[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20130328193957.GH3337@linux.vnet.ibm.com>
Date: Thu, 28 Mar 2013 12:39:57 -0700
From: "Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>
To: Eric Dumazet <eric.dumazet@...il.com>
Cc: Balakumaran Kannan <kumaran.4353@...il.com>,
yoshfuji@...ux-ipv6.org, davem@...emloft.net,
Patrick McHardy <kaber@...sh.net>,
Alexey Kuznetsov <kuznet@....inr.ac.ru>, jmorris@...ei.org,
Balakumaran.Kannan@...sony.com, maruthi.thotad@...sony.com,
netdev@...r.kernel.org, jamshed.a@...sony.com,
amit.agarwal@...sony.com, takuzo.ohara@...sony.com,
aaditya.kumar@...sony.com
Subject: Re: [PATCH v2] net IPv6 : Fix broken IPv6 routing table after
loopback down-up
On Thu, Mar 28, 2013 at 11:49:37AM -0700, Eric Dumazet wrote:
> On Thu, 2013-03-28 at 11:23 -0700, Paul E. McKenney wrote:
>
> >
> > Agreed, there is no point in using RCU when it is not needed.
> >
> > That said...
> >
> > Since v3.1, in CONFIG_PREEMPT=y kernels, rcu_read_lock() does not
> > disable preemption. In CONFIG_PREEMPT=n kernels, rcu_read_lock() does
> > disable preemption, but to no effect because preemption is already
> > disabled anyway.
> >
> > The net effect is that rcu_read_lock() has no effect on preemption.
> >
>
> Good point, but this patch might be a stable candidate.
>
> Rule of thumb for networking is
>
> 1) Control path : RTNL mutex
>
> 2) Data path : RTNL cant be taken (from softirq), so use RCU if
> possible.
Stable candidate and rules of thumb sound good to me!
Thanx, Paul
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists