[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20101201130421.0ac08bcd@nehalam>
Date: Wed, 1 Dec 2010 13:04:21 -0800
From: Stephen Hemminger <shemminger@...tta.com>
To: Lorenzo Colitti <lorenzo@...gle.com>
Cc: netdev@...r.kernel.org, David Miller <davem@...emloft.net>
Subject: Re: [PATCH] ipv6: slightly simplify keeping IPv6 addresses on link
down
On Wed, 1 Dec 2010 12:52:42 -0800
Lorenzo Colitti <lorenzo@...gle.com> wrote:
> On Wed, Dec 1, 2010 at 12:22 PM, Stephen Hemminger
> <shemminger@...tta.com> wrote:
> > > --- a/net/ipv6/addrconf.c
> > > +++ b/net/ipv6/addrconf.c
> > > @@ -2663,7 +2663,8 @@ static int addrconf_ifdown(struct net_device *dev, int how)
> > > {
> > > struct net *net = dev_net(dev);
> > > struct inet6_dev *idev;
> > > - struct inet6_ifaddr *ifa, *ifn;
> > > + struct inet6_ifaddr *ifa;
> > > + LIST_HEAD(keep_list);
> > > int state;
> >
> > Your patch is backwards? The existing code is:
>
> Oops, yes. Wrong order of arguments. Another one coming up.
>
> > Also, the addrconf_ifdown can race with other updates to idev->addr_list
> > from addrconf timers etc. Therefore even list_for_each_entry_safe is not safe.
>
> No, wait... The loop is protected by idev->lock, and the code just
> before it that clears the temporary address list is essentially
> identical (except it looks over tempaddr_list instead). Wouldn't that
> blow up as well?
The old code walked the list until it was empty. New code could
get confused if list changed by other changes during the
period when idev->lock is dropped and notifier is called.
--
--
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