Re: [GIT PULL] notification tree - try 37!

[Matt Helsley <matthltc@us.ibm.com>]

On Sat, Aug 07, 2010 at 03:15:14PM -0400, Eric Paris wrote:
> On Fri, 2010-08-06 at 20:06 -0400, Christoph Hellwig wrote:
> > On Fri, Aug 06, 2010 at 04:34:31PM -0700, Matt Helsley wrote:
> > > On Fri, Aug 06, 2010 at 11:58:39AM -0400, Eric Paris wrote:
> > > > Here it is again!  Another notification pull request!  There is still
> > > > future work to be done on notification, but nothing that I believe
> > > > others would call blocking or functional.  The work I plan to do
> > > > includes:
> > > > 
> > > > 1) Al has discussed the addition of a file_clone() call in the VFS which
> > > > would eliminate my use of __dentry_open() and would allow the removal of
> > > > the rather hideous FMODE_/O_NONOTIFY.
> > > 
> > > I did a quick search and can't find a mailing list post on this. Was
> > > it a private discussion or is there something I can read about what
> > > file_clone() will do?
> 
> No, it was from a face to face meeting and a couple of irc conversations
> talk about all of this stuff.  My understanding was that it was going to
> be a lot like dentry_open() only it was going to require a valid struct
> file and would return a new struct file.  One of the purposes of the new
> interface being the ability to set f_mode at a better time to eliminate
> the FMODE/O_ overlapping horror that fanotify requires to prevent
> recursion and deadlock.

Thanks Eric, that's the information I was looking for. I was curious
because there's a chance file_clone() as you described it may also be useful
for checkpoint/restart.

Cheers,
	-Matt Helsley