xref: /linux/net/openvswitch/flow.h (revision ca55b2fef3a9373fcfc30f82fd26bc7fccbda732)
1 /*
2  * Copyright (c) 2007-2014 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 <net/inet_ecn.h>
35 #include <net/ip_tunnels.h>
36 #include <net/dst_metadata.h>
37 
38 struct sk_buff;
39 
40 /* Store options at the end of the array if they are less than the
41  * maximum size. This allows us to get the benefits of variable length
42  * matching for small options.
43  */
44 #define TUN_METADATA_OFFSET(opt_len) \
45 	(FIELD_SIZEOF(struct sw_flow_key, tun_opts) - opt_len)
46 #define TUN_METADATA_OPTS(flow_key, opt_len) \
47 	((void *)((flow_key)->tun_opts + TUN_METADATA_OFFSET(opt_len)))
48 
49 struct ovs_tunnel_info {
50 	struct metadata_dst	*tun_dst;
51 };
52 
53 #define OVS_SW_FLOW_KEY_METADATA_SIZE			\
54 	(offsetof(struct sw_flow_key, recirc_id) +	\
55 	FIELD_SIZEOF(struct sw_flow_key, recirc_id))
56 
57 struct sw_flow_key {
58 	u8 tun_opts[255];
59 	u8 tun_opts_len;
60 	struct ip_tunnel_key tun_key;	/* Encapsulating tunnel key. */
61 	struct {
62 		u32	priority;	/* Packet QoS priority. */
63 		u32	skb_mark;	/* SKB mark. */
64 		u16	in_port;	/* Input switch port (or DP_MAX_PORTS). */
65 	} __packed phy; /* Safe when right after 'tun_key'. */
66 	u32 ovs_flow_hash;		/* Datapath computed hash value.  */
67 	u32 recirc_id;			/* Recirculation ID.  */
68 	struct {
69 		u8     src[ETH_ALEN];	/* Ethernet source address. */
70 		u8     dst[ETH_ALEN];	/* Ethernet destination address. */
71 		__be16 tci;		/* 0 if no VLAN, VLAN_TAG_PRESENT set otherwise. */
72 		__be16 type;		/* Ethernet frame type. */
73 	} eth;
74 	union {
75 		struct {
76 			__be32 top_lse;	/* top label stack entry */
77 		} mpls;
78 		struct {
79 			u8     proto;	/* IP protocol or lower 8 bits of ARP opcode. */
80 			u8     tos;	    /* IP ToS. */
81 			u8     ttl;	    /* IP TTL/hop limit. */
82 			u8     frag;	/* One of OVS_FRAG_TYPE_*. */
83 		} ip;
84 	};
85 	struct {
86 		__be16 src;		/* TCP/UDP/SCTP source port. */
87 		__be16 dst;		/* TCP/UDP/SCTP destination port. */
88 		__be16 flags;		/* TCP flags. */
89 	} tp;
90 	union {
91 		struct {
92 			struct {
93 				__be32 src;	/* IP source address. */
94 				__be32 dst;	/* IP destination address. */
95 			} addr;
96 			struct {
97 				u8 sha[ETH_ALEN];	/* ARP source hardware address. */
98 				u8 tha[ETH_ALEN];	/* ARP target hardware address. */
99 			} arp;
100 		} ipv4;
101 		struct {
102 			struct {
103 				struct in6_addr src;	/* IPv6 source address. */
104 				struct in6_addr dst;	/* IPv6 destination address. */
105 			} addr;
106 			__be32 label;			/* IPv6 flow label. */
107 			struct {
108 				struct in6_addr target;	/* ND target address. */
109 				u8 sll[ETH_ALEN];	/* ND source link layer address. */
110 				u8 tll[ETH_ALEN];	/* ND target link layer address. */
111 			} nd;
112 		} ipv6;
113 	};
114 	struct {
115 		/* Connection tracking fields. */
116 		u16 zone;
117 		u32 mark;
118 		u8 state;
119 		struct ovs_key_ct_labels labels;
120 	} ct;
121 
122 } __aligned(BITS_PER_LONG/8); /* Ensure that we can do comparisons as longs. */
123 
124 struct sw_flow_key_range {
125 	unsigned short int start;
126 	unsigned short int end;
127 };
128 
129 struct sw_flow_mask {
130 	int ref_count;
131 	struct rcu_head rcu;
132 	struct list_head list;
