diff --git a/Makefile b/Makefile
index 32918c3..18800cb 100644
--- a/Makefile
+++ b/Makefile
@@ -141,6 +141,15 @@
         $(error Unknown BRANCH_PROTECTION value ${BRANCH_PROTECTION})
 endif
 
+# USE_SPINLOCK_CAS requires AArch64 build
+ifeq (${USE_SPINLOCK_CAS},1)
+ifneq (${ARCH},aarch64)
+        $(error USE_SPINLOCK_CAS requires AArch64)
+else
+        $(info USE_SPINLOCK_CAS is an experimental feature)
+endif
+endif
+
 ################################################################################
 # Toolchain
 ################################################################################
@@ -690,6 +699,7 @@
 $(eval $(call assert_boolean,BL2_AT_EL3))
 $(eval $(call assert_boolean,BL2_IN_XIP_MEM))
 $(eval $(call assert_boolean,BL2_INV_DCACHE))
+$(eval $(call assert_boolean,USE_SPINLOCK_CAS))
 
 $(eval $(call assert_numeric,ARM_ARCH_MAJOR))
 $(eval $(call assert_numeric,ARM_ARCH_MINOR))
@@ -755,6 +765,7 @@
 $(eval $(call add_define,BL2_AT_EL3))
 $(eval $(call add_define,BL2_IN_XIP_MEM))
 $(eval $(call add_define,BL2_INV_DCACHE))
+$(eval $(call add_define,USE_SPINLOCK_CAS))
 
 ifeq (${SANITIZE_UB},trap)
         $(eval $(call add_define,MONITOR_TRAPS))
diff --git a/docs/design/firmware-design.rst b/docs/design/firmware-design.rst
index dc08208..2cbd9c9 100644
--- a/docs/design/firmware-design.rst
+++ b/docs/design/firmware-design.rst
@@ -2540,8 +2540,11 @@
 This Architecture Extension is targeted when ``ARM_ARCH_MAJOR`` >= 8, or when
 ``ARM_ARCH_MAJOR`` == 8 and ``ARM_ARCH_MINOR`` >= 1.
 
--  The Compare and Swap instruction is used to implement spinlocks. Otherwise,
-   the load-/store-exclusive instruction pair is used.
+-  By default, a load-/store-exclusive instruction pair is used to implement
+   spinlocks. The ``USE_SPINLOCK_CAS`` build option when set to 1 selects the
+   spinlock implementation using the ARMv8.1-LSE Compare and Swap instruction.
+   Notice this instruction is only available in AArch64 execution state, so
+   the option is only available to AArch64 builds.
 
 Armv8.2-A
 ~~~~~~~~~
diff --git a/docs/getting_started/user-guide.rst b/docs/getting_started/user-guide.rst
index 44bfb7a..ae276f2 100644
--- a/docs/getting_started/user-guide.rst
+++ b/docs/getting_started/user-guide.rst
@@ -820,6 +820,10 @@
    reduces SRAM usage. Refer to `Library at ROM`_ for further details. Default
    is 0.
 
+-  ``USE_SPINLOCK_CAS``: Setting this build flag to 1 selects the spinlock
+    implementation variant using the ARMv8.1-LSE compare-and-swap instruction.
+    Notice this option is experimental and only available to AArch64 builds.
+
 -  ``V``: Verbose build. If assigned anything other than 0, the build commands
    are printed. Default is 0.
 
diff --git a/lib/locks/exclusive/aarch64/spinlock.S b/lib/locks/exclusive/aarch64/spinlock.S
index d0569f1..e941b8a 100644
--- a/lib/locks/exclusive/aarch64/spinlock.S
+++ b/lib/locks/exclusive/aarch64/spinlock.S
@@ -9,56 +9,38 @@
 	.globl	spin_lock
 	.globl	spin_unlock
 
-#if ARM_ARCH_AT_LEAST(8, 1)
+#if USE_SPINLOCK_CAS
+#if !ARM_ARCH_AT_LEAST(8, 1)
+#error USE_SPINLOCK_CAS option requires at least an ARMv8.1 platform
+#endif
 
 /*
  * When compiled for ARMv8.1 or later, choose spin locks based on Compare and
  * Swap instruction.
  */
-# define USE_CAS	1
-
-/*
- * Lock contenders using CAS, upon failing to acquire the lock, wait with the
- * monitor in open state. Therefore, a normal store upon unlocking won't
- * generate an SEV. Use explicit SEV instruction with CAS unlock.
- */
-# define COND_SEV()	sev
-
-#else
-
-# define USE_CAS	0
-
-/*
- * Lock contenders using exclusive pairs, upon failing to acquire the lock, wait
- * with the monitor in exclusive state. A normal store upon unlocking will
- * implicitly generate an envent; so, no explicit SEV with unlock is required.
- */
-# define COND_SEV()
-
-#endif
-
-#if USE_CAS
 
 /*
  * Acquire lock using Compare and Swap instruction.
  *
- * Compare for 0 with acquire semantics, and swap 1. Wait until CAS returns
- * 0.
+ * Compare for 0 with acquire semantics, and swap 1. If failed to acquire, use
+ * load exclusive semantics to monitor the address and enter WFE.
  *
  * void spin_lock(spinlock_t *lock);
  */
 func spin_lock
 	mov	w2, #1
-	sevl
-1:
+1:	mov	w1, wzr
+2:	casa	w1, w2, [x0]
+	cbz	w1, 3f
+	ldxr	w1, [x0]
+	cbz	w1, 2b
 	wfe
-	mov	w1, wzr
-	casa	w1, w2, [x0]
-	cbnz	w1, 1b
+	b	1b
+3:
 	ret
 endfunc spin_lock
 
-#else /* !USE_CAS */
+#else /* !USE_SPINLOCK_CAS */
 
 /*
  * Acquire lock using load-/store-exclusive instruction pair.
@@ -76,17 +58,18 @@
 	ret
 endfunc spin_lock
 
-#endif /* USE_CAS */
+#endif /* USE_SPINLOCK_CAS */
 
 /*
  * Release lock previously acquired by spin_lock.
  *
- * Unconditionally write 0, and conditionally generate an event.
+ * Use store-release to unconditionally clear the spinlock variable.
+ * Store operation generates an event to all cores waiting in WFE
+ * when address is monitored by the global monitor.
  *
  * void spin_unlock(spinlock_t *lock);
  */
 func spin_unlock
 	stlr	wzr, [x0]
-	COND_SEV()
 	ret
 endfunc spin_unlock
diff --git a/make_helpers/defaults.mk b/make_helpers/defaults.mk
index b6f7655..b7fb173 100644
--- a/make_helpers/defaults.mk
+++ b/make_helpers/defaults.mk
@@ -234,3 +234,8 @@
 endif
 
 SANITIZE_UB := off
+
+# For ARMv8.1 (AArch64) platforms, enabling this option selects the spinlock
+# implementation variant using the ARMv8.1-LSE compare-and-swap instruction.
+# Default: disabled
+USE_SPINLOCK_CAS := 0