cgroups: fix race between userspace and kernelspace

Notify userspace about cgroup removing only after rmdir of cgroup
directory to avoid race between userspace and kernelspace.

eventfd are used to notify about two types of event:
 - control file-specific, like crossing memory threshold;
 - cgroup removing.

To understand what really happen, userspace can check if the cgroup still
exists.  To avoid race beetween userspace and kernelspace we have to
notify userspace about cgroup removing only after rmdir of cgroup
directory.

Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Cc: Dan Malek <dan@embeddedalley.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kernel/cgroup.c

@@ -795,28 +795,15 @@ static struct inode *cgroup_new_inode(mode_t mode, struct super_block *sb)
 static int cgroup_call_pre_destroy(struct cgroup *cgrp)
 {
 	struct cgroup_subsys *ss;
-	struct cgroup_event *event, *tmp;
 	int ret = 0;
 
 	for_each_subsys(cgrp->root, ss)
 		if (ss->pre_destroy) {
 			ret = ss->pre_destroy(ss, cgrp);
 			if (ret)
-				goto out;
+				break;
 		}
 
-	/*
-	 * Unregister events and notify userspace.
-	 */
-	spin_lock(&cgrp->event_list_lock);
-	list_for_each_entry_safe(event, tmp, &cgrp->event_list, list) {
-		list_del(&event->list);
-		eventfd_signal(event->eventfd, 1);
-		schedule_work(&event->remove);
-	}
-	spin_unlock(&cgrp->event_list_lock);
-
-out:
 	return ret;
 }
 
@@ -3006,7 +2993,6 @@ static void cgroup_event_remove(struct work_struct *work)
 	event->cft->unregister_event(cgrp, event->cft, event->eventfd);
 
 	eventfd_ctx_put(event->eventfd);
-	remove_wait_queue(event->wqh, &event->wait);
 	kfree(event);
 }
 
@@ -3024,6 +3010,7 @@ static int cgroup_event_wake(wait_queue_t *wait, unsigned mode,
 	unsigned long flags = (unsigned long)key;
 
 	if (flags & POLLHUP) {
+		remove_wait_queue_locked(event->wqh, &event->wait);
 		spin_lock(&cgrp->event_list_lock);
 		list_del(&event->list);
 		spin_unlock(&cgrp->event_list_lock);
@@ -3472,6 +3459,7 @@ static int cgroup_rmdir(struct inode *unused_dir, struct dentry *dentry)
 	struct dentry *d;
 	struct cgroup *parent;
 	DEFINE_WAIT(wait);
+	struct cgroup_event *event, *tmp;
 	int ret;
 
 	/* the vfs holds both inode->i_mutex already */
@@ -3555,6 +3543,20 @@ again:
 	set_bit(CGRP_RELEASABLE, &parent->flags);
 	check_for_release(parent);
 
+	/*
+	 * Unregister events and notify userspace.
+	 * Notify userspace about cgroup removing only after rmdir of cgroup
+	 * directory to avoid race between userspace and kernelspace
+	 */
+	spin_lock(&cgrp->event_list_lock);
+	list_for_each_entry_safe(event, tmp, &cgrp->event_list, list) {
+		list_del(&event->list);
+		remove_wait_queue(event->wqh, &event->wait);
+		eventfd_signal(event->eventfd, 1);
+		schedule_work(&event->remove);
+	}
+	spin_unlock(&cgrp->event_list_lock);
+
 	mutex_unlock(&cgroup_mutex);
 	return 0;
 }