Re: [PATCH 1/4] fanotify: flush outstanding perm requests on group destroy

[Tvrtko Ursulin]

On Tuesday 24 Aug 2010 10:36:29 Andreas Gruenbacher wrote:
> On Tuesday 24 August 2010 10:49:45 Tvrtko Ursulin wrote:
> > I think just switching to interruptible sleep in
> > fanotify_get_response_from_access should be fine. And it should probably
> > deny the current event when signal is received.
>
> Well the result would be -EINTR from the system call that blocked on the
> perm event, the same as with an interruptible nfs mount.  The process would
> never get -EPERM.  Processes may
> not be prepared to handle -EINTR in all cases, and so it may make more
> sense to use the same behavior as NFS and only allow SIGKILL to kill a
> process blocked on a perm event (which the blocked process will never
> see).

That would be wait_event_killable then, even simpler change.

I agree that hiding -EINTR from open, from userspace code is probably a more
compatible way of doing it. If I remember correctly POSIX does allow EINTR
from open but form our experience there are indeed applications which do not
handle it. Some version of X immediately spring to mind who used to set up a
periodic signal delivery (something like internal jiffy, I think they called
it smart scheduler) to themselves and would not handle -EINTR from open. Samba
also had a problem here in specific circumstances.

Tvrtko


Sophos Plc, The Pentagon, Abingdon Science Park, Abingdon, OX14 3YP, United Kingdom.
Company Reg No 2096520. VAT Reg No GB 348 3873 20.
--
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/