ClickHouse/src/Access/ReplicatedAccessStorage.h

97 lines
3.7 KiB
C++
Raw Normal View History

#pragma once
#include <atomic>
#include <list>
#include <memory>
#include <mutex>
#include <unordered_map>
Support for Clang Thread Safety Analysis (TSA) - TSA is a static analyzer build by Google which finds race conditions and deadlocks at compile time. - It works by associating a shared member variable with a synchronization primitive that protects it. The compiler can then check at each access if proper locking happened before. A good introduction are [0] and [1]. - TSA requires some help by the programmer via annotations. Luckily, LLVM's libcxx already has annotations for std::mutex, std::lock_guard, std::shared_mutex and std::scoped_lock. This commit enables them (--> contrib/libcxx-cmake/CMakeLists.txt). - Further, this commit adds convenience macros for the low-level annotations for use in ClickHouse (--> base/defines.h). For demonstration, they are leveraged in a few places. - As we compile with "-Wall -Wextra -Weverything", the required compiler flag "-Wthread-safety-analysis" was already enabled. Negative checks are an experimental feature of TSA and disabled (--> cmake/warnings.cmake). Compile times did not increase noticeably. - TSA is used in a few places with simple locking. I tried TSA also where locking is more complex. The problem was usually that it is unclear which data is protected by which lock :-(. But there was definitely some weird code where locking looked broken. So there is some potential to find bugs. *** Limitations of TSA besides the ones listed in [1]: - The programmer needs to know which lock protects which piece of shared data. This is not always easy for large classes. - Two synchronization primitives used in ClickHouse are not annotated in libcxx: (1) std::unique_lock: A releaseable lock handle often together with std::condition_variable, e.g. in solve producer-consumer problems. (2) std::recursive_mutex: A re-entrant mutex variant. Its usage can be considered a design flaw + typically it is slower than a standard mutex. In this commit, one std::recursive_mutex was converted to std::mutex and annotated with TSA. - For free-standing functions (e.g. helper functions) which are passed shared data members, it can be tricky to specify the associated lock. This is because the annotations use the normal C++ rules for symbol resolution. [0] https://clang.llvm.org/docs/ThreadSafetyAnalysis.html [1] https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/42958.pdf
2022-06-14 22:35:55 +00:00
#include <base/defines.h>
2021-10-08 08:48:08 +00:00
#include <base/scope_guard.h>
#include <Common/ThreadPool.h>
#include <Common/ZooKeeper/Common.h>
#include <Common/ZooKeeper/ZooKeeper.h>
#include <Common/ConcurrentBoundedQueue.h>
#include <Access/IAccessStorage.h>
namespace DB
{
class AccessChangesNotifier;
/// Implementation of IAccessStorage which keeps all data in zookeeper.
class ReplicatedAccessStorage : public IAccessStorage
{
public:
static constexpr char STORAGE_TYPE[] = "replicated";
ReplicatedAccessStorage(const String & storage_name, const String & zookeeper_path, zkutil::GetZooKeeper get_zookeeper, AccessChangesNotifier & changes_notifier_, bool allow_backup);
virtual ~ReplicatedAccessStorage() override;
const char * getStorageType() const override { return STORAGE_TYPE; }
void startPeriodicReloading() override { startWatchingThread(); }
void stopPeriodicReloading() override { stopWatchingThread(); }
bool exists(const UUID & id) const override;
bool isBackupAllowed() const override { return backup_allowed; }
void backup(BackupEntriesCollector & backup_entries_collector, const String & data_path_in_backup, AccessEntityType type) const override;
void restoreFromBackup(RestorerFromBackup & restorer) override;
private:
String zookeeper_path;
zkutil::GetZooKeeper get_zookeeper;
std::atomic<bool> initialized = false;
std::atomic<bool> watching = false;
ThreadFromGlobalPool watching_thread;
std::shared_ptr<ConcurrentBoundedQueue<UUID>> watched_queue;
std::optional<UUID> insertImpl(const AccessEntityPtr & entity, bool replace_if_exists, bool throw_if_exists) override;
bool removeImpl(const UUID & id, bool throw_if_not_exists) override;
bool updateImpl(const UUID & id, const UpdateFunc & update_func, bool throw_if_not_exists) override;
