Skip to contentSkip to conversation

bonding: rejoin multicast groups on VLANs

PATCH Superseded Flavio Leitner 4 participants 17 messages Sep 29, 2010 – Oct 6, 2010

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 bonding: rejoin multicast groups on VLANs

It fixes bonding to rejoin multicast groups added
to VLAN devices on top of bonding when a failover
happens.

The first packet may be discarded, so the timer
assure that at least 3 Reports are sent.

---
 drivers/net/bonding/bond_main.c |   59 +++++++++++++++++++++++++++++++++-----
 drivers/net/bonding/bonding.h   |    2 +
 2 files changed, 53 insertions(+), 8 deletions(-)
drivers/net/bonding/bond_main.c +51 -8
index 3b16f62..a23a5fa 100644
@@ -865,18 +865,14 @@ static void bond_mc_del(struct bonding *bond, void *addr)
 }
 
 
-/*
- * Retrieve the list of registered multicast addresses for the bonding
- * device and retransmit an IGMP JOIN request to the current active
- * slave.
- */
-static void bond_resend_igmp_join_requests(struct bonding *bond)
+static void __bond_resend_igmp_join_requests(struct net_device *dev)
 {
 	struct in_device *in_dev;
 	struct ip_mc_list *im;
 
 	rcu_read_lock();
-	in_dev = __in_dev_get_rcu(bond->dev);
+
+	in_dev = __in_dev_get_rcu(dev);
 	if (in_dev) {
 		for (im = in_dev->mc_list; im; im = im->next)
 			ip_mc_rejoin_group(im);
@@ -885,6 +881,42 @@ static void bond_resend_igmp_join_requests(struct bonding *bond)
 	rcu_read_unlock();
 }
 
+
+/*
+ * Retrieve the list of registered multicast addresses for the bonding
+ * device and retransmit an IGMP JOIN request to the current active
+ * slave.
+ */
+static void bond_resend_igmp_join_requests(struct bonding *bond)
+{
+	struct net_device *vlan_dev;
+	struct vlan_entry *vlan;
+
+	read_lock(&bond->lock);
+	if (bond->kill_timers)
+		goto out;
+
+	/* rejoin all groups on bond device */
+	__bond_resend_igmp_join_requests(bond->dev);
+
+	if (!bond->vlgrp)
+		goto reschedule;
+
+	/* rejoin all groups on vlan devices */
+	list_for_each_entry(vlan, &bond->vlan_list, vlan_list) {
+		vlan_dev = vlan_group_get_device(bond->vlgrp, vlan->vlan_id);
+		if (vlan_dev)
+			__bond_resend_igmp_join_requests(vlan_dev);
+	}
+
+reschedule:
+	if (--bond->resend_igmp > 0)
+		mod_timer(&bond->mc_timer, jiffies + HZ/5);
+
+out:
+	read_unlock(&bond->lock);
+}
+
 /*
  * flush all members of flush->mc_list from device dev->mc_list
  */
@@ -944,7 +976,10 @@ static void bond_mc_swap(struct bonding *bond, struct slave *new_active,
 
 		netdev_for_each_mc_addr(ha, bond->dev)
 			dev_mc_add(new_active->dev, ha->addr);
-		bond_resend_igmp_join_requests(bond);
+
+		/* rejoin multicast groups */
+		bond->resend_igmp = 3;
+		mod_timer(&bond->mc_timer, jiffies + 1);
 	}
 }
 
@@ -3741,9 +3776,15 @@ static int bond_xmit_hash_policy_l2(struct sk_buff *skb, int count)
 static int bond_open(struct net_device *bond_dev)
 {
 	struct bonding *bond = netdev_priv(bond_dev);
+	struct timer_list *mc_timer = &bond->mc_timer;
 
 	bond->kill_timers = 0;
 
+	/* multicast */
+	init_timer(mc_timer);
+	mc_timer->data = (unsigned long)bond;
+	mc_timer->function = (void *)&bond_resend_igmp_join_requests;
+
 	if (bond_is_lb(bond)) {
 		/* bond_alb_initialize must be called before the timer
 		 * is started.
@@ -3808,6 +3849,8 @@ static int bond_close(struct net_device *bond_dev)
 
 	write_unlock_bh(&bond->lock);
 
+	del_timer_sync(&bond->mc_timer);
+
 	if (bond->params.miimon) {  /* link check interval, in milliseconds. */
 		cancel_delayed_work(&bond->mii_work);
 	}
drivers/net/bonding/bonding.h +2 -0
index c6fdd85..5fd4164 100644
@@ -198,6 +198,8 @@ struct bonding {
 	s32      slave_cnt; /* never change this value outside the attach/detach wrappers */
 	rwlock_t lock;
 	rwlock_t curr_slave_lock;
+	struct   timer_list mc_timer;
+	s8       resend_igmp;
 	s8       kill_timers;
 	s8	 send_grat_arp;
 	s8	 send_unsol_na;
Signature · 1 line
1.7.2.3
#1 lore raw

forgot CC to bonding-devel

Quoted 139 lines · Flavio Leitner

It fixes bonding to rejoin multicast groups added
to VLAN devices on top of bonding when a failover
happens.

The first packet may be discarded, so the timer
assure that at least 3 Reports are sent.

---
 drivers/net/bonding/bond_main.c |   59 +++++++++++++++++++++++++++++++++-----
 drivers/net/bonding/bonding.h   |    2 +
 2 files changed, 53 insertions(+), 8 deletions(-)
drivers/net/bonding/bond_main.c +51 -8
index 3b16f62..a23a5fa 100644
@@ -865,18 +865,14 @@ static void bond_mc_del(struct bonding *bond, void *addr)
 }
 
 
-/*
- * Retrieve the list of registered multicast addresses for the bonding
- * device and retransmit an IGMP JOIN request to the current active
- * slave.
- */
-static void bond_resend_igmp_join_requests(struct bonding *bond)
+static void __bond_resend_igmp_join_requests(struct net_device *dev)
 {
 	struct in_device *in_dev;
 	struct ip_mc_list *im;
 
 	rcu_read_lock();
-	in_dev = __in_dev_get_rcu(bond->dev);
+
+	in_dev = __in_dev_get_rcu(dev);
 	if (in_dev) {
 		for (im = in_dev->mc_list; im; im = im->next)
 			ip_mc_rejoin_group(im);
@@ -885,6 +881,42 @@ static void bond_resend_igmp_join_requests(struct bonding *bond)
 	rcu_read_unlock();
 }
 
+
+/*
+ * Retrieve the list of registered multicast addresses for the bonding
+ * device and retransmit an IGMP JOIN request to the current active
+ * slave.
+ */
+static void bond_resend_igmp_join_requests(struct bonding *bond)
+{
+	struct net_device *vlan_dev;
+	struct vlan_entry *vlan;
+
+	read_lock(&bond->lock);
+	if (bond->kill_timers)
+		goto out;
+
+	/* rejoin all groups on bond device */
+	__bond_resend_igmp_join_requests(bond->dev);
+
+	if (!bond->vlgrp)
+		goto reschedule;
+
+	/* rejoin all groups on vlan devices */
+	list_for_each_entry(vlan, &bond->vlan_list, vlan_list) {
+		vlan_dev = vlan_group_get_device(bond->vlgrp, vlan->vlan_id);
+		if (vlan_dev)
+			__bond_resend_igmp_join_requests(vlan_dev);
+	}
+
+reschedule:
+	if (--bond->resend_igmp > 0)
+		mod_timer(&bond->mc_timer, jiffies + HZ/5);
+
+out:
+	read_unlock(&bond->lock);
+}
+
 /*
  * flush all members of flush->mc_list from device dev->mc_list
  */
@@ -944,7 +976,10 @@ static void bond_mc_swap(struct bonding *bond, struct slave *new_active,
 
 		netdev_for_each_mc_addr(ha, bond->dev)
 			dev_mc_add(new_active->dev, ha->addr);
-		bond_resend_igmp_join_requests(bond);
+
+		/* rejoin multicast groups */
+		bond->resend_igmp = 3;
+		mod_timer(&bond->mc_timer, jiffies + 1);
 	}
 }
 
@@ -3741,9 +3776,15 @@ static int bond_xmit_hash_policy_l2(struct sk_buff *skb, int count)
 static int bond_open(struct net_device *bond_dev)
 {
 	struct bonding *bond = netdev_priv(bond_dev);
+	struct timer_list *mc_timer = &bond->mc_timer;
 
 	bond->kill_timers = 0;
 
