ChatGPT解决这个技术问题 Extra ChatGPT

What's the difference between deadlock and livelock?

Can somebody please explain with examples (of code) what is the difference between deadlock and livelock?


C
Community

Taken from http://en.wikipedia.org/wiki/Deadlock:

In concurrent computing, a deadlock is a state in which each member of a group of actions, is waiting for some other member to release a lock A livelock is similar to a deadlock, except that the states of the processes involved in the livelock constantly change with regard to one another, none progressing. Livelock is a special case of resource starvation; the general definition only states that a specific process is not progressing. A real-world example of livelock occurs when two people meet in a narrow corridor, and each tries to be polite by moving aside to let the other pass, but they end up swaying from side to side without making any progress because they both repeatedly move the same way at the same time. Livelock is a risk with some algorithms that detect and recover from deadlock. If more than one process takes action, the deadlock detection algorithm can be repeatedly triggered. This can be avoided by ensuring that only one process (chosen randomly or by priority) takes action.


I found it already, but they don't have examples there as You could see, thanks anyway
I won't provide a code example, but consider two processes each waiting for a resource the other has but waiting in a non-blocking manner. When each learns they cannot continue they release their held resource and sleep for 30 seconds, then they retrieve their original resource followed by trying to the resource the other process held, then left, then reaquired. Since both processes are trying to cope (just badly), this is a livelock.
can You give me the same example but with deadlock, thanks in advance
A deadlock example is much easier... assume two processes A and B, and each wants resource r1 and resource r2. Assume A receives (or already has) r1, and B receives (or already has) r2. Now each try to get the resource the other has, without any timeout. A is blocked because B holds r2, and B is blocked because A holds r1. Each process is blocked and thus cannot release the resource the other wants, causing deadlock.
Within the context of Transactional memory there is a great video demonstrating deadlock and livelock: youtube.com/watch?v=_IxsOEEzf-c
C
Community

Livelock

A thread often acts in response to the action of another thread. If the other thread's action is also a response to the action of another thread, then livelock may result. As with deadlock, livelocked threads are unable to make further progress. However, the threads are not blocked — they are simply too busy responding to each other to resume work. This is comparable to two people attempting to pass each other in a corridor: Alphonse moves to his left to let Gaston pass, while Gaston moves to his right to let Alphonse pass. Seeing that they are still blocking each other, Alphonse moves to his right, while Gaston moves to his left. They're still blocking each other, and so on...

The main difference between livelock and deadlock is that threads are not going to be blocked, instead they will try to respond to each other continuously.

In this image, both circles (threads or processes) will try to give space to the other by moving left and right. But they can't move any further.

https://i.stack.imgur.com/OAda8.jpg


This thing has a name. A slang word perhaps, but still: schlumperdink :P
D
Daniel Frederico Lins Leite

All the content and examples here are from

Operating Systems: Internals and Design Principles William Stallings 8º Edition

Deadlock: A situation in which two or more processes are unable to proceed because each is waiting for one the others to do something.

For example, consider two processes, P1 and P2, and two resources, R1 and R2. Suppose that each process needs access to both resources to perform part of its function. Then it is possible to have the following situation: the OS assigns R1 to P2, and R2 to P1. Each process is waiting for one of the two resources. Neither will release the resource that it already owns until it has acquired the other resource and performed the function requiring both resources. The two processes are deadlocked

Livelock: A situation in which two or more processes continuously change their states in response to changes in the other process(es) without doing any useful work:

Starvation: A situation in which a runnable process is overlooked indefinitely by the scheduler; although it is able to proceed, it is never chosen.

Suppose that three processes (P1, P2, P3) each require periodic access to resource R. Consider the situation in which P1 is in possession of the resource, and both P2 and P3 are delayed, waiting for that resource. When P1 exits its critical section, either P2 or P3 should be allowed access to R. Assume that the OS grants access to P3 and that P1 again requires access before P3 completes its critical section. If the OS grants access to P1 after P3 has finished, and subsequently alternately grants access to P1 and P3, then P2 may indefinitely be denied access to the resource, even though there is no deadlock situation.