bool insertWithID(const UUID & id, const AccessEntityPtr & new_entity, bool replace_if_exists, bool throw_if_exists);
bool insertZooKeeper(const zkutil::ZooKeeperPtr & zookeeper, const UUID & id, const AccessEntityPtr & entity, bool replace_if_exists, bool throw_if_exists);
bool removeZooKeeper(const zkutil::ZooKeeperPtr & zookeeper, const UUID & id, bool throw_if_not_exists);
bool updateZooKeeper(const zkutil::ZooKeeperPtr & zookeeper, const UUID & id, const UpdateFunc & update_func, bool throw_if_not_exists);
void initializeZookeeper();
void createRootNodes(const zkutil::ZooKeeperPtr & zookeeper);
void startWatchingThread();
void stopWatchingThread();
void runWatchingThread();
void resetAfterError();
bool refresh();
void refreshEntities(const zkutil::ZooKeeperPtr & zookeeper);
void refreshEntity(const zkutil::ZooKeeperPtr & zookeeper, const UUID & id);
Support for Clang Thread Safety Analysis (TSA) - TSA is a static analyzer build by Google which finds race conditions and deadlocks at compile time. - It works by associating a shared member variable with a synchronization primitive that protects it. The compiler can then check at each access if proper locking happened before. A good introduction are [0] and [1]. - TSA requires some help by the programmer via annotations. Luckily, LLVM's libcxx already has annotations for std::mutex, std::lock_guard, std::shared_mutex and std::scoped_lock. This commit enables them (--> contrib/libcxx-cmake/CMakeLists.txt). - Further, this commit adds convenience macros for the low-level annotations for use in ClickHouse (--> base/defines.h). For demonstration, they are leveraged in a few places. - As we compile with "-Wall -Wextra -Weverything", the required compiler flag "-Wthread-safety-analysis" was already enabled. Negative checks are an experimental feature of TSA and disabled (--> cmake/warnings.cmake). Compile times did not increase noticeably. - TSA is used in a few places with simple locking. I tried TSA also where locking is more complex. The problem was usually that it is unclear which data is protected by which lock :-(. But there was definitely some weird code where locking looked broken. So there is some potential to find bugs. *** Limitations of TSA besides the ones listed in [1]: - The programmer needs to know which lock protects which piece of shared data. This is not always easy for large classes. - Two synchronization primitives used in ClickHouse are not annotated in libcxx: (1) std::unique_lock: A releaseable lock handle often together with std::condition_variable, e.g. in solve producer-consumer problems. (2) std::recursive_mutex: A re-entrant mutex variant. Its usage can be considered a design flaw + typically it is slower than a standard mutex. In this commit, one std::recursive_mutex was converted to std::mutex and annotated with TSA. - For free-standing functions (e.g. helper functions) which are passed shared data members, it can be tricky to specify the associated lock. This is because the annotations use the normal C++ rules for symbol resolution. [0] https://clang.llvm.org/docs/ThreadSafetyAnalysis.html [1] https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/42958.pdf
2022-06-14 22:35:55 +00:00
void refreshEntityNoLock(const zkutil::ZooKeeperPtr & zookeeper, const UUID & id) TSA_REQUIRES(mutex);
Support for Clang Thread Safety Analysis (TSA) - TSA is a static analyzer build by Google which finds race conditions and deadlocks at compile time. - It works by associating a shared member variable with a synchronization primitive that protects it. The compiler can then check at each access if proper locking happened before. A good introduction are [0] and [1]. - TSA requires some help by the programmer via annotations. Luckily, LLVM's libcxx already has annotations for std::mutex, std::lock_guard, std::shared_mutex and std::scoped_lock. This commit enables them (--> contrib/libcxx-cmake/CMakeLists.txt). - Further, this commit adds convenience macros for the low-level annotations for use in ClickHouse (--> base/defines.h). For demonstration, they are leveraged in a few places. - As we compile with "-Wall -Wextra -Weverything", the required compiler flag "-Wthread-safety-analysis" was already enabled. Negative checks are an experimental feature of TSA and disabled (--> cmake/warnings.cmake). Compile times did not increase noticeably. - TSA is used in a few places with