+	/* multicast */
+	init_timer(mc_timer);
+	mc_timer->data = (unsigned long)bond;
+	mc_timer->function = (void *)&bond_resend_igmp_join_requests;
+
 	if (bond_is_lb(bond)) {
 		/* bond_alb_initialize must be called before the timer
 		 * is started.
@@ -3808,6 +3849,8 @@ static int bond_close(struct net_device *bond_dev)
 
 	write_unlock_bh(&bond->lock);
 
+	del_timer_sync(&bond->mc_timer);
+
 	if (bond->params.miimon) {  /* link check interval, in milliseconds. */
 		cancel_delayed_work(&bond->mii_work);
 	}
drivers/net/bonding/bonding.h +2 -0
index c6fdd85..5fd4164 100644
@@ -198,6 +198,8 @@ struct bonding {
 	s32      slave_cnt; /* never change this value outside the attach/detach wrappers */
 	rwlock_t lock;
 	rwlock_t curr_slave_lock;
+	struct   timer_list mc_timer;
+	s8       resend_igmp;
 	s8       kill_timers;
 	s8	 send_grat_arp;
 	s8	 send_unsol_na;
Signature · 6 lines
1.7.2.3

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Signature · 1 line
Flavio
#2 lore raw

PATCH v2 1/1 bonding: rejoin multicast groups on VLANs

It fixes bonding to rejoin multicast groups added
to VLAN devices on top of bonding when a failover
happens.

---
 drivers/net/bonding/bond_main.c |   61 +++++++++++++++++++++++++++++++++-----
 drivers/net/bonding/bonding.h   |    1 +
 2 files changed, 54 insertions(+), 8 deletions(-)
drivers/net/bonding/bond_main.c +53 -8
index 3b16f62..1016c09 100644
@@ -865,18 +865,14 @@ static void bond_mc_del(struct bonding *bond, void *addr)
 }
 
 
-/*
- * Retrieve the list of registered multicast addresses for the bonding
- * device and retransmit an IGMP JOIN request to the current active
- * slave.
- */
-static void bond_resend_igmp_join_requests(struct bonding *bond)
+static void __bond_resend_igmp_join_requests(struct net_device *dev)
 {
 	struct in_device *in_dev;
 	struct ip_mc_list *im;
 
 	rcu_read_lock();
-	in_dev = __in_dev_get_rcu(bond->dev);
+
+	in_dev = __in_dev_get_rcu(dev);
 	if (in_dev) {
 		for (im = in_dev->mc_list; im; im = im->next)
 			ip_mc_rejoin_group(im);
@@ -885,6 +881,45 @@ static void bond_resend_igmp_join_requests(struct bonding *bond)
 	rcu_read_unlock();
 }
 
+
+/*
+ * Retrieve the list of registered multicast addresses for the bonding
+ * device and retransmit an IGMP JOIN request to the current active
+ * slave.
+ */
+static void bond_resend_igmp_join_requests(struct bonding *bond)
+{
+	struct net_device *vlan_dev;
+	struct vlan_entry *vlan;
+
+	read_lock(&bond->lock);
+	if (bond->kill_timers)
+		goto out;
+
+	/* rejoin all groups on bond device */
+	__bond_resend_igmp_join_requests(bond->dev);
+
+	/* rejoin all groups on vlan devices */
+	if (bond->vlgrp) {
+		list_for_each_entry(vlan, &bond->vlan_list, vlan_list) {
+			vlan_dev = vlan_group_get_device(bond->vlgrp,
+							 vlan->vlan_id);
+			if (vlan_dev)
+				__bond_resend_igmp_join_requests(vlan_dev);
+		}
+	}
+
+out:
+	read_unlock(&bond->lock);
+}
+
+void bond_resend_igmp_join_requests_delayed(struct work_struct *work)
+{
+	struct bonding *bond = container_of(work, struct bonding,
+					    mcast_work.work);
+	bond_resend_igmp_join_requests(bond);
+}
+
 /*
  * flush all members of flush->mc_list from device dev->mc_list
  */
@@ -944,7 +979,9 @@ static void bond_mc_swap(struct bonding *bond, struct slave *new_active,
 
 		netdev_for_each_mc_addr(ha, bond->dev)
 			dev_mc_add(new_active->dev, ha->addr);
-		bond_resend_igmp_join_requests(bond);
+
+		/* rejoin multicast groups */
+		queue_delayed_work(bond->wq, &bond->mcast_work, 0);
 	}
 }
 
@@ -3744,6 +3781,9 @@ static int bond_open(struct net_device *bond_dev)
 
 	bond->kill_timers = 0;
 
+	/* multicast retrans */
+	INIT_DELAYED_WORK(&bond->mcast_work, bond_resend_igmp_join_requests_delayed);
+
 	if (bond_is_lb(bond)) {
 		/* bond_alb_initialize must be called before the timer
 		 * is started.
@@ -3828,6 +3868,8 @@ static int bond_close(struct net_device *bond_dev)
 		break;
 	}
 
+	if (delayed_work_pending(&bond->mcast_work))
+		cancel_delayed_work(&bond->mcast_work);
 
 	if (bond_is_lb(bond)) {
 		/* Must be called only after all
@@ -4699,6 +4741,9 @@ static void bond_work_cancel_all(struct bonding *bond)
 	if (bond->params.mode == BOND_MODE_8023AD &&
 	    delayed_work_pending(&bond->ad_work))
 		cancel_delayed_work(&bond->ad_work);
+
+	if (delayed_work_pending(&bond->mcast_work))
+		cancel_delayed_work(&bond->mcast_work);
 }
 
 /*
drivers/net/bonding/bonding.h +1 -0
index c6fdd85..308ed10 100644
@@ -223,6 +223,7 @@ struct bonding {
 	struct   delayed_work arp_work;
 	struct   delayed_work alb_work;
 	struct   delayed_work ad_work;
+	struct   delayed_work mcast_work;
 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 	struct   in6_addr master_ipv6;
 #endif
Signature · 1 line
1.7.3
#3 lore raw
Quoted 10 lines · Flavio Leitner

It fixes bonding to rejoin multicast groups added
to VLAN devices on top of bonding when a failover
happens.

---
 drivers/net/bonding/bond_main.c |   61 +++++++++++++++++++++++++++++++++-----
 drivers/net/bonding/bonding.h   |    1 +
 2 files changed, 54 insertions(+), 8 deletions(-)

[...]

drivers/net/bonding/bond_main.c static void bond_mc_swap(struct bonding *bond, struct slave *new_active, · line 987

Quoted 11 lines
drivers/net/bonding/bond_main.c +3 -1
@@ -944,7 +979,9 @@ static void bond_mc_swap(struct bonding *bond, struct slave *new_active,
 
 		netdev_for_each_mc_addr(ha, bond->dev)
 			dev_mc_add(new_active->dev, ha->addr);
-		bond_resend_igmp_join_requests(bond);
+
+		/* rejoin multicast groups */
+		queue_delayed_work(bond->wq, &bond->mcast_work, 0);
 	}
 }
 

I was hoping that one patch would just make the changes so retransmission was supported on VLANs and the second patch would queue the work and add the tunable for multiple retransmissions, but I guess it wasn't clear enough.

I felt like that would divide the patches up into the bug-fix (VLANs + multicast not working) and the new feature (multiple retransmissions from the workqueue).

I will test these out this morning and make sure things look good.

#4 lore raw
Quoted 29 lines · Andy Gospodarek
Quoted 10 lines · Flavio Leitner

It fixes bonding to rejoin multicast groups added
to VLAN devices on top of bonding when a failover
happens.

---
 drivers/net/bonding/bond_main.c |   61 +++++++++++++++++++++++++++++++++-----
 drivers/net/bonding/bonding.h   |    1 +
 2 files changed, 54 insertions(+), 8 deletions(-)

[...]

drivers/net/bonding/bond_main.c static void bond_mc_swap(struct bonding *bond, struct slave *new_active, · line 987

Quoted 11 lines
drivers/net/bonding/bond_main.c +3 -1
@@ -944,7 +979,9 @@ static void bond_mc_swap(struct bonding *bond, struct slave *new_active,
 
 		netdev_for_each_mc_addr(ha, bond->dev)
 			dev_mc_add(new_active->dev, ha->addr);
-		bond_resend_igmp_join_requests(bond);
+
+		/* rejoin multicast groups */
+		queue_delayed_work(bond->wq, &bond->mcast_work, 0);
 	}
 }
 

I was hoping that one patch would just make the changes so retransmission was supported on VLANs and the second patch would queue the work and add the tunable for multiple retransmissions, but I guess it wasn't clear enough.

