[deliverable/linux.git] / net / bridge / br_if.c

/*
 *	Userspace interface
 *	Linux ethernet bridge
 *
 *	Authors:
 *	Lennert Buytenhek		<buytenh@gnu.org>
 *
 *	This program is free software; you can redistribute it and/or
 *	modify it under the terms of the GNU General Public License
 *	as published by the Free Software Foundation; either version
 *	2 of the License, or (at your option) any later version.
 */

#include <linux/kernel.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include <linux/netpoll.h>
#include <linux/ethtool.h>
#include <linux/if_arp.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/rtnetlink.h>
#include <linux/if_ether.h>
#include <linux/slab.h>
#include <net/sock.h>
#include <linux/if_vlan.h>
#include <net/switchdev.h>

#include "br_private.h"

/*
 * Determine initial path cost based on speed.
 * using recommendations from 802.1d standard
 *
 * Since driver might sleep need to not be holding any locks.
 */
static int port_cost(struct net_device *dev)
{
	struct ethtool_link_ksettings ecmd;

	if (!__ethtool_get_link_ksettings(dev, &ecmd)) {
		switch (ecmd.base.speed) {
		case SPEED_10000:
			return 2;
		case SPEED_1000:
			return 4;
		case SPEED_100:
			return 19;
		case SPEED_10:
			return 100;
		}
	}

	/* Old silly heuristics based on name */
	if (!strncmp(dev->name, "lec", 3))
		return 7;

	if (!strncmp(dev->name, "plip", 4))
		return 2500;

	return 100;	/* assume old 10Mbps */
}


/* Check for port carrier transitions. */
void br_port_carrier_check(struct net_bridge_port *p)
{
	struct net_device *dev = p->dev;
	struct net_bridge *br = p->br;

	if (!(p->flags & BR_ADMIN_COST) &&
	    netif_running(dev) && netif_oper_up(dev))
		p->path_cost = port_cost(dev);

	if (!netif_running(br->dev))
		return;

	spin_lock_bh(&br->lock);
	if (netif_running(dev) && netif_oper_up(dev)) {
		if (p->state == BR_STATE_DISABLED)
			br_stp_enable_port(p);
	} else {
		if (p->state != BR_STATE_DISABLED)
			br_stp_disable_port(p);
	}
	spin_unlock_bh(&br->lock);
}

static void br_port_set_promisc(struct net_bridge_port *p)
{
	int err = 0;

	if (br_promisc_port(p))
		return;

	err = dev_set_promiscuity(p->dev, 1);
	if (err)
		return;

	br_fdb_unsync_static(p->br, p);
	p->flags |= BR_PROMISC;
}

static void br_port_clear_promisc(struct net_bridge_port *p)
{
	int err;

	/* Check if the port is already non-promisc or if it doesn't
	 * support UNICAST filtering.  Without unicast filtering support
	 * we'll end up re-enabling promisc mode anyway, so just check for
	 * it here.
	 */
	if (!br_promisc_port(p) || !(p->dev->priv_flags & IFF_UNICAST_FLT))
		return;

	/* Since we'll be clearing the promisc mode, program the port
	 * first so that we don't have interruption in traffic.
	 */
	err = br_fdb_sync_static(p->br, p);
	if (err)
		return;

	dev_set_promiscuity(p->dev, -1);
	p->flags &= ~BR_PROMISC;
}

/* When a port is added or removed or when certain port flags
 * change, this function is called to automatically manage
 * promiscuity setting of all the bridge ports.  We are always called
 * under RTNL so can skip using rcu primitives.
 */
void br_manage_promisc(struct net_bridge *br)
{
	struct net_bridge_port *p;
	bool set_all = false;

	/* If vlan filtering is disabled or bridge interface is placed
	 * into promiscuous mode, place all ports in promiscuous mode.
	 */
	if ((br->dev->flags & IFF_PROMISC) || !br_vlan_enabled(br))
		set_all = true;

	list_for_each_entry(p, &br->port_list, list) {
		if (set_all) {
			br_port_set_promisc(p);
		} else {
			/* If the number of auto-ports is <= 1, then all other
			 * ports will have their output configuration
			 * statically specified through fdbs.  Since ingress
			 * on the auto-port becomes forwarding/egress to other
			 * ports and egress configuration is statically known,
			 * we can say that ingress configuration of the
			 * auto-port is also statically known.
			 * This lets us disable promiscuous mode and write
			 * this config to hw.
			 */
			if (br->auto_cnt == 0 ||
			    (br->auto_cnt == 1 && br_auto_port(p)))
				br_port_clear_promisc(p);
			else
				br_port_set_promisc(p);
		}
	}
}

static void nbp_update_port_count(struct net_bridge *br)
{
	struct net_bridge_port *p;
	u32 cnt = 0;

	list_for_each_entry(p, &br->port_list, list) {
		if (br_auto_port(p))
			cnt++;
	}
	if (br->auto_cnt != cnt) {
		br->auto_cnt = cnt;
		br_manage_promisc(br);
	}
}

static void nbp_delete_promisc(struct net_bridge_port *p)
{
	/* If port is currently promiscuous, unset promiscuity.
	 * Otherwise, it is a static port so remove all addresses
	 * from it.
	 */
	dev_set_allmulti(p->dev, -1);
	if (br_promisc_port(p))
		dev_set_promiscuity(p->dev, -1);
	else
		br_fdb_unsync_static(p->br, p);
}

static void release_nbp(struct kobject *kobj)
{
	struct net_bridge_port *p
		= container_of(kobj, struct net_bridge_port, kobj);
	kfree(p);
}

static struct kobj_type brport_ktype = {
#ifdef CONFIG_SYSFS
	.sysfs_ops = &brport_sysfs_ops,
#endif
	.release = release_nbp,
};

static void destroy_nbp(struct net_bridge_port *p)
{
	struct net_device *dev = p->dev;

