datapath.c

/*
 * Copyright (c) 2007-2014 Nicira, Inc.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of version 2 of the GNU General Public
 * License as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
 * 02110-1301, USA
 */

#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

#include <linux/init.h>
#include <linux/module.h>
#include <linux/if_arp.h>
#include <linux/if_vlan.h>
#include <linux/in.h>
#include <linux/ip.h>
#include <linux/jhash.h>
#include <linux/delay.h>
#include <linux/time.h>
#include <linux/etherdevice.h>
#include <linux/genetlink.h>
#include <linux/kernel.h>
#include <linux/kthread.h>
#include <linux/mutex.h>
#include <linux/percpu.h>
#include <linux/rcupdate.h>
#include <linux/tcp.h>
#include <linux/udp.h>
#include <linux/ethtool.h>
#include <linux/wait.h>
#include <asm/div64.h>
#include <linux/highmem.h>
#include <linux/netfilter_bridge.h>
#include <linux/netfilter_ipv4.h>
#include <linux/inetdevice.h>
#include <linux/list.h>
#include <linux/openvswitch.h>
#include <linux/rculist.h>
#include <linux/dmi.h>
#include <net/genetlink.h>
#include <net/net_namespace.h>
#include <net/netns/generic.h>

#include "datapath.h"
#include "flow.h"
#include "flow_table.h"
#include "flow_netlink.h"
#include "vport-internal_dev.h"
#include "vport-netdev.h"

int ovs_net_id __read_mostly;
EXPORT_SYMBOL_GPL(ovs_net_id);

static struct genl_family dp_packet_genl_family;
static struct genl_family dp_flow_genl_family;
static struct genl_family dp_datapath_genl_family;

static const struct nla_policy flow_policy[];

static const struct genl_multicast_group ovs_dp_flow_multicast_group = {
	.name = OVS_FLOW_MCGROUP,
};

static const struct genl_multicast_group ovs_dp_datapath_multicast_group = {
	.name = OVS_DATAPATH_MCGROUP,
};

static const struct genl_multicast_group ovs_dp_vport_multicast_group = {
	.name = OVS_VPORT_MCGROUP,
};

/* Check if need to build a reply message.
 * OVS userspace sets the NLM_F_ECHO flag if it needs the reply. */
static bool ovs_must_notify(struct genl_family *family, struct genl_info *info,
			    unsigned int group)
{
	return info->nlhdr->nlmsg_flags & NLM_F_ECHO ||
	       genl_has_listeners(family, genl_info_net(info), group);
}

static void ovs_notify(struct genl_family *family,
		       struct sk_buff *skb, struct genl_info *info)
{
	genl_notify(family, skb, genl_info_net(info), info->snd_portid,
		    0, info->nlhdr, GFP_KERNEL);
}

/**
 * DOC: Locking:
 *
 * All writes e.g. Writes to device state (add/remove datapath, port, set
 * operations on vports, etc.), Writes to other state (flow table
 * modifications, set miscellaneous datapath parameters, etc.) are protected
 * by ovs_lock.
 *
 * Reads are protected by RCU.
 *
 * There are a few special cases (mostly stats) that have their own
 * synchronization but they nest under all of above and don't interact with
 * each other.
 *
 * The RTNL lock nests inside ovs_mutex.
 */

static DEFINE_MUTEX(ovs_mutex);

void ovs_lock(void)
{
	mutex_lock(&ovs_mutex);
}

void ovs_unlock(void)
{
	mutex_unlock(&ovs_mutex);
}

#ifdef CONFIG_LOCKDEP
int lockdep_ovsl_is_held(void)
{
	if (debug_locks)
		return lockdep_is_held(&ovs_mutex);
	else
		return 1;
}
EXPORT_SYMBOL_GPL(lockdep_ovsl_is_held);
#endif

static struct vport *new_vport(const struct vport_parms *);
static int queue_gso_packets(struct datapath *dp, struct sk_buff *,
			     const struct sw_flow_key *,
			     const struct dp_upcall_info *);
static int queue_userspace_packet(struct datapath *dp, struct sk_buff *,
				  const struct sw_flow_key *,
				  const struct dp_upcall_info *);

/* Must be called with rcu_read_lock. */
static struct datapath *get_dp_rcu(struct net *net, int dp_ifindex)
{
	struct net_device *dev = dev_get_by_index_rcu(net, dp_ifindex);

	if (dev) {
		struct vport *vport = ovs_internal_dev_get_vport(dev);
		if (vport)
			return vport->dp;
	}

	return NULL;
}

/* The caller must hold either ovs_mutex or rcu_read_lock to keep the
 * returned dp pointer valid.
 */
static inline struct datapath *get_dp(struct net *net, int dp_ifindex)
{
	struct datapath *dp;

	WARN_ON_ONCE(!rcu_read_lock_held() && !lockdep_ovsl_is_held());
	rcu_read_lock();
	dp = get_dp_rcu(net, dp_ifindex);
	rcu_read_unlock();

	return dp;
}

/* Must be called with rcu_read_lock or ovs_mutex. */
const char *ovs_dp_name(const struct datapath *dp)
{
	struct vport *vport = ovs_vport_ovsl_rcu(dp, OVSP_LOCAL);
	return ovs_vport_name(vport);
}

static int get_dpifindex(const struct datapath *dp)
{
	struct vport *local;
	int ifindex;

	rcu_read_lock();

	local = ovs_vport_rcu(dp, OVSP_LOCAL);
	if (local)
		ifindex = local->dev->ifindex;
	else
		ifindex = 0;

	rcu_read_unlock();

	return ifindex;
}

static void destroy_dp_rcu(struct rcu_head *rcu)
{
	struct datapath *dp = container_of(rcu, struct datapath, rcu);

	ovs_flow_tbl_destroy(&dp->table);
	free_percpu(dp->stats_percpu);
	kfree(dp->ports);
	kfree(dp);
}

static struct hlist_head *vport_hash_bucket(const struct datapath *dp,
					    u16 port_no)
{
	return &dp->ports[port_no & (DP_VPORT_HASH_BUCKETS - 1)];
}

/* Called with ovs_mutex or RCU read lock. */
struct vport *ovs_lookup_vport(const struct datapath *dp, u16 port_no)
{
	struct vport *vport;
	struct hlist_head *head;

	head = vport_hash_bucket(dp, port_no);
	hlist_for_each_entry_rcu(vport, head, dp_hash_node) {
		if (vport->port_no == port_no)
			return vport;
	}
	return NULL;
}

/* Called with ovs_mutex. */
static struct vport *new_vport(const struct vport_parms *parms)
{
	struct vport *vport;

	vport = ovs_vport_add(parms);
	if (!IS_ERR(vport)) {
		struct datapath *dp = parms->dp;
		struct hlist_head *head = vport_hash_bucket(dp, vport->port_no);

		hlist_add_head_rcu(&vport->dp_hash_node, head);
	}
	return vport;
}

void ovs_dp_detach_port(struct vport *p)
{
	ASSERT_OVSL();

	/* First drop references to device. */
	hlist_del_rcu(&p->dp_hash_node);

	/* Then destroy it. */
	ovs_vport_del(p);
}

/* Must be called with rcu_read_lock. */
void ovs_dp_process_packet(struct sk_buff *skb, struct sw_flow_key *key)
{
	const struct vport *p = OVS_CB(skb)->input_vport;
	struct datapath *dp = p->dp;
	struct sw_flow *flow;
	struct sw_flow_actions *sf_acts;
	struct dp_stats_percpu *stats;
	u64 *stats_counter;
	u32 n_mask_hit;

	stats = this_cpu_ptr(dp->stats_percpu);

