Re: Alternative "dead port listener" implementations under intel?
site_archiver@lists.apple.com Delivered-To: darwin-dev@lists.apple.com On Feb 15, 2006, at 2:39 PM, Jeffrey Johnson wrote: Hello, -tim _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.appl... Now that the use of task_for_pid() on intel boxes has been tightened to use by root or processes belonging to procmod: <http://developer.apple.com/documentation/MacOSX/Conceptual/ universal_binary/index.html#//apple_ref/doc/uid/TP40002217> those of us relying on the helpful snippet in TN2050 have been chopped off at the knees. Does anyone have a workaround for getting notified when a process terminates without resorting to polling? (I only care about processes owned by my UID.) In addition to the kqueue suggestion (which is better for this situation, probably), there are bootstrap namespaces. OmniObjectMeter, for example, needs to be able to scrounge though the target process's memory. We do this by setting up a private bootstrap namespace and then having the target process register its task port there-in under a well-known name. This email sent to site_archiver@lists.apple.com
participants (1)
-
Timothy J. Wood