I have some trouble understanding the unique_lock and condition_variable in the code below.
#define _GLIBCXX_USE_NANOSLEEP
#include <chrono>
#include <mutex>
#include <thread>
#include <iostream>
#include <condition_variable>
using namespace std;
mutex m;
int i=1;
condition_variable cv;
void funcA()
{
cout << "FuncA Before lock" << endl;
unique_lock<mutex> lk(m);
cout << "FuncA After lock" << endl;
while(i != 2) { cv.wait(lk); }
//std::chrono::milliseconds dura(500);//make sure thread is running
//std::this_thread::sleep_for(dura); //this_thread::sleep_for(dura);
cout << "FuncA After sleep" << endl;
lk.unlock();
cv.notify_all();
}
void funcB()
{
cout << "FuncB Before lock" << endl;
unique_lock<mutex> lk(m);
cout << "FuncB After lock" << endl;
while(i != 1) { cv.wait(lk); }
//std::chrono::milliseconds dura(500);//make sure thread is running
//std::this_thread::sleep_for(dura); //this_thread::sleep_for(dura);
cout << "FuncB After sleep" << endl;
++i;
lk.unlock();
cv.notify_all();
}
int main(int argc, char* argv[])
{
auto a = thread(funcA), b=thread(funcB);
a.join(), b.join();
return 0;
}
As per https://en.cppreference.com/w/cpp/thread/unique_lock/unique_lock (when the constructor is called with a mutex) the constructor locks the associated mutex by calling m.lock()
Let us say, threadA/funcA gets scheduled first by the operating system. So, the following lines will get executed
cout << "FuncA Before lock" << endl;
unique_lock<mutex> lk(m);
cout << "FuncA After lock" << endl;
while(i != 2) { cv.wait(lk); }
Now, threadA has the mutex since the constructor for the unique_lock has called lock on it. However, since i!=2, the thread will stay in the while-loop waiting to be notified by the other thread.
Now, let us say the operating system switches the threads and funcB is called.
cout << "FuncB Before lock" << endl;
unique_lock<mutex> lk(m);
Now, this thread should get stuck here on the second line right? Because mutex m is already owned by threadA (since it has called m.lock() on it), and therefore the m.lock() call by unique_lock constructor should block since the mutex is not available.
So, there's a deadlock? The threadA is waiting in the infinite loop waiting for i to be incremented to 2, whereas threadB is waiting for the mutex owned by threadA.
However, nothing like that happens, and the code works fine.
Related
Is it clear what would happen if a thread blocked on a std::condition_variable receives a notification but the lock on the associated mutex is not yet released, and the lock will be released 10 seconds later? Would the thread wait for the lock to be released or is the situation undefined?
I added a 15seconds sleep on purpose to make the lock unavailable during that time to see how waiting thread will be doing. It is doing OK but just wanted to be sure about it.
#include <iostream>
#include <condition_variable>
#include <thread>
#include <chrono>
struct SharedResource
{
SharedResource() :
cv_mutex(), cv(), counter(0)
{
}
/*
This mutex is used for three purposes:
1) to synchronize accesses to counter
2) to synchronize accesses to std::cerr
3) for the condition variable cv
*/
std::mutex cv_mutex;
std::condition_variable cv;
int counter;
};
void waits(SharedResource& sharedRes)
{
std::unique_lock<std::mutex> lk(sharedRes.cv_mutex);
std::cerr << "Waiting... \n";
while (sharedRes.counter != 1)
{
sharedRes.cv.wait_for(lk,3s);
std::cerr << "Thread ID: " << std::this_thread::get_id() << " wakes up every 3 seconds.\n";
}
std::cerr << "...finished waiting." << "counter: " << sharedRes.counter << std::endl;
} //The lk object will be unlocked after this scope ends.
void signals(SharedResource& sharedRes)
{
std::this_thread::sleep_for(std::chrono::seconds(1));
{
std::lock_guard<std::mutex> lk(sharedRes.cv_mutex);
std::cerr << "Notifying...\n";
} // The lk object will be unlocked after this scope ends.
sharedRes.cv.notify_all();
std::this_thread::sleep_for(std::chrono::seconds(6));
{
std::lock_guard<std::mutex> lk(sharedRes.cv_mutex);
sharedRes.counter = 1;
std::cerr << "Notifying again...\n";
sharedRes.cv.notify_all();
std::this_thread::sleep_for(std::chrono::seconds(15));
}// The lk object will be unlocked after this scope ends.
