public inbox for netdev@vger.kernel.org 
 help / color / mirror / Atom feed
From: Hannes Frederic Sowa <hannes@stressinduktion•org>
To: Gao feng <gaofeng@cn•fujitsu.com>
Cc: chenweilong <chenweilong@huawei•com>,
	David Miller <davem@davemloft•net>,
	kumaran.4353@gmail•com, netdev@vger•kernel.org
Subject: Re: [PATCH] ipv6: don't call addrconf_dst_alloc again when enable lo
Date: Wed, 8 Jan 2014 09:05:28 +0100	[thread overview]
Message-ID: <20140108080528.GD9007@order.stressinduktion.org> (raw)
In-Reply-To: <52CD0331.8040204@cn.fujitsu.com>

On Wed, Jan 08, 2014 at 03:50:09PM +0800, Gao feng wrote:
> On 01/03/2014 02:53 PM, Hannes Frederic Sowa wrote:
> > On Thu, Jan 02, 2014 at 05:33:15PM +0800, chenweilong wrote:
> >> diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
> >> index 62d1799..d2f8c0a 100644
> >> --- a/net/ipv6/addrconf.c
> >> +++ b/net/ipv6/addrconf.c
> >> @@ -2422,8 +2422,9 @@ static void init_loopback(struct net_device *dev)
> >>  			if (sp_ifa->flags & (IFA_F_DADFAILED | IFA_F_TENTATIVE))
> >>  				continue;
> >>
> >> -			if (sp_ifa->rt)
> >> -				continue;
> >> +			if (sp_ifa->rt && sp_ifa->rt->dst.dev == dev) {
> >> +				ip6_del_rt(sp_ifa->rt);
> >> +			}
> >>
> >>  			sp_rt = addrconf_dst_alloc(idev, &sp_ifa->addr, 0);
> >>
> > 
> > Maybe this change would not be that bad after all, as those ifa attached dsts
> > are already dead and queued up for gc and should not get inserted back.
> 
> I like this idea, maybe the below patch is better. we only need to delete this
> route when it has been added to garbage list.
> 
> diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
> index 1a341f7..4dca886 100644
> --- a/net/ipv6/addrconf.c
> +++ b/net/ipv6/addrconf.c
> @@ -2610,8 +2610,16 @@ static void init_loopback(struct net_device *dev)
>                         if (sp_ifa->flags & (IFA_F_DADFAILED | IFA_F_TENTATIVE))
>                                 continue;
> 
> -                       if (sp_ifa->rt)
> -                               continue;
> +                       if (sp_ifa->rt) {
> +                               /* This dst has been added to garbage list when
> +                                * lo device down, delete this obsolete dst and
> +                                * reallocate new router for ifa. */
> +                               if (sp_ifa->rt->dst.obsolete > 0) {
> +                                       ip6_del_rt(sp_ifa->rt);
> +                                       sp_ifa->rt = NULL;
> +                               } else
> +                                       continue;
> +                       }
> 
>                         sp_rt = addrconf_dst_alloc(idev, &sp_ifa->addr, false);

It looks like it can work but I don't know if we should just fix this the
clean way (see below).

> > I'll try to just disable routes without removing them at all when we set an
> > interface to down at the weekend.
> > 
> 
> How do you decide which route should be disabled?  use rt6_flags? I don't know
> if your way will cause miscarriage.

What I did so far is that I added a new function next to rt6_ifdown that
only gets called if interface gets shutdown but not unregistered (from
addrconf_ifdown).

fib6_clean_all then iterates over the whole routing table with a new predicate
function which checks in the same way like fib6_ifdown, if it is a matching route
(the interfaces match up) and if so, toggles a new "DEAD" flag in rt6i_flags.

When bringing up the interface I distinguish between up and register and just
clear this death flag from the routes on bringing it up.

fib lookup code then does not honour those routes.

I had no time to test this thoroughly at the weekend and still have some code
paths were I am unsure. Do you see any problems with this so far? We could
then delete the special cases on loopback interface init.

Thanks,

  Hannes

  reply	other threads:[~2014-01-08  8:05 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-06-16  3:14 [PATCH] ipv6: don't call addrconf_dst_alloc again when enable lo Gao feng
2013-06-20  6:05 ` David Miller
2013-12-31  3:57   ` Hannes Frederic Sowa
2014-01-02  5:48     ` chenweilong
2014-01-02  6:03       ` Hannes Frederic Sowa
2014-01-02  8:13         ` chenweilong
2014-01-02  8:32           ` Hannes Frederic Sowa
2014-01-02  9:08             ` chenweilong
2014-01-02  6:54       ` Hannes Frederic Sowa
2014-01-02  7:58         ` chenweilong
2014-01-02  8:23           ` Hannes Frederic Sowa
2014-01-02  9:33             ` chenweilong
2014-01-03  6:53               ` Hannes Frederic Sowa
2014-01-08  7:50                 ` Gao feng
2014-01-08  8:05                   ` Hannes Frederic Sowa [this message]
2014-01-08  8:42                     ` Gao feng
2014-01-08  8:55                       ` Hannes Frederic Sowa
2014-01-17  2:02                         ` chenweilong
2014-01-17  4:09                           ` Hannes Frederic Sowa

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20140108080528.GD9007@order.stressinduktion.org \
    --to=hannes@stressinduktion$(echo .)org \
    --cc=chenweilong@huawei$(echo .)com \
    --cc=davem@davemloft$(echo .)net \
    --cc=gaofeng@cn$(echo .)fujitsu.com \
    --cc=kumaran.4353@gmail$(echo .)com \
    --cc=netdev@vger$(echo .)kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox