/* clear slave from tx_hashtbl */
tx_hash_table = BOND_ALB_INFO(bond).tx_hashtbl;
/* skip this if we've already freed the tx hash table */ if (tx_hash_table) {
index = SLAVE_TLB_INFO(slave).head; while (index != TLB_NULL_INDEX) {
u32 next_index = tx_hash_table[index].next;
tlb_init_table_entry(&tx_hash_table[index], save_load);
index = next_index;
}
}
/* Must be called before starting the monitor timer */ staticint tlb_initialize(struct bonding *bond)
{ struct alb_bond_info *bond_info = &(BOND_ALB_INFO(bond)); int size = TLB_HASH_TABLE_SIZE * sizeof(struct tlb_client_info); struct tlb_client_info *new_hashtbl; int i;
new_hashtbl = kzalloc(size, GFP_KERNEL); if (!new_hashtbl) return -ENOMEM;
spin_lock_bh(&bond->mode_lock);
bond_info->tx_hashtbl = new_hashtbl;
for (i = 0; i < TLB_HASH_TABLE_SIZE; i++)
tlb_init_table_entry(&bond_info->tx_hashtbl[i], 0);
spin_unlock_bh(&bond->mode_lock);
return0;
}
/* Must be called only after all slaves have been released */ staticvoid tlb_deinitialize(struct bonding *bond)
{ struct alb_bond_info *bond_info = &(BOND_ALB_INFO(bond));
/* Find the slave with the largest gap */
bond_for_each_slave_rcu(bond, slave, iter) { if (bond_slave_can_tx(slave)) { longlong gap = compute_gap(slave);
/* We don't need to disable softirq here, because *tlb_choose_channel()isonlycalledbybond_alb_xmit() *whichalreadyhassoftirqdisabled.
*/
spin_lock(&bond->mode_lock);
tx_slave = __tlb_choose_channel(bond, hash_index, skb_len);
spin_unlock(&bond->mode_lock);
return tx_slave;
}
/*********************** rlb specific functions ***************************/
/* when an ARP REPLY is received from a client update its info *intherx_hashtbl
*/ staticvoid rlb_update_entry_from_arp(struct bonding *bond, struct arp_pkt *arp)
{ struct alb_bond_info *bond_info = &(BOND_ALB_INFO(bond)); struct rlb_client_info *client_info;
u32 hash_index;
if (skb->protocol != cpu_to_be16(ETH_P_ARP)) goto out;
arp = skb_header_pointer(skb, 0, sizeof(_arp), &_arp); if (!arp) goto out;
/* We received an ARP from arp->ip_src. *WemighthaveusedthisIPaddresspreviously(onthebondinghost *itselforonasystemthatisbridgedtogetherwiththebond). *However,ifarp->mac_srcisdifferentthanwhatisstoredin *rx_hashtbl,someotherhostisnowusingtheIPandwemustprevent *sendingoutclientupdateswiththisIPaddressandtheoldMAC *address. *Cleanupallhashtableentriesthathavethisaddressasip_srcbut *haveadifferentmac_src.
*/
rlb_purge_src_ip(bond, arp);
if (arp->op_code == htons(ARPOP_REPLY)) { /* update rx hash table for this ARP */
rlb_update_entry_from_arp(bond, arp);
slave_dbg(bond->dev, slave->dev, "Server received an ARP Reply from client\n");
}
out: return RX_HANDLER_ANOTHER;
}
bond_for_each_slave_rcu(bond, slave, iter) { if (!bond_slave_can_tx(slave)) continue; if (!found) { if (!before || before->speed < slave->speed)
before = slave;
} else { if (!rx_slave || rx_slave->speed < slave->speed)
rx_slave = slave;
} if (slave == bond_info->rx_slave)
found = true;
} /* we didn't find anything after the current or we have something *betterbeforeanduptothecurrentslave
*/ if (!rx_slave || (before && rx_slave->speed < before->speed))
rx_slave = before;
if (rx_slave)
bond_info->rx_slave = rx_slave;
return rx_slave;
}
/* Caller must hold RTNL, rcu_read_lock is obtained only to silence checkers */ staticstruct slave *rlb_next_rx_slave(struct bonding *bond)
{ struct slave *rx_slave;
/* slave being removed should not be active at this point * *Callermustholdrtnl.