APPENDIX A - TOPICS IN CONCURRENCY

Deadlock Example

If both processes set their flags to true before either has executed the while statement, then each will think that the other has entered its critical section, causing deadlock.

/* PROCESS 0 */
flag[0] = true;            // <- get lock 0
while (flag[1])            // <- is lock 1 free?
    /* do nothing */;      // <- no? so I wait 1 second, for example
                           // and test again.
                           // on more sophisticated setups we can ask
                           // to be woken when lock 1 is freed
/* critical section*/;     // <- do what we need (this will never happen)
flag[0] = false;           // <- releasing our lock

 /* PROCESS 1 */
flag[1] = true;
while (flag[0])
    /* do nothing */;
/* critical section*/;
flag[1] = false;

Livelock Example

/* PROCESS 0 */
flag[0] = true;          // <- get lock 0
while (flag[1]){         
    flag[0] = false;     // <- instead of sleeping, we do useless work
                         //    needed by the lock mechanism
    /*delay */;          // <- wait for a second
    flag[0] = true;      // <- and restart useless work again.
}
/*critical section*/;    // <- do what we need (this will never happen)
flag[0] = false; 

/* PROCESS 1 */
flag[1] = true;
while (flag[0]) {
    flag[1] = false;
    /*delay */;
    flag[1] = true;
}
/* critical section*/;
flag[1] = false;

[...] consider the following sequence of events:

P0 sets flag[0] to true.

P1 sets flag[1] to true.

P0 checks flag[1].

P1 checks flag[0].

P0 sets flag[0] to false.

P1 sets flag[1] to false.

P0 sets flag[0] to true.

P1 sets flag[1] to true.

This sequence could be extended indefinitely, and neither process could enter its critical section. Strictly speaking, this is not deadlock, because any alteration in the relative speed of the two processes will break this cycle and allow one to enter the critical section. This condition is referred to as livelock. Recall that deadlock occurs when a set of processes wishes to enter their critical sections but no process can succeed. With livelock, there are possible sequences of executions that succeed, but it is also possible to describe one or more execution sequences in which no process ever enters its critical section.

Not content from the book anymore.

And what about spinlocks?

Spinlock is a technique to avoid the cost of the OS lock mechanism. Typically you would do:

try
{
   lock = beginLock();
   doSomething();
}
finally
{
   endLock();
}

A problem start to appear when beginLock() costs much more than doSomething(). In very exagerated terms, imagine what happens when the beginLock costs 1 second, but doSomething cost just 1 millisecond.

In this case if you waited 1 millisecond, you would avoid being hindered for 1 second.

Why the beginLock would cost so much? If the lock is free is does not cost a lot (see https://stackoverflow.com/a/49712993/5397116), but if the lock is not free the OS will "freeze" your thread, setup a mechanism to wake you when the lock is freed, and then wake you again in the future.

All of this is much more expensive than some loops checking the lock. That is why sometimes is better to do a "spinlock".

For example:

void beginSpinLock(lock)
{
   if(lock) loopFor(1 milliseconds);
   else 
   {
     lock = true;
     return;
   }

   if(lock) loopFor(2 milliseconds);
   else 
   {
     lock = true;
     return;
   }

   // important is that the part above never 
   // cause the thread to sleep.
   // It is "burning" the time slice of this thread.
   // Hopefully for good.

   // some implementations fallback to OS lock mechanism
   // after a few tries
   if(lock) return beginLock(lock);
   else 
   {
     lock = true;
     return;
   }
}

If your implementation is not careful, you can fall on livelock, spending all CPU on the lock mechanism.

Also see:

https://preshing.com/20120226/roll-your-own-lightweight-mutex/
Is my spin lock implementation correct and optimal?

Summary:

Deadlock: situation where nobody progress, doing nothing (sleeping, waiting etc..). CPU usage will be low;

Livelock: situation where nobody progress, but CPU is spent on the lock mechanism and not on your calculation;

Starvation: situation where one procress never gets the chance to run; by pure bad luck or by some of its property (low priority, for example);

Spinlock: technique of avoiding the cost waiting the lock to be freed.


