+ }
+ return NULL;
+}
+
+void ovs_flow_tbl_insert(struct flow_table *table, struct sw_flow *flow)
+{
+ struct hlist_head *head;
+
+ head = find_bucket(table, flow->hash);
+ hlist_add_head_rcu(&flow->hash_node[table->node_ver], head);
+ table->count++;
+}
+
+void ovs_flow_tbl_remove(struct flow_table *table, struct sw_flow *flow)
+{
+ hlist_del_rcu(&flow->hash_node[table->node_ver]);
+ table->count--;
+ BUG_ON(table->count < 0);
+}
+
+/* The size of the argument for each %OVS_KEY_ATTR_* Netlink attribute. */
+const int ovs_key_lens[OVS_KEY_ATTR_MAX + 1] = {
+ [OVS_KEY_ATTR_ENCAP] = -1,
+ [OVS_KEY_ATTR_PRIORITY] = sizeof(u32),
+ [OVS_KEY_ATTR_IN_PORT] = sizeof(u32),
+ [OVS_KEY_ATTR_ETHERNET] = sizeof(struct ovs_key_ethernet),
+ [OVS_KEY_ATTR_VLAN] = sizeof(__be16),
+ [OVS_KEY_ATTR_ETHERTYPE] = sizeof(__be16),
+ [OVS_KEY_ATTR_IPV4] = sizeof(struct ovs_key_ipv4),
+ [OVS_KEY_ATTR_IPV6] = sizeof(struct ovs_key_ipv6),
+ [OVS_KEY_ATTR_TCP] = sizeof(struct ovs_key_tcp),
+ [OVS_KEY_ATTR_UDP] = sizeof(struct ovs_key_udp),
+ [OVS_KEY_ATTR_ICMP] = sizeof(struct ovs_key_icmp),
+ [OVS_KEY_ATTR_ICMPV6] = sizeof(struct ovs_key_icmpv6),
+ [OVS_KEY_ATTR_ARP] = sizeof(struct ovs_key_arp),
+ [OVS_KEY_ATTR_ND] = sizeof(struct ovs_key_nd),
+
+ /* Not upstream. */
+ [OVS_KEY_ATTR_TUN_ID] = sizeof(__be64),
+};
+
+static int ipv4_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_len,
+ const struct nlattr *a[], u64 *attrs)
+{
+ const struct ovs_key_icmp *icmp_key;
+ const struct ovs_key_tcp *tcp_key;
+ const struct ovs_key_udp *udp_key;
+
+ switch (swkey->ip.proto) {
+ case IPPROTO_TCP:
+ if (!(*attrs & (1 << OVS_KEY_ATTR_TCP)))
+ return -EINVAL;
+ *attrs &= ~(1 << OVS_KEY_ATTR_TCP);
+
+ *key_len = SW_FLOW_KEY_OFFSET(ipv4.tp);
+ tcp_key = nla_data(a[OVS_KEY_ATTR_TCP]);
+ swkey->ipv4.tp.src = tcp_key->tcp_src;
+ swkey->ipv4.tp.dst = tcp_key->tcp_dst;
+ break;
+
+ case IPPROTO_UDP:
+ if (!(*attrs & (1 << OVS_KEY_ATTR_UDP)))
+ return -EINVAL;
+ *attrs &= ~(1 << OVS_KEY_ATTR_UDP);
+
+ *key_len = SW_FLOW_KEY_OFFSET(ipv4.tp);
+ udp_key = nla_data(a[OVS_KEY_ATTR_UDP]);
+ swkey->ipv4.tp.src = udp_key->udp_src;
+ swkey->ipv4.tp.dst = udp_key->udp_dst;
+ break;
+
+ case IPPROTO_ICMP:
+ if (!(*attrs & (1 << OVS_KEY_ATTR_ICMP)))
+ return -EINVAL;
+ *attrs &= ~(1 << OVS_KEY_ATTR_ICMP);
+
+ *key_len = SW_FLOW_KEY_OFFSET(ipv4.tp);
+ icmp_key = nla_data(a[OVS_KEY_ATTR_ICMP]);
+ swkey->ipv4.tp.src = htons(icmp_key->icmp_type);
+ swkey->ipv4.tp.dst = htons(icmp_key->icmp_code);
