]> git.proxmox.com Git - mirror_ovs.git/blob - datapath/flow.h
compat: Fixup ipv6 fragmentation on 4.9.135+ kernels
[mirror_ovs.git] / datapath / flow.h
1 /*
2 * Copyright (c) 2007-2017 Nicira, Inc.
3 *
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.
7 *
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.
12 *
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
16 * 02110-1301, USA
17 */
18
19 #ifndef FLOW_H
20 #define FLOW_H 1
21
22 #include <linux/cache.h>
23 #include <linux/kernel.h>
24 #include <linux/netlink.h>
25 #include <linux/openvswitch.h>
26 #include <linux/spinlock.h>
27 #include <linux/types.h>
28 #include <linux/rcupdate.h>
29 #include <linux/if_ether.h>
30 #include <linux/in6.h>
31 #include <linux/jiffies.h>
32 #include <linux/time.h>
33 #include <linux/flex_array.h>
34 #include <linux/cpumask.h>
35 #include <net/inet_ecn.h>
36 #include <net/ip_tunnels.h>
37 #include <net/dst_metadata.h>
38 #include <net/nsh.h>
39
40 struct sk_buff;
41
42 enum sw_flow_mac_proto {
43 MAC_PROTO_NONE = 0,
44 MAC_PROTO_ETHERNET,
45 };
46 #define SW_FLOW_KEY_INVALID 0x80
47
48 /* Store options at the end of the array if they are less than the
49 * maximum size. This allows us to get the benefits of variable length
50 * matching for small options.
51 */
52 #define TUN_METADATA_OFFSET(opt_len) \
53 (FIELD_SIZEOF(struct sw_flow_key, tun_opts) - opt_len)
54 #define TUN_METADATA_OPTS(flow_key, opt_len) \
55 ((void *)((flow_key)->tun_opts + TUN_METADATA_OFFSET(opt_len)))
56
57 struct ovs_tunnel_info {
58 struct metadata_dst *tun_dst;
59 };
60
61 struct vlan_head {
62 __be16 tpid; /* Vlan type. Generally 802.1q or 802.1ad.*/
63 __be16 tci; /* 0 if no VLAN, VLAN_TAG_PRESENT set otherwise. */
64 };
65
66 #define OVS_SW_FLOW_KEY_METADATA_SIZE \
67 (offsetof(struct sw_flow_key, recirc_id) + \
68 FIELD_SIZEOF(struct sw_flow_key, recirc_id))
69
70 struct ovs_key_nsh {
71 struct ovs_nsh_key_base base;
72 __be32 context[NSH_MD1_CONTEXT_SIZE];
73 };
74
75 struct sw_flow_key {
76 u8 tun_opts[255];
77 u8 tun_opts_len;
78 struct ip_tunnel_key tun_key; /* Encapsulating tunnel key. */
79 struct {
80 u32 priority; /* Packet QoS priority. */
81 u32 skb_mark; /* SKB mark. */
82 u16 in_port; /* Input switch port (or DP_MAX_PORTS). */
83 } __packed phy; /* Safe when right after 'tun_key'. */
84 u8 mac_proto; /* MAC layer protocol (e.g. Ethernet). */
85 u8 tun_proto; /* Protocol of encapsulating tunnel. */
86 u32 ovs_flow_hash; /* Datapath computed hash value. */
87 u32 recirc_id; /* Recirculation ID. */
88 struct {
89 u8 src[ETH_ALEN]; /* Ethernet source address. */
90 u8 dst[ETH_ALEN]; /* Ethernet destination address. */
91 struct vlan_head vlan;
92 struct vlan_head cvlan;
93 __be16 type; /* Ethernet frame type. */
94 } eth;
95 /* Filling a hole of two bytes. */
96 u8 ct_state;
97 u8 ct_orig_proto; /* CT original direction tuple IP
98 * protocol.
99 */
100 union {
101 struct {
102 __be32 top_lse; /* top label stack entry */
103 } mpls;
104 struct {
105 u8 proto; /* IP protocol or lower 8 bits of ARP opcode. */
106 u8 tos; /* IP ToS. */
107 u8 ttl; /* IP TTL/hop limit. */
108 u8 frag; /* One of OVS_FRAG_TYPE_*. */
109 } ip;
110 };
111 u16 ct_zone; /* Conntrack zone. */
112 struct {
113 __be16 src; /* TCP/UDP/SCTP source port. */
114 __be16 dst; /* TCP/UDP/SCTP destination port. */
115 __be16 flags; /* TCP flags. */
116 } tp;
117 union {
118 struct {
119 struct {
120 __be32 src; /* IP source address. */
121 __be32 dst; /* IP destination address. */
122 } addr;
123 union {
124 struct {
125 __be32 src;
126 __be32 dst;
127 } ct_orig; /* Conntrack original direction fields. */
128 struct {
129 u8 sha[ETH_ALEN]; /* ARP source hardware address. */
130 u8 tha[ETH_ALEN]; /* ARP target hardware address. */
131 } arp;
132 };
133 } ipv4;
134 struct {
135 struct {
136 struct in6_addr src; /* IPv6 source address. */
137 struct in6_addr dst; /* IPv6 destination address. */
138 } addr;
139 __be32 label; /* IPv6 flow label. */
140 union {
141 struct {
142 struct in6_addr src;
143 struct in6_addr dst;
144 } ct_orig; /* Conntrack original direction fields. */
145 struct {
146 struct in6_addr target; /* ND target address. */
147 u8 sll[ETH_ALEN]; /* ND source link layer address. */
148 u8 tll[ETH_ALEN]; /* ND target link layer address. */
149 } nd;
150 };
151 } ipv6;
152 struct ovs_key_nsh nsh; /* network service header */
153 };
154 struct {
155 /* Connection tracking fields not packed above. */
156 struct {
157 __be16 src; /* CT orig tuple tp src port. */
158 __be16 dst; /* CT orig tuple tp dst port. */
159 } orig_tp;
160 u32 mark;
161 struct ovs_key_ct_labels labels;
162 } ct;
163
164 } __aligned(BITS_PER_LONG/8); /* Ensure that we can do comparisons as longs. */
165
166 static inline bool sw_flow_key_is_nd(const struct sw_flow_key *key)
167 {
168 return key->eth.type == htons(ETH_P_IPV6) &&
169 key->ip.proto == NEXTHDR_ICMP &&
170 key->tp.dst == 0 &&
171 (key->tp.src == htons(NDISC_NEIGHBOUR_SOLICITATION) ||
172 key->tp.src == htons(NDISC_NEIGHBOUR_ADVERTISEMENT));
173 }
174
175 struct sw_flow_key_range {
176 unsigned short int start;
177 unsigned short int end;
178 };
179
180 struct sw_flow_mask {
181 int ref_count;
182 struct rcu_head rcu;
183 struct sw_flow_key_range range;
184 struct sw_flow_key key;
185 };
186
187 struct sw_flow_match {
188 struct sw_flow_key *key;
189 struct sw_flow_key_range range;
190 struct sw_flow_mask *mask;
191 };
192
193 #define MAX_UFID_LENGTH 16 /* 128 bits */
194
195 struct sw_flow_id {
196 u32 ufid_len;
197 union {
198 u32 ufid[MAX_UFID_LENGTH / 4];
199 struct sw_flow_key *unmasked_key;
200 };
201 };
202
203 struct sw_flow_actions {
204 struct rcu_head rcu;
205 size_t orig_len; /* From flow_cmd_new netlink actions size */
206 u32 actions_len;
207 struct nlattr actions[];
208 };
209
210 struct flow_stats {
211 u64 packet_count; /* Number of packets matched. */
212 u64 byte_count; /* Number of bytes matched. */
213 unsigned long used; /* Last used time (in jiffies). */
214 spinlock_t lock; /* Lock for atomic stats update. */
215 __be16 tcp_flags; /* Union of seen TCP flags. */
216 };
217
218 struct sw_flow {
219 struct rcu_head rcu;
220 struct {
221 struct hlist_node node[2];
222 u32 hash;
223 } flow_table, ufid_table;
224 int stats_last_writer; /* CPU id of the last writer on
225 * 'stats[0]'.
226 */
227 struct sw_flow_key key;
228 struct sw_flow_id id;
229 struct cpumask cpu_used_mask;
230 struct sw_flow_mask *mask;
231 struct sw_flow_actions __rcu *sf_acts;
232 struct flow_stats __rcu *stats[]; /* One for each CPU. First one
233 * is allocated at flow creation time,
234 * the rest are allocated on demand
235 * while holding the 'stats[0].lock'.
236 */
237 };
238
239 struct arp_eth_header {
240 __be16 ar_hrd; /* format of hardware address */
241 __be16 ar_pro; /* format of protocol address */
242 unsigned char ar_hln; /* length of hardware address */
243 unsigned char ar_pln; /* length of protocol address */
244 __be16 ar_op; /* ARP opcode (command) */
245
246 /* Ethernet+IPv4 specific members. */
247 unsigned char ar_sha[ETH_ALEN]; /* sender hardware address */
248 unsigned char ar_sip[4]; /* sender IP address */
249 unsigned char ar_tha[ETH_ALEN]; /* target hardware address */
250 unsigned char ar_tip[4]; /* target IP address */
251 } __packed;
252
253 static inline u8 ovs_key_mac_proto(const struct sw_flow_key *key)
254 {
255 return key->mac_proto & ~SW_FLOW_KEY_INVALID;
256 }
257
258 static inline u16 __ovs_mac_header_len(u8 mac_proto)
259 {
260 return mac_proto == MAC_PROTO_ETHERNET ? ETH_HLEN : 0;
261 }
262
263 static inline u16 ovs_mac_header_len(const struct sw_flow_key *key)
264 {
265 return __ovs_mac_header_len(ovs_key_mac_proto(key));
266 }
267
268 static inline bool ovs_identifier_is_ufid(const struct sw_flow_id *sfid)
269 {
270 return sfid->ufid_len;
271 }
272
273 static inline bool ovs_identifier_is_key(const struct sw_flow_id *sfid)
274 {
275 return !ovs_identifier_is_ufid(sfid);
276 }
277
278 void ovs_flow_stats_update(struct sw_flow *, __be16 tcp_flags,
279 const struct sk_buff *);
280 void ovs_flow_stats_get(const struct sw_flow *, struct ovs_flow_stats *,
281 unsigned long *used, __be16 *tcp_flags);
282 void ovs_flow_stats_clear(struct sw_flow *);
283 u64 ovs_flow_used_time(unsigned long flow_jiffies);
284
285 /* Update the non-metadata part of the flow key using skb. */
286 int ovs_flow_key_update(struct sk_buff *skb, struct sw_flow_key *key);
287 int ovs_flow_key_extract(const struct ip_tunnel_info *tun_info,
288 struct sk_buff *skb,
289 struct sw_flow_key *key);
290 /* Extract key from packet coming from userspace. */
291 int ovs_flow_key_extract_userspace(struct net *net, const struct nlattr *attr,
292 struct sk_buff *skb,
293 struct sw_flow_key *key, bool log);
294
295 #endif /* flow.h */