On Tue, 27 Apr 2010 16:28:11 -0700 (PDT)
David Miller <davem@davemloft.net> wrote:

quoted text
> From: Stephen Hemminger <shemminger@vyatta.com>
> Date: Tue, 27 Apr 2010 16:25:30 -0700
> 
> > On Wed, 28 Apr 2010 01:11:52 +0200
> > Michał Mirosław <mirqus@gmail.com> wrote:
> > 
> >> 2010/4/27 Stephen Hemminger <shemminger@vyatta.com>:
> >> > I prefer that the hlist be only accessed through the hlist macro
> >> > objects. Explicit twiddling of links (especially with RCU) exposes
> >> > the code to future bugs.
> >> [...]
> >> > -       port->rlist.pprev = h;
> >> > -       port->rlist.next = p;
> >> > -       rcu_assign_pointer(*h, &port->rlist);
> >> > -       if (p)
> >> > -               p->pprev = &port->rlist.next;
> >> > +       hlist_for_each_entry(p, n, &br->router_list, rlist) {
> >> 
> >> Shouldn't this be hlist_for_each_entry_rcu?
> >> 
> > 
> > This code should already be protected by br->multicast_lock
> 
> But the adds et al. use RCU already, I think we should be consistent
> one way or another.
> 
> I've already made Michał's suggested change to net-next-2.6, if you
> think _rcu() isn't necessary then trim it from all the hlist calls
> in this function.

Code that doesn't need rcu for traversal should not use it.
It just confuses things and implies that rcu_read_lock is held
which it is not in this code.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html