	p->br = NULL;
	p->dev = NULL;
	dev_put(dev);

	kobject_put(&p->kobj);
}

static void destroy_nbp_rcu(struct rcu_head *head)
{
	struct net_bridge_port *p =
			container_of(head, struct net_bridge_port, rcu);
	destroy_nbp(p);
}

static unsigned get_max_headroom(struct net_bridge *br)
{
	unsigned max_headroom = 0;
	struct net_bridge_port *p;

	list_for_each_entry(p, &br->port_list, list) {
		unsigned dev_headroom = netdev_get_fwd_headroom(p->dev);

		if (dev_headroom > max_headroom)
			max_headroom = dev_headroom;
	}

	return max_headroom;
}

static void update_headroom(struct net_bridge *br, int new_hr)
{
	struct net_bridge_port *p;

	list_for_each_entry(p, &br->port_list, list)
		netdev_set_rx_headroom(p->dev, new_hr);

	br->dev->needed_headroom = new_hr;
}

/* Delete port(interface) from bridge is done in two steps.
 * via RCU. First step, marks device as down. That deletes
 * all the timers and stops new packets from flowing through.
 *
 * Final cleanup doesn't occur until after all CPU's finished
 * processing packets.
 *
 * Protected from multiple admin operations by RTNL mutex
 */
static void del_nbp(struct net_bridge_port *p)
{
	struct net_bridge *br = p->br;
	struct net_device *dev = p->dev;

	sysfs_remove_link(br->ifobj, p->dev->name);

	nbp_delete_promisc(p);

	spin_lock_bh(&br->lock);
	br_stp_disable_port(p);
	spin_unlock_bh(&br->lock);

	br_ifinfo_notify(RTM_DELLINK, p);

	list_del_rcu(&p->list);
	if (netdev_get_fwd_headroom(dev) == br->dev->needed_headroom)
		update_headroom(br, get_max_headroom(br));
	netdev_reset_rx_headroom(dev);

	nbp_vlan_flush(p);
	br_fdb_delete_by_port(br, p, 0, 1);
	switchdev_deferred_process();

	nbp_update_port_count(br);

	netdev_upper_dev_unlink(dev, br->dev);

	dev->priv_flags &= ~IFF_BRIDGE_PORT;

	netdev_rx_handler_unregister(dev);

	br_multicast_del_port(p);

	kobject_uevent(&p->kobj, KOBJ_REMOVE);
	kobject_del(&p->kobj);

	br_netpoll_disable(p);

	call_rcu(&p->rcu, destroy_nbp_rcu);
}

/* Delete bridge device */
void br_dev_delete(struct net_device *dev, struct list_head *head)
{
	struct net_bridge *br = netdev_priv(dev);
	struct net_bridge_port *p, *n;

	list_for_each_entry_safe(p, n, &br->port_list, list) {
		del_nbp(p);
	}

	br_fdb_delete_by_port(br, NULL, 0, 1);

	br_vlan_flush(br);
	br_multicast_dev_del(br);
	del_timer_sync(&br->gc_timer);

	br_sysfs_delbr(br->dev);
	unregister_netdevice_queue(br->dev, head);
}

/* find an available port number */
static int find_portno(struct net_bridge *br)
{
	int index;
	struct net_bridge_port *p;
	unsigned long *inuse;

	inuse = kcalloc(BITS_TO_LONGS(BR_MAX_PORTS), sizeof(unsigned long),
			GFP_KERNEL);
	if (!inuse)
		return -ENOMEM;

	set_bit(0, inuse);	/* zero is reserved */
	list_for_each_entry(p, &br->port_list, list) {
		set_bit(p->port_no, inuse);
	}
	index = find_first_zero_bit(inuse, BR_MAX_PORTS);
	kfree(inuse);

	return (index >= BR_MAX_PORTS) ? -EXFULL : index;
}

/* called with RTNL but without bridge lock */
static struct net_bridge_port *new_nbp(struct net_bridge *br,
				       struct net_device *dev)
{
	struct net_bridge_port *p;
	int index, err;

	index = find_portno(br);
	if (index < 0)
		return ERR_PTR(index);

	p = kzalloc(sizeof(*p), GFP_KERNEL);
	if (p == NULL)
		return ERR_PTR(-ENOMEM);

	p->br = br;
	dev_hold(dev);
	p->dev = dev;
	p->path_cost = port_cost(dev);
	p->priority = 0x8000 >> BR_PORT_BITS;
	p->port_no = index;
	p->flags = BR_LEARNING | BR_FLOOD | BR_MCAST_FLOOD;
	br_init_port(p);
	br_set_state(p, BR_STATE_DISABLED);
	br_stp_port_timer_init(p);
	err = br_multicast_add_port(p);
	if (err) {
		dev_put(dev);
		kfree(p);
		p = ERR_PTR(err);
	}

	return p;
}

int br_add_bridge(struct net *net, const char *name)
{
	struct net_device *dev;
	int res;

	dev = alloc_netdev(sizeof(struct net_bridge), name, NET_NAME_UNKNOWN,
			   br_dev_setup);

	if (!dev)
		return -ENOMEM;

	dev_net_set(dev, net);
	dev->rtnl_link_ops = &br_link_ops;

	res = register_netdev(dev);
	if (res)
		free_netdev(dev);
	return res;
}

int br_del_bridge(struct net *net, const char *name)
{
	struct net_device *dev;
	int ret = 0;

	rtnl_lock();
	dev = __dev_get_by_name(net, name);
	if (dev == NULL)
		ret =  -ENXIO; 	/* Could not find device */

	else if (!(dev->priv_flags & IFF_EBRIDGE)) {
		/* Attempt to delete non bridge device! */
		ret = -EPERM;
	}

	else if (dev->flags & IFF_UP) {
		/* Not shutdown yet. */
		ret = -EBUSY;
	}

	else
		br_dev_delete(dev, NULL);

	rtnl_unlock();
	return ret;
}

/* MTU of the bridge pseudo-device: ETH_DATA_LEN or the minimum of the ports */
int br_min_mtu(const struct net_bridge *br)
{
	const struct net_bridge_port *p;
	int mtu = 0;