}
int main()
{
SharedResource sharedRes;
std::thread
t1(waits, std::ref(sharedRes)),
t2(signals, std::ref(sharedRes));
t1.join();
t2.join();
}
If a thread that is blocked on a std::condition_variable is notified, but the lock on the associated mutex has not been released what would be result?
It will continue to wait / wait_for until it can reacquire the lock.
When std::condition_variable::wait and wait_for returns (for whatever reason), the lock is held again, so you don't have to worry about that.
It can even return from wait without having gotten any notifications (spurious wake-ups) - but, no matter what, the lock is reacquired when the call returns.
According to several documentation examples the thread can't unlock a mutex unless it locked it explicitly. Here is an excerpt from man page for pthread_mutex_unlock at IBM.
The pthread_mutex_unlock() function unlocks the mutex specified. If
the calling thread does not currently hold the mutex (via a previous
call to pthread_mutex_lock(), pthread_mutex_trylock(), or
pthread_mutex_timedlock_np()) the unlock request fails with the EPERM
error.
Even the new C++ standard says something similar about the thread ownership, yet the following program was able to unlock a mutex locked on a different thread. On gcc & Linux systems the same behavior is seen both on pthread mutex as well as std::mutex (which I believe is implemented based on pthread_mutex anyway).
#include <iostream>
#include <thread>
#include <mutex>
#include <pthread.h>
std::mutex stmutex;
pthread_mutex_t pthrmutex = PTHREAD_MUTEX_INITIALIZER;
void thread1(int i)
{
stmutex.unlock();
std::cout << "Un Locked in thread 1" << std::this_thread::get_id() << std::endl;
}
void pthread1(int i)
{
pthread_mutex_unlock(&pthrmutex);
std::cout << "Un Locked Pthread mutex in thread 1" << std::this_thread::get_id() << std::endl;
}
void thread2(int i)
{
stmutex.lock();
std::cout << "Locked in thread 2" << std::this_thread::get_id() << std::endl;
}
void thread3(int i)
{
stmutex.unlock();
std::cout << "UNLocked in thread 3" << std::this_thread::get_id() << std::endl;
}
int main(int argc, char **argv)
{
try {
stmutex.lock();
std::cout << "Locked in main thread : " << std::this_thread::get_id() << std::endl;
std::thread t1(thread1,1);
t1.join();
stmutex.lock();
std::cout << "Locked in main thread after unlocking in thread1" << std::endl;
stmutex.unlock();
std::cout << "Un Locked in main thread " << std::endl;
pthread_mutex_lock(&pthrmutex);
std::cout << "Locked pthread mutex in main thread : " << std::this_thread::get_id() << std::endl;
std::thread t2(pthread1,1);
t2.join();
pthread_mutex_lock(&pthrmutex);
std::cout << "Locked Pthread mutext in main thread after unlocking in thread1" << std::endl;
pthread_mutex_unlock(&pthrmutex);
std::cout << "Un Locked Pthread mutext in main thread " << std::endl;
std::thread t3(thread2,1);
t3.join();
std::thread t4(thread3,1);
t4.join();
} catch (std::exception& ex)
{
std::cerr << "Exception In main thread: " << ex.what() << std::endl;
}
}
Am I missing anything in my understanding of mutex "Ownership" ?
The request fails with the EPERM error only for mutexes created with PTHREAD_MUTEX_ERRORCHECK.
See pthread_mutex_lock section RATIONALE:
... while being able to extract the thread ID of the owner of a mutex might be desirable, it would require storing the current thread ID when each mutex is locked, and this could incur unacceptable levels of overhead.
I.e. initialize your mutex with PTHREAD_ERRORCHECK_MUTEX_INITIALIZER_NP.
