| |||||||||||||
Toward a safer fput()Toward a safer fput()Posted Apr 26, 2012 4:38 UTC (Thu) by josh (subscriber, #17465)Parent article: Toward a safer fput()
task_work_add also seems like a great way for the kernel to get work done on behalf of a given process, charging the work to that process and letting it get scheduled with the priority of that process.
(Log in to post comments)
Toward a safer fput() Posted Apr 26, 2012 12:54 UTC (Thu) by valyala (guest, #41196) [Link]
Actually, task_work_add() resembles kernel-mode APC (Asynchronous Procedure Call) [1].
[1] http://msdn.microsoft.com/en-us/library/windows/desktop/m... .
Toward a safer fput() Posted Apr 27, 2012 8:09 UTC (Fri) by pbonzini (subscriber, #60935) [Link]
It does, except that it doesn't wreak havoc on synchronization primitives (APCs can be queued on processes that are waiting on a semaphore or event; if that happens, and a concurrent wakeup happens, something else will be woken up with potential for starvation or even deadlocks). Search for PulseEvent on MSDN or Raymond Chen's blog.
|
|||||||||||||