diff --git a/arch/arm/src/armv7-r/arm_head.S b/arch/arm/src/armv7-r/arm_head.S
index 0ec9a7d306c..95ea23456f4 100644
--- a/arch/arm/src/armv7-r/arm_head.S
+++ b/arch/arm/src/armv7-r/arm_head.S
@@ -134,7 +134,17 @@ __start:
 	and		r0, r0, #0x3
 	cmp		r0, #0
 	beq		__cpu0_start
+
+#ifdef CONFIG_ARM_BUSY_WAIT
+	ldr		r2, =CONFIG_ARM_BUSY_WAIT_FLAG_ADDR
+1:
+	ldr		r1, [r2, #0]
+	cmp		r1, #0
+	beq		1b
+#else
 	wfe
+#endif
+
 	cmp		r0, #1
 	beq		__cpu1_start
 #  if CONFIG_SMP_NCPUS > 2
@@ -395,6 +405,15 @@ __cpu0_start:
 
 	b		arm_boot
 
+	/* finish busy wait */
+
+#if defined(CONFIG_ARM_BUSY_WAIT) && defined(CONFIG_SMP)
+	ldr		r0, =CONFIG_ARM_BUSY_WAIT_FLAG_ADDR
+	mov		r1, #1
+	str		r1, [r0]
+	dsb		sy
+#endif
+
 	/* .text Data */
 
 .Lstackpointer:
diff --git a/arch/arm/src/armv8-r/arm_head.S b/arch/arm/src/armv8-r/arm_head.S
index 2caded54464..6d3eec5ff4e 100644
--- a/arch/arm/src/armv8-r/arm_head.S
+++ b/arch/arm/src/armv8-r/arm_head.S
@@ -142,7 +142,17 @@ __start:
 	cmp		r0, #0
 #if defined(CONFIG_SMP) && CONFIG_SMP_NCPUS > 1
 	beq		__cpu0_start
+
+#ifdef CONFIG_ARM_BUSY_WAIT
+	ldr		r2, =CONFIG_ARM_BUSY_WAIT_FLAG_ADDR
+1:
+	ldr		r1, [r2, #0]
+	cmp		r1, #0
+	beq		1b
+#else
 	wfe
+#endif
+
 	cmp		r0, #1
 	beq		__cpu1_start
 #  if CONFIG_SMP_NCPUS > 2
@@ -247,6 +257,13 @@ __cpu0_start:
 	mov		lr, #0				/* LR = return address (none) */
 	b		nx_start			/* Branch to nx_start */
 
+#if defined(CONFIG_ARM_BUSY_WAIT) && defined(CONFIG_SMP)
+	ldr		r0, =CONFIG_ARM_BUSY_WAIT_FLAG_ADDR
+	mov		r1, #1
+	str		r1, [r0]
+	dsb		sy
+#endif
+
 	/* .text Data */
 
 .Lstackpointer: