On Tue, Dec 06, 2011 at 09:23:25AM +0800, boyd yang wrote:
> fanotify: to differ file access event from different threads
> When fanotify is monitoring the whole mount point "/", and multiple
> threads iterate the same direcotry, some thread will hang.
> This patch let fanotify to differ access events from different
> threads, prevent fanotify from merging access events from different
> threads.
I could reproduce the problem you described. Actually it may occur each time
when more than one thread of a process is waiting in fanotify_get_response()
at the same time. This is since we are not prepared to wake up more than one
waiter: We do
wait_event(group->fanotify_data.access_waitq, event->response ||
atomic_read(&group->fanotify_data.bypass_perm));
and after that
event->response = 0;
which is the reason that even if we woke up other waiters on the same waitqueue
they may see event->response being already 0, go back to sleep and then possibly
hang forever.
> It also hide overflow events to reach user space.
This is not part of this patch and also should not be, since overflow
events are supposted to reach user space.
> - if (event->mask & FAN_ALL_PERM_EVENTS) {
> - /* if we merged we need to wait on the new event */
> - if (notify_event)
> - event = notify_event;
> - ret = fanotify_get_response_from_access(group, event);
> + /*if overflow, do not wait for response*/
> + if (event->mask&FS_Q_OVERFLOW) {
> + pr_debug("fanotify overflow!\n");
> + } else {
> + if (event->mask & FAN_ALL_PERM_EVENTS) {
> + /* if we merged we need to wait on the new event */
> + if (notify_event)
> + event = notify_event;
> + ret = fanotify_get_response_from_access(group, event);
> + }
> }
What is this for? All you do is introduce a debug message for no real reason.
However your fix (avoid merging of events from the same process) seems to work.
Regards,
Lino
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
