aboutsummaryrefslogtreecommitdiff
path: root/lib/locks
diff options
context:
space:
mode:
authorJeenu Viswambharan <jeenu.viswambharan@arm.com>2017-01-16 16:52:35 +0000
committerJeenu Viswambharan <jeenu.viswambharan@arm.com>2017-02-14 09:26:11 +0000
commitc877b414870bb3e421518caf9c7652a9807419b5 (patch)
treef7d2837cb46edd4f3578d39985c4d13fa97e419d /lib/locks
parent4d07e7821e19dc1ebc640f5264c2a769354c8b2d (diff)
downloadtrusted-firmware-a-c877b414870bb3e421518caf9c7652a9807419b5.tar.gz
Introduce locking primitives using CAS instruction
The ARMv8v.1 architecture extension has introduced support for far atomics, which includes compare-and-swap. Compare and Swap instruction is only available for AArch64. Introduce build options to choose the architecture versions to target ARM Trusted Firmware: - ARM_ARCH_MAJOR: selects the major version of target ARM Architecture. Default value is 8. - ARM_ARCH_MINOR: selects the minor version of target ARM Architecture. Default value is 0. When: (ARM_ARCH_MAJOR > 8) || ((ARM_ARCH_MAJOR == 8) && (ARM_ARCH_MINOR >= 1)), for AArch64, Compare and Swap instruction is used to implement spin locks. Otherwise, the implementation falls back to using load-/store-exclusive instructions. Update user guide, and introduce a section in Firmware Design guide to summarize support for features introduced in ARMv8 Architecture Extensions. Change-Id: I73096a0039502f7aef9ec6ab3ae36680da033f16 Signed-off-by: Jeenu Viswambharan <jeenu.viswambharan@arm.com>
Diffstat (limited to 'lib/locks')
-rw-r--r--lib/locks/exclusive/aarch64/spinlock.S70
1 files changed, 69 insertions, 1 deletions
diff --git a/lib/locks/exclusive/aarch64/spinlock.S b/lib/locks/exclusive/aarch64/spinlock.S
index 1ca5912328..bdc9ea0f7d 100644
--- a/lib/locks/exclusive/aarch64/spinlock.S
+++ b/lib/locks/exclusive/aarch64/spinlock.S
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2013-2016, ARM Limited and Contributors. All rights reserved.
+ * Copyright (c) 2013-2017, ARM Limited and Contributors. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
@@ -33,7 +33,66 @@
.globl spin_lock
.globl spin_unlock
+#if (ARM_ARCH_MAJOR > 8) || ((ARM_ARCH_MAJOR == 8) && (ARM_ARCH_MINOR >= 1))
+/*
+ * 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
+
+ .arch armv8.1-a
+
+/*
+ * Acquire lock using Compare and Swap instruction.
+ *
+ * Compare for 0 with acquire semantics, and swap 1. Wait until CAS returns
+ * 0.
+ *
+ * void spin_lock(spinlock_t *lock);
+ */
+func spin_lock
+ mov w2, #1
+ sevl
+1:
+ wfe
+ mov w1, wzr
+ casa w1, w2, [x0]
+ cbnz w1, 1b
+ ret
+endfunc spin_lock
+
+ .arch armv8-a
+
+#else /* !USE_CAS */
+
+/*
+ * Acquire lock using load-/store-exclusive instruction pair.
+ *
+ * void spin_lock(spinlock_t *lock);
+ */
func spin_lock
mov w2, #1
sevl
@@ -45,8 +104,17 @@ l2: ldaxr w1, [x0]
ret
endfunc spin_lock
+#endif /* USE_CAS */
+/*
+ * Release lock previously acquired by spin_lock.
+ *
+ * Unconditionally write 0, and conditionally generate an event.
+ *
+ * void spin_unlock(spinlock_t *lock);
+ */
func spin_unlock
stlr wzr, [x0]
+ COND_SEV()
ret
endfunc spin_unlock