Date: Tue, 27 Dec 2011 23:44:40 +0400 From: Eygene Ryabinkin <rea@freebsd.org> To: Doug Barton <dougb@FreeBSD.org>, bz@freebsd.org Cc: Pyun Yong-Hyeon <pyunyh@gmail.com>, Brooks Davis <brooks@freebsd.org>, freebsd-rc@FreeBSD.ORG, Garrett Cooper <yanegomi@gmail.com>, Gleb Smirnoff <glebius@FreeBSD.org>, Dag-Erling Smorgrav <des@des.no>, d@delphij.net, Xin LI <delphij@delphij.net> Subject: Re: Annoying ERROR: 'wlan0' is not a DHCP-enabled interface Message-ID: <CsXBJZZMvEvCyHeet8CyDXTxqHc@YnbH/K3/Y1Z96RV2jTofcGuSPJI> In-Reply-To: <4EF99DBC.6020204@FreeBSD.org> References: <LKVlrdfIBdRPFfTmZOlaU48u3P0@g5jH1yj%2BTnAiSdLOy3xs5Jutvhc> <4EF99DBC.6020204@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
--Q68bSM7Ycu6FN28Q
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
Note for Bjoern: you can skip some parts of this mail, the real
question is near the bottom of the letter. In a nutshell, Doug wants
to arm the devd machinery that looks for IFUP events to do IPv6
autoconfiguration as addition to the current invocation of a DHCP
client, but I have some questions about IPv6 autoconfiguration that
I can't answer myself.
Tue, Dec 27, 2011 at 02:28:12AM -0800, Doug Barton wrote:
> Ok, you've convinced me that this approach isn't viable. That's why
> I asked for review before committing it. Thanks for looking at it in
> more detail.
By the way, it seems that your approach won't work at all for
DHCP-enabled interfaces that aren't set to the synchronous DHCP mode
(rc.conf knob synchronous_dhclient, default value is "NO"): netif has
the following code:
{{{
if dhcpif $1; then
if [ $_cfg -ne 0 ] ; then
ifconfig $1 up
fi
if syncdhcpif $1; then
/etc/rc.d/dhclient start $1
fi
_cfg=3D0
fi
}}}
So, it relies on the devd to invoke dhclient when interface will be up
and configured. I imagine that its mostly for the WiFi interfaces
that should associate themselves with the AP or other host, up the
encryption and do other stuff that takes time. DHCP client should
start only after these negotiations will be complete and the link will
really be up. Thus, DHCP-enabled interfaces should react to the
linkup events by anything that ends up calling dhclient, not netif.
I was just thinking of how can one implement the wrapper script for
doing DHCP and IPv6 autoconfiguration that you want to implement and
my initial thoughts were that we can use a simple one-shot invocation
of netif
{{{
if dhcpif $ifn || ipv6_autoconfif $ifn; then
service netif quietstart $ifn
fi
}}}
but this won't work for DHCP.
On the topic of IPv6 autoconfiguration. The need of restart of the
DHCP client upon linkdown/linkup event is simple: dhclient uses
bpf(4) and it dies when link goes down. So it needs to be restarted.
<IPv6 disclaimer>
Since I have no IPv6 machines to test on and IPv6 will (hopefully)
come to my realm only next year, I can't test my assertions
'bout it, since I have no practical experience with it. Thus I can
be wrong in all my IPv6-related sayings.
</IPv6 disclaimer>
=46rom the other hand, reading RFC 2462,
ftp://ftp.ripe.net/rfc/rfc2462.txt
and crawling over the ifconfig, netif and other stuff, I have a strong
impression that interface flags AUTO_LINKLOCAL and ACCEPT_RTADV (that,
respectively, judge if the link local address should be automatically
generated and if router advertisments should be processed) are persistent
in respect to the link flaps.
Moreover, manual page for rtsold(8) says
{{{
o The interface is up after a temporary interface failure. rtsold
detects such failures by periodically probing to see if the status=
of
the interface is active or not. Note that some network cards and
drivers do not allow the extraction of link state. In such cases,
rtsold cannot detect the change of the interface status.
}}}
so the thing that generates router solicitations is more-or-less
persistent and doesn't die on link down events.
Thus, it may be well so that only the initial linkup event is the
one that will really initiate IPv6 autoconfiguration and all other
linkup events won't really need such a processing.
I am CC'ing bz@freebsd.org, since he is the person who is known to mess
with IPv6 in FreeBSD a lot. Bjoern, could you, please, say if
IPv6-autoconfigured interfaces react to the linkdown/linkup events
in a way that they will require some reconfiguration after the flap
to autoconfigure themselves again. Of course, if there are other
IPv6 wizards nearby, they are welcome to educate me on this topic.
Thanks.
--=20
Eygene Ryabinkin ,,,^..^,,,
[ Life's unfair - but root password helps! | codelabs.ru ]
[ 82FE 06BC D497 C0DE 49EC 4FF0 16AF 9EAE 8152 ECFB | freebsd.org ]
--Q68bSM7Ycu6FN28Q
Content-Type: application/pgp-signature
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.18 (FreeBSD)
iF4EAREIAAYFAk76ICgACgkQFq+eroFS7PuIOAD5ATJCFsboqq1paiuJnZ9WvKtL
1eP2UlYQnM27wZfb+fcA/Auvta258XTENhNgJdguOelpEAvSk+c0n2ePs9onIY5r
=7aZX
-----END PGP SIGNATURE-----
--Q68bSM7Ycu6FN28Q--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CsXBJZZMvEvCyHeet8CyDXTxqHc>