Livelock vs deadlock code#

Clearly the worst type of a hang is one that never comes back-in other words, a deadlock.Įngineering large concurrent applications using the standard Windows® shared-memory toolkit of threads and locks is significantly more involved than it might appear at first glance.

Livelock vs deadlock code#

Code that blocks a GUI thread's execution can prevent a window from processing incoming user input and system-generated messages, resulting in an app that is 'Not Responding.' Even non-GUI programs can suffer from responsiveness problems when using shared resources or when performing interthread, interprocess, or intermachine communication. They can result from slow I/O, a computationally expensive algorithm, or mutually exclusive access to resources, in any case reducing the overall responsiveness of an application. Hangs can range from temporary to permanent. Users frequently shut down a frozen application before such information can be captured, meaning there's no tell-tale stack trace to help you find the source of the problem.

Unlike a crash, an application hang may not produce a crash dump or trigger custom failure logic. They're terribly difficult to find before shipping, and even more difficult to debug after an application has been deployed. This article uses the following technologies:Ĭode download available at: Deadlocks.exe(188 KB)Īpplication hangs are one of the most frustrating situations a user can experience. Examining a custom CLR host for deadlock detection.Advanced Techniques To Avoid And Detect Deadlocks In.