Locking with a Monitor construct
This recipe illustrates another common multithreaded error called a deadlock. Since a deadlock will cause a program to stop working, the first piece in this example is a new Monitor
construct that allows us to avoid a deadlock. Then, the previously described lock
keyword is used to get a deadlock.
Getting ready
To work through this recipe, you will need Visual Studio 2012. There are no other prerequisites. The source code for this recipe can be found at BookSamples\Chapter1\Recipe10
.
How to do it...
To understand the multithreaded error deadlock, perform the following steps:
- Start Visual Studio 2012. Create a new C# Console Application project.
- In the
Program.cs
file, add the followingusing
directives:using System; using System.Threading;
- Add the following code snippet below the
Main
method:static void LockTooMuch(object lock1, object lock2) { lock (lock1) { Thread.Sleep(1000); lock (lock2); } }
- Add the following code snippet inside the
Main
method:object lock1 = new object(); object lock2 = new object(); new Thread(() => LockTooMuch(lock1, lock2)).Start(); lock (lock2) { Thread.Sleep(1000); Console.WriteLine("Monitor.TryEnter allows not to get stuck, returning false after a specified timeout is elapsed"); if (Monitor.TryEnter(lock1, TimeSpan.FromSeconds(5))) { Console.WriteLine("Acquired a protected resource succesfully"); } else { Console.WriteLine("Timeout acquiring a resource!"); } } new Thread(() => LockTooMuch(lock1, lock2)).Start(); Console.WriteLine("----------------------------------"); lock (lock2) { Console.WriteLine("This will be a deadlock!"); Thread.Sleep(1000); lock (lock1) { Console.WriteLine("Acquired a protected resource succesfully"); } }
- Run the program.
How it works...
Let's start with the LockTooMuch
method. In this method, we just lock
the first object, wait a second and then lock
the second object. Then we start this method in another thread and try to lock
the second object and then the first object from the main thread.
If we use the lock
keyword like in the second part of this demo, it would be a deadlock. The first thread holds a lock
on the lock1
object and waits while the lock2
object gets free; the main thread holds a lock
on the lock2
object and waits for the lock1
object to become free, which in this situation will never happen.
Actually, the lock
keyword is a syntactic sugar for Monitor
class usage. If we were to disassemble a code with lock
, we would see that it turns into the following code snippet:
bool acquiredLock = false; try { Monitor.Enter(lockObject, ref acquiredLock); // Code that accesses resources that are protected by the lock. } finally { if (acquiredLock) { Monitor.Exit(lockObject); } }
Therefore, we can use the Monitor
class directly; it has the TryEnter
method, which accepts a timeout parameter and returns false
if this timeout parameter expires before we can acquire the resource protected by lock
.