	ASSERT_RTNL();

	if (list_empty(&br->port_list))
		mtu = ETH_DATA_LEN;
	else {
		list_for_each_entry(p, &br->port_list, list) {
			if (!mtu  || p->dev->mtu < mtu)
				mtu = p->dev->mtu;
		}
	}
	return mtu;
}

static void br_set_gso_limits(struct net_bridge *br)
{
	unsigned int gso_max_size = GSO_MAX_SIZE;
	u16 gso_max_segs = GSO_MAX_SEGS;
	const struct net_bridge_port *p;

	list_for_each_entry(p, &br->port_list, list) {
		gso_max_size = min(gso_max_size, p->dev->gso_max_size);
		gso_max_segs = min(gso_max_segs, p->dev->gso_max_segs);
	}
	br->dev->gso_max_size = gso_max_size;
	br->dev->gso_max_segs = gso_max_segs;
}

/*
 * Recomputes features using slave's features
 */
netdev_features_t br_features_recompute(struct net_bridge *br,
	netdev_features_t features)
{
	struct net_bridge_port *p;
	netdev_features_t mask;

	if (list_empty(&br->port_list))
		return features;

	mask = features;
	features &= ~NETIF_F_ONE_FOR_ALL;

	list_for_each_entry(p, &br->port_list, list) {
		features = netdev_increment_features(features,
						     p->dev->features, mask);
	}
	features = netdev_add_tso_features(features, mask);

	return features;
}

/* called with RTNL */
int br_add_if(struct net_bridge *br, struct net_device *dev)
{
	struct net_bridge_port *p;
	int err = 0;
	unsigned br_hr, dev_hr;
	bool changed_addr;

	/* Don't allow bridging non-ethernet like devices, or DSA-enabled
	 * master network devices since the bridge layer rx_handler prevents
	 * the DSA fake ethertype handler to be invoked, so we do not strip off
	 * the DSA switch tag protocol header and the bridge layer just return
	 * RX_HANDLER_CONSUMED, stopping RX processing for these frames.
	 */
	if ((dev->flags & IFF_LOOPBACK) ||
	    dev->type != ARPHRD_ETHER || dev->addr_len != ETH_ALEN ||
	    !is_valid_ether_addr(dev->dev_addr) ||
	    netdev_uses_dsa(dev))
		return -EINVAL;

	/* No bridging of bridges */
	if (dev->netdev_ops->ndo_start_xmit == br_dev_xmit)
		return -ELOOP;

	/* Device is already being bridged */
	if (br_port_exists(dev))
		return -EBUSY;

	/* No bridging devices that dislike that (e.g. wireless) */
	if (dev->priv_flags & IFF_DONT_BRIDGE)
		return -EOPNOTSUPP;

	p = new_nbp(br, dev);
	if (IS_ERR(p))
		return PTR_ERR(p);

	call_netdevice_notifiers(NETDEV_JOIN, dev);

	err = dev_set_allmulti(dev, 1);
	if (err)
		goto put_back;

	err = kobject_init_and_add(&p->kobj, &brport_ktype, &(dev->dev.kobj),
				   SYSFS_BRIDGE_PORT_ATTR);
	if (err)
		goto err1;

	err = br_sysfs_addif(p);
	if (err)
		goto err2;

	err = br_netpoll_enable(p);
	if (err)
		goto err3;

	err = netdev_rx_handler_register(dev, br_handle_frame, p);
	if (err)
		goto err4;

	dev->priv_flags |= IFF_BRIDGE_PORT;

	err = netdev_master_upper_dev_link(dev, br->dev, NULL, NULL);
	if (err)
		goto err5;

	err = nbp_switchdev_mark_set(p);
	if (err)
		goto err6;

	dev_disable_lro(dev);

	list_add_rcu(&p->list, &br->port_list);

	nbp_update_port_count(br);

	netdev_update_features(br->dev);

	br_hr = br->dev->needed_headroom;
	dev_hr = netdev_get_fwd_headroom(dev);
	if (br_hr < dev_hr)
		update_headroom(br, dev_hr);
	else
		netdev_set_rx_headroom(dev, br_hr);

	if (br_fdb_insert(br, p, dev->dev_addr, 0))
		netdev_err(dev, "failed insert local address bridge forwarding table\n");

	err = nbp_vlan_init(p);
	if (err) {
		netdev_err(dev, "failed to initialize vlan filtering on this port\n");
		goto err7;
	}

	spin_lock_bh(&br->lock);
	changed_addr = br_stp_recalculate_bridge_id(br);

	if (netif_running(dev) && netif_oper_up(dev) &&
	    (br->dev->flags & IFF_UP))
		br_stp_enable_port(p);
	spin_unlock_bh(&br->lock);

	br_ifinfo_notify(RTM_NEWLINK, p);

	if (changed_addr)
		call_netdevice_notifiers(NETDEV_CHANGEADDR, br->dev);

	dev_set_mtu(br->dev, br_min_mtu(br));
	br_set_gso_limits(br);

	kobject_uevent(&p->kobj, KOBJ_ADD);

	return 0;

err7:
	list_del_rcu(&p->list);
	br_fdb_delete_by_port(br, p, 0, 1);
	nbp_update_port_count(br);
err6:
	netdev_upper_dev_unlink(dev, br->dev);
err5:
	dev->priv_flags &= ~IFF_BRIDGE_PORT;
	netdev_rx_handler_unregister(dev);
err4:
	br_netpoll_disable(p);
err3:
	sysfs_remove_link(br->ifobj, p->dev->name);
err2:
	kobject_put(&p->kobj);
	p = NULL; /* kobject_put frees */
err1:
	dev_set_allmulti(dev, -1);
put_back:
	dev_put(dev);
	kfree(p);
	return err;
}

/* called with RTNL */
int br_del_if(struct net_bridge *br, struct net_device *dev)
{
	struct net_bridge_port *p;
	bool changed_addr;

	p = br_port_get_rtnl(dev);
	if (!p || p->br != br)
		return -EINVAL;

	/* Since more than one interface can be attached to a bridge,
	 * there still maybe an alternate path for netconsole to use;
	 * therefore there is no reason for a NETDEV_RELEASE event.
	 */
	del_nbp(p);

	dev_set_mtu(br->dev, br_min_mtu(br));
	br_set_gso_limits(br);

	spin_lock_bh(&br->lock);
	changed_addr = br_stp_recalculate_bridge_id(br);
	spin_unlock_bh(&br->lock);

	if (changed_addr)
		call_netdevice_notifiers(NETDEV_CHANGEADDR, br->dev);

	netdev_update_features(br->dev);

	return 0;
}

void br_port_flags_change(struct net_bridge_port *p, unsigned long mask)
{
	struct net_bridge *br = p->br;

	if (mask & BR_AUTO_MASK)
		nbp_update_port_count(br);
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* Userspace interface
	3	* Linux ethernet bridge
	4	*
	5	* Authors:
	6	* Lennert Buytenhek <buytenh@gnu.org>
	7	*
1da177e4 LT	8	* This program is free software; you can redistribute it and/or
	9	* modify it under the terms of the GNU General Public License
	10	* as published by the Free Software Foundation; either version
	11	* 2 of the License, or (at your option) any later version.
	12	*/
	13
	14	#include <linux/kernel.h>
	15	#include <linux/netdevice.h>
77f98598	16	#include <linux/etherdevice.h>
c06ee961	17	#include <linux/netpoll.h>
1da177e4 LT	18	#include <linux/ethtool.h>
	19	#include <linux/if_arp.h>
	20	#include <linux/module.h>
	21	#include <linux/init.h>
	22	#include <linux/rtnetlink.h>
46f25dff	23	#include <linux/if_ether.h>
5a0e3ad6	24	#include <linux/slab.h>
1da177e4	25	#include <net/sock.h>
407af329	26	#include <linux/if_vlan.h>
56607386	27	#include <net/switchdev.h>
1da177e4 LT	28
	29	#include "br_private.h"
	30
	31	/*
	32	* Determine initial path cost based on speed.
	33	* using recommendations from 802.1d standard
	34	*
61a44b9c	35	* Since driver might sleep need to not be holding any locks.
1da177e4	36	*/
4433f420	37	static int port_cost(struct net_device *dev)
1da177e4	38	{
702b26a2	39	struct ethtool_link_ksettings ecmd;
b4a488d1	40
702b26a2 DD	41	if (!__ethtool_get_link_ksettings(dev, &ecmd)) {
702b26a2 DD	42	switch (ecmd.base.speed) {
fa3df928 JP	43	case SPEED_10000:
	44	return 2;
	45	case SPEED_1000:
	46	return 4;
	47	case SPEED_100:
	48	return 19;
	49	case SPEED_10:
	50	return 100;
1da177e4 LT	51	}
	52	}
	53
	54	/* Old silly heuristics based on name */
	55	if (!strncmp(dev->name, "lec", 3))
	56	return 7;
	57
	58	if (!strncmp(dev->name, "plip", 4))
	59	return 2500;
	60
	61	return 100; /* assume old 10Mbps */
	62	}
	63
4433f420	64
1a81a2e0	65	/* Check for port carrier transitions. */
269def7c	66	void br_port_carrier_check(struct net_bridge_port *p)
4433f420	67	{
269def7c SH	68	struct net_device *dev = p->dev;
269def7c SH	69	struct net_bridge *br = p->br;
6e86b890	70
8f3359bd	71	if (!(p->flags & BR_ADMIN_COST) &&
8f3359bd	72	netif_running(dev) && netif_oper_up(dev))
6e86b890 SH	73	p->path_cost = port_cost(dev);
6e86b890 SH	74
aa7c6e5f	75	if (!netif_running(br->dev))
	76	return;
	77
	78	spin_lock_bh(&br->lock);
576eb625	79	if (netif_running(dev) && netif_oper_up(dev)) {
aa7c6e5f	80	if (p->state == BR_STATE_DISABLED)
	81	br_stp_enable_port(p);
	82	} else {
	83	if (p->state != BR_STATE_DISABLED)
	84	br_stp_disable_port(p);
4433f420	85	}
aa7c6e5f	86	spin_unlock_bh(&br->lock);
4433f420 SH	87	}
4433f420 SH	88
2796d0c6 VY	89	static void br_port_set_promisc(struct net_bridge_port *p)
	90	{
	91	int err = 0;
	92
	93	if (br_promisc_port(p))
	94	return;
	95
	96	err = dev_set_promiscuity(p->dev, 1);
	97	if (err)
	98	return;
	99
	100	br_fdb_unsync_static(p->br, p);
	101	p->flags \|= BR_PROMISC;
	102	}
	103
	104	static void br_port_clear_promisc(struct net_bridge_port *p)
	105	{
	106	int err;
	107
	108	/* Check if the port is already non-promisc or if it doesn't
	109	* support UNICAST filtering. Without unicast filtering support
	110	* we'll end up re-enabling promisc mode anyway, so just check for
	111	* it here.
	112	*/
	113	if (!br_promisc_port(p) \|\| !(p->dev->priv_flags & IFF_UNICAST_FLT))
	114	return;
	115
	116	/* Since we'll be clearing the promisc mode, program the port
	117	* first so that we don't have interruption in traffic.
	118	*/
	119	err = br_fdb_sync_static(p->br, p);
	120	if (err)
	121	return;
	122
	123	dev_set_promiscuity(p->dev, -1);
	124	p->flags &= ~BR_PROMISC;
	125	}
	126
	127	/* When a port is added or removed or when certain port flags
	128	* change, this function is called to automatically manage
	129	* promiscuity setting of all the bridge ports. We are always called
	130	* under RTNL so can skip using rcu primitives.
	131	*/
	132	void br_manage_promisc(struct net_bridge *br)
	133	{
	134	struct net_bridge_port *p;
	135	bool set_all = false;
	136
	137	/* If vlan filtering is disabled or bridge interface is placed
	138	* into promiscuous mode, place all ports in promiscuous mode.
	139	*/
	140	if ((br->dev->flags & IFF_PROMISC) \|\| !br_vlan_enabled(br))
	141	set_all = true;
	142
	143	list_for_each_entry(p, &br->port_list, list) {
	144	if (set_all) {
	145	br_port_set_promisc(p);
	146	} else {
	147	/* If the number of auto-ports is <= 1, then all other
	148	* ports will have their output configuration
	149	* statically specified through fdbs. Since ingress
	150	* on the auto-port becomes forwarding/egress to other
	151	* ports and egress configuration is statically known,
	152	* we can say that ingress configuration of the
153	* auto-port is also statically known.
154	* This lets us disable promiscuous mode and write
155	* this config to hw.
156	*/
e0a47d1f TM	157	if (br->auto_cnt == 0 \|\|
e0a47d1f TM	158	(br->auto_cnt == 1 && br_auto_port(p)))
2796d0c6 VY	159	br_port_clear_promisc(p);
	160	else
	161	br_port_set_promisc(p);
	162	}
	163	}
	164	}
	165
e028e4b8 VY	166	static void nbp_update_port_count(struct net_bridge *br)
	167	{
	168	struct net_bridge_port *p;
	169	u32 cnt = 0;
	170
	171	list_for_each_entry(p, &br->port_list, list) {
	172	if (br_auto_port(p))
	173	cnt++;
	174	}
2796d0c6 VY	175	if (br->auto_cnt != cnt) {
	176	br->auto_cnt = cnt;
	177	br_manage_promisc(br);
	178	}
	179	}
	180
	181	static void nbp_delete_promisc(struct net_bridge_port *p)
	182	{
025559ee	183	/* If port is currently promiscuous, unset promiscuity.
2796d0c6 VY	184	* Otherwise, it is a static port so remove all addresses
	185	* from it.
	186	*/
	187	dev_set_allmulti(p->dev, -1);
	188	if (br_promisc_port(p))
	189	dev_set_promiscuity(p->dev, -1);
	190	else
	191	br_fdb_unsync_static(p->br, p);
e028e4b8 VY	192	}
e028e4b8 VY	193
bab1deea SH	194	static void release_nbp(struct kobject *kobj)
	195	{
	196	struct net_bridge_port *p
	197	= container_of(kobj, struct net_bridge_port, kobj);
	198	kfree(p);
	199	}
	200
	201	static struct kobj_type brport_ktype = {
	202	#ifdef CONFIG_SYSFS
	203	.sysfs_ops = &brport_sysfs_ops,
	204	#endif
	205	.release = release_nbp,
	206	};
	207
1da177e4 LT	208	static void destroy_nbp(struct net_bridge_port *p)
	209	{
	210	struct net_device *dev = p->dev;
	211
1da177e4 LT	212	p->br = NULL;
	213	p->dev = NULL;
	214	dev_put(dev);
	215
bab1deea	216	kobject_put(&p->kobj);
1da177e4 LT	217	}
	218
	219	static void destroy_nbp_rcu(struct rcu_head *head)
	220	{
	221	struct net_bridge_port *p =
	222	container_of(head, struct net_bridge_port, rcu);
	223	destroy_nbp(p);
	224	}
	225
45493d47 PA	226	static unsigned get_max_headroom(struct net_bridge *br)
	227	{
	228	unsigned max_headroom = 0;
	229	struct net_bridge_port *p;
	230
	231	list_for_each_entry(p, &br->port_list, list) {
	232	unsigned dev_headroom = netdev_get_fwd_headroom(p->dev);
	233
	234	if (dev_headroom > max_headroom)
	235	max_headroom = dev_headroom;
	236	}
	237
	238	return max_headroom;
	239	}
	240
	241	static void update_headroom(struct net_bridge *br, int new_hr)
	242	{
	243	struct net_bridge_port *p;
	244
	245	list_for_each_entry(p, &br->port_list, list)
	246	netdev_set_rx_headroom(p->dev, new_hr);
	247
	248	br->dev->needed_headroom = new_hr;
	249	}
	250
3f4cfc2d SH	251	/* Delete port(interface) from bridge is done in two steps.
	252	* via RCU. First step, marks device as down. That deletes
	253	* all the timers and stops new packets from flowing through.
	254	*
	255	* Final cleanup doesn't occur until after all CPU's finished
	256	* processing packets.
	257	*
	258	* Protected from multiple admin operations by RTNL mutex
	259	*/
1da177e4 LT	260	static void del_nbp(struct net_bridge_port *p)
	261	{
	262	struct net_bridge *br = p->br;
	263	struct net_device *dev = p->dev;
	264
b3bcb72e	265	sysfs_remove_link(br->ifobj, p->dev->name);
bab1deea	266
2796d0c6	267	nbp_delete_promisc(p);
1da177e4 LT	268
	269	spin_lock_bh(&br->lock);
	270	br_stp_disable_port(p);
	271	spin_unlock_bh(&br->lock);
	272
b86c4503 SH	273	br_ifinfo_notify(RTM_DELLINK, p);
b86c4503 SH	274
1da177e4	275	list_del_rcu(&p->list);
45493d47 PA	276	if (netdev_get_fwd_headroom(dev) == br->dev->needed_headroom)
	277	update_headroom(br, get_max_headroom(br));
	278	netdev_reset_rx_headroom(dev);
1da177e4	279
f409d0ed	280	nbp_vlan_flush(p);
1ea2d020	281	br_fdb_delete_by_port(br, p, 0, 1);
56607386 JP	282	switchdev_deferred_process();
56607386 JP	283
e028e4b8 VY	284	nbp_update_port_count(br);
e028e4b8 VY	285
0f49579a JP	286	netdev_upper_dev_unlink(dev, br->dev);
0f49579a JP	287
f350a0a8 JP	288	dev->priv_flags &= ~IFF_BRIDGE_PORT;
f350a0a8 JP	289
ab95bfe0	290	netdev_rx_handler_unregister(dev);
b3f1be4b	291
3fe2d7c7 HX	292	br_multicast_del_port(p);
3fe2d7c7 HX	293
125a12cc	294	kobject_uevent(&p->kobj, KOBJ_REMOVE);
bab1deea SH	295	kobject_del(&p->kobj);
bab1deea SH	296
91d2c34a HX	297	br_netpoll_disable(p);
91d2c34a HX	298
1da177e4 LT	299	call_rcu(&p->rcu, destroy_nbp_rcu);
	300	}
	301
1ce5cce8	302	/* Delete bridge device */
1ce5cce8	303	void br_dev_delete(struct net_device dev, struct list_head head)
1da177e4	304	{
1ce5cce8	305	struct net_bridge *br = netdev_priv(dev);
1da177e4 LT	306	struct net_bridge_port p, n;
	307
	308	list_for_each_entry_safe(p, n, &br->port_list, list) {
1da177e4 LT	309	del_nbp(p);
	310	}
	311
1ea2d020	312	br_fdb_delete_by_port(br, NULL, 0, 1);
f8730420	313
b4e09b29	314	br_vlan_flush(br);
e10177ab	315	br_multicast_dev_del(br);
1da177e4 LT	316	del_timer_sync(&br->gc_timer);
	317
	318	br_sysfs_delbr(br->dev);
8c56ba05	319	unregister_netdevice_queue(br->dev, head);
1da177e4 LT	320	}
1da177e4 LT	321
1da177e4 LT	322	/* find an available port number */
	323	static int find_portno(struct net_bridge *br)
	324	{
	325	int index;
	326	struct net_bridge_port *p;
	327	unsigned long *inuse;
	328
3b781fa1	329	inuse = kcalloc(BITS_TO_LONGS(BR_MAX_PORTS), sizeof(unsigned long),
1da177e4 LT	330	GFP_KERNEL);
	331	if (!inuse)
	332	return -ENOMEM;
	333
1da177e4 LT	334	set_bit(0, inuse); /* zero is reserved */
	335	list_for_each_entry(p, &br->port_list, list) {
	336	set_bit(p->port_no, inuse);
	337	}
	338	index = find_first_zero_bit(inuse, BR_MAX_PORTS);
	339	kfree(inuse);
	340
	341	return (index >= BR_MAX_PORTS) ? -EXFULL : index;
	342	}
	343
4433f420	344	/* called with RTNL but without bridge lock */
9d6f229f	345	static struct net_bridge_port new_nbp(struct net_bridge br,
4433f420	346	struct net_device *dev)
1da177e4	347	{
1da177e4	348	struct net_bridge_port *p;
1080ab95	349	int index, err;
9d6f229f	350
1da177e4 LT	351	index = find_portno(br);
	352	if (index < 0)
	353	return ERR_PTR(index);
	354
cee48541	355	p = kzalloc(sizeof(*p), GFP_KERNEL);
1da177e4 LT	356	if (p == NULL)
	357	return ERR_PTR(-ENOMEM);
	358
1da177e4 LT	359	p->br = br;
	360	dev_hold(dev);
	361	p->dev = dev;
4433f420	362	p->path_cost = port_cost(dev);
9d6f229f	363	p->priority = 0x8000 >> BR_PORT_BITS;
1da177e4	364	p->port_no = index;
b6cb5ac8	365	p->flags = BR_LEARNING \| BR_FLOOD \| BR_MCAST_FLOOD;
1da177e4	366	br_init_port(p);
775dd692	367	br_set_state(p, BR_STATE_DISABLED);
d32439c0	368	br_stp_port_timer_init(p);
1080ab95 NA	369	err = br_multicast_add_port(p);
	370	if (err) {
	371	dev_put(dev);
	372	kfree(p);
	373	p = ERR_PTR(err);
	374	}
1da177e4 LT	375
	376	return p;
	377	}
	378
4aa678ba	379	int br_add_bridge(struct net net, const char name)
1da177e4 LT	380	{
1da177e4 LT	381	struct net_device *dev;
11f3a6bd	382	int res;
1da177e4	383
c835a677	384	dev = alloc_netdev(sizeof(struct net_bridge), name, NET_NAME_UNKNOWN,
bb900b27	385	br_dev_setup);
bb900b27	386
9d6f229f	387	if (!dev)
1da177e4 LT	388	return -ENOMEM;
1da177e4 LT	389
bb900b27	390	dev_net_set(dev, net);
149ddd83	391	dev->rtnl_link_ops = &br_link_ops;
e340a90e	392
11f3a6bd ED	393	res = register_netdev(dev);
	394	if (res)
