Cleanup after namespace associated resources
Posted Jul 25, 2012 11:41 UTC (Wed) by levonshe
Parent article: A new approach to user namespaces
The only problem with this approach is cleanup of kernel resources. In user space the cloned process just exits and associated resources are freed. But the kernel module may allocate resource valid for the specific namespace but it can not release this resource. The patch track namespaces using kref counting and free_user_ns() function is called at the and I do not see how to chain extra functions to clean kernel resources associated with the given user namespace.
This is real the issue I stumbled developing real kernel module which make use of the proposed kernel
to post comments)