Quoted 3 lines

I felt like that would divide the patches up into the bug-fix (VLANs + multicast not working) and the new feature (multiple retransmissions from the workqueue).

I'm not seeing how to that because the first patch changes to send the membership directly and not using timers anymore, so the call trace usually is:

write_lock_bh(&bond->curr_slave_lock); <-- hold lock bond_select_active_slave()
->bond_change_active_slave()

     ->bond_mc_swap()
        ->tx membership reports

then trying to send the IGMP packet directly, it will end up at:
bond_start_xmit()
->bond_xmit_activebackup() (mode=1, for example)

      ->read_lock(&bond->lock);
      ->read_lock(&bond->curr_slave_lock); <-- deadlock
Quoted 1 line

I will test these out this morning and make sure things look good.

I have a better patch sequence which I'm planning post replying to this thread as soon as I finish up testing them. It is still using workqueue though.

--
Flavio

#5 lore raw

PATCH 1/3 bonding: rejoin multicast groups on VLANs

During a failover, the IGMP membership is sent to update the switch restoring the traffic, but it misses groups added to VLAN devices running on top of bonding devices.

This patch changes it to iterate over all VLAN devices on top of it sending IGMP memberships too.

---
 drivers/net/bonding/bond_main.c |   60 +++++++++++++++++++++++++++++++-------
 drivers/net/bonding/bonding.h   |    1 +
 2 files changed, 50 insertions(+), 11 deletions(-)
drivers/net/bonding/bond_main.c +49 -11
index 3b16f62..6554b47 100644
@@ -865,18 +865,13 @@ static void bond_mc_del(struct bonding *bond, void *addr)
 }
 
 
-/*
- * Retrieve the list of registered multicast addresses for the bonding
- * device and retransmit an IGMP JOIN request to the current active
- * slave.
- */
-static void bond_resend_igmp_join_requests(struct bonding *bond)
+static void __bond_resend_igmp_join_requests(struct net_device *dev)
 {
 	struct in_device *in_dev;
 	struct ip_mc_list *im;
 
 	rcu_read_lock();
-	in_dev = __in_dev_get_rcu(bond->dev);
+	in_dev = __in_dev_get_rcu(dev);
 	if (in_dev) {
 		for (im = in_dev->mc_list; im; im = im->next)
 			ip_mc_rejoin_group(im);
@@ -886,6 +881,41 @@ static void bond_resend_igmp_join_requests(struct bonding *bond)
 }
 
 /*
+ * Retrieve the list of registered multicast addresses for the bonding
+ * device and retransmit an IGMP JOIN request to the current active
+ * slave.
+ */
+static void bond_resend_igmp_join_requests(struct bonding *bond)
+{
+	struct net_device *vlan_dev;
+	struct vlan_entry *vlan;
+
+	read_lock(&bond->lock);
+
+	/* rejoin all groups on bond device */
+	__bond_resend_igmp_join_requests(bond->dev);
+
+	/* rejoin all groups on vlan devices */
+	if (bond->vlgrp) {
+		list_for_each_entry(vlan, &bond->vlan_list, vlan_list) {
+			vlan_dev = vlan_group_get_device(bond->vlgrp,
+							 vlan->vlan_id);
+			if (vlan_dev)
+				__bond_resend_igmp_join_requests(vlan_dev);
+		}
+	}
+
+	read_unlock(&bond->lock);
+}
+
+void bond_resend_igmp_join_requests_delayed(struct work_struct *work)
+{
+	struct bonding *bond = container_of(work, struct bonding,
+							mcast_work.work);
+	bond_resend_igmp_join_requests(bond);
+}
+
+/*
  * flush all members of flush->mc_list from device dev->mc_list
  */
 static void bond_mc_list_flush(struct net_device *bond_dev,
@@ -944,7 +974,6 @@ static void bond_mc_swap(struct bonding *bond, struct slave *new_active,
 
 		netdev_for_each_mc_addr(ha, bond->dev)
 			dev_mc_add(new_active->dev, ha->addr);
-		bond_resend_igmp_join_requests(bond);
 	}
 }
 
@@ -1180,9 +1209,11 @@ void bond_change_active_slave(struct bonding *bond, struct slave *new_active)
 		}
 	}
 
-	/* resend IGMP joins since all were sent on curr_active_slave */
-	if (bond->params.mode == BOND_MODE_ROUNDROBIN) {
-		bond_resend_igmp_join_requests(bond);
+	/* resend IGMP joins since active slave has changed or
+	 * all were sent on curr_active_slave */
+	if ((USES_PRIMARY(bond->params.mode) && new_active) ||
+	    bond->params.mode == BOND_MODE_ROUNDROBIN) {
+		queue_delayed_work(bond->wq, &bond->mcast_work, 0);
 	}
 }
 
@@ -3744,6 +3775,8 @@ static int bond_open(struct net_device *bond_dev)
 
 	bond->kill_timers = 0;
 
+	INIT_DELAYED_WORK(&bond->mcast_work, bond_resend_igmp_join_requests_delayed);
+
 	if (bond_is_lb(bond)) {
 		/* bond_alb_initialize must be called before the timer
 		 * is started.
@@ -3828,6 +3861,8 @@ static int bond_close(struct net_device *bond_dev)
 		break;
 	}
 
+	if (delayed_work_pending(&bond->mcast_work))
+		cancel_delayed_work(&bond->mcast_work);
 
 	if (bond_is_lb(bond)) {
 		/* Must be called only after all
@@ -4699,6 +4734,9 @@ static void bond_work_cancel_all(struct bonding *bond)
 	if (bond->params.mode == BOND_MODE_8023AD &&
 	    delayed_work_pending(&bond->ad_work))
 		cancel_delayed_work(&bond->ad_work);
+
+	if (delayed_work_pending(&bond->mcast_work))
+		cancel_delayed_work(&bond->mcast_work);
 }
 
 /*
drivers/net/bonding/bonding.h +1 -0
index c6fdd85..308ed10 100644
@@ -223,6 +223,7 @@ struct bonding {
 	struct   delayed_work arp_work;
 	struct   delayed_work alb_work;
 	struct   delayed_work ad_work;
+	struct   delayed_work mcast_work;
 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 	struct   in6_addr master_ipv6;
 #endif
Signature · 1 line
1.7.0.1
maintainer
#6 lore raw

From: Flavio Leitner <[email protected]>
Date: Tue, 5 Oct 2010 21:23:57 -0300

Quoted 8 lines

During a failover, the IGMP membership is sent to update the switch restoring the traffic, but it misses groups added to VLAN devices running on top of bonding devices.

This patch changes it to iterate over all VLAN devices on top of it sending IGMP memberships too.

This seems to address Andy's feedback properly, and otherwise looks good to me, so applied to net-next-2.6

#7 lore raw
Quoted 7 lines · David Miller

From: Flavio Leitner <[email protected]>
Date: Tue, 5 Oct 2010 21:23:57 -0300

Quoted 4 lines

During a failover, the IGMP membership is sent to update the switch restoring the traffic, but it misses groups added to VLAN devices running on top of bonding devices.

Quoted 7 lines
Quoted 4 lines

This patch changes it to iterate over all VLAN devices on top of it sending IGMP memberships too.

This seems to address Andy's feedback properly, and otherwise looks good to me, so applied to net-next-2.6

Agreed. Thanks for making those changes, Flavio.

#8 lore raw

PATCH 2/3 bonding: fix to rejoin multicast groups immediately

In reply to Flavio Leitner

The IGMP specs states that if the system receives a membership report, it shouldn't send another for the next minute. However, if a link failure happens right after that, the backup slave and the switch connected to this slave will not know about the multicast and the traffic will hang for about a minute.

This patch fixes it to rejoin multicast groups immediately after a failover restoring the multicast traffic.