*/ staticvoid rlb_clear_slave(struct bonding *bond, struct slave *slave)
{ struct alb_bond_info *bond_info = &(BOND_ALB_INFO(bond)); struct rlb_client_info *rx_hash_table;
u32 index, next_index;
/* clear slave from rx_hashtbl */
spin_lock_bh(&bond->mode_lock);
rx_hash_table = bond_info->rx_hashtbl;
index = bond_info->rx_hashtbl_used_head; for (; index != RLB_NULL_INDEX; index = next_index) {
next_index = rx_hash_table[index].used_next; if (rx_hash_table[index].slave == slave) { struct slave *assigned_slave = rlb_next_rx_slave(bond);
if (assigned_slave) {
rx_hash_table[index].slave = assigned_slave; if (is_valid_ether_addr(rx_hash_table[index].mac_dst)) {
bond_info->rx_hashtbl[index].ntt = 1;
bond_info->rx_ntt = 1; /* A slave has been removed from the *tablebecauseitiseitherdisabled *orbeingreleased.Wemustretrythe *updatetoavoidclientsfromnot *beingupdated&disconnectingwhen *thereisstress
*/
bond_info->rlb_update_retry_counter =
RLB_UPDATE_RETRY;
}
} else { /* there is no active slave */
rx_hash_table[index].slave = NULL;
}
}
}
spin_unlock_bh(&bond->mode_lock);
if (slave != rtnl_dereference(bond->curr_active_slave))
rlb_teach_disabled_mac_on_primary(bond, slave->dev->dev_addr);
}
staticvoid rlb_update_client(struct rlb_client_info *client_info)
{ int i;
if (!client_info->slave || !is_valid_ether_addr(client_info->mac_dst)) return;
for (i = 0; i < RLB_ARP_BURST_SIZE; i++) { struct sk_buff *skb;
skb = arp_create(ARPOP_REPLY, ETH_P_ARP,
client_info->ip_dst,
client_info->slave->dev,
client_info->ip_src,
client_info->mac_dst,
client_info->slave->dev->dev_addr,
client_info->mac_dst); if (!skb) {
slave_err(client_info->slave->bond->dev,
client_info->slave->dev, "failed to create an ARP packet\n"); continue;
}
skb->dev = client_info->slave->dev;
if (client_info->vlan_id) {
__vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q),
client_info->vlan_id);
}
arp_xmit(skb);
}
}
/* sends ARP REPLIES that update the clients that need updating */ staticvoid rlb_update_rx_clients(struct bonding *bond)
{ struct alb_bond_info *bond_info = &(BOND_ALB_INFO(bond)); struct rlb_client_info *client_info;
u32 hash_index;
spin_lock_bh(&bond->mode_lock);
hash_index = bond_info->rx_hashtbl_used_head; for (; hash_index != RLB_NULL_INDEX;
hash_index = client_info->used_next) {
client_info = &(bond_info->rx_hashtbl[hash_index]); if (client_info->ntt) {
rlb_update_client(client_info); if (bond_info->rlb_update_retry_counter == 0)
client_info->ntt = 0;
}
}
/* do not update the entries again until this counter is zero so that *nottoconfusetheclients.
*/
bond_info->rlb_update_delay_counter = RLB_UPDATE_DELAY;
spin_unlock_bh(&bond->mode_lock);
}
/* The slave was assigned a new mac address - update the clients */ staticvoid rlb_req_update_slave_clients(struct bonding *bond, struct slave *slave)
{ struct alb_bond_info *bond_info = &(BOND_ALB_INFO(bond)); struct rlb_client_info *client_info; int ntt = 0;
u32 hash_index;
/* update the team's flag only after the whole iteration */ if (ntt) {
bond_info->rx_ntt = 1; /* fasten the change */
bond_info->rlb_update_retry_counter = RLB_UPDATE_RETRY;
}
spin_unlock_bh(&bond->mode_lock);
}
/* mark all clients using src_ip to be updated */ staticvoid rlb_req_update_subnet_clients(struct bonding *bond, __be32 src_ip)
{ struct alb_bond_info *bond_info = &(BOND_ALB_INFO(bond)); struct rlb_client_info *client_info;
u32 hash_index;
if (!client_info->slave) {
netdev_err(bond->dev, "found a client with no channel in the client's hash table\n"); continue;
} /* update all clients using this src_ip, that are not assigned *totheteam'saddress(curr_active_slave)andhaveaknown *unicastmacaddress.
