percpu: fix locking regression in the failure path of pcpu_alloc()

While updating locking, b38d08f3 ("percpu: restructure locking")
broke pcpu_create_chunk() creation path in pcpu_alloc().  It returns
without releasing pcpu_alloc_mutex.  Fix it.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Julia Lawall <julia.lawall@lip6.fr>

mm/percpu.c

@@ -974,6 +974,7 @@ static void __percpu *pcpu_alloc(size_t size, size_t align, bool reserved,
 	if (list_empty(&pcpu_slot[pcpu_nr_slots - 1])) {
 		chunk = pcpu_create_chunk();
 		if (!chunk) {
+			mutex_unlock(&pcpu_alloc_mutex);
 			err = "failed to allocate new chunk";
 			goto fail;
 		}