simple locking. I tried TSA also where locking is more complex. The problem was usually that it is unclear which data is protected by which lock :-(. But there was definitely some weird code where locking looked broken. So there is some potential to find bugs. *** Limitations of TSA besides the ones listed in [1]: - The programmer needs to know which lock protects which piece of shared data. This is not always easy for large classes. - Two synchronization primitives used in ClickHouse are not annotated in libcxx: (1) std::unique_lock: A releaseable lock handle often together with std::condition_variable, e.g. in solve producer-consumer problems. (2) std::recursive_mutex: A re-entrant mutex variant. Its usage can be considered a design flaw + typically it is slower than a standard mutex. In this commit, one std::recursive_mutex was converted to std::mutex and annotated with TSA. - For free-standing functions (e.g. helper functions) which are passed shared data members, it can be tricky to specify the associated lock. This is because the annotations use the normal C++ rules for symbol resolution. [0] https://clang.llvm.org/docs/ThreadSafetyAnalysis.html [1] https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/42958.pdf
2022-06-14 22:35:55 +00:00
void setEntityNoLock(const UUID & id, const AccessEntityPtr & entity) TSA_REQUIRES(mutex);
void removeEntityNoLock(const UUID & id) TSA_REQUIRES(mutex);
struct Entry
{
UUID id;
AccessEntityPtr entity;
};
std::optional<UUID> findImpl(AccessEntityType type, const String & name) const override;
std::vector<UUID> findAllImpl(AccessEntityType type) const override;
AccessEntityPtr readImpl(const UUID & id, bool throw_if_not_exists) const override;
mutable std::mutex mutex;
Support for Clang Thread Safety Analysis (TSA) - TSA is a static analyzer build by Google which finds race conditions and deadlocks at compile time. - It works by associating a shared member variable with a synchronization primitive that protects it. The compiler can then check at each access if proper locking happened before. A good introduction are [0] and [1]. - TSA requires some help by the programmer via annotations. Luckily, LLVM's libcxx already has annotations for std::mutex, std::lock_guard, std::shared_mutex and std::scoped_lock. This commit enables them (--> contrib/libcxx-cmake/CMakeLists.txt). - Further, this commit adds convenience macros for the low-level annotations for use in ClickHouse (--> base/defines.h). For demonstration, they are leveraged in a few places. - As we compile with "-Wall -Wextra -Weverything", the required compiler flag "-Wthread-safety-analysis" was already enabled. Negative checks are an experimental feature of TSA and disabled (--> cmake/warnings.cmake). Compile times did not increase noticeably. - TSA is used in a few places with simple locking. I tried TSA also where locking is more complex. The problem was usually that it is unclear which data is protected by which lock :-(. But there was definitely some weird code where locking looked broken. So there is some potential to find bugs. *** Limitations of TSA besides the ones listed in [1]: - The programmer needs to know which lock protects which piece of shared data. This is not always easy for large classes. - Two synchronization primitives used in ClickHouse are not annotated in libcxx: (1) std::unique_lock: A releaseable lock handle often together with std::condition_variable, e.g. in solve producer-consumer problems. (2) std::recursive_mutex: A re-entrant mutex variant. Its usage can be considered a design flaw + typically it is slower than a standard mutex. In this commit, one std::recursive_mutex was converted to std::mutex and annotated with TSA. - For free-standing functions (e.g. helper functions) which are passed shared data members, it can be tricky to specify the associated lock. This is because the annotations use the normal C++ rules for symbol resolution. [0] https://clang.llvm.org/docs/ThreadSafetyAnalysis.html [1] https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/42958.pdf
2022-06-14 22:35:55 +00:00
std::unordered_map<UUID, Entry> entries_by_id TSA_GUARDED_BY(mutex);
std::unordered_map<String, Entry *> entries_by_name_and_type[static_cast<size_t>(AccessEntityType::MAX)] TSA_GUARDED_BY(mutex);
AccessChangesNotifier & changes_notifier;
bool backup_allowed = false;
};
}