On 07/31/2013 08:38 PM, Stephen Hemminger wrote:

On Wed, 31 Jul 2013 17:12:29 +0200
Nikolay Aleksandrov [off-list ref] wrote:

quoted

-	bond_for_each_slave(bond, tmp, i)
+	list_for_each_entry(tmp, &bond->slave_list, list)

It would be nice to keep the wrapper macro's bond_for_each_slave
and introduce a bond_for_each_slave_rcu() for the cases where you
are using RCU.

I had done it like that in the first versions of these patches but since
this patchset is changing parts that people have relied on for a long time
I'd prefer to keep the list_for_each_entry(_rcu) explicit at least in the
beginning so no one can get confused about what's happening or a person who
hasn't looked in the code recently would find it easier to get up to speed.
Also I don't think the full writing of list_for_each_entry has lead to
breaking any lines right now.
In the end I don't have strong feelings about how this should look, so
either way is fine with me, these are my arguments for doing it this way.

Nik