diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index ad0cd439223b..c56479b4884b 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -5495,8 +5495,6 @@ int __init workqueue_init_early(void)
 
 	pwq_cache = KMEM_CACHE(pool_workqueue, SLAB_PANIC);
 
-	wq_numa_init();
-
 	/* initialize CPU pools */
 	for_each_possible_cpu(cpu) {
 		struct worker_pool *pool;
@@ -5566,9 +5564,32 @@ int __init workqueue_init_early(void)
  */
 int __init workqueue_init(void)
 {
+	struct workqueue_struct *wq;
 	struct worker_pool *pool;
 	int cpu, bkt;
 
+	/*
+	 * It'd be simpler to initialize NUMA in workqueue_init_early() but
+	 * CPU to node mapping may not be available that early on some
+	 * archs such as power and arm64.  As per-cpu pools created
+	 * previously could be missing node hint and unbound pools NUMA
+	 * affinity, fix them up.
+	 */
+	wq_numa_init();
+
+	mutex_lock(&wq_pool_mutex);
+
+	for_each_possible_cpu(cpu) {
+		for_each_cpu_worker_pool(pool, cpu) {
+			pool->node = cpu_to_node(cpu);
+		}
+	}
+
+	list_for_each_entry(wq, &workqueues, list)
+		wq_update_unbound_numa(wq, smp_processor_id(), true);
+
+	mutex_unlock(&wq_pool_mutex);
+
 	/* create the initial workers */
 	for_each_online_cpu(cpu) {
 		for_each_cpu_worker_pool(pool, cpu) {