---
 net/ipv4/igmp.c |   16 ++++++++--------
 1 files changed, 8 insertions(+), 8 deletions(-)
net/ipv4/igmp.c +8 -8
index 2a4bb76..25f3396 100644
@@ -1269,14 +1269,14 @@ void ip_mc_rejoin_group(struct ip_mc_list *im)
 	if (im->multiaddr == IGMP_ALL_HOSTS)
 		return;
 
-	if (IGMP_V1_SEEN(in_dev) || IGMP_V2_SEEN(in_dev)) {
-		igmp_mod_timer(im, IGMP_Initial_Report_Delay);
-		return;
-	}
-	/* else, v3 */
-	im->crcount = in_dev->mr_qrv ? in_dev->mr_qrv :
-		IGMP_Unsolicited_Report_Count;
-	igmp_ifc_event(in_dev);
+	/* a failover is happening and switches
+	 * must be notified immediately */
+	if (IGMP_V1_SEEN(in_dev))
+		igmp_send_report(in_dev, im, IGMP_HOST_MEMBERSHIP_REPORT);
+	else if (IGMP_V2_SEEN(in_dev))
+		igmp_send_report(in_dev, im, IGMPV2_HOST_MEMBERSHIP_REPORT);
+	else
+		igmp_send_report(in_dev, im, IGMPV3_HOST_MEMBERSHIP_REPORT);
 #endif
 }
 EXPORT_SYMBOL(ip_mc_rejoin_group);
Signature · 1 line
1.7.0.1
maintainer Applied
#9 lore raw

Re: [PATCH 2/3] bonding: fix to rejoin multicast groups immediately

From: Flavio Leitner <[email protected]>
Date: Tue, 5 Oct 2010 21:23:58 -0300

Quoted 11 lines

The IGMP specs states that if the system receives a membership report, it shouldn't send another for the next minute. However, if a link failure happens right after that, the backup slave and the switch connected to this slave will not know about the multicast and the traffic will hang for about a minute.

This patch fixes it to rejoin multicast groups immediately after a failover restoring the multicast traffic.

Much better commit message :-)

Applied to net-next-2.6

#10 lore raw

PATCH 3/3 bonding: add retransmit membership reports tunable

In reply to Flavio Leitner

Allow sysadmins to configure the number of multicast membership report sent on a link failure event.

---
 Documentation/networking/bonding.txt |    8 ++++++
 drivers/net/bonding/bond_main.c      |   15 +++++++++++
 drivers/net/bonding/bond_sysfs.c     |   44 ++++++++++++++++++++++++++++++++++
 drivers/net/bonding/bonding.h        |    2 +
 include/linux/if_bonding.h           |    3 ++
 5 files changed, 72 insertions(+), 0 deletions(-)
Documentation/networking/bonding.txt +8 -0
index d2b62b7..5dc6387 100644
@@ -765,6 +765,14 @@ xmit_hash_policy
 	does not exist, and the layer2 policy is the only policy.  The
 	layer2+3 value was added for bonding version 3.2.2.
 
+resend_igmp
+
+	Specifies the number of IGMP membership reports to be issued after
+	a failover event. One membership report is issued immediately after
+	the failover, subsequent packets are sent in each 200ms interval.
+
+	The valid range is 0 - 255; the default value is 1. This option
+	was added for bonding version 3.7.0.
 
 3. Configuring Bonding Devices
 ==============================
drivers/net/bonding/bond_main.c +15 -0
index 6554b47..19206ba 100644
@@ -109,6 +109,7 @@ static char *arp_validate;
 static char *fail_over_mac;
 static int all_slaves_active = 0;
 static struct bond_params bonding_defaults;
+static int resend_igmp = BOND_DEFAULT_RESEND_IGMP;
 
 module_param(max_bonds, int, 0);
 MODULE_PARM_DESC(max_bonds, "Max number of bonded devices");
@@ -163,6 +164,8 @@ module_param(all_slaves_active, int, 0);
 MODULE_PARM_DESC(all_slaves_active, "Keep all frames received on an interface"
 				     "by setting active flag for all slaves.  "
 				     "0 for never (default), 1 for always.");
+module_param(resend_igmp, int, 0);
+MODULE_PARM_DESC(resend_igmp, "Number of IGMP membership reports to send on link failure");
 
 /*----------------------------- Global variables ----------------------------*/
 
@@ -905,6 +908,9 @@ static void bond_resend_igmp_join_requests(struct bonding *bond)
 		}
 	}
 
+	if (--bond->igmp_retrans > 0)
+		queue_delayed_work(bond->wq, &bond->mcast_work, HZ/5);
+
 	read_unlock(&bond->lock);
 }
 
@@ -1213,6 +1219,7 @@ void bond_change_active_slave(struct bonding *bond, struct slave *new_active)
 	 * all were sent on curr_active_slave */
 	if ((USES_PRIMARY(bond->params.mode) && new_active) ||
 	    bond->params.mode == BOND_MODE_ROUNDROBIN) {
+		bond->igmp_retrans = bond->params.resend_igmp;
 		queue_delayed_work(bond->wq, &bond->mcast_work, 0);
 	}
 }
@@ -4929,6 +4936,13 @@ static int bond_check_params(struct bond_params *params)
 		all_slaves_active = 0;
 	}
 
+	if (resend_igmp < 0 || resend_igmp > 255) {
+		pr_warning("Warning: resend_igmp (%d) should be between "
+			   "0 and 255, resetting to %d\n",
+			   resend_igmp, BOND_DEFAULT_RESEND_IGMP);
+		resend_igmp = BOND_DEFAULT_RESEND_IGMP;
+	}
+
 	/* reset values for TLB/ALB */
 	if ((bond_mode == BOND_MODE_TLB) ||
 	    (bond_mode == BOND_MODE_ALB)) {
@@ -5101,6 +5115,7 @@ static int bond_check_params(struct bond_params *params)
 	params->fail_over_mac = fail_over_mac_value;
 	params->tx_queues = tx_queues;
 	params->all_slaves_active = all_slaves_active;
+	params->resend_igmp = resend_igmp;
 
 	if (primary) {
 		strncpy(params->primary, primary, IFNAMSIZ);
drivers/net/bonding/bond_sysfs.c +44 -0

56 lines not on this page · raw message

drivers/net/bonding/bonding.h +2 -0
index 308ed10..c15f213 100644
@@ -136,6 +136,7 @@ struct bond_params {
 	__be32 arp_targets[BOND_MAX_ARP_TARGETS];
 	int tx_queues;
 	int all_slaves_active;
+	int resend_igmp;
 };
 
 struct bond_parm_tbl {
@@ -202,6 +203,7 @@ struct bonding {
 	s8	 send_grat_arp;
 	s8	 send_unsol_na;
 	s8	 setup_by_slave;
+	s8       igmp_retrans;
 #ifdef CONFIG_PROC_FS
 	struct   proc_dir_entry *proc_entry;
 	char     proc_file_name[IFNAMSIZ];
include/linux/if_bonding.h +3 -0
index 2c79943..a17edda 100644
@@ -84,6 +84,9 @@
 #define BOND_DEFAULT_MAX_BONDS  1   /* Default maximum number of devices to support */
 
 #define BOND_DEFAULT_TX_QUEUES 16   /* Default number of tx queues per device */
+
+#define BOND_DEFAULT_RESEND_IGMP	1 /* Default number of IGMP membership reports */
+
 /* hashing types */
 #define BOND_XMIT_POLICY_LAYER2		0 /* layer 2 (MAC only), default */
 #define BOND_XMIT_POLICY_LAYER34	1 /* layer 3+4 (IP ^ (TCP || UDP)) */
Signature · 1 line
1.7.0.1
maintainer
#11 lore raw

Re: [PATCH 3/3] bonding: add retransmit membership reports tunable

From: Flavio Leitner <[email protected]>
Date: Tue, 5 Oct 2010 21:23:59 -0300

Quoted 4 lines

Allow sysadmins to configure the number of multicast membership report sent on a link failure event.

Also applied to net-next-2.6, thanks.

#12 lore raw

PATCH 1/1 bonding: add retransmit membership reports tunable

In reply to Flavio Leitner

Allow admins to adjust the number of multicast membership reports sent on a link failure.

---
 drivers/net/bonding/bond_main.c  |   14 ++++++++++++
 drivers/net/bonding/bond_sysfs.c |   44 ++++++++++++++++++++++++++++++++++++++
 drivers/net/bonding/bonding.h    |    2 +
 include/linux/if_bonding.h       |    3 ++
 4 files changed, 63 insertions(+), 0 deletions(-)
drivers/net/bonding/bond_main.c +14 -0
index 1016c09..60bfd33 100644
@@ -109,6 +109,7 @@ static char *arp_validate;
 static char *fail_over_mac;
 static int all_slaves_active = 0;
 static struct bond_params bonding_defaults;
+static int resend_igmp = BOND_DEFAULT_RESEND_IGMP;
 
 module_param(max_bonds, int, 0);
 MODULE_PARM_DESC(max_bonds, "Max number of bonded devices");
@@ -163,6 +164,8 @@ module_param(all_slaves_active, int, 0);
 MODULE_PARM_DESC(all_slaves_active, "Keep all frames received on an interface"
 				     "by setting active flag for all slaves.  "
 				     "0 for never (default), 1 for always.");
+module_param(resend_igmp, int, 0);
+MODULE_PARM_DESC(resend_igmp, "Number of IGMP membership reports to send on link failure");
 
 /*----------------------------- Global variables ----------------------------*/
 
@@ -909,6 +912,9 @@ static void bond_resend_igmp_join_requests(struct bonding *bond)
 		}
 	}
 
+	if (--bond->igmp_retrans > 0)
+		queue_delayed_work(bond->wq, &bond->mcast_work, HZ/5);
+
 out:
 	read_unlock(&bond->lock);
 }
@@ -981,6 +987,7 @@ static void bond_mc_swap(struct bonding *bond, struct slave *new_active,
 			dev_mc_add(new_active->dev, ha->addr);
 
 		/* rejoin multicast groups */
+		bond->igmp_retrans = bond->params.resend_igmp;
 		queue_delayed_work(bond->wq, &bond->mcast_work, 0);
 	}
 }
@@ -4936,6 +4943,12 @@ static int bond_check_params(struct bond_params *params)
 		all_slaves_active = 0;
 	}
 
+	if (resend_igmp < 0 || resend_igmp > 255) {
+		pr_warning("Warning: resend_igmp (%d) should be between "
+			   "0 and 255, resetting to %d\n",
+			   resend_igmp, BOND_DEFAULT_RESEND_IGMP);
+		resend_igmp = BOND_DEFAULT_RESEND_IGMP;
+	}
 	/* reset values for TLB/ALB */
 	if ((bond_mode == BOND_MODE_TLB) ||
 	    (bond_mode == BOND_MODE_ALB)) {
@@ -5108,6 +5121,7 @@ static int bond_check_params(struct bond_params *params)
 	params->fail_over_mac = fail_over_mac_value;
 	params->tx_queues = tx_queues;
 	params->all_slaves_active = all_slaves_active;
+	params->resend_igmp = resend_igmp;
 
 	if (primary) {
 		strncpy(params->primary, primary, IFNAMSIZ);
drivers/net/bonding/bond_sysfs.c +44 -0

56 lines not on this page · raw message

drivers/net/bonding/bonding.h +2 -0
index 308ed10..c49bdb0 100644
@@ -136,6 +136,7 @@ struct bond_params {
 	__be32 arp_targets[BOND_MAX_ARP_TARGETS];
 	int tx_queues;
 	int all_slaves_active;
+	int resend_igmp;
 };
 
 struct bond_parm_tbl {
@@ -198,6 +199,7 @@ struct bonding {
 	s32      slave_cnt; /* never change this value outside the attach/detach wrappers */
 	rwlock_t lock;
 	rwlock_t curr_slave_lock;
+	s8       igmp_retrans;
 	s8       kill_timers;
 	s8	 send_grat_arp;
 	s8	 send_unsol_na;
include/linux/if_bonding.h +3 -0
index 2c799437..d2f78b7 100644
@@ -84,6 +84,9 @@
 #define BOND_DEFAULT_MAX_BONDS  1   /* Default maximum number of devices to support */
 
 #define BOND_DEFAULT_TX_QUEUES 16   /* Default number of tx queues per device */
+
+#define BOND_DEFAULT_RESEND_IGMP	1 /* Default number of IGMP membership reports
+					     to resend on link failure. */
 /* hashing types */
 #define BOND_XMIT_POLICY_LAYER2		0 /* layer 2 (MAC only), default */
 #define BOND_XMIT_POLICY_LAYER34	1 /* layer 3+4 (IP ^ (TCP || UDP)) */
Signature · 1 line
1.7.3
#13 lore raw
In reply to Flavio Leitner
Quoted 7 lines · Flavio Leitner

It fixes bonding to rejoin multicast groups added
to VLAN devices on top of bonding when a failover
happens.

The first packet may be discarded, so the timer
assure that at least 3 Reports are sent.

Good find, Flavio. Clearly the fact that multicast membership is broken needs to be fixed, but I would rather not see timers used at all. We worked hard in the past to eliminate timers for several reasons, so I would rather see a workqueue used.

I also don't like retransmitting the membership report 3 times when it may not be needed. Though many switches can handle it, the cost of receiving and processing what might be a large list of multicast addresses every 200ms for 600ms doesn't seem ideal. It also feels like a hack. :)

If retransmission of the membership reports is a requirement for some users, I would rather see it as a configuration option.

Maybe something like this?

drivers/net/bonding/bond_main.c +67 -8
index 3b16f62..b7b4a74 100644
@@ -109,6 +109,7 @@ static char *arp_validate;
 static char *fail_over_mac;
 static int all_slaves_active = 0;
 static struct bond_params bonding_defaults;
+static int resend_igmp = BOND_DEFAULT_RESEND_IGMP;
 
 module_param(max_bonds, int, 0);
 MODULE_PARM_DESC(max_bonds, "Max number of bonded devices");
@@ -163,6 +164,8 @@ module_param(all_slaves_active, int, 0);
 MODULE_PARM_DESC(all_slaves_active, "Keep all frames received on an interface"
 				     "by setting active flag for all slaves.  "
 				     "0 for never (default), 1 for always.");
+module_param(resend_igmp, int, 0);
+MODULE_PARM_DESC(resend_igmp, "Number of IGMP membership reports to send on link failure");
 
 /*----------------------------- Global variables ----------------------------*/
 
@@ -865,18 +868,14 @@ static void bond_mc_del(struct bonding *bond, void *addr)
 }
 
 
-/*
- * Retrieve the list of registered multicast addresses for the bonding
- * device and retransmit an IGMP JOIN request to the current active
- * slave.
- */
-static void bond_resend_igmp_join_requests(struct bonding *bond)
+static void __bond_resend_igmp_join_requests(struct net_device *dev)
 {
 	struct in_device *in_dev;
 	struct ip_mc_list *im;
 
 	rcu_read_lock();
-	in_dev = __in_dev_get_rcu(bond->dev);
+
+	in_dev = __in_dev_get_rcu(dev);
 	if (in_dev) {
 		for (im = in_dev->mc_list; im; im = im->next)
 			ip_mc_rejoin_group(im);
@@ -885,6 +884,48 @@ static void bond_resend_igmp_join_requests(struct bonding *bond)
 	rcu_read_unlock();
 }
 
+
+/*
+ * Retrieve the list of registered multicast addresses for the bonding
+ * device and retransmit an IGMP JOIN request to the current active
+ * slave.
+ */
+static void bond_resend_igmp_join_requests(struct bonding *bond)
+{
+	struct net_device *vlan_dev;
+	struct vlan_entry *vlan;
+
+	read_lock(&bond->lock);
+	if (bond->kill_timers)
+		goto out;
+
+	/* rejoin all groups on bond device */
+	__bond_resend_igmp_join_requests(bond->dev);
+
+	/* rejoin all groups on vlan devices */
+	if (bond->vlgrp) {
+		list_for_each_entry(vlan, &bond->vlan_list, vlan_list) {
+			vlan_dev = vlan_group_get_device(bond->vlgrp,
+							 vlan->vlan_id);
+			if (vlan_dev)
+				__bond_resend_igmp_join_requests(vlan_dev);
+		}
+	}
+
+	if (--bond->igmp_retrans > 0)
+		queue_delayed_work(bond->wq, &bond->mcast_work, HZ/5);
+
+out:
+	read_unlock(&bond->lock);
+}
+
+void bond_resend_igmp_join_requests_delayed(struct work_struct *work)
+{
+	struct bonding *bond = container_of(work, struct bonding,
+					    mcast_work.work);
+	bond_resend_igmp_join_requests(bond);
+}
+
 /*
  * flush all members of flush->mc_list from device dev->mc_list
  */
@@ -944,7 +985,10 @@ static void bond_mc_swap(struct bonding *bond, struct slave *new_active,
 
 		netdev_for_each_mc_addr(ha, bond->dev)
 			dev_mc_add(new_active->dev, ha->addr);
-		bond_resend_igmp_join_requests(bond);
+
+		/* rejoin multicast groups */
+		bond->igmp_retrans = bond->params.resend_igmp;
+		queue_delayed_work(bond->wq, &bond->mcast_work, 0);
 	}
 }
 