	/* Look up flow. */
	flow = ovs_flow_tbl_lookup_stats(&dp->table, key, &n_mask_hit);
	if (unlikely(!flow)) {
		struct dp_upcall_info upcall;
		int error;

		memset(&upcall, 0, sizeof(upcall));
		upcall.cmd = OVS_PACKET_CMD_MISS;
		upcall.portid = ovs_vport_find_upcall_portid(p, skb);
		error = ovs_dp_upcall(dp, skb, key, &upcall);
		if (unlikely(error))
			kfree_skb(skb);
		else
			consume_skb(skb);
		stats_counter = &stats->n_missed;
		goto out;
	}

	ovs_flow_stats_update(flow, key->tp.flags, skb);
	sf_acts = rcu_dereference(flow->sf_acts);
	ovs_execute_actions(dp, skb, sf_acts, key);

	stats_counter = &stats->n_hit;

out:
	/* Update datapath statistics. */
	u64_stats_update_begin(&stats->syncp);
	(*stats_counter)++;
	stats->n_mask_hit += n_mask_hit;
	u64_stats_update_end(&stats->syncp);
}

int ovs_dp_upcall(struct datapath *dp, struct sk_buff *skb,
		  const struct sw_flow_key *key,
		  const struct dp_upcall_info *upcall_info)
{
	struct dp_stats_percpu *stats;
	int err;

	if (upcall_info->portid == 0) {
		err = -ENOTCONN;
		goto err;
	}

	if (!skb_is_gso(skb))
		err = queue_userspace_packet(dp, skb, key, upcall_info);
	else
		err = queue_gso_packets(dp, skb, key, upcall_info);
	if (err)
		goto err;

	return 0;

err:
	stats = this_cpu_ptr(dp->stats_percpu);

	u64_stats_update_begin(&stats->syncp);
	stats->n_lost++;
	u64_stats_update_end(&stats->syncp);

	return err;
}

static int queue_gso_packets(struct datapath *dp, struct sk_buff *skb,
			     const struct sw_flow_key *key,
			     const struct dp_upcall_info *upcall_info)
{
	unsigned short gso_type = skb_shinfo(skb)->gso_type;
	struct sw_flow_key later_key;
	struct sk_buff *segs, *nskb;
	struct ovs_skb_cb ovs_cb;
	int err;

	ovs_cb = *OVS_CB(skb);
	segs = __skb_gso_segment(skb, NETIF_F_SG, false);
	*OVS_CB(skb) = ovs_cb;
	if (IS_ERR(segs))
		return PTR_ERR(segs);
	if (segs == NULL)
		return -EINVAL;

	if (gso_type & SKB_GSO_UDP) {
		/* The initial flow key extracted by ovs_flow_key_extract()
		 * in this case is for a first fragment, so we need to
		 * properly mark later fragments.
		 */
		later_key = *key;
		later_key.ip.frag = OVS_FRAG_TYPE_LATER;
	}

	/* Queue all of the segments. */
	skb = segs;
	do {
		*OVS_CB(skb) = ovs_cb;
		if (gso_type & SKB_GSO_UDP && skb != segs)
			key = &later_key;

		err = queue_userspace_packet(dp, skb, key, upcall_info);
		if (err)
			break;

	} while ((skb = skb->next));

	/* Free all of the segments. */
	skb = segs;
	do {
		nskb = skb->next;
		if (err)
			kfree_skb(skb);
		else
			consume_skb(skb);
	} while ((skb = nskb));
	return err;
}

static size_t upcall_msg_size(const struct dp_upcall_info *upcall_info,
			      unsigned int hdrlen)
{
	size_t size = NLMSG_ALIGN(sizeof(struct ovs_header))
		+ nla_total_size(hdrlen) /* OVS_PACKET_ATTR_PACKET */
		+ nla_total_size(ovs_key_attr_size()); /* OVS_PACKET_ATTR_KEY */

	/* OVS_PACKET_ATTR_USERDATA */
	if (upcall_info->userdata)
		size += NLA_ALIGN(upcall_info->userdata->nla_len);

	/* OVS_PACKET_ATTR_EGRESS_TUN_KEY */
	if (upcall_info->egress_tun_info)
		size += nla_total_size(ovs_tun_key_attr_size());

	/* OVS_PACKET_ATTR_ACTIONS */
	if (upcall_info->actions_len)
		size += nla_total_size(upcall_info->actions_len);

	return size;
}

static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
				  const struct sw_flow_key *key,
				  const struct dp_upcall_info *upcall_info)
{
	struct ovs_header *upcall;
	struct sk_buff *nskb = NULL;
	struct sk_buff *user_skb = NULL; /* to be queued to userspace */
	struct nlattr *nla;
	struct genl_info info = {
		.dst_sk = ovs_dp_get_net(dp)->genl_sock,
		.snd_portid = upcall_info->portid,
	};
	size_t len;
	unsigned int hlen;
	int err, dp_ifindex;

	dp_ifindex = get_dpifindex(dp);
	if (!dp_ifindex)
		return -ENODEV;

	if (skb_vlan_tag_present(skb)) {
		nskb = skb_clone(skb, GFP_ATOMIC);
		if (!nskb)
			return -ENOMEM;

		nskb = __vlan_hwaccel_push_inside(nskb);
		if (!nskb)
			return -ENOMEM;

		skb = nskb;
	}

	if (nla_attr_size(skb->len) > USHRT_MAX) {
		err = -EFBIG;
		goto out;
	}

	/* Complete checksum if needed */
	if (skb->ip_summed == CHECKSUM_PARTIAL &&
	    (err = skb_checksum_help(skb)))
		goto out;

	/* Older versions of OVS user space enforce alignment of the last
	 * Netlink attribute to NLA_ALIGNTO which would require extensive
	 * padding logic. Only perform zerocopy if padding is not required.
	 */
	if (dp->user_features & OVS_DP_F_UNALIGNED)
		hlen = skb_zerocopy_headlen(skb);
	else
		hlen = skb->len;

	len = upcall_msg_size(upcall_info, hlen);
	user_skb = genlmsg_new_unicast(len, &info, GFP_ATOMIC);
	if (!user_skb) {
		err = -ENOMEM;
		goto out;
	}

	upcall = genlmsg_put(user_skb, 0, 0, &dp_packet_genl_family,
			     0, upcall_info->cmd);
	upcall->dp_ifindex = dp_ifindex;

	err = ovs_nla_put_key(key, key, OVS_PACKET_ATTR_KEY, false, user_skb);
	BUG_ON(err);

	if (upcall_info->userdata)
		__nla_put(user_skb, OVS_PACKET_ATTR_USERDATA,
			  nla_len(upcall_info->userdata),
			  nla_data(upcall_info->userdata));

	if (upcall_info->egress_tun_info) {
		nla = nla_nest_start(user_skb, OVS_PACKET_ATTR_EGRESS_TUN_KEY);
		err = ovs_nla_put_egress_tunnel_key(user_skb,
						    upcall_info->egress_tun_info);
		BUG_ON(err);
		nla_nest_end(user_skb, nla);
	}

	if (upcall_info->actions_len) {
		nla = nla_nest_start(user_skb, OVS_PACKET_ATTR_ACTIONS);
		err = ovs_nla_put_actions(upcall_info->actions,
					  upcall_info->actions_len,
					  user_skb);
		if (!err)
			nla_nest_end(user_skb, nla);
		else
			nla_nest_cancel(user_skb, nla);
	}

	/* Only reserve room for attribute header, packet data is added
	 * in skb_zerocopy() */
	if (!(nla = nla_reserve(user_skb, OVS_PACKET_ATTR_PACKET, 0))) {
		err = -ENOBUFS;
		goto out;
	}
	nla->nla_len = nla_attr_size(skb->len);

	err = skb_zerocopy(user_skb, skb, skb->len, hlen);
	if (err)
		goto out;

	/* Pad OVS_PACKET_ATTR_PACKET if linear copy was performed */
	if (!(dp->user_features & OVS_DP_F_UNALIGNED)) {
		size_t plen = NLA_ALIGN(user_skb->len) - user_skb->len;

		if (plen > 0)
			memset(skb_put(user_skb, plen), 0, plen);
	}

	((struct nlmsghdr *) user_skb->data)->nlmsg_len = user_skb->len;

	err = genlmsg_unicast(ovs_dp_get_net(dp), user_skb, upcall_info->portid);
	user_skb = NULL;
out:
	if (err)
		skb_tx_error(skb);
	kfree_skb(user_skb);
	kfree_skb(nskb);
	return err;
}

static int ovs_packet_cmd_execute(struct sk_buff *skb, struct genl_info *info)
{
	struct ovs_header *ovs_header = info->userhdr;
	struct nlattr **a = info->attrs;
	struct sw_flow_actions *acts;
	struct sk_buff *packet;
	struct sw_flow *flow;
	struct sw_flow_actions *sf_acts;
	struct datapath *dp;
	struct ethhdr *eth;
	struct vport *input_vport;
	int len;
	int err;
	bool log = !a[OVS_PACKET_ATTR_PROBE];

	err = -EINVAL;
	if (!a[OVS_PACKET_ATTR_PACKET] || !a[OVS_PACKET_ATTR_KEY] ||
	    !a[OVS_PACKET_ATTR_ACTIONS])
		goto err;

	len = nla_len(a[OVS_PACKET_ATTR_PACKET]);
	packet = __dev_alloc_skb(NET_IP_ALIGN + len, GFP_KERNEL);
	err = -ENOMEM;
	if (!packet)
		goto err;
	skb_reserve(packet, NET_IP_ALIGN);

	nla_memcpy(__skb_put(packet, len), a[OVS_PACKET_ATTR_PACKET], len);

	skb_reset_mac_header(packet);
	eth = eth_hdr(packet);

	/* Normally, setting the skb 'protocol' field would be handled by a
	 * call to eth_type_trans(), but it assumes there's a sending
	 * device, which we may not have. */
	if (eth_proto_is_802_3(eth->h_proto))
		packet->protocol = eth->h_proto;
	else
		packet->protocol = htons(ETH_P_802_2);

	/* Build an sw_flow for sending this packet. */
	flow = ovs_flow_alloc();
	err = PTR_ERR(flow);
	if (IS_ERR(flow))
		goto err_kfree_skb;

	err = ovs_flow_key_extract_userspace(a[OVS_PACKET_ATTR_KEY], packet,
					     &flow->key, log);
	if (err)
		goto err_flow_free;

	err = ovs_nla_copy_actions(a[OVS_PACKET_ATTR_ACTIONS],
				   &flow->key, &acts, log);
	if (err)
		goto err_flow_free;

	rcu_assign_pointer(flow->sf_acts, acts);
	OVS_CB(packet)->egress_tun_info = NULL;
	packet->priority = flow->key.phy.priority;
	packet->mark = flow->key.phy.skb_mark;

	rcu_read_lock();
	dp = get_dp_rcu(sock_net(skb->sk), ovs_header->dp_ifindex);
	err = -ENODEV;
	if (!dp)
		goto err_unlock;

	input_vport = ovs_vport_rcu(dp, flow->key.phy.in_port);
	if (!input_vport)
		input_vport = ovs_vport_rcu(dp, OVSP_LOCAL);

	if (!input_vport)
		goto err_unlock;

	OVS_CB(packet)->input_vport = input_vport;
	sf_acts = rcu_dereference(flow->sf_acts);

	local_bh_disable();
	err = ovs_execute_actions(dp, packet, sf_acts, &flow->key);
	local_bh_enable();
	rcu_read_unlock();

	ovs_flow_free(flow, false);
	return err;

err_unlock:
	rcu_read_unlock();
err_flow_free:
	ovs_flow_free(flow, false);
err_kfree_skb:
	kfree_skb(packet);
err:
	return err;
}

static const struct nla_policy packet_policy[OVS_PACKET_ATTR_MAX + 1] = {
	[OVS_PACKET_ATTR_PACKET] = { .len = ETH_HLEN },
	[OVS_PACKET_ATTR_KEY] = { .type = NLA_NESTED },
	[OVS_PACKET_ATTR_ACTIONS] = { .type = NLA_NESTED },
	[OVS_PACKET_ATTR_PROBE] = { .type = NLA_FLAG },
};

static const struct genl_ops dp_packet_genl_ops[] = {
	{ .cmd = OVS_PACKET_CMD_EXECUTE,
	  .flags = GENL_ADMIN_PERM, /* Requires CAP_NET_ADMIN privilege. */
	  .policy = packet_policy,
	  .doit = ovs_packet_cmd_execute
	}
};

static struct genl_family dp_packet_genl_family = {
	.id = GENL_ID_GENERATE,
	.hdrsize = sizeof(struct ovs_header),
	.name = OVS_PACKET_FAMILY,
	.version = OVS_PACKET_VERSION,
	.maxattr = OVS_PACKET_ATTR_MAX,
	.netnsok = true,
	.parallel_ops = true,
	.ops = dp_packet_genl_ops,
	.n_ops = ARRAY_SIZE(dp_packet_genl_ops),
};

static void get_dp_stats(const struct datapath *dp, struct ovs_dp_stats *stats,
			 struct ovs_dp_megaflow_stats *mega_stats)
{
	int i;

	memset(mega_stats, 0, sizeof(*mega_stats));

	stats->n_flows = ovs_flow_tbl_count(&dp->table);
	mega_stats->n_masks = ovs_flow_tbl_num_masks(&dp->table);

	stats->n_hit = stats->n_missed = stats->n_lost = 0;

	for_each_possible_cpu(i) {
		const struct dp_stats_percpu *percpu_stats;
		struct dp_stats_percpu local_stats;
		unsigned int start;

		percpu_stats = per_cpu_ptr(dp->stats_percpu, i);

		do {
			start = u64_stats_fetch_begin_irq(&percpu_stats->syncp);
			local_stats = *percpu_stats;
		} while (u64_stats_fetch_retry_irq(&percpu_stats->syncp, start));

		stats->n_hit += local_stats.n_hit;
		stats->n_missed += local_stats.n_missed;
		stats->n_lost += local_stats.n_lost;
		mega_stats->n_mask_hit += local_stats.n_mask_hit;
	}
}

static bool should_fill_key(const struct sw_flow_id *sfid, uint32_t ufid_flags)
{
	return ovs_identifier_is_ufid(sfid) &&
	       !(ufid_flags & OVS_UFID_F_OMIT_KEY);
}

static bool should_fill_mask(uint32_t ufid_flags)
{
	return !(ufid_flags & OVS_UFID_F_OMIT_MASK);
}

static bool should_fill_actions(uint32_t ufid_flags)
{
	return !(ufid_flags & OVS_UFID_F_OMIT_ACTIONS);
}

static size_t ovs_flow_cmd_msg_size(const struct sw_flow_actions *acts,
				    const struct sw_flow_id *sfid,
				    uint32_t ufid_flags)
{
	size_t len = NLMSG_ALIGN(sizeof(struct ovs_header));

	/* OVS_FLOW_ATTR_UFID */
	if (sfid && ovs_identifier_is_ufid(sfid))
		len += nla_total_size(sfid->ufid_len);

	/* OVS_FLOW_ATTR_KEY */
	if (!sfid || should_fill_key(sfid, ufid_flags))
		len += nla_total_size(ovs_key_attr_size());

	/* OVS_FLOW_ATTR_MASK */
	if (should_fill_mask(ufid_flags))
		len += nla_total_size(ovs_key_attr_size());

	/* OVS_FLOW_ATTR_ACTIONS */
	if (should_fill_actions(ufid_flags))
		len += nla_total_size(acts->actions_len);

	return len
		+ nla_total_size(sizeof(struct ovs_flow_stats)) /* OVS_FLOW_ATTR_STATS */
		+ nla_total_size(1) /* OVS_FLOW_ATTR_TCP_FLAGS */
		+ nla_total_size(8); /* OVS_FLOW_ATTR_USED */
}

/* Called with ovs_mutex or RCU read lock. */
static int ovs_flow_cmd_fill_stats(const struct sw_flow *flow,
				   struct sk_buff *skb)
{
	struct ovs_flow_stats stats;
	__be16 tcp_flags;
	unsigned long used;

	ovs_flow_stats_get(flow, &stats, &used, &tcp_flags);

	if (used &&
	    nla_put_u64(skb, OVS_FLOW_ATTR_USED, ovs_flow_used_time(used)))
		return -EMSGSIZE;

	if (stats.n_packets &&
	    nla_put(skb, OVS_FLOW_ATTR_STATS, sizeof(struct ovs_flow_stats), &stats))
		return -EMSGSIZE;

	if ((u8)ntohs(tcp_flags) &&
	     nla_put_u8(skb, OVS_FLOW_ATTR_TCP_FLAGS, (u8)ntohs(tcp_flags)))
		return -EMSGSIZE;

	return 0;
}

/* Called with ovs_mutex or RCU read lock. */
static int ovs_flow_cmd_fill_actions(const struct sw_flow *flow,
				     struct sk_buff *skb, int skb_orig_len)
{
	struct nlattr *start;
	int err;

