On Fri, 16 Aug 2013 05:17:26 -0700 Christoph Hellwig <hch@xxxxxxxxxxxxx> wrote: > On Fri, Aug 16, 2013 at 12:41:14AM +0800, Feng Shuo wrote: > > much easier. Especially, if there > > Copy to Jeff and Arnd who were working on that lock. Also Miklos, yes, > > it's still me, the one was > > worked on fuse adaptive readdir_plus....now I'm working for GPFS.... > > People will start giving a shit once your filesystem in the Linux source > tree. People that violate our license are generally not overly liked. > I concur. I'm not sure you'll get much traction on this since no existing ->setlease caller in the tree needs it. The only current ->setlease caller that actually does anything is cifs_setlease, and it will only succeed if there's already an oplock on the file. Eventually we might want to add support for upgrading SMB2/3 leases there, and at that point we probably will need to do something that allows that function to sleep. I know that Steve Whitehouse had also mentioned implementing leases in GFS2 at some point, but no idea whether that will involve operations that need to sleep (I imagine it will since you'll probably need to talk to DLM). So, I have no specific objection to sane patches that allow ->setlease to be called without the lock held, but you do need a way to ensure that the returned lease will stick around until you can do the fasync stuff. I don't think we want to convert all of this to mutexes, so you'll probably need to get more clever with how the spinlocking is handled. -- Jeff Layton <jlayton@xxxxxxxxxx> -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