@@ -3744,6 +3788,9 @@ static int bond_open(struct net_device *bond_dev)
 
 	bond->kill_timers = 0;
 
+	/* multicast retrans */
+	INIT_DELAYED_WORK(&bond->mcast_work, bond_resend_igmp_join_requests_delayed);
+
 	if (bond_is_lb(bond)) {
 		/* bond_alb_initialize must be called before the timer
 		 * is started.
@@ -3828,6 +3875,8 @@ static int bond_close(struct net_device *bond_dev)
 		break;
 	}
 
+	if (delayed_work_pending(&bond->mcast_work))
+		cancel_delayed_work(&bond->ad_work);
 
 	if (bond_is_lb(bond)) {
 		/* Must be called only after all
@@ -4699,6 +4748,9 @@ static void bond_work_cancel_all(struct bonding *bond)
 	if (bond->params.mode == BOND_MODE_8023AD &&
 	    delayed_work_pending(&bond->ad_work))
 		cancel_delayed_work(&bond->ad_work);
+
+	if (delayed_work_pending(&bond->mcast_work))
+		cancel_delayed_work(&bond->ad_work);
 }
 
 /*
@@ -4891,6 +4943,12 @@ static int bond_check_params(struct bond_params *params)
 		all_slaves_active = 0;
 	}
 
+	if (resend_igmp < 0 || resend_igmp > 255) {
+		pr_warning("Warning: resend_igmp (%d) should be between "
+			   "0 and 255, resetting to %d\n",
+			   resend_igmp, BOND_DEFAULT_RESEND_IGMP);
+		resend_igmp = BOND_DEFAULT_RESEND_IGMP;
+	}
 	/* reset values for TLB/ALB */
 	if ((bond_mode == BOND_MODE_TLB) ||
 	    (bond_mode == BOND_MODE_ALB)) {
@@ -5063,6 +5121,7 @@ static int bond_check_params(struct bond_params *params)
 	params->fail_over_mac = fail_over_mac_value;
 	params->tx_queues = tx_queues;
 	params->all_slaves_active = all_slaves_active;
+	params->resend_igmp = resend_igmp;
 
 	if (primary) {
 		strncpy(params->primary, primary, IFNAMSIZ);
drivers/net/bonding/bonding.h +3 -0
index c6fdd85..c49bdb0 100644
@@ -136,6 +136,7 @@ struct bond_params {
 	__be32 arp_targets[BOND_MAX_ARP_TARGETS];
 	int tx_queues;
 	int all_slaves_active;
+	int resend_igmp;
 };
 
 struct bond_parm_tbl {
@@ -198,6 +199,7 @@ struct bonding {
 	s32      slave_cnt; /* never change this value outside the attach/detach wrappers */
 	rwlock_t lock;
 	rwlock_t curr_slave_lock;
+	s8       igmp_retrans;
 	s8       kill_timers;
 	s8	 send_grat_arp;
 	s8	 send_unsol_na;
@@ -223,6 +225,7 @@ struct bonding {
 	struct   delayed_work arp_work;
 	struct   delayed_work alb_work;
 	struct   delayed_work ad_work;
+	struct   delayed_work mcast_work;
 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 	struct   in6_addr master_ipv6;
 #endif
include/linux/if_bonding.h +3 -0
index 2c79943..d2f78b7 100644
@@ -84,6 +84,9 @@
 #define BOND_DEFAULT_MAX_BONDS  1   /* Default maximum number of devices to support */
 
 #define BOND_DEFAULT_TX_QUEUES 16   /* Default number of tx queues per device */
+
+#define BOND_DEFAULT_RESEND_IGMP	1 /* Default number of IGMP membership reports
+					     to resend on link failure. */
 /* hashing types */
 #define BOND_XMIT_POLICY_LAYER2		0 /* layer 2 (MAC only), default */
 #define BOND_XMIT_POLICY_LAYER34	1 /* layer 3+4 (IP ^ (TCP || UDP)) */
#14 lore raw
Quoted 5 lines · Andy Gospodarek
Quoted 4 lines · Flavio Leitner

It fixes bonding to rejoin multicast groups added
to VLAN devices on top of bonding when a failover
happens.

Quoted 8 lines
Quoted 3 lines

The first packet may be discarded, so the timer
assure that at least 3 Reports are sent.

Good find, Flavio. Clearly the fact that multicast membership is broken needs to be fixed, but I would rather not see timers used at all. We worked hard in the past to eliminate timers for several reasons, so I would rather see a workqueue used.

I noticed that the code is using workqueues now, just thought a simple thing which may run couple times would fit perfectly with a simple timer.

Quoted 5 lines

I also don't like retransmitting the membership report 3 times when it may not be needed. Though many switches can handle it, the cost of receiving and processing what might be a large list of multicast addresses every 200ms for 600ms doesn't seem ideal. It also feels like a hack. :)

Definitely a parameter is much better, but I wasn't sure about the patch approach so I was expecting a review like this and then do the refinements needed. Better to post early, right? :)

I see your point to change the default to one membership report, but we can't assure during a failover if everything has been received. Also, it isn't supposed to keep failing flooding the network, so I would rather have couple membership reports being send than watch an important multicast application failing.

Perhaps 3 is too much, but one sounds too few to me.

what you think?

Flavio

Quoted 202 lines

If retransmission of the membership reports is a requirement for some users, I would rather see it as a configuration option.

Maybe something like this?

drivers/net/bonding/bond_main.c +67 -8
index 3b16f62..b7b4a74 100644
@@ -109,6 +109,7 @@ static char *arp_validate;
 static char *fail_over_mac;
 static int all_slaves_active = 0;
 static struct bond_params bonding_defaults;
+static int resend_igmp = BOND_DEFAULT_RESEND_IGMP;
 
 module_param(max_bonds, int, 0);
 MODULE_PARM_DESC(max_bonds, "Max number of bonded devices");
@@ -163,6 +164,8 @@ module_param(all_slaves_active, int, 0);
 MODULE_PARM_DESC(all_slaves_active, "Keep all frames received on an interface"
 				     "by setting active flag for all slaves.  "
 				     "0 for never (default), 1 for always.");
+module_param(resend_igmp, int, 0);
+MODULE_PARM_DESC(resend_igmp, "Number of IGMP membership reports to send on link failure");
 
 /*----------------------------- Global variables ----------------------------*/
 
@@ -865,18 +868,14 @@ static void bond_mc_del(struct bonding *bond, void *addr)
 }
 
 
-/*
- * Retrieve the list of registered multicast addresses for the bonding
- * device and retransmit an IGMP JOIN request to the current active
- * slave.
- */
-static void bond_resend_igmp_join_requests(struct bonding *bond)
+static void __bond_resend_igmp_join_requests(struct net_device *dev)
 {
 	struct in_device *in_dev;
 	struct ip_mc_list *im;
 
 	rcu_read_lock();
-	in_dev = __in_dev_get_rcu(bond->dev);
+
+	in_dev = __in_dev_get_rcu(dev);
 	if (in_dev) {
 		for (im = in_dev->mc_list; im; im = im->next)
 			ip_mc_rejoin_group(im);
@@ -885,6 +884,48 @@ static void bond_resend_igmp_join_requests(struct bonding *bond)
 	rcu_read_unlock();
 }
 
+
+/*
+ * Retrieve the list of registered multicast addresses for the bonding
+ * device and retransmit an IGMP JOIN request to the current active
+ * slave.
+ */
+static void bond_resend_igmp_join_requests(struct bonding *bond)
+{
+	struct net_device *vlan_dev;
+	struct vlan_entry *vlan;
+
+	read_lock(&bond->lock);
+	if (bond->kill_timers)
+		goto out;
+
+	/* rejoin all groups on bond device */
+	__bond_resend_igmp_join_requests(bond->dev);
+
+	/* rejoin all groups on vlan devices */
+	if (bond->vlgrp) {
+		list_for_each_entry(vlan, &bond->vlan_list, vlan_list) {
+			vlan_dev = vlan_group_get_device(bond->vlgrp,
+							 vlan->vlan_id);
+			if (vlan_dev)
+				__bond_resend_igmp_join_requests(vlan_dev);
+		}
+	}
+
+	if (--bond->igmp_retrans > 0)
+		queue_delayed_work(bond->wq, &bond->mcast_work, HZ/5);
+
+out:
+	read_unlock(&bond->lock);
+}
+
+void bond_resend_igmp_join_requests_delayed(struct work_struct *work)
+{
+	struct bonding *bond = container_of(work, struct bonding,
+					    mcast_work.work);
+	bond_resend_igmp_join_requests(bond);
+}
+
 /*
  * flush all members of flush->mc_list from device dev->mc_list
  */
@@ -944,7 +985,10 @@ static void bond_mc_swap(struct bonding *bond, struct slave *new_active,
 
 		netdev_for_each_mc_addr(ha, bond->dev)
 			dev_mc_add(new_active->dev, ha->addr);
-		bond_resend_igmp_join_requests(bond);
+
+		/* rejoin multicast groups */
+		bond->igmp_retrans = bond->params.resend_igmp;
+		queue_delayed_work(bond->wq, &bond->mcast_work, 0);
 	}
 }
 