+ break;
+ }
+
+ return 0;
+}
+
+static int ipv6_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_len,
+ const struct nlattr *a[], u64 *attrs)
+{
+ const struct ovs_key_icmpv6 *icmpv6_key;
+ const struct ovs_key_tcp *tcp_key;
+ const struct ovs_key_udp *udp_key;
+
+ switch (swkey->ip.proto) {
+ case IPPROTO_TCP:
+ if (!(*attrs & (1 << OVS_KEY_ATTR_TCP)))
+ return -EINVAL;
+ *attrs &= ~(1 << OVS_KEY_ATTR_TCP);
+
+ *key_len = SW_FLOW_KEY_OFFSET(ipv6.tp);
+ tcp_key = nla_data(a[OVS_KEY_ATTR_TCP]);
+ swkey->ipv6.tp.src = tcp_key->tcp_src;
+ swkey->ipv6.tp.dst = tcp_key->tcp_dst;
+ break;
+
+ case IPPROTO_UDP:
+ if (!(*attrs & (1 << OVS_KEY_ATTR_UDP)))
+ return -EINVAL;
+ *attrs &= ~(1 << OVS_KEY_ATTR_UDP);
+
+ *key_len = SW_FLOW_KEY_OFFSET(ipv6.tp);
+ udp_key = nla_data(a[OVS_KEY_ATTR_UDP]);
+ swkey->ipv6.tp.src = udp_key->udp_src;
+ swkey->ipv6.tp.dst = udp_key->udp_dst;
+ break;
+
+ case IPPROTO_ICMPV6:
+ if (!(*attrs & (1 << OVS_KEY_ATTR_ICMPV6)))
+ return -EINVAL;
+ *attrs &= ~(1 << OVS_KEY_ATTR_ICMPV6);
+
+ *key_len = SW_FLOW_KEY_OFFSET(ipv6.tp);
+ icmpv6_key = nla_data(a[OVS_KEY_ATTR_ICMPV6]);
+ swkey->ipv6.tp.src = htons(icmpv6_key->icmpv6_type);
+ swkey->ipv6.tp.dst = htons(icmpv6_key->icmpv6_code);
+
+ if (swkey->ipv6.tp.src == htons(NDISC_NEIGHBOUR_SOLICITATION) ||
+ swkey->ipv6.tp.src == htons(NDISC_NEIGHBOUR_ADVERTISEMENT)) {
+ const struct ovs_key_nd *nd_key;
+
+ if (!(*attrs & (1 << OVS_KEY_ATTR_ND)))
+ return -EINVAL;
+ *attrs &= ~(1 << OVS_KEY_ATTR_ND);
+
+ *key_len = SW_FLOW_KEY_OFFSET(ipv6.nd);
+ nd_key = nla_data(a[OVS_KEY_ATTR_ND]);
+ memcpy(&swkey->ipv6.nd.target, nd_key->nd_target,
+ sizeof(swkey->ipv6.nd.target));
+ memcpy(swkey->ipv6.nd.sll, nd_key->nd_sll, ETH_ALEN);
+ memcpy(swkey->ipv6.nd.tll, nd_key->nd_tll, ETH_ALEN);
+ }
+ break;
+ }
+
+ return 0;
+}
+
+static int parse_flow_nlattrs(const struct nlattr *attr,
+ const struct nlattr *a[], u64 *attrsp)
+{
+ const struct nlattr *nla;
+ u64 attrs;
+ int rem;
+
+ attrs = 0;
+ nla_for_each_nested(nla, attr, rem) {
+ u16 type = nla_type(nla);
+ int expected_len;
+
+ if (type > OVS_KEY_ATTR_MAX || attrs & (1ULL << type))
+ return -EINVAL;
+
+ expected_len = ovs_key_lens[type];
+ if (nla_len(nla) != expected_len && expected_len != -1)
+ return -EINVAL;
+
+ attrs |= 1ULL << type;
+ a[type] = nla;
+ }
+ if (rem)
+ return -EINVAL;
+
+ *attrsp = attrs;
+ return 0;
+}
+
+/**
+ * ovs_flow_from_nlattrs - parses Netlink attributes into a flow key.
