exit: avoid sig->count in __exit_signal() to detect the group-dead case

Change __exit_signal() to check thread_group_leader() instead of
atomic_dec_and_test(&sig->count).  This must be equivalent, the group
leader must be released only after all other threads have exited and
passed __exit_signal().

Henceforth sig->count is not actually used, except in fs/proc for
get_nr_threads/etc.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Roland McGrath <roland@redhat.com>
Cc: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kernel/exit.c

@@ -88,11 +88,12 @@ static void __exit_signal(struct task_struct *tsk)
 					rcu_read_lock_held() ||
 					lockdep_tasklist_lock_is_held());
 	spin_lock(&sighand->siglock);
+	atomic_dec(&sig->count);
 
 	posix_cpu_timers_exit(tsk);
-	if (atomic_dec_and_test(&sig->count))
+	if (thread_group_leader(tsk)) {
 		posix_cpu_timers_exit_group(tsk);
-	else {
+	} else {
 		/*
 		 * If there is any task waiting for the group exit
 		 * then notify it: