2 * Copyright (c) 2007-2013 Nicira, Inc.
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of version 2 of the GNU General Public
6 * License as published by the Free Software Foundation.
8 * This program is distributed in the hope that it will be useful, but
9 * WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
13 * You should have received a copy of the GNU General Public License
14 * along with this program; if not, write to the Free Software
15 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
20 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
23 #include <linux/in_route.h>
24 #include <linux/inetdevice.h>
25 #include <linux/jhash.h>
26 #include <linux/list.h>
27 #include <linux/kernel.h>
28 #include <linux/version.h>
29 #include <linux/workqueue.h>
30 #include <linux/rculist.h>
31 #include <net/ip_tunnels.h>
32 #include <net/route.h>
37 #include "vport-netdev.h"
39 #if LINUX_VERSION_CODE < KERNEL_VERSION(3,18,0)
40 int rpl_iptunnel_xmit(struct sock *sk, struct rtable *rt, struct sk_buff *skb,
41 __be32 src, __be32 dst, __u8 proto, __u8 tos, __u8 ttl,
44 int pkt_len = skb->len;
48 skb_scrub_packet(skb, xnet);
51 skb_dst_set(skb, &rt_dst(rt));
54 /* Do not clear ovs_skb_cb. It will be done in gso code. */
55 memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
58 /* Push down and install the IP header. */
59 __skb_push(skb, sizeof(struct iphdr));
60 skb_reset_network_header(skb);
65 iph->ihl = sizeof(struct iphdr) >> 2;
67 iph->protocol = proto;
73 #ifdef HAVE_IP_SELECT_IDENT_USING_DST_ENTRY
74 __ip_select_ident(iph, &rt_dst(rt), (skb_shinfo(skb)->gso_segs ?: 1) - 1);
75 #elif defined(HAVE_IP_SELECT_IDENT_USING_NET)
76 __ip_select_ident(dev_net(rt->dst.dev), iph,
77 skb_shinfo(skb)->gso_segs ?: 1);
79 __ip_select_ident(iph, skb_shinfo(skb)->gso_segs ?: 1);
82 err = ip_local_out(skb);
83 if (unlikely(net_xmit_eval(err)))
87 EXPORT_SYMBOL_GPL(rpl_iptunnel_xmit);
89 struct sk_buff *ovs_iptunnel_handle_offloads(struct sk_buff *skb,
90 bool csum_help, int gso_type_mask,
91 void (*fix_segment)(struct sk_buff *))
95 if (likely(!skb_is_encapsulated(skb))) {
96 skb_reset_inner_headers(skb);
97 #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,8,0)
98 skb->encapsulation = 1;
100 } else if (skb_is_gso(skb)) {
108 OVS_GSO_CB(skb)->fix_segment = fix_segment;
110 if (skb_is_gso(skb)) {
111 err = skb_unclone(skb, GFP_ATOMIC);
114 skb_shinfo(skb)->gso_type |= gso_type_mask;
118 #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,8,0)
119 /* If packet is not gso and we are resolving any partial checksum,
120 * clear encapsulation flag. This allows setting CHECKSUM_PARTIAL
121 * on the outer header without confusing devices that implement
122 * NETIF_F_IP_CSUM with encapsulation.