133 	struct sw_flow_key_range range;
134 	struct sw_flow_key key;
135 };
136 
137 struct sw_flow_match {
138 	struct sw_flow_key *key;
139 	struct sw_flow_key_range range;
140 	struct sw_flow_mask *mask;
141 };
142 
143 #define MAX_UFID_LENGTH 16 /* 128 bits */
144 
145 struct sw_flow_id {
146 	u32 ufid_len;
147 	union {
148 		u32 ufid[MAX_UFID_LENGTH / 4];
149 		struct sw_flow_key *unmasked_key;
150 	};
151 };
152 
153 struct sw_flow_actions {
154 	struct rcu_head rcu;
155 	size_t orig_len;	/* From flow_cmd_new netlink actions size */
156 	u32 actions_len;
157 	struct nlattr actions[];
158 };
159 
160 struct flow_stats {
161 	u64 packet_count;		/* Number of packets matched. */
162 	u64 byte_count;			/* Number of bytes matched. */
163 	unsigned long used;		/* Last used time (in jiffies). */
164 	spinlock_t lock;		/* Lock for atomic stats update. */
165 	__be16 tcp_flags;		/* Union of seen TCP flags. */
166 };
167 
168 struct sw_flow {
169 	struct rcu_head rcu;
170 	struct {
171 		struct hlist_node node[2];
172 		u32 hash;
173 	} flow_table, ufid_table;
174 	int stats_last_writer;		/* NUMA-node id of the last writer on
175 					 * 'stats[0]'.
176 					 */
177 	struct sw_flow_key key;
178 	struct sw_flow_id id;
179 	struct sw_flow_mask *mask;
180 	struct sw_flow_actions __rcu *sf_acts;
181 	struct flow_stats __rcu *stats[]; /* One for each NUMA node.  First one
182 					   * is allocated at flow creation time,
183 					   * the rest are allocated on demand
184 					   * while holding the 'stats[0].lock'.
185 					   */
186 };
187 
188 struct arp_eth_header {
189 	__be16      ar_hrd;	/* format of hardware address   */
190 	__be16      ar_pro;	/* format of protocol address   */
191 	unsigned char   ar_hln;	/* length of hardware address   */
192 	unsigned char   ar_pln;	/* length of protocol address   */
193 	__be16      ar_op;	/* ARP opcode (command)     */
194 
195 	/* Ethernet+IPv4 specific members. */
196 	unsigned char       ar_sha[ETH_ALEN];	/* sender hardware address  */
197 	unsigned char       ar_sip[4];		/* sender IP address        */
198 	unsigned char       ar_tha[ETH_ALEN];	/* target hardware address  */
199 	unsigned char       ar_tip[4];		/* target IP address        */
200 } __packed;
201 
202 static inline bool ovs_identifier_is_ufid(const struct sw_flow_id *sfid)
203 {
204 	return sfid->ufid_len;
205 }
206 
207 static inline bool ovs_identifier_is_key(const struct sw_flow_id *sfid)
208 {
209 	return !ovs_identifier_is_ufid(sfid);
210 }
211 
212 void ovs_flow_stats_update(struct sw_flow *, __be16 tcp_flags,
213 			   const struct sk_buff *);
214 void ovs_flow_stats_get(const struct sw_flow *, struct ovs_flow_stats *,
215 			unsigned long *used, __be16 *tcp_flags);
216 void ovs_flow_stats_clear(struct sw_flow *);
217 u64 ovs_flow_used_time(unsigned long flow_jiffies);
218 
219 int ovs_flow_key_update(struct sk_buff *skb, struct sw_flow_key *key);
220 int ovs_flow_key_extract(const struct ip_tunnel_info *tun_info,
221 			 struct sk_buff *skb,
222 			 struct sw_flow_key *key);
223 /* Extract key from packet coming from userspace. */
224 int ovs_flow_key_extract_userspace(struct net *net, const struct nlattr *attr,
225 				   struct sk_buff *skb,
226 				   struct sw_flow_key *key, bool log);
227 
228 #endif /* flow.h */
229