*/ if ((client_info->ip_src == src_ip) &&
!ether_addr_equal_64bits(client_info->slave->dev->dev_addr,
bond->dev->dev_addr) &&
is_valid_ether_addr(client_info->mac_dst)) {
client_info->ntt = 1;
bond_info->rx_ntt = 1;
}
}
if (client_info->assigned) { if ((client_info->ip_src == arp->ip_src) &&
(client_info->ip_dst == arp->ip_dst)) { /* the entry is already assigned to this client */ if (!is_broadcast_ether_addr(arp->mac_dst)) { /* update mac address from arp */
ether_addr_copy(client_info->mac_dst, arp->mac_dst);
}
ether_addr_copy(client_info->mac_src, arp->mac_src);
assigned_slave = client_info->slave; if (assigned_slave) {
spin_unlock(&bond->mode_lock); return assigned_slave;
}
} else { /* the entry is already assigned to some other client, *movetheoldclienttoprimary(curr_active_slave)so *thatthenewclientcanbeassignedtothisentry.
*/ if (curr_active_slave &&
client_info->slave != curr_active_slave) {
client_info->slave = curr_active_slave;
rlb_update_client(client_info);
}
}
} /* assign a new slave */
assigned_slave = __rlb_next_rx_slave(bond);
if (assigned_slave) { if (!(client_info->assigned &&
client_info->ip_src == arp->ip_src)) { /* ip_src is going to be updated, *fixthesrchashlist
*/
u32 hash_src = _simple_hash((u8 *)&arp->ip_src, sizeof(arp->ip_src));
rlb_src_unlink(bond, hash_index);
rlb_src_link(bond, hash_src, hash_index);
}
if (!pskb_network_may_pull(skb, sizeof(*arp))) return NULL;
arp = (struct arp_pkt *)skb_network_header(skb);
/* Don't modify or load balance ARPs that do not originate *fromthebonditselforaVLANdirectlyabovethebond.
*/ if (!bond_slave_has_mac_rcu(bond, arp->mac_src)) return NULL;
dev = ip_dev_find(dev_net(bond->dev), arp->ip_src); if (dev) { if (netif_is_any_bridge_master(dev)) {
dev_put(dev); return NULL;
}
dev_put(dev);
}
if (arp->op_code == htons(ARPOP_REPLY)) { /* the arp must be sent on the selected rx channel */
tx_slave = rlb_choose_channel(skb, bond, arp); if (tx_slave)
bond_hw_addr_copy(arp->mac_src, tx_slave->dev->dev_addr,
tx_slave->dev->addr_len);
netdev_dbg(bond->dev, "(slave %s): Server sent ARP Reply packet\n",
tx_slave ? tx_slave->dev->name : "NULL");
} elseif (arp->op_code == htons(ARPOP_REQUEST)) { /* Create an entry in the rx_hashtbl for this client as a *placeholder. *Whenthearpreplyisreceivedtheentrywillbeupdated *withthecorrectunicastaddressoftheclient.
*/
tx_slave = rlb_choose_channel(skb, bond, arp);
/* The ARP reply packets must be delayed so that *theycancancelouttheinfluenceoftheARPrequest.
*/
bond->alb_info.rlb_update_delay_counter = RLB_UPDATE_DELAY;
/* arp requests are broadcast and are sent on the primary *thearprequestwillcollapseallclientsonthesubnetto *theprimaryslave.Wemustregistertheseclientstobe *updatedwiththeirassignedmac.
*/
rlb_req_update_subnet_clients(bond, arp->ip_src);
netdev_dbg(bond->dev, "(slave %s): Server sent ARP Request packet\n",
tx_slave ? tx_slave->dev->name : "NULL");
}
if (next_index != RLB_NULL_INDEX)
bond_info->rx_hashtbl[next_index].src_prev = prev_index;
if (prev_index == RLB_NULL_INDEX) return;
/* is prev_index pointing to the head of this list? */ if (bond_info->rx_hashtbl[prev_index].src_first == index)
bond_info->rx_hashtbl[prev_index].src_first = next_index; else
bond_info->rx_hashtbl[prev_index].src_next = next_index;
/* If this is a macvlan device, then only send updates *whenstrict_matchisturnedoff.
