Re: [PATCH 1/1 linux-next] fanotify: fix a lock in fanotify_should_send_event()

[Jan Kara <jack@suse.cz>]

On Sun 22-03-15 11:48:20, Lino Sanfilippo wrote:
> On 22.03.2015 10:46, Fabian Frederick wrote:
> 
> > Let's hope it only breaks ltp tests and no _real_ userland stuff
> > (search systems ...)
> > 
> > Regards,
> > Fabian
> > 
> 
> Hi Fabian,
> 
> yes, that worries me too. I know that there have been discussions on
> lkml in which it was made clear that userspace breakage is frowned upon.
> And it is obvious, that the latest changes concerning the handling of
> FAN_ONDIR are also visible to userspace. But since the concerning
> patches have been accepted I think it is ok. I could be wrong though.
> 
> Maybe someone with a deeper knowledge of kernel policy/guidelines could
> comment on this?
  For now I'd stay calm. Events on directories with fanotify are hopelessly
useless anyway so I doubt anyone uses them. I just tested those in fanotify
LTP testcase for completeness (and screwed that up as you noted). If we
find a real application that tries to use directory events with fanotify
and which is broken by this patch, we'll have to revert it, that's for
sure.

								Honza
-- 
Jan Kara <jack@suse.cz>
SUSE Labs, CR