On 05/18/2012 10:18 PM, Christoph Hellwig wrote: > On Fri, May 18, 2012 at 10:16:32PM +0800, Liu Yuan wrote: >> Compared to actual disk IO, the locking overhead can be neglected. I >> think use kernel's strcut flock is simplest method, otherwise we would >> keep track of a huge memory for lock structures. > > That's not nessecarily true with fast enough storage devices. I'm fine > with using lockf for now to get things going, but doing something > smarted is something I'll put in my todo list. > Okay, if there is a relative easy and efficient user land locking mechanism out, I'd be really happy to take it, I also think lockf is just working but not optimal. Thanks, Yuan |