Sir, the example you gave for Deadlock is actually an example of Spinlock. Deadlock occurs when a set of processes are blocked which are not in ready or running state and waiting for some resources. But in our example each one is performing some task i.e., checking the condition again and again. Correct me if I am wrong.
The example is so minimal that do open chance for this interpretation, so I improved it tring to be a little more explicit about their difference. Hope that helps.
Thank you for adding about spinlocks, according to you spinlocks is a technique and u justified it as well and I understood. But what about that priority inversion problem when one process P1 is in Critical Section and other high priority process P2 gets scheduled preempting P1 then in this case CPU is with P2 and our Synchronisation mechanism is with P1. This is called Spinlock as P1 is in ready state and P2 is in run state. Here spinlock is a problem. Am I getting the things right? I'm not able to get the intricacies right. Please help
My suggestion to you is to create another question stating your problem more clearly. Now, if you are in "user space", and P1 is inside a critical session protected with a SpinLock implemented with a infinite loop and its preempted; then P2 will try to enter it, wil fail and will burn all of its time-slice. You created a livelock (one CPU will be at 100%). (a bad use would be to protect a sync IO with spinlock. You can easily try this example) On the "kernel space" maybe this note can this help you: lxr.linux.no/linux+v3.6.6/Documentation/…
Thank you very much for the clarification. Anyway, your answer was quite descriptive and helpful unlike others
D
Deepak Lamichhane

DEADLOCK Deadlock is a condition in which a task waits indefinitely for conditions that can never be satisfied - task claims exclusive control over shared resources - task holds resources while waiting for other resources to be released - tasks cannot be forced to relinguish resources - a circular waiting condition exists

LIVELOCK Livelock conditions can arise when two or more tasks depend on and use the some resource causing a circular dependency condition where those tasks continue running forever, thus blocking all lower priority level tasks from running (these lower priority tasks experience a condition called starvation)


If the 'livelocked' tasks are following resource arbitration protocols which include 'backoff' delays, and spend most of their time in sleep state as result, then other tasks will not be starved.
m
mmirwaldt

Maybe these two examples illustrate you the difference between a deadlock and a livelock:

Java-Example for a deadlock:

import java.util.concurrent.locks.Lock;
import java.util.concurrent.locks.ReentrantLock;

public class DeadlockSample {

    private static final Lock lock1 = new ReentrantLock(true);
    private static final Lock lock2 = new ReentrantLock(true);

    public static void main(String[] args) {
        Thread threadA = new Thread(DeadlockSample::doA,"Thread A");
        Thread threadB = new Thread(DeadlockSample::doB,"Thread B");
        threadA.start();
        threadB.start();
    }

    public static void doA() {
        System.out.println(Thread.currentThread().getName() + " : waits for lock 1");
        lock1.lock();
        System.out.println(Thread.currentThread().getName() + " : holds lock 1");

        try {
            System.out.println(Thread.currentThread().getName() + " : waits for lock 2");
            lock2.lock();
            System.out.println(Thread.currentThread().getName() + " : holds lock 2");

            try {
                System.out.println(Thread.currentThread().getName() + " : critical section of doA()");
            } finally {
                lock2.unlock();
                System.out.println(Thread.currentThread().getName() + " : does not hold lock 2 any longer");
            }
        } finally {
            lock1.unlock();
            System.out.println(Thread.currentThread().getName() + " : does not hold lock 1 any longer");
        }
    }

    public static void doB() {
        System.out.println(Thread.currentThread().getName() + " : waits for lock 2");
        lock2.lock();
        System.out.println(Thread.currentThread().getName() + " : holds lock 2");

        try {
            System.out.println(Thread.currentThread().getName() + " : waits for lock 1");
            lock1.lock();
            System.out.println(Thread.currentThread().getName() + " : holds lock 1");

            try {
                System.out.println(Thread.currentThread().getName() + " : critical section of doB()");
            } finally {
                lock1.unlock();
                System.out.println(Thread.currentThread().getName() + " : does not hold lock 1 any longer");
            }
        } finally {
            lock2.unlock();
            System.out.println(Thread.currentThread().getName() + " : does not hold lock 2 any longer");
        }
    }
}