*/ if (netif_is_macvlan(upper) && !strict_match) {
tags = bond_verify_device_path(bond->dev, upper, 0); if (IS_ERR_OR_NULL(tags)) return -ENOMEM;
/* loop through all devices and see if we need to send a packet *forthatdevice.
*/
rcu_read_lock();
netdev_walk_all_upper_dev_rcu(bond->dev, alb_upper_dev_walk, &priv);
rcu_read_unlock();
}
if (BOND_MODE(slave->bond) == BOND_MODE_TLB) {
__dev_addr_set(dev, addr, len); return0;
}
/* for rlb each slave must have a unique hw mac addresses so that *eachslavewillreceivepacketsdestinedtoadifferentmac
*/
memcpy(ss.__data, addr, len);
ss.ss_family = dev->type; if (dev_set_mac_address(dev, &ss, NULL)) {
slave_err(slave->bond->dev, dev, "dev_set_mac_address on slave failed! ALB mode requires that the base driver support setting the hw address also when the network device's interface is open\n"); return -EOPNOTSUPP;
} return0;
}
/* Swap MAC addresses between two slaves. * *CalledwithRTNLheld,andnootherlocks.
*/ staticvoid alb_swap_mac_addr(struct slave *slave1, struct slave *slave2)
{
u8 tmp_mac_addr[MAX_ADDR_LEN];
/* fasten the change in the switch */ if (bond_slave_can_tx(slave1)) {
alb_send_learning_packets(slave1, slave1->dev->dev_addr, false); if (bond->alb_info.rlb_enabled) { /* inform the clients that the mac address *haschanged
*/
rlb_req_update_slave_clients(bond, slave1);
}
} else {
disabled_slave = slave1;
}
if (bond_slave_can_tx(slave2)) {
alb_send_learning_packets(slave2, slave2->dev->dev_addr, false); if (bond->alb_info.rlb_enabled) { /* inform the clients that the mac address *haschanged
*/
rlb_req_update_slave_clients(bond, slave2);
}
} else {
disabled_slave = slave2;
}
if (bond->alb_info.rlb_enabled && slaves_state_differ) { /* A disabled slave was assigned an active mac addr */
rlb_teach_disabled_mac_on_primary(bond,
disabled_slave->dev->dev_addr);
}
}
if (!bond_has_slaves(bond)) { /* this is the first slave */ return0;
}
/* if slave's mac address differs from bond's mac address *checkuniquenessofslave'smacaddressagainsttheother *slavesinthebond.
*/ if (!ether_addr_equal_64bits(slave->perm_hwaddr, bond->dev->dev_addr)) { if (!bond_slave_has_mac(bond, slave->dev->dev_addr)) return0;
/* Try setting slave mac to bond address and fall-through *tocodehandlingthatsituationbelow...
*/
alb_set_slave_mac_addr(slave, bond->dev->dev_addr,
bond->dev->addr_len);
}
/* The slave's address is equal to the address of the bond. *Searchforaspareaddressinthebondforthisslave.