+ * @swkey: receives the extracted flow key.
+ * @key_lenp: number of bytes used in @swkey.
+ * @attr: Netlink attribute holding nested %OVS_KEY_ATTR_* Netlink attribute
+ * sequence.
+ */
+int ovs_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp,
+ const struct nlattr *attr)
+{
+ const struct nlattr *a[OVS_KEY_ATTR_MAX + 1];
+ const struct ovs_key_ethernet *eth_key;
+ int key_len;
+ u64 attrs;
+ int err;
+
+ memset(swkey, 0, sizeof(struct sw_flow_key));
+ key_len = SW_FLOW_KEY_OFFSET(eth);
+
+ err = parse_flow_nlattrs(attr, a, &attrs);
+ if (err)
+ return err;
+
+ /* Metadata attributes. */
+ if (attrs & (1 << OVS_KEY_ATTR_PRIORITY)) {
+ swkey->phy.priority = nla_get_u32(a[OVS_KEY_ATTR_PRIORITY]);
+ attrs &= ~(1 << OVS_KEY_ATTR_PRIORITY);
+ }
+ if (attrs & (1 << OVS_KEY_ATTR_IN_PORT)) {
+ u32 in_port = nla_get_u32(a[OVS_KEY_ATTR_IN_PORT]);
+ if (in_port >= DP_MAX_PORTS)
+ return -EINVAL;
+ swkey->phy.in_port = in_port;
+ attrs &= ~(1 << OVS_KEY_ATTR_IN_PORT);
+ } else {
+ swkey->phy.in_port = DP_MAX_PORTS;
+ }
+
+ if (attrs & (1ULL << OVS_KEY_ATTR_TUN_ID)) {
+ swkey->phy.tun_id = nla_get_be64(a[OVS_KEY_ATTR_TUN_ID]);
+ attrs &= ~(1ULL << OVS_KEY_ATTR_TUN_ID);
+ }
+
+ /* Data attributes. */
+ if (!(attrs & (1 << OVS_KEY_ATTR_ETHERNET)))
+ return -EINVAL;
+ attrs &= ~(1 << OVS_KEY_ATTR_ETHERNET);
+
+ eth_key = nla_data(a[OVS_KEY_ATTR_ETHERNET]);
+ memcpy(swkey->eth.src, eth_key->eth_src, ETH_ALEN);
+ memcpy(swkey->eth.dst, eth_key->eth_dst, ETH_ALEN);
+
+ if (attrs & (1u << OVS_KEY_ATTR_ETHERTYPE) &&
+ nla_get_be16(a[OVS_KEY_ATTR_ETHERTYPE]) == htons(ETH_P_8021Q)) {
+ const struct nlattr *encap;
+ __be16 tci;
+
+ if (attrs != ((1 << OVS_KEY_ATTR_VLAN) |
+ (1 << OVS_KEY_ATTR_ETHERTYPE) |
+ (1 << OVS_KEY_ATTR_ENCAP)))
+ return -EINVAL;
+
+ encap = a[OVS_KEY_ATTR_ENCAP];
+ tci = nla_get_be16(a[OVS_KEY_ATTR_VLAN]);
+ if (tci & htons(VLAN_TAG_PRESENT)) {
+ swkey->eth.tci = tci;
+
+ err = parse_flow_nlattrs(encap, a, &attrs);
+ if (err)
+ return err;
+ } else if (!tci) {
+ /* Corner case for truncated 802.1Q header. */
+ if (nla_len(encap))
+ return -EINVAL;
+
+ swkey->eth.type = htons(ETH_P_8021Q);
+ *key_lenp = key_len;
+ return 0;
+ } else {
+ return -EINVAL;
+ }
+ }
+
+ if (attrs & (1 << OVS_KEY_ATTR_ETHERTYPE)) {
+ swkey->eth.type = nla_get_be16(a[OVS_KEY_ATTR_ETHERTYPE]);
+ if (ntohs(swkey->eth.type) < 1536)
+ return -EINVAL;
+ attrs &= ~(1 << OVS_KEY_ATTR_ETHERTYPE);
