever used it. This diff resolves 453 of the 476.

Approach, picked per class from least to most invasive:
1. An empty user-provided destructor becomes `= default`. This covers most
   interface classes, such as `virtual ~X() {}` -> `= default`. If the
   class is still copied, that would only rename the deprecated-copy
   warning, so the destructor is removed instead when nothing needs it
   (`Customizable`, `TableFactory` and `InternalKeyComparator` inherit a
   virtual one; `IteratorWrapperBase` isn't polymorphic), and
   `MemTableRep::KeyComparator` gets defaulted copy and move.
2. Classes whose copy is already deleted get explicitly deleted move
   operations too. Move was already unavailable, so this only documents
   it.
3. Resource-owning classes that are never copied get deleted copy and move
   operations. Examples: the C API wrapper structs, which run a
   `destructor_` callback (edited in `tools/c_api_gen/c_base.cc` and
   regenerated into `db/c.cc`); `ShardedCache`, which frees its shards;
   iterator wrappers; and RAII handles.
4. Value types that are copied or moved get the missing operations as
   `= default`, which keeps their current behavior.
5. Exception specifications of existing functions are unchanged, except
   for `WriteBatch`'s move assignment (below): no other `noexcept` is
   added or removed. Only the move operations this diff
   newly declares (`Comparator`, `FileSampledStats` and the thread-local
   `Entry`) are `noexcept`; they copy only scalars or atomics.

`WriteBatch`'s copy and move assignment destroyed `*this` and
placement-new'ed a new `WriteBatch` over it. That is undefined behavior
when `*this` is a subclass, and if the copy constructor threw, `*this`
was left destroyed and was later destroyed again. Move assignment now
assigns member-wise and ends in the same state as the move constructor,
which leaves `has_key_with_ts_`, the timestamp-size tracking and the
other fields it doesn't copy at their defaults. `rep_` is swapped through
a temporary, because a plain string move-assignment can keep the old
buffer (e.g. a large batch reset with `batch = WriteBatch()` would keep
its memory). Nothing in it can throw, so it is now `noexcept`. Copy
assignment copies into a temporary and moves it in, so a throwing copy
leaves `*this` unchanged.

Bugs this exposed and fixed:
- `RestoreAfterCopyOrCreateWorkItem` (backup engine): the hand-written
  move constructor and assignment moved only some fields, so `from_file`
  and `to_file` were empty after `push_back(std::move(...))`. A restore
  that hit a checksum mismatch then reported "While restoring  -> : ..."
  without file names. It now uses the compiler-generated moves, which
  are still `noexcept`.
- `ExternalSstFileIngestionJob` lived in a `std::vector`, and its
  destructor frees raw-owned pointers, so the implicit copy used on
  vector reallocation would double-free. Its file batches also point into
  its own inline file storage, so neither a copy nor a move is safe after
  `Prepare()`. Copy and move are now deleted, and `DBImpl` holds jobs
  through `std::unique_ptr`.

Compatibility: no public class loses an operation users can call. Public
interfaces mostly only get `virtual ~X() = default;`.
`WriteBatch`'s move assignment becomes `noexcept`, which is not part of
the mangled name, so the ABI doesn't change.
`SubcompactionJobInfo` and `CompactionJobInfo` intentionally stay C++20
aggregates, so users' aggregate and designated initializers keep
compiling. `Configurable`, `CompareInterface`, `WriteBatchBase`,
`ObjectLibrary::Entry` and `MemTableRep::KeyComparator` declare defaulted
copy and move, and `Customizable` no longer declares a destructor. As a
result, `Configurable` and `Customizable` subclasses that declare no
destructor or copy operations of their own now move their
registered-options list instead of copying it. In-tree, these are the
mutable-options adapters, the table factories, the memory allocators and
test classes; all are held through smart pointers or used in place, and
`Clone()` copies, so none is moved by value.

Not changed (23): `IODebugContext` (custom deep copy plus a
`std::shared_mutex`), `DBWithColumnFamilies` in db_bench (stored in a
`std::vector` with non-movable members), `Snapshot`'s protected virtual
destructor (intended), `ManagedSnapshot`, the `SubcompactionJobInfo` and
`CompactionJobInfo` aggregates, and the 16 noexcept-move-constructor and
noexcept-swap findings on existing functions, including `autovector`'s
move assignment (adding `noexcept` would change their exception
specifications). Copying `ManagedSnapshot` already double-releases the
snapshot. Fixing that needs a real move implementation and is a behavior
change, so it is left for a separate diff.

