bonding: replicate IGMP traffic in activebackup mode
Subsystems (3)
PATCH 1/1 bonding: replicate IGMP traffic in activebackup mode
Replicate IGMP frames across all slaves in activebackup mode. This ensures fail-over is rapid for multicast traffic as well. Otherwise, multicast traffic will be lost until the next IGMP membership report poll timeout.
---
This is conceptually similar to the treatment of IGMP traffic in bond_alb_xmit. In that case, IGMP traffic transmitted on any slave is re-routed to the active slave in order to ensure that multicast traffic continues to be directed to the active receiver.
drivers/net/bonding/bond_main.c | 53 ++++++++++++++++++++++++++++++++++++++-- 1 files changed, 51 insertions(+), 2 deletions(-)
drivers/net/bonding/bond_main.c +51 -2
@@ -4391,6 +4391,39 @@ out:return 0 ; } +static void bond_activebackup_xmit_copy (struct sk_buff *skb, +struct bonding *bond, +struct slave *slave) +{ +struct sk_buff *skb2 =skb_copy (skb,GFP_ATOMIC ); +struct ethhdr *eth_data; +u8 *hwaddr; +int res; + +if (!skb2) { + printk(KERN_ERR DRV_NAME ": Error: " + "bond_activebackup_xmit_copy(): skb_copy() failed\n"); +return ; + } + + skb2->mac.raw = (unsigned char *)skb2->data; + eth_data =eth_hdr (skb2); + +/* Pick an appropriate source MAC address */ + hwaddr = slave->perm_hwaddr; +if (!memcmp (eth_data->h_source, hwaddr,ETH_ALEN )) + hwaddr = bond->curr_active_slave->perm_hwaddr; + +/* Set source MAC address appropriately */ +memcpy (eth_data->h_source, hwaddr,ETH_ALEN ); + + res =bond_dev_queue_xmit (bond, skb2, slave->dev); +if (res) +dev_kfree_skb (skb2); + +return ; +} +/* * in active-backup mode, we know that bond->curr_active_slave is always valid if * the bond has a usable interface.
@@ -4407,10 +4440,26 @@ static int bond_xmit_activebackup(struct goto out; } -if (bond->curr_active_slave) {/* one usable interface */ - res =bond_dev_queue_xmit (bond, skb, bond->curr_active_slave->dev); +if (!bond->curr_active_slave) + goto out; + +/* Xmit IGMP frames on all slaves to ensure rapid fail-over +for multicast traffic on snooping switches */ +if (skb->protocol ==__constant_htons (ETH_P_IP ) && + skb->nh.iph->protocol ==IPPROTO_IGMP ) { +struct slave *slave, *active_slave; +int i; + + active_slave = bond->curr_active_slave; +bond_for_each_slave_from_to (bond, slave, i, active_slave->next, + active_slave->prev) +if (IS_UP (slave->dev) && + (slave->link ==BOND_LINK_UP )) +bond_activebackup_xmit_copy (skb, bond, slave); } + res =bond_dev_queue_xmit (bond, skb, bond->curr_active_slave->dev); + out:if (res) {/* no suitable interface, frame not sent */
Jeff Garzik
AppliedQuoted 6 lines · John W. Linville
Replicate IGMP frames across all slaves in activebackup mode. This ensures fail-over is rapid for multicast traffic as well. Otherwise, multicast traffic will be lost until the next IGMP membership report poll timeout.
Signed-off-by John W. Linville <[email protected]>
Quoted 8 lines
---
This is conceptually similar to the treatment of IGMP traffic inbond_alb_xmit. In that case, IGMP traffic transmitted on any slave is re-routed to the active slave in order to ensure that multicast traffic continues to be directed to the active receiver.drivers/net/bonding/bond_main.c | 53 ++++++++++++++++++++++++++++++++++++++-- 1 files changed, 51 insertions(+), 2 deletions(-)
Applied. For future patches, please
* don't include description following the signed-off-by line
* if you are willing to special-case me, I actually don't like seeing a diffstat; I have to manually remove it before applying