I just sent v2’s of the first patches discussed so far.
Jarno
On Jul 4, 2014, at 6:15 AM, Jarno Rajahalme wrote:
> Additional data point: I used
> http://en.cppreference.com/w/cpp/atomic/memory_order as a reference, and it
> reads:
>
> “Typical use for relaxed memory ordering is updating c
Additional data point: I used
http://en.cppreference.com/w/cpp/atomic/memory_order as a reference, and it
reads:
“Typical use for relaxed memory ordering is updating counters, such as the
reference counters of std::shared_ptr, since this only requires atomicity, but
not ordering or synchroniza
It seems I was too invested in the combined refcount/RCU case here. I still
think that with RCU postponed destruction relaxed is the proper memory model.
So maybe we should add a relaxed variant of the unref function to be used with
RCU objects and make the normal unref use release to guarantee
On Mon, Jun 30, 2014 at 08:17:18AM -0700, Jarno Rajahalme wrote:
> Updating the reference count only requires atomicity, but no memory
> ordering with respect to any other loads or stores. Avoiding the
> overhead of the default memory_order_seq_cst can make these more
> efficient.
>
> Signed-off-
Updating the reference count only requires atomicity, but no memory
ordering with respect to any other loads or stores. Avoiding the
overhead of the default memory_order_seq_cst can make these more
efficient.
Signed-off-by: Jarno Rajahalme
---
lib/ovs-atomic.h |8 +---
1 file changed, 5