Sample output:

Thread A : waits for lock 1
Thread B : waits for lock 2
Thread A : holds lock 1
Thread B : holds lock 2
Thread B : waits for lock 1
Thread A : waits for lock 2

Java-Example for a livelock:


import java.util.concurrent.locks.Lock;
import java.util.concurrent.locks.ReentrantLock;

public class LivelockSample {

    private static final Lock lock1 = new ReentrantLock(true);
    private static final Lock lock2 = new ReentrantLock(true);

    public static void main(String[] args) {
        Thread threadA = new Thread(LivelockSample::doA, "Thread A");
        Thread threadB = new Thread(LivelockSample::doB, "Thread B");
        threadA.start();
        threadB.start();
    }

    public static void doA() {
        try {
            while (!lock1.tryLock()) {
                System.out.println(Thread.currentThread().getName() + " : waits for lock 1");
                Thread.sleep(100);
            }
            System.out.println(Thread.currentThread().getName() + " : holds lock 1");

            try {
                while (!lock2.tryLock()) {
                    System.out.println(Thread.currentThread().getName() + " : waits for lock 2");
                    Thread.sleep(100);
                }
                System.out.println(Thread.currentThread().getName() + " : holds lock 2");

                try {
                    System.out.println(Thread.currentThread().getName() + " : critical section of doA()");
                } finally {
                    lock2.unlock();
                    System.out.println(Thread.currentThread().getName() + " : does not hold lock 2 any longer");
                }
            } finally {
                lock1.unlock();
                System.out.println(Thread.currentThread().getName() + " : does not hold lock 1 any longer");
            }
        } catch (InterruptedException e) {
            // can be ignored here for this sample
        }
    }

    public static void doB() {
        try {
            while (!lock2.tryLock()) {
                System.out.println(Thread.currentThread().getName() + " : waits for lock 2");
                Thread.sleep(100);
            }
            System.out.println(Thread.currentThread().getName() + " : holds lock 2");

            try {
                while (!lock1.tryLock()) {
                    System.out.println(Thread.currentThread().getName() + " : waits for lock 1");
                    Thread.sleep(100);
                }
                System.out.println(Thread.currentThread().getName() + " : holds lock 1");

                try {
                    System.out.println(Thread.currentThread().getName() + " : critical section of doB()");
                } finally {
                    lock1.unlock();
                    System.out.println(Thread.currentThread().getName() + " : does not hold lock 1 any longer");
                }
            } finally {
                lock2.unlock();
                System.out.println(Thread.currentThread().getName() + " : does not hold lock 2 any longer");
            }
        } catch (InterruptedException e) {
            // can be ignored here for this sample
        }
    }
}

Sample output:

Thread B : holds lock 2
Thread A : holds lock 1
Thread A : waits for lock 2
Thread B : waits for lock 1
Thread B : waits for lock 1
Thread A : waits for lock 2
Thread A : waits for lock 2
Thread B : waits for lock 1
Thread B : waits for lock 1
Thread A : waits for lock 2
Thread A : waits for lock 2
Thread B : waits for lock 1
...

Both examples force the threads to aquire the locks in different orders. While the deadlock waits for the other lock, the livelock does not really wait - it desperately tries to acquire the lock without the chance of getting it. Every try consumes CPU cycles.


