Loading...
1/*
2 * Copyright (c) 2007-2011 Nicira Networks.
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/kernel.h>
23#include <linux/netlink.h>
24#include <linux/openvswitch.h>
25#include <linux/spinlock.h>
26#include <linux/types.h>
27#include <linux/rcupdate.h>
28#include <linux/if_ether.h>
29#include <linux/in6.h>
30#include <linux/jiffies.h>
31#include <linux/time.h>
32#include <linux/flex_array.h>
33#include <net/inet_ecn.h>
34
35struct sk_buff;
36
37struct sw_flow_actions {
38 struct rcu_head rcu;
39 u32 actions_len;
40 struct nlattr actions[];
41};
42
43struct sw_flow_key {
44 struct {
45 u32 priority; /* Packet QoS priority. */
46 u16 in_port; /* Input switch port (or USHRT_MAX). */
47 } phy;
48 struct {
49 u8 src[ETH_ALEN]; /* Ethernet source address. */
50 u8 dst[ETH_ALEN]; /* Ethernet destination address. */
51 __be16 tci; /* 0 if no VLAN, VLAN_TAG_PRESENT set otherwise. */
52 __be16 type; /* Ethernet frame type. */
53 } eth;
54 struct {
55 u8 proto; /* IP protocol or lower 8 bits of ARP opcode. */
56 u8 tos; /* IP ToS. */
57 u8 ttl; /* IP TTL/hop limit. */
58 u8 frag; /* One of OVS_FRAG_TYPE_*. */
59 } ip;
60 union {
61 struct {
62 struct {
63 __be32 src; /* IP source address. */
64 __be32 dst; /* IP destination address. */
65 } addr;
66 union {
67 struct {
68 __be16 src; /* TCP/UDP source port. */
69 __be16 dst; /* TCP/UDP destination port. */
70 } tp;
71 struct {
72 u8 sha[ETH_ALEN]; /* ARP source hardware address. */
73 u8 tha[ETH_ALEN]; /* ARP target hardware address. */
74 } arp;
75 };
76 } ipv4;
77 struct {
78 struct {
79 struct in6_addr src; /* IPv6 source address. */
80 struct in6_addr dst; /* IPv6 destination address. */
81 } addr;
82 __be32 label; /* IPv6 flow label. */
83 struct {
84 __be16 src; /* TCP/UDP source port. */
85 __be16 dst; /* TCP/UDP destination port. */
86 } tp;
87 struct {
88 struct in6_addr target; /* ND target address. */
89 u8 sll[ETH_ALEN]; /* ND source link layer address. */
90 u8 tll[ETH_ALEN]; /* ND target link layer address. */
91 } nd;
92 } ipv6;
93 };
94};
95
96struct sw_flow {
97 struct rcu_head rcu;
98 struct hlist_node hash_node[2];
99 u32 hash;
100
101 struct sw_flow_key key;
102 struct sw_flow_actions __rcu *sf_acts;
103
104 spinlock_t lock; /* Lock for values below. */
105 unsigned long used; /* Last used time (in jiffies). */
106 u64 packet_count; /* Number of packets matched. */
107 u64 byte_count; /* Number of bytes matched. */
108 u8 tcp_flags; /* Union of seen TCP flags. */
109};
110
111struct arp_eth_header {
112 __be16 ar_hrd; /* format of hardware address */
113 __be16 ar_pro; /* format of protocol address */
114 unsigned char ar_hln; /* length of hardware address */
115 unsigned char ar_pln; /* length of protocol address */
116 __be16 ar_op; /* ARP opcode (command) */
117
118 /* Ethernet+IPv4 specific members. */
119 unsigned char ar_sha[ETH_ALEN]; /* sender hardware address */
120 unsigned char ar_sip[4]; /* sender IP address */
121 unsigned char ar_tha[ETH_ALEN]; /* target hardware address */
122 unsigned char ar_tip[4]; /* target IP address */
123} __packed;
124
125int ovs_flow_init(void);
126void ovs_flow_exit(void);
127
128struct sw_flow *ovs_flow_alloc(void);
129void ovs_flow_deferred_free(struct sw_flow *);
130void ovs_flow_free(struct sw_flow *flow);
131
132struct sw_flow_actions *ovs_flow_actions_alloc(const struct nlattr *);
133void ovs_flow_deferred_free_acts(struct sw_flow_actions *);
134
135int ovs_flow_extract(struct sk_buff *, u16 in_port, struct sw_flow_key *,
136 int *key_lenp);
137void ovs_flow_used(struct sw_flow *, struct sk_buff *);
138u64 ovs_flow_used_time(unsigned long flow_jiffies);
139
140/* Upper bound on the length of a nlattr-formatted flow key. The longest
141 * nlattr-formatted flow key would be:
142 *
143 * struct pad nl hdr total
144 * ------ --- ------ -----
145 * OVS_KEY_ATTR_PRIORITY 4 -- 4 8
146 * OVS_KEY_ATTR_IN_PORT 4 -- 4 8
147 * OVS_KEY_ATTR_ETHERNET 12 -- 4 16
148 * OVS_KEY_ATTR_8021Q 4 -- 4 8
149 * OVS_KEY_ATTR_ETHERTYPE 2 2 4 8
150 * OVS_KEY_ATTR_IPV6 40 -- 4 44
151 * OVS_KEY_ATTR_ICMPV6 2 2 4 8
152 * OVS_KEY_ATTR_ND 28 -- 4 32
153 * -------------------------------------------------
154 * total 132
155 */
156#define FLOW_BUFSIZE 132
157
158int ovs_flow_to_nlattrs(const struct sw_flow_key *, struct sk_buff *);
159int ovs_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp,
160 const struct nlattr *);
161int ovs_flow_metadata_from_nlattrs(u32 *priority, u16 *in_port,
162 const struct nlattr *);
163
164#define TBL_MIN_BUCKETS 1024
165
166struct flow_table {
167 struct flex_array *buckets;
168 unsigned int count, n_buckets;
169 struct rcu_head rcu;
170 int node_ver;
171 u32 hash_seed;
172 bool keep_flows;
173};
174
175static inline int ovs_flow_tbl_count(struct flow_table *table)
176{
177 return table->count;
178}
179
180static inline int ovs_flow_tbl_need_to_expand(struct flow_table *table)
181{
182 return (table->count > table->n_buckets);
183}
184
185struct sw_flow *ovs_flow_tbl_lookup(struct flow_table *table,
186 struct sw_flow_key *key, int len);
187void ovs_flow_tbl_destroy(struct flow_table *table);
188void ovs_flow_tbl_deferred_destroy(struct flow_table *table);
189struct flow_table *ovs_flow_tbl_alloc(int new_size);
190struct flow_table *ovs_flow_tbl_expand(struct flow_table *table);
191struct flow_table *ovs_flow_tbl_rehash(struct flow_table *table);
192void ovs_flow_tbl_insert(struct flow_table *table, struct sw_flow *flow);
193void ovs_flow_tbl_remove(struct flow_table *table, struct sw_flow *flow);
194u32 ovs_flow_hash(const struct sw_flow_key *key, int key_len);
195
196struct sw_flow *ovs_flow_tbl_next(struct flow_table *table, u32 *bucket, u32 *idx);
197extern const int ovs_key_lens[OVS_KEY_ATTR_MAX + 1];
198
199#endif /* flow.h */
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
38struct 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
49struct 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
57struct sw_flow_key {
58 u8 tun_opts[IP_TUNNEL_OPTS_MAX];
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 u8 tun_proto; /* Protocol of encapsulating tunnel. */
67 u32 ovs_flow_hash; /* Datapath computed hash value. */
68 u32 recirc_id; /* Recirculation ID. */
69 struct {
70 u8 src[ETH_ALEN]; /* Ethernet source address. */
71 u8 dst[ETH_ALEN]; /* Ethernet destination address. */
72 __be16 tci; /* 0 if no VLAN, VLAN_TAG_PRESENT set otherwise. */
73 __be16 type; /* Ethernet frame type. */
74 } eth;
75 union {
76 struct {
77 __be32 top_lse; /* top label stack entry */
78 } mpls;
79 struct {
80 u8 proto; /* IP protocol or lower 8 bits of ARP opcode. */
81 u8 tos; /* IP ToS. */
82 u8 ttl; /* IP TTL/hop limit. */
83 u8 frag; /* One of OVS_FRAG_TYPE_*. */
84 } ip;
85 };
86 struct {
87 __be16 src; /* TCP/UDP/SCTP source port. */
88 __be16 dst; /* TCP/UDP/SCTP destination port. */
89 __be16 flags; /* TCP flags. */
90 } tp;
91 union {
92 struct {
93 struct {
94 __be32 src; /* IP source address. */
95 __be32 dst; /* IP destination address. */
96 } addr;
97 struct {
98 u8 sha[ETH_ALEN]; /* ARP source hardware address. */
99 u8 tha[ETH_ALEN]; /* ARP target hardware address. */
100 } arp;
101 } ipv4;
102 struct {
103 struct {
104 struct in6_addr src; /* IPv6 source address. */
105 struct in6_addr dst; /* IPv6 destination address. */
106 } addr;
107 __be32 label; /* IPv6 flow label. */