+ } else {
+ swkey->eth.type = htons(ETH_P_802_2);
+ }
+
+ if (swkey->eth.type == htons(ETH_P_IP)) {
+ const struct ovs_key_ipv4 *ipv4_key;
+
+ if (!(attrs & (1 << OVS_KEY_ATTR_IPV4)))
+ return -EINVAL;
+ attrs &= ~(1 << OVS_KEY_ATTR_IPV4);
+
+ key_len = SW_FLOW_KEY_OFFSET(ipv4.addr);
+ ipv4_key = nla_data(a[OVS_KEY_ATTR_IPV4]);
+ if (ipv4_key->ipv4_frag > OVS_FRAG_TYPE_MAX)
+ return -EINVAL;
+ swkey->ip.proto = ipv4_key->ipv4_proto;
+ swkey->ip.tos = ipv4_key->ipv4_tos;
+ swkey->ip.ttl = ipv4_key->ipv4_ttl;
+ swkey->ip.frag = ipv4_key->ipv4_frag;
+ swkey->ipv4.addr.src = ipv4_key->ipv4_src;
+ swkey->ipv4.addr.dst = ipv4_key->ipv4_dst;
+
+ if (swkey->ip.frag != OVS_FRAG_TYPE_LATER) {
+ err = ipv4_flow_from_nlattrs(swkey, &key_len, a, &attrs);
+ if (err)
+ return err;
+ }
+ } else if (swkey->eth.type == htons(ETH_P_IPV6)) {
+ const struct ovs_key_ipv6 *ipv6_key;
+
+ if (!(attrs & (1 << OVS_KEY_ATTR_IPV6)))
+ return -EINVAL;
+ attrs &= ~(1 << OVS_KEY_ATTR_IPV6);
+
+ key_len = SW_FLOW_KEY_OFFSET(ipv6.label);
+ ipv6_key = nla_data(a[OVS_KEY_ATTR_IPV6]);
+ if (ipv6_key->ipv6_frag > OVS_FRAG_TYPE_MAX)
+ return -EINVAL;
+ swkey->ipv6.label = ipv6_key->ipv6_label;
+ swkey->ip.proto = ipv6_key->ipv6_proto;
+ swkey->ip.tos = ipv6_key->ipv6_tclass;
+ swkey->ip.ttl = ipv6_key->ipv6_hlimit;
+ swkey->ip.frag = ipv6_key->ipv6_frag;
+ memcpy(&swkey->ipv6.addr.src, ipv6_key->ipv6_src,
+ sizeof(swkey->ipv6.addr.src));
+ memcpy(&swkey->ipv6.addr.dst, ipv6_key->ipv6_dst,
+ sizeof(swkey->ipv6.addr.dst));
+
+ if (swkey->ip.frag != OVS_FRAG_TYPE_LATER) {
+ err = ipv6_flow_from_nlattrs(swkey, &key_len, a, &attrs);
+ if (err)
+ return err;
+ }
+ } else if (swkey->eth.type == htons(ETH_P_ARP)) {
+ const struct ovs_key_arp *arp_key;
+
+ if (!(attrs & (1 << OVS_KEY_ATTR_ARP)))
+ return -EINVAL;
+ attrs &= ~(1 << OVS_KEY_ATTR_ARP);
+
+ key_len = SW_FLOW_KEY_OFFSET(ipv4.arp);
+ arp_key = nla_data(a[OVS_KEY_ATTR_ARP]);
+ swkey->ipv4.addr.src = arp_key->arp_sip;
+ swkey->ipv4.addr.dst = arp_key->arp_tip;
+ if (arp_key->arp_op & htons(0xff00))
+ return -EINVAL;
+ swkey->ip.proto = ntohs(arp_key->arp_op);
+ memcpy(swkey->ipv4.arp.sha, arp_key->arp_sha, ETH_ALEN);
+ memcpy(swkey->ipv4.arp.tha, arp_key->arp_tha, ETH_ALEN);
+ }
+
+ if (attrs)
+ return -EINVAL;
+ *key_lenp = key_len;
+
+ return 0;
+}
+
+/**
+ * ovs_flow_metadata_from_nlattrs - parses Netlink attributes into a flow key.
