Fix potential race over inotify mark list access.

Accessing mark entries chain should be done with protection.
At least we need @mark_lock to be taken when we're dumping
things.

This patch should eliminate possible "race" candidate for

https://jira.sw.ru/browse/PSBM-31567

QA: Please run LTP inotify tests together with suspend/resume.

Signed-off-by: Cyrill Gorcunov <gorcunov@parallels.com>
CC: Andrew Vagin <avagin@parallels.com>
CC: Kirill Tkhai <ktkhai@parallels.com>
CC: Kir Kolyshkin <kir@parallels.com>
CC: Pavel Emelyanov <xemul@parallels.com>
---
 kernel/cpt/cpt_inotify.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

Index: linux-rhel6.git/kernel/cpt/cpt_inotify.c
===================================================================
--- linux-rhel6.git.orig/kernel/cpt/cpt_inotify.c
+++ linux-rhel6.git/kernel/cpt/cpt_inotify.c
@@ -77,7 +77,7 @@ static int cpt_dump_watches(struct fsnot
 	struct cpt_inotify_wd_image wi;
 	loff_t saved_obj;
 
-	/* FIXME locking */
+	spin_lock(&g->mark_lock);
 	list_for_each_entry(fse, &g->mark_entries, g_list) {
 		struct nameidata nd;
 
@@ -127,6 +127,7 @@ static int cpt_dump_watches(struct fsnot
 		if (err)
 			break;
 	}
+	spin_unlock(&g->mark_lock);
 
 	return err;
 }