[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20150324180628.278017fb@griffin>
Date: Tue, 24 Mar 2015 18:06:28 +0100
From: Jiri Benc <jbenc@...hat.com>
To: David Miller <davem@...emloft.net>
Cc: netdev@...r.kernel.org, maheshb@...gle.com, dcbw@...hat.com
Subject: Re: [PATCH net] ipvlan: fix addr hash list corruption
On Tue, 24 Mar 2015 13:00:37 -0400 (EDT), David Miller wrote:
> From: Jiri Benc <jbenc@...hat.com>
> Date: Mon, 23 Mar 2015 22:10:19 +0100
>
> > @@ -504,7 +504,8 @@ static void ipvlan_link_delete(struct net_device *dev, struct list_head *head)
> >
> > if (ipvlan->ipv6cnt > 0 || ipvlan->ipv4cnt > 0) {
> > list_for_each_entry_safe(addr, next, &ipvlan->addrs, anode) {
> > - ipvlan_ht_addr_del(addr, !dev->dismantle);
> > + if (netif_running(dev))
> > + ipvlan_ht_addr_del(addr, !dev->dismantle);
> > list_del_rcu(&addr->anode);
> > }
> > }
>
> This is so error prone, because you are depending upon so many implementation
> details to infer a boolean state "is this address hashed".
>
> So just add the boolean state to struct ipvl_addr, and manage it in
> the ipvlan_ht_addr_{add,del}() code.
I had that originally but then decided to go with smaller memory
footprint. Which obviously doesn't matter much here. I'll send v2 with
the boolean state.
Jiri
--
Jiri Benc
--
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