If I understand the existing code correctly, a server process (I don't know which one -- fileserver?) keeps track of clients that have exclusive or non-exclusive locks on a file. Clients are responsible for keeping track of which processes have such locks.
So it looks as though it should be easy to add local byte-range locking for processes on a single machine. Read locks would be propagated to the server as shared locks, and write locks would propagate as exclusive locks, and the byte ranges would be handled in the local client kernel code.
Does someone already have patches for this perhaps? -- Joe Buehler
_______________________________________________ OpenAFS-devel mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-devel
