#include <linux/in.h>
#include <linux/ip.h>
-#include <linux/list.h>
#include <linux/net.h>
+#include <linux/rculist.h>
#include <linux/udp.h>
#include <net/icmp.h>
#define LISP_HLEN (sizeof(struct udphdr) + sizeof(struct lisphdr))
-static inline int lisp_hdr_len(const struct tnl_mutable_config *mutable,
- const struct ovs_key_ipv4_tunnel *tun_key)
-{
- return LISP_HLEN;
-}
-
/**
* struct lisp_port - Keeps track of open UDP ports
- * @list: list element.
- * @port: The UDP port number in network byte order.
- * @socket: The socket created for this port number.
- * @count: How many ports are using this socket/port.
+ * @dst_port: lisp UDP port no.
+ * @list: list element in @lisp_ports.
+ * @lisp_rcv_socket: The socket created for this port number.
+ * @name: vport name.
*/
struct lisp_port {
+ __be16 dst_port;
struct list_head list;
- __be16 port;
struct socket *lisp_rcv_socket;
- int count;
+ char name[IFNAMSIZ];
};
static LIST_HEAD(lisp_ports);
-static struct lisp_port *lisp_port_exists(struct net *net, __be16 port)
+static inline struct lisp_port *lisp_vport(const struct vport *vport)
+{
+ return vport_priv(vport);
+}
+
+static struct lisp_port *lisp_find_port(struct net *net, __be16 port)
{
struct lisp_port *lisp_port;
- list_for_each_entry(lisp_port, &lisp_ports, list) {
- if (lisp_port->port == port &&
+ list_for_each_entry_rcu(lisp_port, &lisp_ports, list) {
+ if (lisp_port->dst_port == port &&
net_eq(sock_net(lisp_port->lisp_rcv_socket->sk), net))
return lisp_port;
}
return (struct lisphdr *)(udp_hdr(skb) + 1);
}
-static int lisp_tnl_send(struct vport *vport, struct sk_buff *skb)
-{
- int tnl_len;
- int network_offset = skb_network_offset(skb);
-
- /* We only encapsulate IPv4 and IPv6 packets */
- switch (skb->protocol) {
- case htons(ETH_P_IP):
- case htons(ETH_P_IPV6):
- /* Pop off "inner" Ethernet header */
- skb_pull(skb, network_offset);
- tnl_len = ovs_tnl_send(vport, skb);
- return tnl_len > 0 ? tnl_len + network_offset : tnl_len;
- default:
- kfree_skb(skb);
- return 0;
- }
-}
-
/* Convert 64 bit tunnel ID to 24 bit Instance ID. */
static void tunnel_id_to_instance_id(__be64 tun_id, __u8 *iid)
{
#endif
}
-static struct sk_buff *lisp_build_header(const struct vport *vport,
- const struct tnl_mutable_config *mutable,
- struct dst_entry *dst,
- struct sk_buff *skb,
- int tunnel_hlen)
+static void lisp_build_header(const struct vport *vport,
+ struct sk_buff *skb,
+ int tunnel_hlen)
{
+ struct lisp_port *lisp_port = lisp_vport(vport);
struct udphdr *udph = udp_hdr(skb);
struct lisphdr *lisph = (struct lisphdr *)(udph + 1);
const struct ovs_key_ipv4_tunnel *tun_key = OVS_CB(skb)->tun_key;
- __be64 out_key;
- u32 flags;
- tnl_get_param(mutable, tun_key, &flags, &out_key);
-
- udph->dest = mutable->dst_port;
+ udph->dest = lisp_port->dst_port;
udph->source = htons(ovs_tnl_get_src_port(skb));
udph->check = 0;
udph->len = htons(skb->len - skb_transport_offset(skb));
lisph->u1.nonce[1] = 0;
lisph->u1.nonce[2] = 0;
- tunnel_id_to_instance_id(out_key, &lisph->u2.word2.instance_id[0]);
+ tunnel_id_to_instance_id(tun_key->tun_id, &lisph->u2.word2.instance_id[0]);
lisph->u2.word2.locator_status_bits = 1;
-
- /*
- * Allow our local IP stack to fragment the outer packet even if the
- * DF bit is set as a last resort. We also need to force selection of
- * an IP ID here because Linux will otherwise leave it at 0 if the
- * packet originally had DF set.
- */
- skb->local_df = 1;
- __ip_select_ident(ip_hdr(skb), dst, 0);
-
- return skb;
}
/* Called with rcu_read_lock and BH disabled. */
static int lisp_rcv(struct sock *sk, struct sk_buff *skb)
{
- struct vport *vport;
+ struct lisp_port *lisp_port;
struct lisphdr *lisph;
- const struct tnl_mutable_config *mutable;
struct iphdr *iph, *inner_iph;
struct ovs_key_ipv4_tunnel tun_key;
__be64 key;
- u32 tunnel_flags = 0;
struct ethhdr *ethh;
__be16 protocol;
+ lisp_port = lisp_find_port(dev_net(skb->dev), udp_hdr(skb)->dest);
+ if (unlikely(!lisp_port))
+ goto error;
+
if (unlikely(!pskb_may_pull(skb, LISP_HLEN)))
goto error;
else
key = instance_id_to_tunnel_id(&lisph->u2.word2.instance_id[0]);
- iph = ip_hdr(skb);
- vport = ovs_tnl_find_port(dev_net(skb->dev), iph->daddr, iph->saddr,
- key, TNL_T_PROTO_LISP, &mutable);
- if (unlikely(!vport))
- goto error;
-
- if (mutable->flags & TNL_F_IN_KEY_MATCH || !mutable->key.daddr)
- tunnel_flags = OVS_TNL_F_KEY;
- else
- key = 0;
-
/* Save outer tunnel values */
- tnl_tun_key_init(&tun_key, iph, key, tunnel_flags);
- OVS_CB(skb)->tun_key = &tun_key;
+ iph = ip_hdr(skb);
+ tnl_tun_key_init(&tun_key, iph, key, OVS_TNL_F_KEY);
/* Drop non-IP inner packets */
inner_iph = (struct iphdr *)(lisph + 1);
ethh->h_source[0] = 0x02;
ethh->h_proto = protocol;
- ovs_tnl_rcv(vport, skb);
+ ovs_skb_postpush_rcsum(skb, skb->data, ETH_HLEN);
+
+ ovs_tnl_rcv(vport_from_priv(lisp_port), skb, &tun_key);
goto out;
error:
#define UDP_ENCAP_LISP 1
static int lisp_socket_init(struct lisp_port *lisp_port, struct net *net)
{
- int err;
struct sockaddr_in sin;
+ int err;
err = sock_create_kern(AF_INET, SOCK_DGRAM, 0,
&lisp_port->lisp_rcv_socket);
sin.sin_family = AF_INET;
sin.sin_addr.s_addr = htonl(INADDR_ANY);
- sin.sin_port = lisp_port->port;
+ sin.sin_port = lisp_port->dst_port;
err = kernel_bind(lisp_port->lisp_rcv_socket, (struct sockaddr *)&sin,
sizeof(struct sockaddr_in));
return err;
}
-static void lisp_tunnel_release(struct lisp_port *lisp_port)
+static int lisp_get_options(const struct vport *vport, struct sk_buff *skb)
{
- lisp_port->count--;
+ struct lisp_port *lisp_port = lisp_vport(vport);
- if (lisp_port->count == 0) {
- /* Release old socket */
- sk_release_kernel(lisp_port->lisp_rcv_socket->sk);
- list_del(&lisp_port->list);
- kfree(lisp_port);
- }
+ if (nla_put_u16(skb, OVS_TUNNEL_ATTR_DST_PORT, ntohs(lisp_port->dst_port)))
+ return -EMSGSIZE;
+ return 0;
}
-static int lisp_tunnel_setup(struct net *net, struct nlattr *options,
- struct lisp_port **lport)
+static void lisp_tnl_destroy(struct vport *vport)
{
+ struct lisp_port *lisp_port = lisp_vport(vport);
+
+ list_del_rcu(&lisp_port->list);
+ /* Release socket */
+ sk_release_kernel(lisp_port->lisp_rcv_socket->sk);
+
+ ovs_vport_deferred_free(vport);
+}
+
+static struct vport *lisp_tnl_create(const struct vport_parms *parms)
+{
+ struct net *net = ovs_dp_get_net(parms->dp);
+ struct nlattr *options = parms->options;
+ struct lisp_port *lisp_port;
+ struct vport *vport;
struct nlattr *a;
int err;
u16 dst_port;
- struct lisp_port *lisp_port = NULL;
-
- *lport = NULL;
if (!options) {
err = -EINVAL;
- goto out;
+ goto error;
}
a = nla_find_nested(options, OVS_TUNNEL_ATTR_DST_PORT);
} else {
/* Require destination port from userspace. */
err = -EINVAL;
- goto out;
+ goto error;
}
/* Verify if we already have a socket created for this port */
- lisp_port = lisp_port_exists(net, htons(dst_port));
- if (lisp_port) {
- lisp_port->count++;
- err = 0;
- *lport = lisp_port;
- goto out;
+ if (lisp_find_port(net, htons(dst_port))) {
+ err = -EEXIST;
+ goto error;
}