*/
bond_for_each_slave(bond, tmp_slave1, iter) { if (!bond_slave_has_mac(bond, tmp_slave1->perm_hwaddr)) { /* no slave has tmp_slave1's perm addr *asitscurraddr
*/
free_mac_slave = tmp_slave1; break;
}
if (!has_bond_addr) { if (ether_addr_equal_64bits(tmp_slave1->dev->dev_addr,
bond->dev->dev_addr)) {
has_bond_addr = tmp_slave1;
}
}
}
if (free_mac_slave) {
alb_set_slave_mac_addr(slave, free_mac_slave->perm_hwaddr,
free_mac_slave->dev->addr_len);
slave_warn(bond->dev, slave->dev, "the slave hw address is in use by the bond; giving it the hw address of %s\n",
free_mac_slave->dev->name);
} elseif (has_bond_addr) {
slave_err(bond->dev, slave->dev, "the slave hw address is in use by the bond; couldn't find a slave with a free hw address to give it (this should not have happened)\n"); return -EFAULT;
}
/* unwind from head to the slave that failed */
bond_for_each_slave(bond, rollback_slave, iter) { if (rollback_slave == slave) break;
bond_hw_addr_copy(tmp_addr, rollback_slave->dev->dev_addr,
rollback_slave->dev->addr_len);
dev_set_mac_address(rollback_slave->dev, &ss, NULL);
dev_addr_set(rollback_slave->dev, tmp_addr);
}
return res;
}
/* determine if the packet is NA or NS */ staticbool alb_determine_nd(struct sk_buff *skb, struct bonding *bond)
{ struct ipv6hdr *ip6hdr; struct icmp6hdr *hdr;
if (!pskb_network_may_pull(skb, sizeof(*ip6hdr))) returntrue;
ip6hdr = ipv6_hdr(skb); if (ip6hdr->nexthdr != IPPROTO_ICMPV6) returnfalse;
if (!pskb_network_may_pull(skb, sizeof(*ip6hdr) + sizeof(*hdr))) returntrue;
if (!tx_slave) { /* unbalanced or unassigned, send through primary */
tx_slave = rcu_dereference(bond->curr_active_slave); if (bond->params.tlb_dynamic_lb)
bond_info->unbalanced_load += skb->len;
}
if (tx_slave && bond_slave_can_tx(tx_slave)) { if (tx_slave != rcu_access_pointer(bond->curr_active_slave)) {
ether_addr_copy(eth_data->h_source,
tx_slave->dev->dev_addr);
}
/* IPv6 doesn't really use broadcast mac address, but leave *thatherejustincase.
*/ if (is_broadcast_ether_addr(eth_data->h_dest)) {
do_tx_balance = false; break;
}
/* IPv6 uses all-nodes multicast as an equivalent to *broadcastsinIPv4.
*/ if (ether_addr_equal_64bits(eth_data->h_dest, mac_v6_allmcast)) {
do_tx_balance = false; break;
}
if (alb_determine_nd(skb, bond)) {
do_tx_balance = false; break;
}
/* The IPv6 header is pulled by alb_determine_nd */ /* Additionally, DAD probes should not be tx-balanced as that *willleadtofalsepositivesforduplicateaddressesand *preventaddressconfigurationfromworking.
*/
ip6hdr = ipv6_hdr(skb); if (ipv6_addr_any(&ip6hdr->saddr)) {
do_tx_balance = false; break;
}
if (bond_info->rlb_enabled) { if (bond_info->primary_is_promisc &&
(++bond_info->rlb_promisc_timeout_counter >= RLB_PROMISC_TIMEOUT)) {
/* dev_set_promiscuity requires rtnl and *nothingelse.Avoidracewithbond_close.
*/
rcu_read_unlock(); if (!rtnl_trylock()) goto re_arm;
bond_info->rlb_promisc_timeout_counter = 0;
/* If the primary was set to promiscuous mode *becauseaslavewasdisabledthen *itcannowleavepromiscuousmode.
*/
dev_set_promiscuity(rtnl_dereference(bond->curr_active_slave)->dev,
-1);
bond_info->primary_is_promisc = 0;
rtnl_unlock();
rcu_read_lock();
}
if (bond_info->rlb_rebalance) {
bond_info->rlb_rebalance = 0;
rlb_rebalance(bond);
}
/* check if clients need updating */ if (bond_info->rx_ntt) { if (bond_info->rlb_update_delay_counter) {
--bond_info->rlb_update_delay_counter;
} else {
rlb_update_rx_clients(bond); if (bond_info->rlb_update_retry_counter)
--bond_info->rlb_update_retry_counter; else
bond_info->rx_ntt = 0;
}
}
}
rcu_read_unlock();
re_arm:
queue_delayed_work(bond->wq, &bond->alb_work, alb_delta_in_ticks);
}
/* assumption: called before the slave is attached to the bond *andnotlockedbythebondlock
