Patchwork arp_notify: unconditionally send gratuitous ARP for NETDEV_NOTIFY_PEERS.

login
register
mail settings
Submitter Ian Campbell
Date Feb. 11, 2011, 5:44 p.m.
Message ID <1297446256-23917-1-git-send-email-ian.campbell@citrix.com>
Download mbox | patch
Permalink /patch/82813/
State Accepted
Delegated to: David Miller
Headers show

Comments

Ian Campbell - Feb. 11, 2011, 5:44 p.m.
NETDEV_NOTIFY_PEER is an explicit request by the driver to send a link
notification while NETDEV_UP/NETDEV_CHANGEADDR generate link
notifications as a sort of side effect.

In the later cases the sysctl option is present because link
notification events can have undesired effects e.g. if the link is
flapping. I don't think this applies in the case of an explicit
request from a driver.

This patch makes NETDEV_NOTIFY_PEER unconditional, if preferred we
could add a new sysctl for this case which defaults to on.

This change causes Xen post-migration ARP notifications (which cause
switches to relearn their MAC tables etc) to be sent by default.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
---
 net/ipv4/devinet.c |   30 ++++++++++++++++++++----------
 1 files changed, 20 insertions(+), 10 deletions(-)
David Miller - Feb. 15, 2011, 1:46 a.m.
From: Ian Campbell <ian.campbell@citrix.com>
Date: Fri, 11 Feb 2011 17:44:16 +0000

> NETDEV_NOTIFY_PEER is an explicit request by the driver to send a link
> notification while NETDEV_UP/NETDEV_CHANGEADDR generate link
> notifications as a sort of side effect.
> 
> In the later cases the sysctl option is present because link
> notification events can have undesired effects e.g. if the link is
> flapping. I don't think this applies in the case of an explicit
> request from a driver.
> 
> This patch makes NETDEV_NOTIFY_PEER unconditional, if preferred we
> could add a new sysctl for this case which defaults to on.
> 
> This change causes Xen post-migration ARP notifications (which cause
> switches to relearn their MAC tables etc) to be sent by default.
> 
> Signed-off-by: Ian Campbell <ian.campbell@citrix.com>

Ok, applied, thanks.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Ian Campbell - Feb. 15, 2011, 9:13 a.m.
On Tue, 2011-02-15 at 01:46 +0000, David Miller wrote:
> From: Ian Campbell <ian.campbell@citrix.com>
> Date: Fri, 11 Feb 2011 17:44:16 +0000
> 
> > NETDEV_NOTIFY_PEER is an explicit request by the driver to send a link
> > notification while NETDEV_UP/NETDEV_CHANGEADDR generate link
> > notifications as a sort of side effect.
> > 
> > In the later cases the sysctl option is present because link
> > notification events can have undesired effects e.g. if the link is
> > flapping. I don't think this applies in the case of an explicit
> > request from a driver.
> > 
> > This patch makes NETDEV_NOTIFY_PEER unconditional, if preferred we
> > could add a new sysctl for this case which defaults to on.
> > 
> > This change causes Xen post-migration ARP notifications (which cause
> > switches to relearn their MAC tables etc) to be sent by default.
> > 
> > Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
> 
> Ok, applied, thanks.

Thanks.

I nearly forgot -- the NETDEV_NOTIFY_PEER stuff was tagged for
stable/longterm backport (it appeared in 2.6.32.19 or so). I think this
change should likewise go back, what do you think?

Ian.


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Miller - Feb. 15, 2011, 6:51 p.m.
From: Ian Campbell <Ian.Campbell@eu.citrix.com>
Date: Tue, 15 Feb 2011 09:13:45 +0000

> On Tue, 2011-02-15 at 01:46 +0000, David Miller wrote:
>> From: Ian Campbell <ian.campbell@citrix.com>
>> Date: Fri, 11 Feb 2011 17:44:16 +0000
>> 
>> > NETDEV_NOTIFY_PEER is an explicit request by the driver to send a link
>> > notification while NETDEV_UP/NETDEV_CHANGEADDR generate link
>> > notifications as a sort of side effect.
 ...
> I nearly forgot -- the NETDEV_NOTIFY_PEER stuff was tagged for
> stable/longterm backport (it appeared in 2.6.32.19 or so). I think this
> change should likewise go back, what do you think?

Sure, I've queued it up for -stable.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Anders Franzen - March 14, 2011, 9:26 a.m.
Nice, so this is a generic way of getting out gratuitous arps.
Can it be used by drivers (macvlan) on top of bonding.
Bonding is sending the BOND_FAILOVER event that could be mapped
to NOTIFY_PEERS in the higher level devices.

And bonding is implementing its own grat arps functions, maybe it could
be broken out to use the NOTIFY_PEERS instead.

And what about IPv6, would it be ok to implement an unsolicited NA on
the event?

I will happily implement the above, if anyone thinks it's worth it?

BR
 Anders


On Tue, 2011-02-15 at 19:51 +0100, David Miller wrote:
> From: Ian Campbell <Ian.Campbell@eu.citrix.com>
> Date: Tue, 15 Feb 2011 09:13:45 +0000
> 
> > On Tue, 2011-02-15 at 01:46 +0000, David Miller wrote:
> >> From: Ian Campbell <ian.campbell@citrix.com>
> >> Date: Fri, 11 Feb 2011 17:44:16 +0000
> >> 
> >> > NETDEV_NOTIFY_PEER is an explicit request by the driver to send a link
> >> > notification while NETDEV_UP/NETDEV_CHANGEADDR generate link
> >> > notifications as a sort of side effect.
>  ...
> > I nearly forgot -- the NETDEV_NOTIFY_PEER stuff was tagged for
> > stable/longterm backport (it appeared in 2.6.32.19 or so). I think this
> > change should likewise go back, what do you think?
> 
> Sure, I've queued it up for -stable.
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index 748cb5b..df4616f 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -1030,6 +1030,21 @@  static inline bool inetdev_valid_mtu(unsigned mtu)
 	return mtu >= 68;
 }
 
+static void inetdev_send_gratuitous_arp(struct net_device *dev,
+					struct in_device *in_dev)
+
+{
+	struct in_ifaddr *ifa = in_dev->ifa_list;
+
+	if (!ifa)
+		return;
+
+	arp_send(ARPOP_REQUEST, ETH_P_ARP,
+		 ifa->ifa_address, dev,
+		 ifa->ifa_address, NULL,
+		 dev->dev_addr, NULL);
+}
+
 /* Called only under RTNL semaphore */
 
 static int inetdev_event(struct notifier_block *this, unsigned long event,
@@ -1082,18 +1097,13 @@  static int inetdev_event(struct notifier_block *this, unsigned long event,
 		}
 		ip_mc_up(in_dev);
 		/* fall through */
-	case NETDEV_NOTIFY_PEERS:
 	case NETDEV_CHANGEADDR:
+		if (!IN_DEV_ARP_NOTIFY(in_dev))
+			break;
+		/* fall through */
+	case NETDEV_NOTIFY_PEERS:
 		/* Send gratuitous ARP to notify of link change */
-		if (IN_DEV_ARP_NOTIFY(in_dev)) {
-			struct in_ifaddr *ifa = in_dev->ifa_list;
-
-			if (ifa)
-				arp_send(ARPOP_REQUEST, ETH_P_ARP,
-					 ifa->ifa_address, dev,
-					 ifa->ifa_address, NULL,
-					 dev->dev_addr, NULL);
-		}
+		inetdev_send_gratuitous_arp(dev, in_dev);
 		break;
 	case NETDEV_DOWN:
 		ip_mc_down(in_dev);