Running the very core of SKL network, SKALE BFT consensus is universal, modern, modular, high-performance, asynchronous, provably-secure, agent-based Proof-of-Stake blockchain consensus engine in C++ 17. Includes provably secure embedded Oracle. Used by SKALE elastic blockchains. Easy and flexible enough to implement your own blockchain or smart contract platform. BLS signatures and Binary Asynchronous Consensus are main building blocks.
The CacheLevelDB class needs review and improvement because of helgrind reports about serious multi-threading problems like unlocked an invalid lock, associated lock is not held by any thread and unlocked a not-locked lock at. This class contains CacheLevelDB::checkForDeadLockRead() which violates lock based data protection strategy:
it unlocks synchronization object in other thread than it was initially locked
it unlocks already unlocked synchronization object
it unlocks dead synchronization object
it violates normal design of std::lock_guard and std::unique_lock strategy
Example of call stack 1:
==23347== Thread skalenetwork/IMA#55 unlocked an invalid lock at 0x19D713F8
==23347== at 0x4D9CCD6: ??? (in /usr/libexec/valgrind/vgpreload_helgrind-amd64-linux.so)
==23347== by 0x2211FF1: std::__glibcxx_rwlock_unlock(pthread_rwlock_t*) (shared_mutex:75)
==23347== by 0x2212049: std::__shared_mutex_pthread::unlock() (shared_mutex:206)
==23347== by 0x22120A5: std::__shared_mutex_pthread::unlock_shared() (shared_mutex:246)
==23347== by 0x25D1E9F: std::shared_timed_mutex::unlock_shared() (shared_mutex:474)
==23347== by 0x25C7C76: CacheLevelDB::checkForDeadLockRead(char const*) (CacheLevelDB.cpp:183)
==23347== by 0x25C99C7: CacheLevelDB::readPrefixRange(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >&) (CacheLevelDB.cpp:296)
==23347== by 0x25EC561: MsgDB::getMessages(block_id) (MsgDB.cpp:87)
==23347== by 0x24BF075: Schain::ifIncompleteConsensusDetectedRestartAndRebroadcastAllMessagesForCurrentBlock() (Schain.cpp:970)
==23347== by 0x24BE125: Schain::bootstrap(block_id, unsigned long, unsigned long) (Schain.cpp:953)
==23347== by 0x23A6C1D: ConsensusEngine::bootStrapAll() (ConsensusEngine.cpp:559)
==23347== by 0x1FC2D8E: SkaleHost::startWorking()::{lambda()#1}::operator()() const (SkaleHost.cpp:755)
Example of call stack 2:
==23347== Thread skalenetwork/IMA#73 unlocked a not-locked lock at 0x1A285E08
==23347== at 0x4D9CCD6: ??? (in /usr/libexec/valgrind/vgpreload_helgrind-amd64-linux.so)
==23347== by 0x2211FF1: std::__glibcxx_rwlock_unlock(pthread_rwlock_t*) (shared_mutex:75)
==23347== by 0x2212049: std::__shared_mutex_pthread::unlock() (shared_mutex:206)
==23347== by 0x22120A5: std::__shared_mutex_pthread::unlock_shared() (shared_mutex:246)
==23347== by 0x25D1E9F: std::shared_timed_mutex::unlock_shared() (shared_mutex:474)
==23347== by 0x25C7C76: CacheLevelDB::checkForDeadLockRead(char const*) (CacheLevelDB.cpp:183)
==23347== by 0x25C6C1A: CacheLevelDB::readString(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >&) (CacheLevelDB.cpp:120)
==23347== by 0x25EE8D8: PriceDB::readPrice(block_id) (PriceDB.cpp:58)
==23347== by 0x267DE79: PricingAgent::readPrice(block_id) (PricingAgent.cpp:105)
==23347== by 0x24F08DF: Schain::getPriceForBlockId(unsigned long) (SchainGettersSetters.cpp:236)
==23347== by 0x23AD5BF: ConsensusEngine::getPriceForBlockId(unsigned long) const (ConsensusEngine.cpp:906)
==23347== by 0x1FC4CC2: SkaleHost::getGasPrice() const (SkaleHost.cpp:906)
==23347== Lock at 0x1A285E08 was first observed
==23347== at 0x4D9C5DF: ??? (in /usr/libexec/valgrind/vgpreload_helgrind-amd64-linux.so)
==23347== by 0x2211FC2: std::__glibcxx_rwlock_wrlock(pthread_rwlock_t*) (shared_mutex:73)
==23347== by 0x2212017: std::__shared_mutex_pthread::lock() (shared_mutex:186)
==23347== by 0x25D1E3F: std::shared_timed_mutex::lock() (shared_mutex:459)
==23347== by 0x25D2587: std::lock_guard<std::shared_timed_mutex>::lock_guard(std::shared_timed_mutex&) (std_mutex.h:159)
==23347== by 0x25C7E46: CacheLevelDB::writeString(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, bool) (CacheLevelDB.cpp:202)
==23347== by 0x25EF20D: PriceDB::savePrice(boost::multiprecision::number<boost::multiprecision::backends::cpp_int_backend<256u, 256u, (boost::multiprecision::cpp_integer_type)0, (boost::multiprecision::cpp_int_check_type)0, void>, (boost::multiprecision::expression_template_option)0> const&, block_id) (PriceDB.cpp:83)
==23347== by 0x267D9F3: PricingAgent::savePrice(boost::multiprecision::number<boost::multiprecision::backends::cpp_int_backend<256u, 256u, (boost::multiprecision::cpp_integer_type)0, (boost::multiprecision::cpp_int_check_type)0, void>, (boost::multiprecision::expression_template_option)0>, block_id) (PricingAgent.cpp:98)
==23347== by 0x267D3F4: PricingAgent::calculatePrice(std::vector<std::vector<unsigned char, std::allocator<unsigned char> >, std::allocator<std::vector<unsigned char, std::allocator<unsigned char> > > > const&, unsigned long, unsigned int, block_id) (PricingAgent.cpp:84)
==23347== by 0x24BE092: Schain::bootstrap(block_id, unsigned long, unsigned long) (Schain.cpp:943)
==23347== by 0x23A6C1D: ConsensusEngine::bootStrapAll() (ConsensusEngine.cpp:559)
==23347== by 0x1FC2D8E: SkaleHost::startWorking()::{lambda()#1}::operator()() const (SkaleHost.cpp:755)
The test was performed by running skaled like valgrind --tool=helgrind ./skaled --colors --config ./config0.json -v 4 --log-value-size-limit 1024000 --performance-timeline-enable --performance-timeline-max-items=16000000 --ssl-key ./key.pem --ssl-cert ./cert.pem
The problem described in this task is top most occurring and repeating while running under helgrind analysis.
kladkogex
commented
2 years ago
The
CacheLevelDBclass needs review and improvement because of helgrind reports about serious multi-threading problems likeunlocked an invalid lock,associated lock is not held by any threadandunlocked a not-locked lock at. This class containsCacheLevelDB::checkForDeadLockRead()which violates lock based data protection strategy:std::lock_guardandstd::unique_lockstrategy Example of call stack 1:Example of call stack 2:
The test was performed by running
skaledlikevalgrind --tool=helgrind ./skaled --colors --config ./config0.json -v 4 --log-value-size-limit 1024000 --performance-timeline-enable --performance-timeline-max-items=16000000 --ssl-key ./key.pem --ssl-cert ./cert.pemThe problem described in this task is top most occurring and repeating while running under helgrind analysis.