diff --git a/sys/arch/arm/include/lock.h b/sys/arch/arm/include/lock.h
index f345f54ba46d..b70af86b0fe2 100644
--- a/sys/arch/arm/include/lock.h
+++ b/sys/arch/arm/include/lock.h
@@ -41,6 +41,10 @@
 #ifndef _ARM_LOCK_H_
 #define	_ARM_LOCK_H_
 
+#ifdef _ARM_ARCH_8
+#error This file is only for armv<=7.
+#endif
+
 static __inline int
 __SIMPLELOCK_LOCKED_P(const __cpu_simple_lock_t *__ptr)
 {
@@ -136,21 +140,12 @@ __swp(int __val, __cpu_simple_lock_t *__ptr)
 }
 #endif /* !_ARM_ARCH_6 */
 
+/* load/dmb implies load-acquire, dmb/store implies store-release */
 static __inline void
-__arm_membar_producer(void)
+__arm_dmb(void)
 {
 #ifdef _ARM_ARCH_7
-	__asm __volatile("dsb" ::: "memory");
-#elif defined(_ARM_ARCH_6)
-	__asm __volatile("mcr\tp15,0,%0,c7,c10,4" :: "r"(0) : "memory");
-#endif
-}
-
-static __inline void
-__arm_membar_consumer(void)
-{
-#ifdef _ARM_ARCH_7
-	__asm __volatile("dmb" ::: "memory");
+	__asm __volatile("dmb ish" ::: "memory");
 #elif defined(_ARM_ARCH_6)
 	__asm __volatile("mcr\tp15,0,%0,c7,c10,5" :: "r"(0) : "memory");
 #endif
@@ -161,7 +156,6 @@ __cpu_simple_lock_init(__cpu_simple_lock_t *__alp)
 {
 
 	*__alp = __SIMPLELOCK_UNLOCKED;
-	__arm_membar_producer();
 }
 
 #if !defined(__thumb__) || defined(_ARM_ARCH_T2)
@@ -169,12 +163,11 @@ static __inline void __unused
 __cpu_simple_lock(__cpu_simple_lock_t *__alp)
 {
 #ifdef _ARM_ARCH_6
-	__arm_membar_consumer();
 	do {
 		/* spin */
 	} while (__arm_load_exclusive(__alp) != __SIMPLELOCK_UNLOCKED
 		 || __arm_store_exclusive(__alp, __SIMPLELOCK_LOCKED));
-	__arm_membar_producer();
+	__arm_dmb();
 #else
 	while (__swp(__SIMPLELOCK_LOCKED, __alp) != __SIMPLELOCK_UNLOCKED)
 		continue;
@@ -189,13 +182,12 @@ static __inline int __unused
 __cpu_simple_lock_try(__cpu_simple_lock_t *__alp)
 {
 #ifdef _ARM_ARCH_6
-	__arm_membar_consumer();
 	do {
 		if (__arm_load_exclusive(__alp) != __SIMPLELOCK_UNLOCKED) {
 			return 0;
 		}
 	} while (__arm_store_exclusive(__alp, __SIMPLELOCK_LOCKED));
-	__arm_membar_producer();
+	__arm_dmb();
 	return 1;
 #else
 	return (__swp(__SIMPLELOCK_LOCKED, __alp) == __SIMPLELOCK_UNLOCKED);
@@ -209,19 +201,8 @@ static __inline void __unused
 __cpu_simple_unlock(__cpu_simple_lock_t *__alp)
 {
 
-#ifdef _ARM_ARCH_8
-	if (sizeof(*__alp) == 1) {
-		__asm __volatile("stlb\t%0, [%1]"
-		    :: "r"(__SIMPLELOCK_UNLOCKED), "r"(__alp) : "memory");
-	} else {
-		__asm __volatile("stl\t%0, [%1]"
-		    :: "r"(__SIMPLELOCK_UNLOCKED), "r"(__alp) : "memory");
-	}
-#else
-	__arm_membar_consumer();
+	__arm_dmb();
 	*__alp = __SIMPLELOCK_UNLOCKED;
-	__arm_membar_producer();
-#endif
 }
 
 #endif /* _ARM_LOCK_H_ */