- /* Add a new socket for this port */
- lisp_port = kzalloc(sizeof(struct lisp_port), GFP_KERNEL);
- if (!lisp_port) {
- err = -ENOMEM;
- goto out;
- }
+ vport = ovs_vport_alloc(sizeof(struct lisp_port),
+ &ovs_lisp_vport_ops, parms);
+ if (IS_ERR(vport))
+ return vport;
- lisp_port->port = htons(dst_port);
- lisp_port->count = 1;
- list_add_tail(&lisp_port->list, &lisp_ports);
+ lisp_port = lisp_vport(vport);
+ lisp_port->dst_port = htons(dst_port);
+ strncpy(lisp_port->name, parms->name, IFNAMSIZ);
err = lisp_socket_init(lisp_port, net);
if (err)
- goto error;
+ goto error_free;
- *lport = lisp_port;
- goto out;
+ list_add_tail_rcu(&lisp_port->list, &lisp_ports);
+ return vport;
+error_free:
+ ovs_vport_free(vport);
error:
- list_del(&lisp_port->list);
- kfree(lisp_port);
-out:
- return err;
+ return ERR_PTR(err);
}
-static int lisp_tnl_set_options(struct vport *vport, struct nlattr *options)
+static int lisp_tnl_send(struct vport *vport, struct sk_buff *skb)
{
- int err;
- struct net *net = ovs_dp_get_net(vport->dp);
- struct tnl_vport *tnl_vport = tnl_vport_priv(vport);
- struct tnl_mutable_config *config;
- struct lisp_port *old_port = NULL;
- struct lisp_port *lisp_port = NULL;
-
- config = rtnl_dereference(tnl_vport->mutable);
-
- old_port = lisp_port_exists(net, config->dst_port);
-
- err = lisp_tunnel_setup(net, options, &lisp_port);
- if (err)
- goto out;
+ int tnl_len;
+ int network_offset = skb_network_offset(skb);
- err = ovs_tnl_set_options(vport, options);
+ if (unlikely(!OVS_CB(skb)->tun_key))
+ return -EINVAL;
- if (err)
- lisp_tunnel_release(lisp_port);
- else {
- /* Release old socket */
- lisp_tunnel_release(old_port);
+ /* We only encapsulate IPv4 and IPv6 packets */
+ switch (skb->protocol) {
+ case htons(ETH_P_IP):
+ case htons(ETH_P_IPV6):
+ /* Pop off "inner" Ethernet header */
+ skb_pull(skb, network_offset);
+ tnl_len = ovs_tnl_send(vport, skb, IPPROTO_UDP,
+ LISP_HLEN, lisp_build_header);
+ return tnl_len > 0 ? tnl_len + network_offset : tnl_len;
+ default:
+ kfree_skb(skb);
+ return 0;
}
-out:
- return err;
-}
-
-static const struct tnl_ops ovs_lisp_tnl_ops = {
- .tunnel_type = TNL_T_PROTO_LISP,
- .ipproto = IPPROTO_UDP,
- .hdr_len = lisp_hdr_len,
- .build_header = lisp_build_header,
-};
-
-static void lisp_tnl_destroy(struct vport *vport)
-{
- struct lisp_port *lisp_port;
- struct tnl_vport *tnl_vport = tnl_vport_priv(vport);
- struct tnl_mutable_config *config;
-
- config = rtnl_dereference(tnl_vport->mutable);
-
- lisp_port = lisp_port_exists(ovs_dp_get_net(vport->dp),
- config->dst_port);
-
- lisp_tunnel_release(lisp_port);
-
- ovs_tnl_destroy(vport);
}
-static struct vport *lisp_tnl_create(const struct vport_parms *parms)
+static const char *lisp_get_name(const struct vport *vport)
{
- int err;
- struct vport *vport;
- struct lisp_port *lisp_port = NULL;
-
- err = lisp_tunnel_setup(ovs_dp_get_net(parms->dp), parms->options,
- &lisp_port);
- if (err)
- return ERR_PTR(err);
-
- vport = ovs_tnl_create(parms, &ovs_lisp_vport_ops, &ovs_lisp_tnl_ops);
-
- if (IS_ERR(vport))
- lisp_tunnel_release(lisp_port);
-
- return vport;
+ struct lisp_port *lisp_port = lisp_vport(vport);
+ return lisp_port->name;
}
const struct vport_ops ovs_lisp_vport_ops = {
.type = OVS_VPORT_TYPE_LISP,
- .flags = VPORT_F_TUN_ID,
.create = lisp_tnl_create,
.destroy = lisp_tnl_destroy,
- .get_name = ovs_tnl_get_name,
- .get_options = ovs_tnl_get_options,
- .set_options = lisp_tnl_set_options,
+ .get_name = lisp_get_name,
+ .get_options = lisp_get_options,
.send = lisp_tnl_send,
};
#else
git://git.onelab.eu / sliver-openvswitch.git / blobdiff