The code is nice. But the live-lock example is not good. Whether a thread is blocked on a value or is polling for a change in value is not conceptually different. An easy change to better illustrate a live-lock is to have threads A and B release the locks they have when they realize they can't get the second lock they need. Then they sleep for a second each, reacquire the lock they originally had, then sleep for another second and try to acquire the other lock again. So the cycle for each would be: 1) acquire-mine, 2) sleep, 3) try to acquire other & fail, 4) release-mine, 5) sleep, 6) Repeat.
I doubt whether the live-locks you think of really exist long enough that they cause trouble. When you always give up all locks you hold when you can't allocate the next lock, the deadlock (and live-lock) condition "hold and wait" is missing because there is actually no wait anymore. ( en.wikipedia.org/wiki/Deadlock )
indeed the dead lock condition missing because these are live locks we are discussing. The example I gave is similar to the standard hallway example given: geeksforgeeks.org/deadlock-starvation-and-livelock, en.wikibooks.org/wiki/Operating_System_Design/Concurrency/…, docs.oracle.com/javase/tutorial/essential/concurrency/…
The code example in the first link misses unlock-statements which makes it very confusing for me somehow. It's not clear where the critical sections start and where they end. My point is the order in which the statements are executed changes all the time with every try and it is actually never the same at the next turn. And not every execution order causes a livelock at the end. Most of them even don't! So you won't observe those livelocks because they just vanish very fast with the next harmless order of execution which is very probable.There is no perfect livelock example you can observe ;-)
It depends on the triggers for the actions & how long they take. It can definitely be an effective lock. If it takes 10 seconds of computation to jump to a state or back and two threads are reacting to each other with a phase difference of 5 seconds then the chance the computational speed varies enough between two threads in the same process enough to push it off by 5 seconds is very low. Try it out for your self. Create two programs that run for 10 seconds and start them 5 seconds apart and see how long it takes for them to get into phase within a certain margin (say 1 second).
s
stdout

Imagine you've thread A and thread B. They are both synchronised on the same object and inside this block there's a global variable they are both updating;

static boolean commonVar = false;
Object lock = new Object;

...

void threadAMethod(){
    ...
    while(commonVar == false){
         synchornized(lock){
              ...
              commonVar = true
         }
    }
}

void threadBMethod(){
    ...
    while(commonVar == true){
         synchornized(lock){
              ...
              commonVar = false
         }
    }
}

So, when thread A enters in the while loop and holds the lock, it does what it has to do and set the commonVar to true. Then thread B comes in, enters in the while loop and since commonVar is true now, it is be able to hold the lock. It does so, executes the synchronised block, and sets commonVar back to false. Now, thread A again gets it's new CPU window, it was about to quit the while loop but thread B has just set it back to false, so the cycle repeats over again. Threads do something (so they're not blocked in the traditional sense) but for pretty much nothing.

It maybe also nice to mention that livelock does not necessarily have to appear here. I'm assuming that the scheduler favours the other thread once the synchronised block finish executing. Most of the time, I think it's a hard-to-hit expectation and depends on many things happening under the hood.


Nice example. It also illustrates why you should always read and write atomically in a concurrent context. If the while loops were inside the synchronize blocks then the above would not be an issue.
R
RCvaram

I just planned to share some knowledge.

Deadlocks A set of threads/processes is deadlocked, if each thread/process in the set is waiting for an event that only another process in the set can cause.

The important thing here is another process is also in the same set. that means another process also blocked and no one can proceed.

Deadlocks occur when processes are granted exclusive access to resources.

These four conditions should be satisfied to have a deadlock.

Mutual exclusion condition (Each resource is assigned to 1 process) Hold and wait condition (Process holding resources and at the same time it can ask other resources). No preemption condition (Previously granted resources can not forcibly be taken away) #This condition depends on the application Circular wait condition (Must be a circular chain of 2 or more processes and each is waiting for resource held by the next member of the chain) # It will happen dynamically

If we found these conditions then we can say there may be occurred a situation like a deadlock.

LiveLock

Each thread/process is repeating the same state again and again but doesn't progress further. Something similar to a deadlock since the process can not enter the critical section. However in a deadlock, processes are wait without doing anything but in livelock, the processes are trying to proceed but processes are repeated to the same state again and again.

(In a deadlocked computation there is no possible execution sequence which succeeds. but In a livelocked computation, there are successful computations, but there are one or more execution sequences in which no process enters its critical section.)

Difference from deadlock and livelock

When deadlock happens, No execution will happen. but in livelock, some executions will happen but those executions are not enough to enter the critical section.