	/* If OVS_FLOW_ATTR_ACTIONS doesn't fit, skip dumping the actions if
	 * this is the first flow to be dumped into 'skb'.  This is unusual for
	 * Netlink but individual action lists can be longer than
	 * NLMSG_GOODSIZE and thus entirely undumpable if we didn't do this.
	 * The userspace caller can always fetch the actions separately if it
	 * really wants them.  (Most userspace callers in fact don't care.)
	 *
	 * This can only fail for dump operations because the skb is always
	 * properly sized for single flows.
	 */
	start = nla_nest_start(skb, OVS_FLOW_ATTR_ACTIONS);
	if (start) {
		const struct sw_flow_actions *sf_acts;

		sf_acts = rcu_dereference_ovsl(flow->sf_acts);
		err = ovs_nla_put_actions(sf_acts->actions,
					  sf_acts->actions_len, skb);

		if (!err)
			nla_nest_end(skb, start);
		else {
			if (skb_orig_len)
				return err;

			nla_nest_cancel(skb, start);
		}
	} else if (skb_orig_len) {
		return -EMSGSIZE;
	}

	return 0;
}

/* Called with ovs_mutex or RCU read lock. */
static int ovs_flow_cmd_fill_info(const struct sw_flow *flow, int dp_ifindex,
				  struct sk_buff *skb, u32 portid,
				  u32 seq, u32 flags, u8 cmd, u32 ufid_flags)
{
	const int skb_orig_len = skb->len;
	struct ovs_header *ovs_header;
	int err;

	ovs_header = genlmsg_put(skb, portid, seq, &dp_flow_genl_family,
				 flags, cmd);
	if (!ovs_header)
		return -EMSGSIZE;

	ovs_header->dp_ifindex = dp_ifindex;

	err = ovs_nla_put_identifier(flow, skb);
	if (err)
		goto error;

	if (should_fill_key(&flow->id, ufid_flags)) {
		err = ovs_nla_put_masked_key(flow, skb);
		if (err)
			goto error;
	}

	if (should_fill_mask(ufid_flags)) {
		err = ovs_nla_put_mask(flow, skb);
		if (err)
			goto error;
	}

	err = ovs_flow_cmd_fill_stats(flow, skb);
	if (err)
		goto error;

	if (should_fill_actions(ufid_flags)) {
		err = ovs_flow_cmd_fill_actions(flow, skb, skb_orig_len);
		if (err)
			goto error;
	}

	genlmsg_end(skb, ovs_header);
	return 0;

error:
	genlmsg_cancel(skb, ovs_header);
	return err;
}

/* May not be called with RCU read lock. */
static struct sk_buff *ovs_flow_cmd_alloc_info(const struct sw_flow_actions *acts,
					       const struct sw_flow_id *sfid,
					       struct genl_info *info,
					       bool always,
					       uint32_t ufid_flags)
{
	struct sk_buff *skb;
	size_t len;

	if (!always && !ovs_must_notify(&dp_flow_genl_family, info, 0))
		return NULL;

	len = ovs_flow_cmd_msg_size(acts, sfid, ufid_flags);
	skb = genlmsg_new_unicast(len, info, GFP_KERNEL);
	if (!skb)
		return ERR_PTR(-ENOMEM);

	return skb;
}

/* Called with ovs_mutex. */
static struct sk_buff *ovs_flow_cmd_build_info(const struct sw_flow *flow,
					       int dp_ifindex,
					       struct genl_info *info, u8 cmd,
					       bool always, u32 ufid_flags)
{
	struct sk_buff *skb;
	int retval;

	skb = ovs_flow_cmd_alloc_info(ovsl_dereference(flow->sf_acts),
				      &flow->id, info, always, ufid_flags);
	if (IS_ERR_OR_NULL(skb))
		return skb;

	retval = ovs_flow_cmd_fill_info(flow, dp_ifindex, skb,
					info->snd_portid, info->snd_seq, 0,
					cmd, ufid_flags);
	BUG_ON(retval < 0);
	return skb;
}

