On 08/12/2014 06:23 PM, Aryeh Gregor wrote:
On Tue, Aug 12, 2014 at 6:16 PM, Benoit Jacob <jacob.benoi...@gmail.com> wrote:
As far as I know, the only downside in replacing already_AddRefed by
nsCOMPtr would be to incur more useless calls to AddRef and Release. In the
case of "threadsafe" i.e. atomic refcounting, these use atomic instructions,
which might be expensive enough on certain ARM CPUs that this might matter.
So if you're interested, you could take a low-end ARM CPU that we care about
and see if replacing already_AddRefed by nsCOMPtr causes any measurable
performance regression.
Bug 1015114 removes those extra addrefs using C++11 move semantics, so
assuming that lands, it's not an issue. (IIRC, Boris has previously
said that excessive addref/release is a real performance problem and
needs to be avoided.)
AddRef/Release calls are performance issue everywhere, in hot code paths,
whether or not
there is threadsafety involved.
(excluding inline non-virtual AddRef/Release)
-Olli
_______________________________________________
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform
