Suppose there are two parallel threads, one executing the increment() and another executing the decrement(). Obviously, steps 1 and 2 in both methods are passable only if the corresponding locks are free, otherwise, the executing thread would have to wait for their release.
Let’s consider in detail how this happens.īoth increment() and decrement() basically consist of the following 5 steps: The methods are intentionally designed to produce deadlocks.
In the following sections, we will not distinguish between the synchronized statements and the lock API, using the term “lock” for reentrant locks of both types, preferring the former in the examples. It is assumed that the reader is already familiar with multithreaded programming and has a good understanding of thread synchronization primitives in Java. The concepts will be illustrated with a simple demo project. We will take a look at the origin of deadlocks, consider a method for discovering potential deadlocks in existing code, and present practical approaches to designing a deadlock-free synchronization. The only surefire way to prevent them is properly designing your code, which is the subject of this article. But, it also has a dark side deadlocks: nasty errors that are difficult to discover, reproduce, and fix.
Thread synchronization is a great tool to overcome race conditions in multithreaded programs.
0 kommentar(er)
