Re: [PATCH v2] rcu: fix a race in hlist_nulls_for_each_entry_rcu macro

From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
To: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Roman Gushchin <klamm@yandex-team.ru>,
	Jesper Dangaard Brouer <brouer@redhat.com>,
	Dipankar Sarma <dipankar@in.ibm.com>,
	zhmurov@yandex-team.ru, linux-kernel@vger.kernel.org,
	netdev@vger.kernel.org, "David S. Miller" <davem@davemloft.net>,
	Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>,
	James Morris <jmorris@namei.org>,
	Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>,
	Patrick McHardy <kaber@trash.net>
Subject: Re: [PATCH v2] rcu: fix a race in hlist_nulls_for_each_entry_rcu macro
Date: Tue, 28 May 2013 18:31:11 -0700	[thread overview]
Message-ID: <20130529013111.GS6172@linux.vnet.ibm.com> (raw)
In-Reply-To: <1369787693.3301.586.camel@edumazet-glaptop>

On Tue, May 28, 2013 at 05:34:53PM -0700, Eric Dumazet wrote:
> On Tue, 2013-05-28 at 13:10 +0400, Roman Gushchin wrote:
> > On 28.05.2013 04:12, Eric Dumazet wrote:
> 
> > > Adding a barrier() is probably what we want.
> > 
> > I agree, inserting barrier() is also a correct and working fix.
> 
> Yeah, but I can not find a clean way to put it inside the "for (;;)"
> 
> for (barrier();;)  ->
> 
> error: expected expression before ‘__asm__’
> 
> No user currently does :
> 
> if (condition)
> 	hlist_nulls_for_each_entry_rcu(tpos, pos, head, member)
> 
> But who knows...

I still have my earlier question, but I suggest "({ barrier(); XXX })"
to put the barrier into the for loop, either in the second or third
clause, where XXX was the original second or third clause.

							Thanx, Paul