Reviewed By: rban1

Differential Revision: D123186309

fbshipit-source-id: 6f8e57e233e12199a007407130bb9a1f6d52e8d1
2026-10-12 16:00:09 -07:00

227 lines
8.5 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).
//
// The implementation of ThreadStatus.
//
// Note that we make get and set access to ThreadStatusData lockless.
// As a result, ThreadStatusData as a whole is not atomic. However,
// we guarantee consistent ThreadStatusData all the time whenever
// user call GetThreadList(). This consistency guarantee is done
// by having the following constraint in the internal implementation
// of set and get order:
//
// 1. When reset any information in ThreadStatusData, always start from
// clearing up the lower-level information first.
// 2. When setting any information in ThreadStatusData, always start from
// setting the higher-level information.
// 3. When returning ThreadStatusData to the user, fields are fetched from
// higher-level to lower-level. In addition, where there's a nullptr
// in one field, then all fields that has lower-level than that field
// should be ignored.
//
// The high to low level information would be:
// thread_id > thread_type > db > cf > operation > state
//
// This means user might not always get full information, but whenever
// returned by the GetThreadList() is guaranteed to be consistent.
#pragma once
#include <atomic>
#include <list>
#include <memory>
#include <mutex>
#include <string>
#include <unordered_map>
#include <unordered_set>
#include <vector>
#include "port/port.h"
#include "rocksdb/status.h"
#include "rocksdb/thread_status.h"
#include "util/thread_operation.h"
namespace ROCKSDB_NAMESPACE {
class ColumnFamilyHandle;
// The structure that keeps constant information about a column family.
struct ConstantColumnFamilyInfo {
#ifndef NROCKSDB_THREAD_STATUS
public:
ConstantColumnFamilyInfo(const void* _db_key, const std::string& _db_name,
const std::string& _cf_name)
: db_key(_db_key), db_name(_db_name), cf_name(_cf_name) {}
const void* db_key;
const std::string db_name;
const std::string cf_name;
#endif // !NROCKSDB_THREAD_STATUS
};
// the internal data-structure that is used to reflect the current
// status of a thread using a set of atomic pointers.
struct ThreadStatusData {
#ifndef NROCKSDB_THREAD_STATUS
explicit ThreadStatusData() {
enable_tracking.store(false);
thread_id.store(0);
thread_type.store(ThreadStatus::USER);
cf_key.store(nullptr);
operation_type.store(ThreadStatus::OP_UNKNOWN);
op_start_time.store(0);
state_type.store(ThreadStatus::STATE_UNKNOWN);
}
// A flag to indicate whether the thread tracking is enabled
// in the current thread.
// If set to false, then SetThreadOperation and SetThreadState
// will be no-op.
std::atomic<bool> enable_tracking;
std::atomic<uint64_t> thread_id;
std::atomic<ThreadStatus::ThreadType> thread_type;
std::atomic<void*> cf_key;
std::atomic<ThreadStatus::OperationType> operation_type;
std::atomic<uint64_t> op_start_time;
std::atomic<ThreadStatus::OperationStage> operation_stage;
std::atomic<uint64_t> op_properties[ThreadStatus::kNumOperationProperties];
std::atomic<ThreadStatus::StateType> state_type;
#endif // !NROCKSDB_THREAD_STATUS
};
// The class that stores and updates the status of the current thread
// using a thread-local ThreadStatusData.
//
// In most of the case, you should use ThreadStatusUtil to update
// the status of the current thread instead of using ThreadSatusUpdater
// directly.
//
// @see ThreadStatusUtil
class ThreadStatusUpdater {
public:
ThreadStatusUpdater() {}
// Releases all ThreadStatusData of all active threads.
virtual ~ThreadStatusUpdater() = default;
// Unregister the current thread.
void UnregisterThread();
// Reset the status of the current thread. This includes resetting
// ColumnFamilyInfoKey, ThreadOperation, and ThreadState.
void ResetThreadStatus();
// Set the id of the current thread.
void SetThreadID(uint64_t thread_id);
// Register the current thread for tracking.
void RegisterThread(ThreadStatus::ThreadType ttype, uint64_t thread_id);
void SetEnableTracking(bool enable_tracking);
// Update the column-family info of the current thread by setting
