The pthread_testcancel() function forces testing for cancellation. This is useful when you need to execute code that runs for long periods without encountering cancellation points; such as a library routine that executes long-running computations without cancellation points. This type of code can block cancellation for unacceptable long periods of time. One strategy for avoiding blocking cancellation for long periods, is to insert calls to pthread_testcancel() in the long-running computation code and to setup a cancellation handler in the library code, if required.

The pthread_testcancel() function returns a void.

The pthread_testcancel() function does not return errors.

See attributes(5) for descriptions of the following attributes:

Intro(3), cancellation(3THR), condition(3THR), pthread_cleanup_pop(3THR), pthread_cleanup_push(3THR), pthread_exit(3THR), pthread_join(3THR), pthread_setcancelstate(3THR), pthread_setcanceltype(3THR), setjmp(3C), attributes(5)

pthread_testcancel() has no effect if cancellation is disabled.

Use pthread_testcancel() with pthread_setcanceltype() called with its canceltype set to PTHREAD_CANCEL_DEFERRED. pthread_testcancel() operation is undefined if pthread_setcanceltype() was called with its canceltype argument set to PTHREAD_CANCEL_ASYNCHRONOUS.

It is possible to kill a thread when it is holding a resource, such as lock or allocated memory. If that thread has not setup a cancellation cleanup handler to release the held resource, the application is "cancel-unsafe". See attributes(5) for a discussion of Cancel-Safety, Deferred-Cancel-Safety, and Asynchronous-Cancel-Safety.