Deadlock Avoiding Based on Future Lockset

Existing dynamic methods for deadlock avoidance have four main drawbacks: limited capability, passive or blind avoiding algorithm, large performance overhead and no guarantee of correctness of target programs. In order to solve these problems, a combined static and dynamic avoiding method based on future lockset is proposed and named as Flider. The key idea is that, for a lock operation, if none of its future locks are occupied, then it makes sure that executing this lock operation won't lead the current thread to trap into a deadlock state. The future lockset for a lock operation is a set of locks that will be requested by the current thread before the corresponding unlock operation is reached. Firstly, Flider statically computes lock effects for lock operations and function call operations, and inserts them before and after the corresponding operations. Secondly, Flider dynamically intercepts lock operations and computes its future lockset using lock effects inserted by static analysis. Flider permits a lock operation to be executed if and only if all locks of its lockset are not held by any other threads. Otherwise, the lock operation waits until the condition is satisfied. Evaluation and comparison experiments verify that this method can efficiently avoid multi-type deadlocks in an active, intelligent, scalable and correctness-guaranteed way.