diff --git a/arch/arm64/kernel/setup.c b/arch/arm64/kernel/setup.c
index 7b1a67582e172f50a7a238cfeb0997dab94f7d28..a4f9791e03b298602e18367b546a1e90b219ad92 100644
--- a/arch/arm64/kernel/setup.c
+++ b/arch/arm64/kernel/setup.c
@@ -312,6 +312,11 @@ void __init setup_arch(char **cmdline_p)
 	early_ioremap_init();
 
 	setup_machine_fdt(__fdt_pointer);
+	/*
+	 * Initialise the static keys early as they may be enabled by the
+	 * early param code.
+	 */
+	jump_label_init();
 
 	parse_early_param();
 
diff --git a/arch/arm64/kernel/smp.c b/arch/arm64/kernel/smp.c
index 52787a83de5588f066eee7c6a5cf9e0218aa209a..6d673db8124e6f1424d22a3b5db435fe716ad668 100644
--- a/arch/arm64/kernel/smp.c
+++ b/arch/arm64/kernel/smp.c
@@ -433,11 +433,6 @@ void __init smp_cpus_done(unsigned int max_cpus)
 void __init smp_prepare_boot_cpu(void)
 {
 	set_my_cpu_offset(per_cpu_offset(smp_processor_id()));
-	/*
-	 * Initialise the static keys early as they may be enabled by the
-	 * cpufeature code.
-	 */
-	jump_label_init();
 	cpuinfo_store_boot_cpu();
 
 	/*
diff --git a/kernel/locking/qspinlock.c b/kernel/locking/qspinlock.c
index 6138ec2b6b2d0ea5158646244573a3c2fcfb46b3..fdd3a64ec0828587e96109cb5bbd685ee614d29c 100644
--- a/kernel/locking/qspinlock.c
+++ b/kernel/locking/qspinlock.c
@@ -584,6 +584,21 @@ EXPORT_SYMBOL(queued_spin_lock_slowpath);
  * Generate the code for NUMA-aware spin locks
  */
 #if !defined(_GEN_CNA_LOCK_SLOWPATH) && defined(CONFIG_NUMA_AWARE_SPINLOCKS)
+#include <linux/jump_label.h>
+
+DEFINE_STATIC_KEY_TRUE(cna_lock_disabled);
+
+static int __init cna_locks_init_jump(char *str)
+{
+	if (!IS_ENABLED(CONFIG_NUMA))
+		return 0;
+
+	static_branch_dec(&cna_lock_disabled);
+	pr_info("NUMA aware qspinlock is enabled\n");
+	return 0;
+}
+early_param("using_numa_aware_qspinlock", cna_locks_init_jump);
+
 #define _GEN_CNA_LOCK_SLOWPATH
 
 #undef pv_init_node
diff --git a/kernel/locking/qspinlock_cna.h b/kernel/locking/qspinlock_cna.h
index 9e6bd9e6d82b1ddf31a8c274cd43d14b4a7f7e8d..97f420be0624aa6b30887a6896dabc2a60cdd6b7 100644
--- a/kernel/locking/qspinlock_cna.h
+++ b/kernel/locking/qspinlock_cna.h
@@ -79,6 +79,9 @@ static void cna_init_node(struct mcs_spinlock *node)
 	struct mcs_spinlock *base_node;
 	int cpuid;
 
+	if (static_branch_likely(&cna_lock_disabled))
+		return;
+
 	BUILD_BUG_ON(sizeof(struct cna_node) > sizeof(struct qnode));
 	/* we store a pointer in the node's @locked field */
 	BUILD_BUG_ON(sizeof(uintptr_t) > sizeof_field(struct mcs_spinlock, locked));
@@ -147,6 +150,9 @@ static struct cna_node *find_successor(struct mcs_spinlock *me)
 static inline bool cna_set_locked_empty_mcs(struct qspinlock *lock, u32 val,
 					struct mcs_spinlock *node)
 {
+	if (static_branch_likely(&cna_lock_disabled))
+		return __set_locked_empty_mcs(lock, val, node);
+
 	/* Check whether the secondary queue is empty. */
 	if (node->locked <= 1) {
 		if (atomic_try_cmpxchg_relaxed(&lock->val, &val,
@@ -177,6 +183,11 @@ static inline void cna_pass_mcs_lock(struct mcs_spinlock *node,
 	u64 *var = &next->locked;
 	u64 val = 1;
 
+	if (static_branch_likely(&cna_lock_disabled)) {
+		__pass_mcs_lock(node, next);
+		return;
+	}
+
 	/*
 	 * Limit thread shuffling when the secondary queue is empty.
 	 * This copes with the overhead the shuffling creates when the