*/ int bond_alb_init_slave(struct bonding *bond, struct slave *slave)
{ int res;
res = alb_set_slave_mac_addr(slave, slave->perm_hwaddr,
slave->dev->addr_len); if (res) return res;
res = alb_handle_addr_collision_on_attach(bond, slave); if (res) return res;
tlb_init_slave(slave);
/* order a rebalance ASAP */
atomic_set(&bond->alb_info.tx_rebalance_counter,
BOND_TLB_REBALANCE_TICKS);
if (bond->alb_info.rlb_enabled)
bond->alb_info.rlb_rebalance = 1;
return0;
}
/* Remove slave from tlb and rlb hash tables, and fix up MAC addresses *ifnecessary. * *CallermustholdRTNLandnootherlocks
*/ void bond_alb_deinit_slave(struct bonding *bond, struct slave *slave)
{ if (bond_has_slaves(bond))
alb_change_hw_addr_on_detach(bond, slave);
tlb_clear_slave(bond, slave, 0);
if (bond->alb_info.rlb_enabled) {
bond->alb_info.rx_slave = NULL;
rlb_clear_slave(bond, slave);
}
if (link == BOND_LINK_DOWN) {
tlb_clear_slave(bond, slave, 0); if (bond->alb_info.rlb_enabled)
rlb_clear_slave(bond, slave);
} elseif (link == BOND_LINK_UP) { /* order a rebalance ASAP */
atomic_set(&bond_info->tx_rebalance_counter,
BOND_TLB_REBALANCE_TICKS); if (bond->alb_info.rlb_enabled) {
bond->alb_info.rlb_rebalance = 1; /* If the updelay module parameter is smaller than the *forwardingdelayoftheswitchtherebalancewill *notworkbecausetherebalancearpreplieswill *notbeforwardedtotheclients..
*/
}
}
if (bond_is_nondyn_tlb(bond)) { if (bond_update_slave_arr(bond, NULL))
pr_err("Failed to build slave-array for TLB mode.\n");
}
}
/* set the new curr_active_slave to the bonds mac address *i.e.swapmacaddressesofoldcurr_active_slaveandnewcurr_active_slave
*/ if (!swap_slave)
swap_slave = bond_slave_has_mac(bond, bond->dev->dev_addr);
/* Arrange for swap_slave and new_slave to temporarily be *ignoredsowecanmesswiththeirMACaddresseswithout *fearofinterferencefromtransmitactivity.
*/ if (swap_slave)
tlb_clear_slave(bond, swap_slave, 1);
tlb_clear_slave(bond, new_slave, 1);
/* in TLB mode, the slave might flip down/up with the old dev_addr, *andthusfilterbond->dev_addr'spackets,soforcebond'smac
*/ if (BOND_MODE(bond) == BOND_MODE_TLB) { struct sockaddr_storage ss;
u8 tmp_addr[MAX_ADDR_LEN];
bond_hw_addr_copy(ss.__data, bond->dev->dev_addr,
bond->dev->addr_len);
ss.ss_family = bond->dev->type; /* we don't care if it can't change its mac, best effort */
dev_set_mac_address(new_slave->dev, &ss, NULL);
dev_addr_set(new_slave->dev, tmp_addr);
}
/* curr_active_slave must be set before calling alb_swap_mac_addr */ if (swap_slave) { /* swap mac address */
alb_swap_mac_addr(swap_slave, new_slave);
alb_fasten_mac_swap(bond, swap_slave, new_slave);
} else { /* set the new_slave to the bond mac address */
alb_set_slave_mac_addr(new_slave, bond->dev->dev_addr,
bond->dev->addr_len);
alb_send_learning_packets(new_slave, bond->dev->dev_addr, false);
}
}
/* Called with RTNL */ int bond_alb_set_mac_address(struct net_device *bond_dev, void *addr)
{ struct bonding *bond = netdev_priv(bond_dev); struct sockaddr_storage *ss = addr; struct slave *curr_active; struct slave *swap_slave; int res;
if (!is_valid_ether_addr(ss->__data)) return -EADDRNOTAVAIL;
res = alb_set_mac_address(bond, addr); if (res) return res;
dev_addr_set(bond_dev, ss->__data);
/* If there is no curr_active_slave there is nothing else to do. *Otherwisewe'llneedtopassthenewaddresstoitandhandle *duplications.
*/
curr_active = rtnl_dereference(bond->curr_active_slave); if (!curr_active) return0;
Die Informationen auf dieser Webseite wurden
nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit,
noch Qualität der bereit gestellten Informationen zugesichert.
Bemerkung:
Die farbliche Syntaxdarstellung und die Messung sind noch experimentell.