Example:
void locker(pthread_mutex_t* mutex) {
if(int e = pthread_mutex_lock(mutex))
fprintf(stderr, "pthread_mutex_lock: (%d)%s\n", e, strerror(e));
}
void unlocker(pthread_mutex_t* mutex) {
if(int e = pthread_mutex_unlock(mutex))
fprintf(stderr, "pthread_mutex_unlock: (%d)%s\n", e, strerror(e));
}
int main() {
pthread_mutex_t a = PTHREAD_MUTEX_INITIALIZER;
pthread_mutex_t b = PTHREAD_ERRORCHECK_MUTEX_INITIALIZER_NP;
std::thread(locker, &a).join();
std::thread(locker, &b).join();
std::thread(unlocker, &a).join();
std::thread(unlocker, &b).join(); // pthread_mutex_unlock: (1)Operation not permitted
}
I'm trying to write a program with c++11 in which multiple threads are run, and, during each cycle the main thread will wait for each thread to be finished. The program below is a testing program for this concept.
Apparently I'm missing something trivial in my implementation as it looks like I'm experiencing a deadlock (Not always, just during some random runs).
#include <iostream>
#include <stdio.h>
#include <thread>
#include <chrono>
#include <condition_variable>
#include <mutex>
using namespace std;
class Producer
{
public:
Producer(int a_id):
m_id(a_id),
m_ready(false),
m_terminate(false)
{
m_id = a_id;
m_thread = thread(&Producer::run, this);
// ensure thread is available before it is started
this_thread::sleep_for(std::chrono::milliseconds(100));
}
~Producer() {
terminate();
m_thread.join();
}
void start() {
//cout << "start " << m_id << endl;
unique_lock<mutex> runLock(m_muRun);
m_ready = true;
runLock.unlock();
m_cond.notify_all();
}
void wait() {
cout << "wait " << m_id << endl;
unique_lock<decltype(m_muRun)> runLock(m_muRun);
m_cond.wait(runLock, [this]{return !m_ready;});
}
void terminate() {
m_terminate = true;
start();
}
void run() {
do {
unique_lock<decltype(m_muRun)> runLock(m_muRun);
m_cond.wait(runLock, [this]{return m_ready;});
if (!m_terminate) {
cout << "running thread: " << m_id << endl;
} else {
cout << "exit thread: " << m_id << endl;
}
runLock.unlock();
m_ready = false;
m_cond.notify_all();
} while (!m_terminate);
}
private:
int m_id;
bool m_ready;
bool m_terminate;
thread m_thread;
mutex m_muRun;
condition_variable m_cond;
};
int main()
{
Producer producer1(1);
Producer producer2(2);
Producer producer3(3);
for (int i=0; i<10000; ++i) {
cout << i << endl;
producer1.start();
producer2.start();
producer3.start();
producer1.wait();
producer2.wait();
producer3.wait();
}
cout << "exit" << endl;
return 0;
}
The program's output when the deadlock is occurring:
....
.......
running thread: 2
running thread: 1
wait 1
wait 2
wait 3
running thread: 3
Looking at the program's output when the deadlock occurs, I suspect the bottleneck of the program is that sometimes the Producer::wait function is called, before the corresponding thread is actually started, i.e. the command Producer::start should have triggered the start, a.k. unlocking of the mutex, however it is not yet picked up by the thread's run method (Producer::run), (NB: I'm not 100% sure of this!). I'm a bit lost here, hopefully somebody can provide some help.
You have race condition in this code:
runLock.unlock();
m_ready = false;
m_ready variable must be always protected by mutex for proper synchronization. And it is completely unnecessary to wait for thread to start this_thread::sleep_for() - proper synchronization would take care of that as well so you can simply remove that line. Note this is pretty inefficient way of doing proper multithreading - there should be thread pool instead of individual object with separate mutex and condition variable each.
In the following codes I suppose mLooperMutex can't be acquired by the children thread. But the program output is quite surprising. It looks that mLooperMutex captured in the std::thread is not the same one in main thread.
But if I changed the detach() call of std::thread to join(), this will lead to a deadlock since the mLooperMutex has been locked by the main thread.
