On Sun, 2003-06-01 at 22:00, Paul E. McKenney wrote:
> The immediate motivation is to avoid the race with zap_page_range()
> when another node writes to the corresponding portion of the file,
> similar to the situation with vmtruncate().  The thought was to
> leverage locking within the distributed filesystem, but if the
> race is solved locally, then, as you say, perhaps this is not 
> necessary.

is said distributed filesystem open source by chance ?