Flock() based locking. This is fairly efficient, but it will not work on NFS and is known to be unreliable on some operating systems including some flavors of the 2.4 Linux kernel.
Release the given lock(s) (queue the specified locks for later release on transactional databases or release them immediately if transactions are not supported)
Reference counts for every lock we're holding so that if we've got a file doubly read locked we don't try to delete it until all read locks are released.
Reference counts for every lock we're holding so that if we've got a file doubly read locked we don't try to delete it until all read locks are released.
API Tags:
Access:
private
Information Tags:
Todo:
When we get rid of double read locks, we can delete this.