+ * @in_port: receives the extracted input port.
+ * @tun_id: receives the extracted tunnel ID.
+ * @key: Netlink attribute holding nested %OVS_KEY_ATTR_* Netlink attribute
+ * sequence.
+ *
+ * This parses a series of Netlink attributes that form a flow key, which must
+ * take the same form accepted by flow_from_nlattrs(), but only enough of it to
+ * get the metadata, that is, the parts of the flow key that cannot be
+ * extracted from the packet itself.
+ */
+int ovs_flow_metadata_from_nlattrs(u32 *priority, u16 *in_port, __be64 *tun_id,
+ const struct nlattr *attr)
+{
+ const struct nlattr *nla;
+ int rem;
+
+ *in_port = DP_MAX_PORTS;
+ *tun_id = 0;
+ *priority = 0;
+
+ nla_for_each_nested(nla, attr, rem) {
+ int type = nla_type(nla);
+
+ if (type <= OVS_KEY_ATTR_MAX && ovs_key_lens[type] > 0) {
+ if (nla_len(nla) != ovs_key_lens[type])
+ return -EINVAL;
+
+ switch (type) {
+ case OVS_KEY_ATTR_PRIORITY:
+ *priority = nla_get_u32(nla);
+ break;
+
+ case OVS_KEY_ATTR_TUN_ID:
+ *tun_id = nla_get_be64(nla);
+ break;
+
+ case OVS_KEY_ATTR_IN_PORT:
+ if (nla_get_u32(nla) >= DP_MAX_PORTS)
+ return -EINVAL;
+ *in_port = nla_get_u32(nla);
+ break;
+ }
+ }
+ }
+ if (rem)
+ return -EINVAL;
+ return 0;
+}
+
+int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb)
+{
+ struct ovs_key_ethernet *eth_key;
+ struct nlattr *nla, *encap;
+
+ if (swkey->phy.priority)
+ NLA_PUT_U32(skb, OVS_KEY_ATTR_PRIORITY, swkey->phy.priority);
+
+ if (swkey->phy.tun_id != cpu_to_be64(0))
+ NLA_PUT_BE64(skb, OVS_KEY_ATTR_TUN_ID, swkey->phy.tun_id);
+
+ if (swkey->phy.in_port != DP_MAX_PORTS)
+ NLA_PUT_U32(skb, OVS_KEY_ATTR_IN_PORT, swkey->phy.in_port);
+
+ nla = nla_reserve(skb, OVS_KEY_ATTR_ETHERNET, sizeof(*eth_key));
+ if (!nla)
+ goto nla_put_failure;
+ eth_key = nla_data(nla);
+ memcpy(eth_key->eth_src, swkey->eth.src, ETH_ALEN);
+ memcpy(eth_key->eth_dst, swkey->eth.dst, ETH_ALEN);
+
+ if (swkey->eth.tci || swkey->eth.type == htons(ETH_P_8021Q)) {
+ NLA_PUT_BE16(skb, OVS_KEY_ATTR_ETHERTYPE, htons(ETH_P_8021Q));
+ NLA_PUT_BE16(skb, OVS_KEY_ATTR_VLAN, swkey->eth.tci);
+ encap = nla_nest_start(skb, OVS_KEY_ATTR_ENCAP);
+ if (!swkey->eth.tci)
+ goto unencap;