@@ -3744,6 +3788,9 @@ static int bond_open(struct net_device *bond_dev)
 
 	bond->kill_timers = 0;
 
+	/* multicast retrans */
+	INIT_DELAYED_WORK(&bond->mcast_work, bond_resend_igmp_join_requests_delayed);
+
 	if (bond_is_lb(bond)) {
 		/* bond_alb_initialize must be called before the timer
 		 * is started.
@@ -3828,6 +3875,8 @@ static int bond_close(struct net_device *bond_dev)
 		break;
 	}
 
+	if (delayed_work_pending(&bond->mcast_work))
+		cancel_delayed_work(&bond->ad_work);
 
 	if (bond_is_lb(bond)) {
 		/* Must be called only after all
@@ -4699,6 +4748,9 @@ static void bond_work_cancel_all(struct bonding *bond)
 	if (bond->params.mode == BOND_MODE_8023AD &&
 	    delayed_work_pending(&bond->ad_work))
 		cancel_delayed_work(&bond->ad_work);
+
+	if (delayed_work_pending(&bond->mcast_work))
+		cancel_delayed_work(&bond->ad_work);
 }
 
 /*
@@ -4891,6 +4943,12 @@ static int bond_check_params(struct bond_params *params)
 		all_slaves_active = 0;
 	}
 
+	if (resend_igmp < 0 || resend_igmp > 255) {
+		pr_warning("Warning: resend_igmp (%d) should be between "
+			   "0 and 255, resetting to %d\n",
+			   resend_igmp, BOND_DEFAULT_RESEND_IGMP);
+		resend_igmp = BOND_DEFAULT_RESEND_IGMP;
+	}
 	/* reset values for TLB/ALB */
 	if ((bond_mode == BOND_MODE_TLB) ||
 	    (bond_mode == BOND_MODE_ALB)) {
@@ -5063,6 +5121,7 @@ static int bond_check_params(struct bond_params *params)
 	params->fail_over_mac = fail_over_mac_value;
 	params->tx_queues = tx_queues;
 	params->all_slaves_active = all_slaves_active;
+	params->resend_igmp = resend_igmp;
 
 	if (primary) {
 		strncpy(params->primary, primary, IFNAMSIZ);
drivers/net/bonding/bonding.h +3 -0
index c6fdd85..c49bdb0 100644
@@ -136,6 +136,7 @@ struct bond_params {
 	__be32 arp_targets[BOND_MAX_ARP_TARGETS];
 	int tx_queues;
 	int all_slaves_active;
+	int resend_igmp;
 };
 
 struct bond_parm_tbl {
@@ -198,6 +199,7 @@ struct bonding {
 	s32      slave_cnt; /* never change this value outside the attach/detach wrappers */
 	rwlock_t lock;
 	rwlock_t curr_slave_lock;
+	s8       igmp_retrans;
 	s8       kill_timers;
 	s8	 send_grat_arp;
 	s8	 send_unsol_na;
@@ -223,6 +225,7 @@ struct bonding {
 	struct   delayed_work arp_work;
 	struct   delayed_work alb_work;
 	struct   delayed_work ad_work;
+	struct   delayed_work mcast_work;
 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 	struct   in6_addr master_ipv6;
 #endif
include/linux/if_bonding.h +3 -0
index 2c79943..d2f78b7 100644
@@ -84,6 +84,9 @@
 #define BOND_DEFAULT_MAX_BONDS  1   /* Default maximum number of devices to support */
 
 #define BOND_DEFAULT_TX_QUEUES 16   /* Default number of tx queues per device */
+
+#define BOND_DEFAULT_RESEND_IGMP	1 /* Default number of IGMP membership reports
+					     to resend on link failure. */
 /* hashing types */
 #define BOND_XMIT_POLICY_LAYER2		0 /* layer 2 (MAC only), default */
 #define BOND_XMIT_POLICY_LAYER34	1 /* layer 3+4 (IP ^ (TCP || UDP)) */
Signature · 1 line
Flavio
#15 lore raw
Quoted 10 lines · Flavio Leitner
Quoted 9 lines · Andy Gospodarek
Quoted 7 lines · Flavio Leitner

It fixes bonding to rejoin multicast groups added
to VLAN devices on top of bonding when a failover
happens.

The first packet may be discarded, so the timer
assure that at least 3 Reports are sent.

Quoted 9 lines
Quoted 4 lines

Good find, Flavio. Clearly the fact that multicast membership is broken needs to be fixed, but I would rather not see timers used at all. We worked hard in the past to eliminate timers for several reasons, so I would rather see a workqueue used.

I noticed that the code is using workqueues now, just thought a simple thing which may run couple times would fit perfectly with a simple timer.

Timers runs in softirq context, so I'd rather not add code that takes locks and runs in softirq context.

Quoted 17 lines
Quoted 6 lines

I also don't like retransmitting the membership report 3 times when it may not be needed. Though many switches can handle it, the cost of receiving and processing what might be a large list of multicast addresses every 200ms for 600ms doesn't seem ideal. It also feels like a hack. :)

Definitely a parameter is much better, but I wasn't sure about the patch approach so I was expecting a review like this and then do the refinements needed. Better to post early, right? :)

I see your point to change the default to one membership report, but we can't assure during a failover if everything has been received. Also, it isn't supposed to keep failing flooding the network, so I would rather have couple membership reports being send than watch an important multicast application failing.

Quoted 4 lines

Perhaps 3 is too much, but one sounds too few to me.

what you think?

Adding a tunable parameter allows the administrator to decide how many is enough. I would rather keep the default at one and add the tunable parameter (which needs to be added to bond_sysfs.c to be effective).

I have not heard loud complaints about only sending one since the code to send retransmits of membership reports was added a few years ago, so I'm inclined to think it is working well for most users (or no one is using bonding).

Maybe it would be best to break this into 2 patches. One that simply fixes the failover code so it works with VLANs (that could be done easily today) and another patch that can add the code to send multiple retransmits. Would you be willing to do that?

Quoted 211 lines

Flavio

Quoted 202 lines

If retransmission of the membership reports is a requirement for some users, I would rather see it as a configuration option.

Maybe something like this?

drivers/net/bonding/bond_main.c +67 -8
index 3b16f62..b7b4a74 100644
@@ -109,6 +109,7 @@ static char *arp_validate;
 static char *fail_over_mac;
 static int all_slaves_active = 0;
 static struct bond_params bonding_defaults;
+static int resend_igmp = BOND_DEFAULT_RESEND_IGMP;
 
 module_param(max_bonds, int, 0);
 MODULE_PARM_DESC(max_bonds, "Max number of bonded devices");
@@ -163,6 +164,8 @@ module_param(all_slaves_active, int, 0);
 MODULE_PARM_DESC(all_slaves_active, "Keep all frames received on an interface"
 				     "by setting active flag for all slaves.  "
 				     "0 for never (default), 1 for always.");
+module_param(resend_igmp, int, 0);
+MODULE_PARM_DESC(resend_igmp, "Number of IGMP membership reports to send on link failure");
 
 /*----------------------------- Global variables ----------------------------*/
 
@@ -865,18 +868,14 @@ static void bond_mc_del(struct bonding *bond, void *addr)
 }
 
