The kernel uses an upcall to request mountd to resolve pathnames into exports, file handles, or junctions. The current upcall mechanism is text-based and somewhat difficult to extend. It also has a limit of one page per request, so a response that includes a significant amount of data (for example, a large fs_locations list) is not possible.
One thought is to replace the current mechanism with a netlink protocol that is purpose-built for handling kernel pathname resolution.
The kernel uses an upcall to request mountd to resolve pathnames into exports, file handles, or junctions. The current upcall mechanism is text-based and somewhat difficult to extend. It also has a limit of one page per request, so a response that includes a significant amount of data (for example, a large fs_locations list) is not possible.
One thought is to replace the current mechanism with a netlink protocol that is purpose-built for handling kernel pathname resolution.