Re: A few concerns about fanotify implementation.

[Eric Paris <eparis-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>]

On Mon, 2011-06-06 at 18:42 +0400, Vasily Novikov wrote:
> On 06/06/2011 05:43 PM, Eric Paris wrote:
> > Well, you are both correct.  If you add a mark with only 'ignored'
> > events set it will not pin the inode into the kernel.  If the system
> > starts to get under memory pressure the kernel will kick unused inodes
> > and any associated ignored marks out of ram.  Inodes with 'real' events
> > attached will be pinned in memory and cannot be evicted under memory
> > pressure.
> 
> So if we use marks with only 'ignored' events then under memory pressure 
> mm subsystem will shrink inode cache that will free our marks and 
> therefore it's safe to use FAN_UNLIMITED_MARKS in this case?
> If it really works then we don't need LRU cache in fanotify because it's 
> already implemented in dentry_cache/inode_cache.

That's how it's supposed to work.  Just remember, if you set a real
event, the inode becomes pinned in core and the mm will be unable to
evict either the inode or the mark.

-Eric