	395	free_netdev(dev);
	396	return res;
1da177e4 LT	397	}
1da177e4 LT	398
4aa678ba	399	int br_del_bridge(struct net net, const char name)
1da177e4 LT	400	{
	401	struct net_device *dev;
	402	int ret = 0;
	403
	404	rtnl_lock();
4aa678ba	405	dev = __dev_get_by_name(net, name);
9d6f229f	406	if (dev == NULL)
1da177e4 LT	407	ret = -ENXIO; /* Could not find device */
	408
	409	else if (!(dev->priv_flags & IFF_EBRIDGE)) {
	410	/* Attempt to delete non bridge device! */
	411	ret = -EPERM;
	412	}
	413
	414	else if (dev->flags & IFF_UP) {
	415	/* Not shutdown yet. */
	416	ret = -EBUSY;
9d6f229f	417	}
1da177e4	418
9d6f229f	419	else
1ce5cce8	420	br_dev_delete(dev, NULL);
1da177e4 LT	421
	422	rtnl_unlock();
	423	return ret;
	424	}
	425
46f25dff	426	/* MTU of the bridge pseudo-device: ETH_DATA_LEN or the minimum of the ports */
1da177e4 LT	427	int br_min_mtu(const struct net_bridge *br)
	428	{
	429	const struct net_bridge_port *p;
	430	int mtu = 0;
	431
	432	ASSERT_RTNL();
	433
	434	if (list_empty(&br->port_list))
46f25dff	435	mtu = ETH_DATA_LEN;
1da177e4 LT	436	else {
	437	list_for_each_entry(p, &br->port_list, list) {
	438	if (!mtu \|\| p->dev->mtu < mtu)
	439	mtu = p->dev->mtu;
	440	}
	441	}
	442	return mtu;
	443	}
	444
ae74f100 ED	445	static void br_set_gso_limits(struct net_bridge *br)
	446	{
	447	unsigned int gso_max_size = GSO_MAX_SIZE;
	448	u16 gso_max_segs = GSO_MAX_SEGS;
	449	const struct net_bridge_port *p;
	450
	451	list_for_each_entry(p, &br->port_list, list) {
	452	gso_max_size = min(gso_max_size, p->dev->gso_max_size);
	453	gso_max_segs = min(gso_max_segs, p->dev->gso_max_segs);
	454	}
	455	br->dev->gso_max_size = gso_max_size;
	456	br->dev->gso_max_segs = gso_max_segs;
	457	}
	458
81d35307 SH	459	/*
	460	* Recomputes features using slave's features
	461	*/
c8f44aff MM	462	netdev_features_t br_features_recompute(struct net_bridge *br,
c8f44aff MM	463	netdev_features_t features)
81d35307 SH	464	{
81d35307 SH	465	struct net_bridge_port *p;
c8f44aff	466	netdev_features_t mask;
81d35307	467
b63365a2	468	if (list_empty(&br->port_list))
c4d27ef9	469	return features;
b63365a2	470
c4d27ef9	471	mask = features;
b63365a2	472	features &= ~NETIF_F_ONE_FOR_ALL;
81d35307 SH	473
81d35307 SH	474	list_for_each_entry(p, &br->port_list, list) {
b63365a2 HX	475	features = netdev_increment_features(features,
b63365a2 HX	476	p->dev->features, mask);
81d35307	477	}
f902e881	478	features = netdev_add_tso_features(features, mask);
81d35307	479
c4d27ef9	480	return features;
81d35307 SH	481	}
81d35307 SH	482
1da177e4 LT	483	/* called with RTNL */
	484	int br_add_if(struct net_bridge br, struct net_device dev)
	485	{
	486	struct net_bridge_port *p;
	487	int err = 0;
45493d47	488	unsigned br_hr, dev_hr;
edf947f1	489	bool changed_addr;
1da177e4	490
8db0a2ee FF	491	/* Don't allow bridging non-ethernet like devices, or DSA-enabled
	492	* master network devices since the bridge layer rx_handler prevents
	493	* the DSA fake ethertype handler to be invoked, so we do not strip off
	494	* the DSA switch tag protocol header and the bridge layer just return
	495	* RX_HANDLER_CONSUMED, stopping RX processing for these frames.
	496	*/
1056bd51	497	if ((dev->flags & IFF_LOOPBACK) \|\|
77f98598	498	dev->type != ARPHRD_ETHER \|\| dev->addr_len != ETH_ALEN \|\|
8db0a2ee FF	499	!is_valid_ether_addr(dev->dev_addr) \|\|
8db0a2ee FF	500	netdev_uses_dsa(dev))
1da177e4 LT	501	return -EINVAL;
1da177e4 LT	502
1056bd51	503	/* No bridging of bridges */
00829823	504	if (dev->netdev_ops->ndo_start_xmit == br_dev_xmit)
1da177e4 LT	505	return -ELOOP;
1da177e4 LT	506
1056bd51	507	/* Device is already being bridged */
f350a0a8	508	if (br_port_exists(dev))
1da177e4 LT	509	return -EBUSY;
1da177e4 LT	510
ad4bb6f8 JB	511	/* No bridging devices that dislike that (e.g. wireless) */
	512	if (dev->priv_flags & IFF_DONT_BRIDGE)
	513	return -EOPNOTSUPP;
	514
bab1deea SH	515	p = new_nbp(br, dev);
bab1deea SH	516	if (IS_ERR(p))
1da177e4 LT	517	return PTR_ERR(p);
1da177e4 LT	518
bb8ed630 AW	519	call_netdevice_notifiers(NETDEV_JOIN, dev);
bb8ed630 AW	520
2796d0c6	521	err = dev_set_allmulti(dev, 1);
bc3f9076 WC	522	if (err)
	523	goto put_back;
	524
e32cc736 GKH	525	err = kobject_init_and_add(&p->kobj, &brport_ktype, &(dev->dev.kobj),