static int ovs_flow_cmd_new(struct sk_buff *skb, struct genl_info *info)
{
	struct nlattr **a = info->attrs;
	struct ovs_header *ovs_header = info->userhdr;
	struct sw_flow *flow = NULL, *new_flow;
	struct sw_flow_mask mask;
	struct sk_buff *reply;
	struct datapath *dp;
	struct sw_flow_key key;
	struct sw_flow_actions *acts;
	struct sw_flow_match match;
	u32 ufid_flags = ovs_nla_get_ufid_flags(a[OVS_FLOW_ATTR_UFID_FLAGS]);
	int error;
	bool log = !a[OVS_FLOW_ATTR_PROBE];

	/* Must have key and actions. */
	error = -EINVAL;
	if (!a[OVS_FLOW_ATTR_KEY]) {
		OVS_NLERR(log, "Flow key attr not present in new flow.");
		goto error;
	}
	if (!a[OVS_FLOW_ATTR_ACTIONS]) {
		OVS_NLERR(log, "Flow actions attr not present in new flow.");
		goto error;
	}

	/* Most of the time we need to allocate a new flow, do it before
	 * locking.
	 */
	new_flow = ovs_flow_alloc();
	if (IS_ERR(new_flow)) {
		error = PTR_ERR(new_flow);
		goto error;
	}

	/* Extract key. */
	ovs_match_init(&match, &key, &mask);
	error = ovs_nla_get_match(&match, a[OVS_FLOW_ATTR_KEY],
				  a[OVS_FLOW_ATTR_MASK], log);
	if (error)
		goto err_kfree_flow;

	ovs_flow_mask_key(&new_flow->key, &key, &mask);

	/* Extract flow identifier. */
	error = ovs_nla_get_identifier(&new_flow->id, a[OVS_FLOW_ATTR_UFID],
				       &key, log);
	if (error)
		goto err_kfree_flow;

	/* Validate actions. */
	error = ovs_nla_copy_actions(a[OVS_FLOW_ATTR_ACTIONS], &new_flow->key,
				     &acts, log);
	if (error) {
		OVS_NLERR(log, "Flow actions may not be safe on all matching packets.");
		goto err_kfree_flow;
	}

	reply = ovs_flow_cmd_alloc_info(acts, &new_flow->id, info, false,
					ufid_flags);
	if (IS_ERR(reply)) {
		error = PTR_ERR(reply);
		goto err_kfree_acts;
	}

	ovs_lock();
	dp = get_dp(sock_net(skb->sk), ovs_header->dp_ifindex);
	if (unlikely(!dp)) {
		error = -ENODEV;
		goto err_unlock_ovs;
	}

	/* Check if this is a duplicate flow */
	if (ovs_identifier_is_ufid(&new_flow->id))
		flow = ovs_flow_tbl_lookup_ufid(&dp->table, &new_flow->id);
	if (!flow)
		flow = ovs_flow_tbl_lookup(&dp->table, &key);
	if (likely(!flow)) {
		rcu_assign_pointer(new_flow->sf_acts, acts);

		/* Put flow in bucket. */
		error = ovs_flow_tbl_insert(&dp->table, new_flow, &mask);
		if (unlikely(error)) {
			acts = NULL;
			goto err_unlock_ovs;
		}

		if (unlikely(reply)) {
			error = ovs_flow_cmd_fill_info(new_flow,
						       ovs_header->dp_ifindex,
						       reply, info->snd_portid,
						       info->snd_seq, 0,
						       OVS_FLOW_CMD_NEW,
						       ufid_flags);
			BUG_ON(error < 0);
		}
		ovs_unlock();
	} else {
		struct sw_flow_actions *old_acts;

		/* Bail out if we're not allowed to modify an existing flow.
		 * We accept NLM_F_CREATE in place of the intended NLM_F_EXCL
		 * because Generic Netlink treats the latter as a dump
		 * request.  We also accept NLM_F_EXCL in case that bug ever
		 * gets fixed.
		 */
		if (unlikely(info->nlhdr->nlmsg_flags & (NLM_F_CREATE
							 | NLM_F_EXCL))) {
			error = -EEXIST;