// its thread-local pointer of ThreadStatusData to the correct entry.
void SetColumnFamilyInfoKey(const void* cf_key);
// returns the column family info key.
const void* GetColumnFamilyInfoKey();
// Update the thread operation of the current thread.
void SetThreadOperation(const ThreadStatus::OperationType type);
// Return the thread operation of the current thread.
ThreadStatus::OperationType GetThreadOperation();
// The start time of the current thread operation. It is in the format
// of micro-seconds since some fixed point in time.
void SetOperationStartTime(const uint64_t start_time);
// Set the "i"th property of the current operation.
//
// NOTE: Our practice here is to set all the thread operation properties
// and stage before we set thread operation, and thread operation
// will be set in std::memory_order_release. This is to ensure
// whenever a thread operation is not OP_UNKNOWN, we will always
// have a consistent information on its properties.
void SetThreadOperationProperty(int i, uint64_t value);
// Increase the "i"th property of the current operation with
// the specified delta.
void IncreaseThreadOperationProperty(int i, uint64_t delta);
// Update the thread operation stage of the current thread.
ThreadStatus::OperationStage SetThreadOperationStage(
const ThreadStatus::OperationStage stage);
// Clear thread operation of the current thread.
void ClearThreadOperation();
// Reset all thread-operation-properties to 0.
void ClearThreadOperationProperties();
// Update the thread state of the current thread.
void SetThreadState(const ThreadStatus::StateType type);
// Clear the thread state of the current thread.
void ClearThreadState();
// Obtain the status of all active registered threads.
Status GetThreadList(std::vector<ThreadStatus>* thread_list);
// Create an entry in the global ColumnFamilyInfo table for the
// specified column family. This function should be called only
// when the current thread does not hold db_mutex.
void NewColumnFamilyInfo(const void* db_key, const std::string& db_name,
const void* cf_key, const std::string& cf_name);
// Erase all ConstantColumnFamilyInfo that is associated with the
// specified db instance. This function should be called only when
// the current thread does not hold db_mutex.
void EraseDatabaseInfo(const void* db_key);
// Erase the ConstantColumnFamilyInfo that is associated with the
// specified ColumnFamilyData. This function should be called only
// when the current thread does not hold db_mutex.
void EraseColumnFamilyInfo(const void* cf_key);
// Verifies whether the input ColumnFamilyHandles matches
// the information stored in the current cf_info_map.
void TEST_VerifyColumnFamilyInfoMap(
const std::vector<ColumnFamilyHandle*>& handles, bool check_exist);
protected:
#ifndef NROCKSDB_THREAD_STATUS
// The thread-local variable for storing thread status.
static thread_local ThreadStatusData* thread_status_data_;
// Returns the pointer to the thread status data only when the
// thread status data is non-null and has enable_tracking == true.
ThreadStatusData* GetLocalThreadStatus();
// Directly returns the pointer to thread_status_data_ without
// checking whether enabling_tracking is true of not.
ThreadStatusData* Get() { return thread_status_data_; }
// The mutex that protects cf_info_map and db_key_map.
std::mutex thread_list_mutex_;
// The current status data of all active threads.
std::unordered_set<ThreadStatusData*> thread_data_set_;
// A global map that keeps the column family information. It is stored
// globally instead of inside DB is to avoid the situation where DB is
// closing while GetThreadList function already get the pointer to its
// CopnstantColumnFamilyInfo.
std::unordered_map<const void*, ConstantColumnFamilyInfo> cf_info_map_;
// A db_key to cf_key map that allows erasing elements in cf_info_map
// associated to the same db_key faster.
std::unordered_map<const void*, std::unordered_set<const void*>> db_key_map_;
#else
static ThreadStatusData* thread_status_data_;
#endif // !NROCKSDB_THREAD_STATUS
};
} // namespace ROCKSDB_NAMESPACE