108 struct {
109 struct in6_addr target; /* ND target address. */
110 u8 sll[ETH_ALEN]; /* ND source link layer address. */
111 u8 tll[ETH_ALEN]; /* ND target link layer address. */
112 } nd;
113 } ipv6;
114 };
115 struct {
116 /* Connection tracking fields. */
117 u16 zone;
118 u32 mark;
119 u8 state;
120 struct ovs_key_ct_labels labels;
121 } ct;
122
123} __aligned(BITS_PER_LONG/8); /* Ensure that we can do comparisons as longs. */
124
125struct sw_flow_key_range {
126 unsigned short int start;
127 unsigned short int end;
128};
129
130struct sw_flow_mask {
131 int ref_count;
132 struct rcu_head rcu;
133 struct list_head list;
134 struct sw_flow_key_range range;
135 struct sw_flow_key key;
136};
137
138struct sw_flow_match {
139 struct sw_flow_key *key;
140 struct sw_flow_key_range range;
141 struct sw_flow_mask *mask;
142};
143
144#define MAX_UFID_LENGTH 16 /* 128 bits */
145
146struct sw_flow_id {
147 u32 ufid_len;
148 union {
149 u32 ufid[MAX_UFID_LENGTH / 4];
150 struct sw_flow_key *unmasked_key;
151 };
152};
153
154struct sw_flow_actions {
155 struct rcu_head rcu;
156 size_t orig_len; /* From flow_cmd_new netlink actions size */
157 u32 actions_len;
158 struct nlattr actions[];
159};
160
161struct flow_stats {
162 u64 packet_count; /* Number of packets matched. */
163 u64 byte_count; /* Number of bytes matched. */
164 unsigned long used; /* Last used time (in jiffies). */
165 spinlock_t lock; /* Lock for atomic stats update. */
166 __be16 tcp_flags; /* Union of seen TCP flags. */
167};
168
169struct sw_flow {
170 struct rcu_head rcu;
171 struct {
172 struct hlist_node node[2];
173 u32 hash;
174 } flow_table, ufid_table;
175 int stats_last_writer; /* NUMA-node id of the last writer on
176 * 'stats[0]'.
177 */
178 struct sw_flow_key key;
179 struct sw_flow_id id;
180 struct sw_flow_mask *mask;
181 struct sw_flow_actions __rcu *sf_acts;
182 struct flow_stats __rcu *stats[]; /* One for each NUMA node. First one
183 * is allocated at flow creation time,
184 * the rest are allocated on demand
185 * while holding the 'stats[0].lock'.
186 */
187};
188
189struct arp_eth_header {
190 __be16 ar_hrd; /* format of hardware address */
191 __be16 ar_pro; /* format of protocol address */
192 unsigned char ar_hln; /* length of hardware address */
193 unsigned char ar_pln; /* length of protocol address */
194 __be16 ar_op; /* ARP opcode (command) */
195
196 /* Ethernet+IPv4 specific members. */
197 unsigned char ar_sha[ETH_ALEN]; /* sender hardware address */
198 unsigned char ar_sip[4]; /* sender IP address */
199 unsigned char ar_tha[ETH_ALEN]; /* target hardware address */
200 unsigned char ar_tip[4]; /* target IP address */
201} __packed;
202
203static inline bool ovs_identifier_is_ufid(const struct sw_flow_id *sfid)
204{
205 return sfid->ufid_len;
206}
207
208static inline bool ovs_identifier_is_key(const struct sw_flow_id *sfid)
209{
210 return !ovs_identifier_is_ufid(sfid);
211}
212
213void ovs_flow_stats_update(struct sw_flow *, __be16 tcp_flags,
214 const struct sk_buff *);
215void ovs_flow_stats_get(const struct sw_flow *, struct ovs_flow_stats *,
216 unsigned long *used, __be16 *tcp_flags);
217void ovs_flow_stats_clear(struct sw_flow *);
218u64 ovs_flow_used_time(unsigned long flow_jiffies);
219
220int ovs_flow_key_update(struct sk_buff *skb, struct sw_flow_key *key);
221int ovs_flow_key_extract(const struct ip_tunnel_info *tun_info,
222 struct sk_buff *skb,
223 struct sw_flow_key *key);
224/* Extract key from packet coming from userspace. */
225int ovs_flow_key_extract_userspace(struct net *net, const struct nlattr *attr,
226 struct sk_buff *skb,
227 struct sw_flow_key *key, bool log);
228
229#endif /* flow.h */