diff --git a/kernel/bpf/cpumap.c b/kernel/bpf/cpumap.c
index e5888d401d79..08a8e8102728 100644
--- a/kernel/bpf/cpumap.c
+++ b/kernel/bpf/cpumap.c
@@ -26,6 +26,7 @@
 #include <linux/workqueue.h>
 #include <linux/kthread.h>
 #include <linux/capability.h>
+#include <linux/completion.h>
 #include <trace/events/xdp.h>
 #include <linux/btf_ids.h>
 
@@ -71,6 +72,7 @@ struct bpf_cpu_map_entry {
 	struct rcu_head rcu;
 
 	struct work_struct kthread_stop_wq;
+	struct completion kthread_running;
 };
 
 struct bpf_cpu_map {
@@ -164,7 +166,6 @@ static void put_cpu_map_entry(struct bpf_cpu_map_entry *rcpu)
 static void cpu_map_kthread_stop(struct work_struct *work)
 {
 	struct bpf_cpu_map_entry *rcpu;
-	int err;
 
 	rcpu = container_of(work, struct bpf_cpu_map_entry, kthread_stop_wq);
 
@@ -174,14 +175,7 @@ static void cpu_map_kthread_stop(struct work_struct *work)
 	rcu_barrier();
 
 	/* kthread_stop will wake_up_process and wait for it to complete */
-	err = kthread_stop(rcpu->kthread);
-	if (err) {
-		/* kthread_stop may be called before cpu_map_kthread_run
-		 * is executed, so we need to release the memory related
-		 * to rcpu.
-		 */
-		put_cpu_map_entry(rcpu);
-	}
+	kthread_stop(rcpu->kthread);
 }
 
 static void cpu_map_bpf_prog_run_skb(struct bpf_cpu_map_entry *rcpu,
@@ -309,11 +303,11 @@ static int cpu_map_bpf_prog_run(struct bpf_cpu_map_entry *rcpu, void **frames,
 	return nframes;
 }
 
-
 static int cpu_map_kthread_run(void *data)
 {
 	struct bpf_cpu_map_entry *rcpu = data;
 
+	complete(&rcpu->kthread_running);
 	set_current_state(TASK_INTERRUPTIBLE);
 
 	/* When kthread gives stop order, then rcpu have been disconnected
@@ -478,6 +472,7 @@ __cpu_map_entry_alloc(struct bpf_map *map, struct bpf_cpumap_val *value,
 		goto free_ptr_ring;
 
 	/* Setup kthread */
+	init_completion(&rcpu->kthread_running);
 	rcpu->kthread = kthread_create_on_node(cpu_map_kthread_run, rcpu, numa,
 					       "cpumap/%d/map:%d", cpu,
 					       map->id);
@@ -491,6 +486,12 @@ __cpu_map_entry_alloc(struct bpf_map *map, struct bpf_cpumap_val *value,
 	kthread_bind(rcpu->kthread, cpu);
 	wake_up_process(rcpu->kthread);
 
+	/* Make sure kthread has been running, so kthread_stop() will not
+	 * stop the kthread prematurely and all pending frames or skbs
+	 * will be handled by the kthread before kthread_stop() returns.
+	 */
+	wait_for_completion(&rcpu->kthread_running);
+
 	return rcpu;
 
 free_prog: