Re: supporting functions missing from inotify patch

[Robert Love]

On Wed, 2005-07-13 at 15:21 -0500, Steve French wrote:

> I did not think that inotify_add_watch called dir_notify.  I don't see a path in which 
> calls to add a new inotify watch end up in a call to fcntl_dirnotify or file->dir_notify 
> This is for the case in which an app only calls inotify ioctl - ie does not [also] do a call 
> to dnotify. 

No, you are right, they do not, right now.

Is dir_notify suitable for inotify and your uses?  In the 10 months of
inotify development, I had hoped that a remote filesystem developer
would add support so we could test it.  But there is no rush to get this
hook added, so its okay.

The problem with dir_notify is that the args parameter is dnotify flags.
Those don't map directly to inotify flags.

What I'd like is

	(a) a patch adding the requisite inotify hook (really, 4 lines)
	(b) a filesystem successfully using the hook

> Without such a call - an app that does your new ioctl to add a watch on a file or directory will
> not cause the network/cluster fs to turn on notification on the server since the watch
> will be not seen by the client filesystem.

It is a system call, now.  ;-)

> OK - you exported a common underlying function
> 	inotify_inode_queue_event
> under the inline functions which the network/cluster fs would call to notify of remote changes.
> That makes sense. I had missed that.

Nod.

	Robert Love


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/