http://bugzilla.opensuse.org/show_bug.cgi?id=1131330 Bug ID: 1131330 Summary: pthread_mutex_destroy occasionally returns EBUSY even though mutex is unlocked Classification: openSUSE Product: openSUSE Distribution Version: Leap 15.0 Hardware: x86-64 OS: SUSE Other Status: NEW Severity: Normal Priority: P5 - None Component: Basesystem Assignee: bnc-team-screening@forge.provo.novell.com Reporter: asbeer@gmail.com QA Contact: qa-bugs@suse.de Found By: --- Blocker: --- User-Agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/73.0.3683.86 Safari/537.36 Build Identifier: I'm running the latest patched/updated OpenSUSE LEAP 15.0. My glibc version is 2.26-lp150.11.9.1. When I compile and run the following program on my system and let it run for a while (normally less than 30 seconds) it exits indicating that pthread_mutex_destroy() returned 16 (EBUSY). However, it's obvious from the code that pthread_mutex_unlock() was called right before pthread_mutex_destroy() and thus the mutex is not locked. So I believe the EBUSY value is being erroneously returned in this case. g++ -g -O2 -o test_ebusy_bug -lpthread test_ebusy_bug.cpp test_ebusy_bug.cpp: #include <vector> #include <thread> #include <pthread.h> #include <iostream> static const size_t tCnt = 10; struct test_struct { pthread_mutex_t mutex; pthread_cond_t cond1; pthread_cond_t cond2; bool run; std::vector<bool> runningList; test_struct() : run(false) , runningList(tCnt) { } void run_func(size_t index) { pthread_mutex_lock(&mutex); runningList[index] = true; pthread_cond_broadcast(&cond1); while (run) { pthread_cond_wait(&cond2, &mutex); } runningList[index] = false; pthread_mutex_unlock(&mutex); } bool spawn_join_threads() { std::vector<std::thread> tList; // Initialize the pthread variables pthread_mutex_init(&mutex, NULL); pthread_cond_init(&cond1, NULL); pthread_cond_init(&cond2, NULL); // Start all of the threads run = true; for (size_t i = 0; i < tCnt; i++) { runningList[i] = false; } for (size_t i = 0; i < tCnt; i++) { tList.push_back(std::thread(&test_struct::run_func, this, i)); } // Wait for all of the threads to start pthread_mutex_lock(&mutex); for (size_t i = 0; i < tCnt; i++) { while (!runningList[i]) { pthread_cond_wait(&cond1, &mutex); } } pthread_mutex_unlock(&mutex); // Stop all of the threads run = false; pthread_cond_broadcast(&cond2); for (int i = 0; i < tCnt; i++) { tList[i].join(); } tList.clear(); // Verify that all of the threads have stopped pthread_mutex_lock(&mutex); for (size_t i = 0; i < tCnt; i++) { if (runningList[i]) { std::cerr << "A thread hasn't stopped" << std::endl; pthread_mutex_unlock(&mutex); return false; } } pthread_mutex_unlock(&mutex); // Destroy the pthread variables int ret = pthread_cond_destroy(&cond2); if (ret) { std::cerr << "pthread_cond_destroy returned " << ret << std::endl; return false; } ret = pthread_cond_destroy(&cond1); if (ret) { std::cerr << "pthread_cond_destroy returned " << ret << std::endl; return false; } ret = pthread_mutex_destroy(&mutex); if (ret) { std::cerr << "pthread_mutex_destroy returned " << ret << std::endl; return false; } return true; } }; int main() { test_struct t; while (t.spawn_join_threads()); return 0; } Reproducible: Always Steps to Reproduce: Compile and run the above program. Actual Results: The program runs for awhile and then exits with the message: pthread_mutex_destroy returned 16 Expected Results: The program runs indefinitely without exiting. -- You are receiving this mail because: You are on the CC list for the bug.