Compiler implementations may provide sub-optimal support for a
memory_order passed in as a run-time value (ref.
https://gcc.gnu.org/onlinedocs/gcc/_005f_005fatomic-Builtins.html).
Document that OVS atomics require the memory order to be passed in as
a compile-time constant.
It should be noted, however, that when inlining is disabled (i.e.,
compiling without optimization) even compile-time constants may be
passed as run-time values to (non-inlined) functions.
Signed-off-by: Jarno Rajahalme <jrajaha...@nicira.com>
---
lib/ovs-atomic-gcc4+.h | 10 ++++------
lib/ovs-atomic.h | 4 ++++
2 files changed, 8 insertions(+), 6 deletions(-)
diff --git a/lib/ovs-atomic-gcc4+.h b/lib/ovs-atomic-gcc4+.h
index 9a79f7e..bb889c6 100644
--- a/lib/ovs-atomic-gcc4+.h
+++ b/lib/ovs-atomic-gcc4+.h
@@ -63,7 +63,7 @@ atomic_thread_fence_if_seq_cst(memory_order order)
}
static inline void
-atomic_signal_fence(memory_order order OVS_UNUSED)
+atomic_signal_fence(memory_order order)
{
if (order != memory_order_relaxed) {
asm volatile("" : : : "memory");
@@ -80,12 +80,11 @@ atomic_signal_fence(memory_order order OVS_UNUSED)
({ \
typeof(DST) dst__ = (DST); \
typeof(SRC) src__ = (SRC); \
- memory_order order__ = (ORDER); \
\
if (IS_LOCKLESS_ATOMIC(*dst__)) { \
- atomic_thread_fence(order__); \
+ atomic_thread_fence(ORDER); \
*dst__ = src__; \
- atomic_thread_fence_if_seq_cst(order__); \
+ atomic_thread_fence_if_seq_cst(ORDER); \
} else { \
atomic_store_locked(dst__, src__); \
} \
@@ -97,10 +96,9 @@ atomic_signal_fence(memory_order order OVS_UNUSED)
({ \
typeof(DST) dst__ = (DST); \
typeof(SRC) src__ = (SRC); \
- memory_order order__ = (ORDER); \
\
if (IS_LOCKLESS_ATOMIC(*src__)) { \
- atomic_thread_fence_if_seq_cst(order__); \
+ atomic_thread_fence_if_seq_cst(ORDER); \
*dst__ = *src__; \
} else { \
atomic_read_locked(src__, dst__); \
diff --git a/lib/ovs-atomic.h b/lib/ovs-atomic.h
index d9c506a..b9a3203 100644
--- a/lib/ovs-atomic.h
+++ b/lib/ovs-atomic.h
@@ -181,6 +181,10 @@
* whole system, providing a total order for stores an all atomic
* variables.
*
+ * OVS atomics require the memory_order to be passed as a compile-time constant
+ * value, as some compiler implementations may perform poorly if the memory
+ * order parameter is passed in as a run-time value.
+ *
* The following functions insert explicit barriers. Most of the other atomic
* functions also include barriers.
*
--
1.7.10.4
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