e32cc736 GKH	526	SYSFS_BRIDGE_PORT_ATTR);
bab1deea SH	527	if (err)
bab1deea SH	528	goto err1;
1da177e4	529
bab1deea SH	530	err = br_sysfs_addif(p);
	531	if (err)
	532	goto err2;
1da177e4	533
a8779ec1	534	err = br_netpoll_enable(p);
93d8bf9f	535	if (err)
91d2c34a HX	536	goto err3;
91d2c34a HX	537
0f49579a	538	err = netdev_rx_handler_register(dev, br_handle_frame, p);
ab95bfe0	539	if (err)
9b1536c4	540	goto err4;
f350a0a8	541
0f49579a JP	542	dev->priv_flags \|= IFF_BRIDGE_PORT;
0f49579a JP	543
29bf24af	544	err = netdev_master_upper_dev_link(dev, br->dev, NULL, NULL);
afc6151a	545	if (err)
9b1536c4	546	goto err5;
afc6151a	547
6bc506b4 IS	548	err = nbp_switchdev_mark_set(p);
	549	if (err)
	550	goto err6;
	551
0187bdfb	552	dev_disable_lro(dev);
bab1deea SH	553
	554	list_add_rcu(&p->list, &br->port_list);
	555
e028e4b8 VY	556	nbp_update_port_count(br);
e028e4b8 VY	557
c4d27ef9 MM	558	netdev_update_features(br->dev);
c4d27ef9 MM	559
45493d47 PA	560	br_hr = br->dev->needed_headroom;
	561	dev_hr = netdev_get_fwd_headroom(dev);
	562	if (br_hr < dev_hr)
	563	update_headroom(br, dev_hr);
	564	else
	565	netdev_set_rx_headroom(dev, br_hr);
fd094808	566
a4b816d8 TM	567	if (br_fdb_insert(br, p, dev->dev_addr, 0))
	568	netdev_err(dev, "failed insert local address bridge forwarding table\n");
	569
08474cc1 ER	570	err = nbp_vlan_init(p);
08474cc1 ER	571	if (err) {
5be5a2df	572	netdev_err(dev, "failed to initialize vlan filtering on this port\n");
6bc506b4	573	goto err7;
08474cc1	574	}
5be5a2df	575
bab1deea	576	spin_lock_bh(&br->lock);
edf947f1	577	changed_addr = br_stp_recalculate_bridge_id(br);
de79059e	578
576eb625	579	if (netif_running(dev) && netif_oper_up(dev) &&
de79059e AS	580	(br->dev->flags & IFF_UP))
de79059e AS	581	br_stp_enable_port(p);
bab1deea SH	582	spin_unlock_bh(&br->lock);
bab1deea SH	583
b86c4503 SH	584	br_ifinfo_notify(RTM_NEWLINK, p);
b86c4503 SH	585
edf947f1	586	if (changed_addr)
56139fc5	587	call_netdevice_notifiers(NETDEV_CHANGEADDR, br->dev);
edf947f1	588
bab1deea	589	dev_set_mtu(br->dev, br_min_mtu(br));
ae74f100	590	br_set_gso_limits(br);
269def7c	591
bab1deea	592	kobject_uevent(&p->kobj, KOBJ_ADD);
1da177e4	593
bab1deea	594	return 0;
afc6151a	595
6bc506b4	596	err7:
08474cc1 ER	597	list_del_rcu(&p->list);
	598	br_fdb_delete_by_port(br, p, 0, 1);
	599	nbp_update_port_count(br);
6bc506b4	600	err6:
08474cc1	601	netdev_upper_dev_unlink(dev, br->dev);
9b1536c4	602	err5:
0f49579a JP	603	dev->priv_flags &= ~IFF_BRIDGE_PORT;
0f49579a JP	604	netdev_rx_handler_unregister(dev);
9b1536c4 G	605	err4:
9b1536c4 G	606	br_netpoll_disable(p);
91d2c34a HX	607	err3:
91d2c34a HX	608	sysfs_remove_link(br->ifobj, p->dev->name);
bab1deea	609	err2:
c587aea9	610	kobject_put(&p->kobj);
30df94f8	611	p = NULL; /* kobject_put frees */
77f98598	612	err1:
019ee792	613	dev_set_allmulti(dev, -1);
43af8532 VL	614	put_back:
43af8532 VL	615	dev_put(dev);
bc3f9076	616	kfree(p);
1da177e4 LT	617	return err;
	618	}
	619
	620	/* called with RTNL */
	621	int br_del_if(struct net_bridge br, struct net_device dev)
	622	{
f350a0a8	623	struct net_bridge_port *p;
9be6dd65	624	bool changed_addr;
f350a0a8	625
ec1e5610	626	p = br_port_get_rtnl(dev);
b5ed54e9	627	if (!p \|\| p->br != br)
1da177e4 LT	628	return -EINVAL;
1da177e4 LT	629
d30362c0 AW	630	/* Since more than one interface can be attached to a bridge,
	631	* there still maybe an alternate path for netconsole to use;
	632	* therefore there is no reason for a NETDEV_RELEASE event.
	633	*/
1da177e4 LT	634	del_nbp(p);
1da177e4 LT	635
4c906c27	636	dev_set_mtu(br->dev, br_min_mtu(br));
ae74f100	637	br_set_gso_limits(br);
4c906c27	638
1da177e4	639	spin_lock_bh(&br->lock);
9be6dd65	640	changed_addr = br_stp_recalculate_bridge_id(br);
1da177e4 LT	641	spin_unlock_bh(&br->lock);
1da177e4 LT	642
9be6dd65 AW	643	if (changed_addr)
	644	call_netdevice_notifiers(NETDEV_CHANGEADDR, br->dev);
	645
c4d27ef9 MM	646	netdev_update_features(br->dev);
c4d27ef9 MM	647
1da177e4 LT	648	return 0;
1da177e4 LT	649	}
e028e4b8 VY	650
	651	void br_port_flags_change(struct net_bridge_port *p, unsigned long mask)
	652	{
	653	struct net_bridge *br = p->br;
	654
	655	if (mask & BR_AUTO_MASK)
	656	nbp_update_port_count(br);
	657	}