Are there anything wrong with this program if I'd like to use the mLooperMutex among different threads?
a.out:
main: wait cond begin
child: acquiring lock begin
child: acquiring lock done
child: notify one begin
child: notify one done
main: wait cond done
#include <iostream>
#include <thread>
#include <mutex>
#include <condition_variable>
using namespace std;
int main()
{
std::condition_variable looperSet;
bool child_done = false;
std::mutex mLooperMutex;
cout << "main: acquiring lock begin" << endl;
std::unique_lock<std::mutex> lock(mLooperMutex);
cout << "main: acquiring lock done" << endl;
std::thread{[&mutex=mLooperMutex, &looperSet, &child_done] () {
cout << "child: acquiring lock begin" << endl;
std::unique_lock<std::mutex> lock(mutex);
cout << "child: acquiring lock done" << endl;
child_done = true;
lock.unlock();
cout << "child: notify one begin" << endl;
looperSet.notify_one();
cout << "child: notify one done" << endl;
}}.detach();
cout << "main: wait cond begin" << endl;
looperSet.wait(lock, [&child_done]{ return child_done; });
cout << "main: wait cond done" << endl;
return 0;
}
The reason why the mLooperMutex can be acquired in the child thread is because the lock is released by looperSet.wait:
// This unlocks "lock", and then locks it again afterwards.
looperSet.wait(lock, [&child_done]{ return child_done; });
The reason why this doesn't work with .join() is because .join() waits for the thread to finish before proceeding, and the thread can't finish until the lock is released, and looperSet.wait() which releases the lock won't run until .join() finishes.
Creating a thread and then immediately calling .join() is not very useful, you might as well run the code directly rather than using a thread.
The code is acquiring the same mutex from two different threads at the same time.
I understand that a deadlock should occur. Why it is not happening?
#include <iostream>
#include <thread>
#include <queue>
#include <mutex>
#include <condition_variable>
template <typename T>
class SafeQueue
{
public:
T pop()
{
std::unique_lock<std::mutex> mlock(mutex_);
std::cout << "lock pop()" << std::endl;
while (queue_.empty())
{
cond_.wait(mlock);
std::cout << "lock pop awake. Items: " << queue_.size() << std::endl;
}
auto item = queue_.front();
queue_.pop();
std::cout << "returning from pop" << std::endl;
return item;
}
void push(const T& item)
{
std::unique_lock<std::mutex> mlock(mutex_);
std::cout << "lock push()" << std::endl;
queue_.push(item);
mlock.unlock();
cond_.notify_one();
}
private:
std::queue<T> queue_;
mutable std::mutex mutex_;
std::condition_variable cond_;
};
SafeQueue<int> queue;
void pop()
{
std::cout << "popping..." << std::endl;
std::cout << "popped: " << queue.pop() << std::endl;
}
int main()
{
std::thread consumerThread(pop);
std::this_thread::sleep_for(std::chrono::seconds(3));
std::cout << "main thread will push" << std::endl;
queue.push(2);
std::cout << "pushed" << std::endl;
consumerThread.join();
std::cout << "end" << std::endl << std::endl;
}
My output is:
popping...
lock pop()
main thread will push
lock push()
pushed
lock pop awake. Items: 1
returning from pop
popped: 2
end
This statement:
cond_.wait(mlock);
actually unlocks the mutex during the wait, and reacquires the lock once signaled. That's why you don't have any deadlock.
What is happening is that std::condition_variable::wait is releasing the mutex. The thread then waits until the notify_one call, that will release the condition and re-acquire the mutex.
http://en.cppreference.com/w/cpp/thread/condition_variable/wait
"There must be at least one thread that is waiting for a condition to become true. The waiting thread must first acquire a unique_lock. This lock is passed to the wait() method, that releases the mutex and suspends the thread until the condition variable is signaled. When that happens the thread is awaken and the lock is re-acquired."
http://www.codeproject.com/Articles/598695/Cplusplus-threads-locks-and-condition-variables
A deadlock requires TWO mutexes to be acquired in different orders in different threads. I only see a single mutex in your code.