 
-/*
- * Retrieve the list of registered multicast addresses for the bonding
- * device and retransmit an IGMP JOIN request to the current active
- * slave.
- */
-static void bond_resend_igmp_join_requests(struct bonding *bond)
+static void __bond_resend_igmp_join_requests(struct net_device *dev)
 {
 	struct in_device *in_dev;
 	struct ip_mc_list *im;
 
 	rcu_read_lock();
-	in_dev = __in_dev_get_rcu(bond->dev);
+
+	in_dev = __in_dev_get_rcu(dev);
 	if (in_dev) {
 		for (im = in_dev->mc_list; im; im = im->next)
 			ip_mc_rejoin_group(im);
@@ -885,6 +884,48 @@ static void bond_resend_igmp_join_requests(struct bonding *bond)
 	rcu_read_unlock();
 }
 
+
+/*
+ * Retrieve the list of registered multicast addresses for the bonding
+ * device and retransmit an IGMP JOIN request to the current active
+ * slave.
+ */
+static void bond_resend_igmp_join_requests(struct bonding *bond)
+{
+	struct net_device *vlan_dev;
+	struct vlan_entry *vlan;
+
+	read_lock(&bond->lock);
+	if (bond->kill_timers)
+		goto out;
+
+	/* rejoin all groups on bond device */
+	__bond_resend_igmp_join_requests(bond->dev);
+
+	/* rejoin all groups on vlan devices */
+	if (bond->vlgrp) {
+		list_for_each_entry(vlan, &bond->vlan_list, vlan_list) {
+			vlan_dev = vlan_group_get_device(bond->vlgrp,
+							 vlan->vlan_id);
+			if (vlan_dev)
+				__bond_resend_igmp_join_requests(vlan_dev);
+		}
+	}
+
+	if (--bond->igmp_retrans > 0)
+		queue_delayed_work(bond->wq, &bond->mcast_work, HZ/5);
+
+out:
+	read_unlock(&bond->lock);
+}
+
+void bond_resend_igmp_join_requests_delayed(struct work_struct *work)
+{
+	struct bonding *bond = container_of(work, struct bonding,
+					    mcast_work.work);
+	bond_resend_igmp_join_requests(bond);
+}
+
 /*
  * flush all members of flush->mc_list from device dev->mc_list
  */
@@ -944,7 +985,10 @@ static void bond_mc_swap(struct bonding *bond, struct slave *new_active,
 
 		netdev_for_each_mc_addr(ha, bond->dev)
 			dev_mc_add(new_active->dev, ha->addr);
-		bond_resend_igmp_join_requests(bond);
+
+		/* rejoin multicast groups */
+		bond->igmp_retrans = bond->params.resend_igmp;
+		queue_delayed_work(bond->wq, &bond->mcast_work, 0);
 	}
 }
 
@@ -3744,6 +3788,9 @@ static int bond_open(struct net_device *bond_dev)
 
 	bond->kill_timers = 0;
 
+	/* multicast retrans */
+	INIT_DELAYED_WORK(&bond->mcast_work, bond_resend_igmp_join_requests_delayed);
+
 	if (bond_is_lb(bond)) {
 		/* bond_alb_initialize must be called before the timer
 		 * is started.
@@ -3828,6 +3875,8 @@ static int bond_close(struct net_device *bond_dev)
 		break;
 	}
 
+	if (delayed_work_pending(&bond->mcast_work))
+		cancel_delayed_work(&bond->ad_work);
 
 	if (bond_is_lb(bond)) {
 		/* Must be called only after all
@@ -4699,6 +4748,9 @@ static void bond_work_cancel_all(struct bonding *bond)
 	if (bond->params.mode == BOND_MODE_8023AD &&
 	    delayed_work_pending(&bond->ad_work))
 		cancel_delayed_work(&bond->ad_work);
+
+	if (delayed_work_pending(&bond->mcast_work))
+		cancel_delayed_work(&bond->ad_work);
 }
 
 /*
@@ -4891,6 +4943,12 @@ static int bond_check_params(struct bond_params *params)
 		all_slaves_active = 0;
 	}
 
+	if (resend_igmp < 0 || resend_igmp > 255) {
+		pr_warning("Warning: resend_igmp (%d) should be between "
+			   "0 and 255, resetting to %d\n",
+			   resend_igmp, BOND_DEFAULT_RESEND_IGMP);
+		resend_igmp = BOND_DEFAULT_RESEND_IGMP;
+	}
 	/* reset values for TLB/ALB */
 	if ((bond_mode == BOND_MODE_TLB) ||
 	    (bond_mode == BOND_MODE_ALB)) {
@@ -5063,6 +5121,7 @@ static int bond_check_params(struct bond_params *params)
 	params->fail_over_mac = fail_over_mac_value;
 	params->tx_queues = tx_queues;
 	params->all_slaves_active = all_slaves_active;
+	params->resend_igmp = resend_igmp;
 
 	if (primary) {
 		strncpy(params->primary, primary, IFNAMSIZ);
drivers/net/bonding/bonding.h +3 -0
index c6fdd85..c49bdb0 100644
@@ -136,6 +136,7 @@ struct bond_params {
 	__be32 arp_targets[BOND_MAX_ARP_TARGETS];
 	int tx_queues;
 	int all_slaves_active;
+	int resend_igmp;
 };
 
 struct bond_parm_tbl {
@@ -198,6 +199,7 @@ struct bonding {
 	s32      slave_cnt; /* never change this value outside the attach/detach wrappers */
 	rwlock_t lock;
 	rwlock_t curr_slave_lock;
+	s8       igmp_retrans;
 	s8       kill_timers;
 	s8	 send_grat_arp;
 	s8	 send_unsol_na;
@@ -223,6 +225,7 @@ struct bonding {
 	struct   delayed_work arp_work;
 	struct   delayed_work alb_work;
 	struct   delayed_work ad_work;
+	struct   delayed_work mcast_work;
 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 	struct   in6_addr master_ipv6;
 #endif
include/linux/if_bonding.h +3 -0
index 2c79943..d2f78b7 100644
@@ -84,6 +84,9 @@
 #define BOND_DEFAULT_MAX_BONDS  1   /* Default maximum number of devices to support */
 
 #define BOND_DEFAULT_TX_QUEUES 16   /* Default number of tx queues per device */
+
+#define BOND_DEFAULT_RESEND_IGMP	1 /* Default number of IGMP membership reports
+					     to resend on link failure. */
 /* hashing types */
 #define BOND_XMIT_POLICY_LAYER2		0 /* layer 2 (MAC only), default */
 #define BOND_XMIT_POLICY_LAYER34	1 /* layer 3+4 (IP ^ (TCP || UDP)) */
Signature · 5 lines
Flavio
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
#16 lore raw
Quoted 55 lines · Andy Gospodarek
Quoted 10 lines · Flavio Leitner
Quoted 9 lines · Andy Gospodarek
Quoted 7 lines · Flavio Leitner

It fixes bonding to rejoin multicast groups added
to VLAN devices on top of bonding when a failover
happens.

The first packet may be discarded, so the timer
assure that at least 3 Reports are sent.

Quoted 9 lines
Quoted 4 lines

Good find, Flavio. Clearly the fact that multicast membership is broken needs to be fixed, but I would rather not see timers used at all. We worked hard in the past to eliminate timers for several reasons, so I would rather see a workqueue used.

I noticed that the code is using workqueues now, just thought a simple thing which may run couple times would fit perfectly with a simple timer.

Timers runs in softirq context, so I'd rather not add code that takes locks and runs in softirq context.

Quoted 17 lines
Quoted 6 lines

I also don't like retransmitting the membership report 3 times when it may not be needed. Though many switches can handle it, the cost of receiving and processing what might be a large list of multicast addresses every 200ms for 600ms doesn't seem ideal. It also feels like a hack. :)

Definitely a parameter is much better, but I wasn't sure about the patch approach so I was expecting a review like this and then do the refinements needed. Better to post early, right? :)

I see your point to change the default to one membership report, but we can't assure during a failover if everything has been received. Also, it isn't supposed to keep failing flooding the network, so I would rather have couple membership reports being send than watch an important multicast application failing.

Quoted 4 lines

Perhaps 3 is too much, but one sounds too few to me.

what you think?

Adding a tunable parameter allows the administrator to decide how many is enough. I would rather keep the default at one and add the tunable parameter (which needs to be added to bond_sysfs.c to be effective).

I have not heard loud complaints about only sending one since the code to send retransmits of membership reports was added a few years ago, so I'm inclined to think it is working well for most users (or no one is using bonding).

Quoted 4 lines

Maybe it would be best to break this into 2 patches. One that simply fixes the failover code so it works with VLANs (that could be done easily today) and another patch that can add the code to send multiple retransmits. Would you be willing to do that?

Sure, I can do it and then start another testing session here.

Signature · 1 line
Flavio