125 skb->encapsulation = 0;
128 if (skb->ip_summed == CHECKSUM_PARTIAL && csum_help) {
129 err = skb_checksum_help(skb);
132 } else if (skb->ip_summed != CHECKSUM_PARTIAL)
133 skb->ip_summed = CHECKSUM_NONE;
140 EXPORT_SYMBOL_GPL(ovs_iptunnel_handle_offloads);
142 int rpl_iptunnel_pull_header(struct sk_buff *skb, int hdr_len, __be16 inner_proto)
144 if (unlikely(!pskb_may_pull(skb, hdr_len)))
147 skb_pull_rcsum(skb, hdr_len);
149 if (inner_proto == htons(ETH_P_TEB)) {
152 if (unlikely(!pskb_may_pull(skb, ETH_HLEN)))
155 eh = (struct ethhdr *)skb->data;
157 if (likely(ntohs(eh->h_proto) >= ETH_P_802_3_MIN))
158 skb->protocol = eh->h_proto;
160 skb->protocol = htons(ETH_P_802_2);
163 skb->protocol = inner_proto;
170 vlan_set_tci(skb, 0);
171 skb_set_queue_mapping(skb, 0);
172 skb->pkt_type = PACKET_HOST;
175 EXPORT_SYMBOL_GPL(rpl_iptunnel_pull_header);
179 bool ovs_skb_is_encapsulated(struct sk_buff *skb)
181 /* checking for inner protocol should be sufficient on newer kernel, but
182 * old kernel just set encapsulation bit.
184 return ovs_skb_get_inner_protocol(skb) || skb_encapsulation(skb);
186 EXPORT_SYMBOL_GPL(ovs_skb_is_encapsulated);
188 /* derived from ip_tunnel_rcv(). */
189 void ovs_ip_tunnel_rcv(struct net_device *dev, struct sk_buff *skb,
190 struct metadata_dst *tun_dst)
192 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,39)
193 struct pcpu_sw_netstats *tstats;
195 tstats = this_cpu_ptr((struct pcpu_sw_netstats __percpu *)dev->tstats);
196 u64_stats_update_begin(&tstats->syncp);
197 tstats->rx_packets++;
198 tstats->rx_bytes += skb->len;
199 u64_stats_update_end(&tstats->syncp);
202 skb_reset_mac_header(skb);
203 skb_scrub_packet(skb, false);
204 skb->protocol = eth_type_trans(skb, dev);
205 skb_postpull_rcsum(skb, eth_hdr(skb), ETH_HLEN);
207 ovs_skb_dst_set(skb, (struct dst_entry *)tun_dst);
209 #ifndef HAVE_METADATA_DST
210 netdev_port_receive(skb, &tun_dst->u.tun_info);
216 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,39)
217 #ifndef HAVE_PCPU_SW_NETSTATS
218 #define netdev_stats_to_stats64 rpl_netdev_stats_to_stats64
219 static void netdev_stats_to_stats64(struct rtnl_link_stats64 *stats64,
220 const struct net_device_stats *netdev_stats)
222 #if BITS_PER_LONG == 64
223 BUILD_BUG_ON(sizeof(*stats64) != sizeof(*netdev_stats));
224 memcpy(stats64, netdev_stats, sizeof(*stats64));
226 size_t i, n = sizeof(*stats64) / sizeof(u64);
227 const unsigned long *src = (const unsigned long *)netdev_stats;
228 u64 *dst = (u64 *)stats64;
230 BUILD_BUG_ON(sizeof(*netdev_stats) / sizeof(unsigned long) !=
231 sizeof(*stats64) / sizeof(u64));
232 for (i = 0; i < n; i++)
237 struct rtnl_link_stats64 *rpl_ip_tunnel_get_stats64(struct net_device *dev,
238 struct rtnl_link_stats64 *tot)
242 netdev_stats_to_stats64(tot, &dev->stats);
244 for_each_possible_cpu(i) {
245 const struct pcpu_sw_netstats *tstats =
246 per_cpu_ptr((struct pcpu_sw_netstats __percpu *)dev->tstats, i);
247 u64 rx_packets, rx_bytes, tx_packets, tx_bytes;
251 start = u64_stats_fetch_begin_irq(&tstats->syncp);
252 rx_packets = tstats->rx_packets;
253 tx_packets = tstats->tx_packets;
254 rx_bytes = tstats->rx_bytes;
255 tx_bytes = tstats->tx_bytes;
256 } while (u64_stats_fetch_retry_irq(&tstats->syncp, start));
258 tot->rx_packets += rx_packets;
259 tot->tx_packets += tx_packets;
260 tot->rx_bytes += rx_bytes;
261 tot->tx_bytes += tx_bytes;