blob: 873d5be7926c1bef17c83ea7c79eecafede9149e [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
2 * IP Virtual Server
3 * data structure and functionality definitions
4 */
5
Julius Volzbc4768e2008-07-31 20:45:24 -07006#ifndef _NET_IP_VS_H
7#define _NET_IP_VS_H
Linus Torvalds1da177e2005-04-16 15:20:36 -07008
Julius Volzbc4768e2008-07-31 20:45:24 -07009#include <linux/ip_vs.h> /* definitions shared with userland */
Linus Torvalds1da177e2005-04-16 15:20:36 -070010
Julius Volzbc4768e2008-07-31 20:45:24 -070011#include <asm/types.h> /* for __uXX types */
12
13#include <linux/sysctl.h> /* for ctl_path */
Linus Torvalds1da177e2005-04-16 15:20:36 -070014#include <linux/list.h> /* for struct list_head */
15#include <linux/spinlock.h> /* for struct rwlock_t */
Arun Sharma600634972011-07-26 16:09:06 -070016#include <linux/atomic.h> /* for struct atomic_t */
Linus Torvalds1da177e2005-04-16 15:20:36 -070017#include <linux/compiler.h>
Arnaldo Carvalho de Melo14c85022005-12-27 02:43:12 -020018#include <linux/timer.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070019
Arnaldo Carvalho de Melo14c85022005-12-27 02:43:12 -020020#include <net/checksum.h>
Julius Volze7ade462008-09-02 15:55:33 +020021#include <linux/netfilter.h> /* for union nf_inet_addr */
KOVACS Krisztian1668e012008-10-01 07:33:10 -070022#include <linux/ip.h>
Julius Volze7ade462008-09-02 15:55:33 +020023#include <linux/ipv6.h> /* for struct ipv6hdr */
24#include <net/ipv6.h> /* for ipv6_addr_copy */
Julian Anastasovcf356d62010-10-17 16:21:07 +030025#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
Julian Anastasovf4bc17c2010-09-21 17:35:41 +020026#include <net/netfilter/nf_conntrack.h>
27#endif
Hans Schillstrom61b1ab42011-01-03 14:44:42 +010028#include <net/net_namespace.h> /* Netw namespace */
29
30/*
31 * Generic access of ipvs struct
32 */
33static inline struct netns_ipvs *net_ipvs(struct net* net)
34{
35 return net->ipvs;
36}
Hans Schillstromfc723252011-01-03 14:44:43 +010037/*
38 * Get net ptr from skb in traffic cases
39 * use skb_sknet when call is from userland (ioctl or netlink)
40 */
Hans Schillstroma0840e22011-01-03 14:44:58 +010041static inline struct net *skb_net(const struct sk_buff *skb)
Hans Schillstromfc723252011-01-03 14:44:43 +010042{
43#ifdef CONFIG_NET_NS
44#ifdef CONFIG_IP_VS_DEBUG
45 /*
46 * This is used for debug only.
47 * Start with the most likely hit
48 * End with BUG
49 */
50 if (likely(skb->dev && skb->dev->nd_net))
51 return dev_net(skb->dev);
Hans Schillstroma09d1972011-04-04 15:25:18 +020052 if (skb_dst(skb) && skb_dst(skb)->dev)
Hans Schillstromfc723252011-01-03 14:44:43 +010053 return dev_net(skb_dst(skb)->dev);
54 WARN(skb->sk, "Maybe skb_sknet should be used in %s() at line:%d\n",
55 __func__, __LINE__);
56 if (likely(skb->sk && skb->sk->sk_net))
57 return sock_net(skb->sk);
58 pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
59 __func__, __LINE__);
60 BUG();
61#else
62 return dev_net(skb->dev ? : skb_dst(skb)->dev);
63#endif
64#else
65 return &init_net;
66#endif
67}
68
Hans Schillstroma0840e22011-01-03 14:44:58 +010069static inline struct net *skb_sknet(const struct sk_buff *skb)
Hans Schillstromfc723252011-01-03 14:44:43 +010070{
71#ifdef CONFIG_NET_NS
72#ifdef CONFIG_IP_VS_DEBUG
73 /* Start with the most likely hit */
74 if (likely(skb->sk && skb->sk->sk_net))
75 return sock_net(skb->sk);
76 WARN(skb->dev, "Maybe skb_net should be used instead in %s() line:%d\n",
77 __func__, __LINE__);
78 if (likely(skb->dev && skb->dev->nd_net))
79 return dev_net(skb->dev);
80 pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
81 __func__, __LINE__);
82 BUG();
83#else
84 return sock_net(skb->sk);
85#endif
86#else
87 return &init_net;
88#endif
89}
Hans Schillstromb17fc992011-01-03 14:44:56 +010090/*
91 * This one needed for single_open_net since net is stored directly in
Lucas De Marchi25985ed2011-03-30 22:57:33 -030092 * private not as a struct i.e. seq_file_net can't be used.
Hans Schillstromb17fc992011-01-03 14:44:56 +010093 */
94static inline struct net *seq_file_single_net(struct seq_file *seq)
95{
96#ifdef CONFIG_NET_NS
97 return (struct net *)seq->private;
98#else
99 return &init_net;
100#endif
101}
Catalin(ux) M. BOIE6f7edb42010-01-05 05:50:24 +0100102
103/* Connections' size value needed by ip_vs_ctl.c */
104extern int ip_vs_conn_tab_size;
105
106
Julius Volz64aae3c2008-09-02 15:55:34 +0200107struct ip_vs_iphdr {
108 int len;
109 __u8 protocol;
110 union nf_inet_addr saddr;
111 union nf_inet_addr daddr;
112};
113
114static inline void
115ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
116{
117#ifdef CONFIG_IP_VS_IPV6
118 if (af == AF_INET6) {
119 const struct ipv6hdr *iph = nh;
120 iphdr->len = sizeof(struct ipv6hdr);
121 iphdr->protocol = iph->nexthdr;
122 ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr);
123 ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr);
124 } else
125#endif
126 {
127 const struct iphdr *iph = nh;
128 iphdr->len = iph->ihl * 4;
129 iphdr->protocol = iph->protocol;
130 iphdr->saddr.ip = iph->saddr;
131 iphdr->daddr.ip = iph->daddr;
132 }
133}
134
135static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
136 const union nf_inet_addr *src)
137{
138#ifdef CONFIG_IP_VS_IPV6
139 if (af == AF_INET6)
140 ipv6_addr_copy(&dst->in6, &src->in6);
141 else
142#endif
143 dst->ip = src->ip;
144}
145
146static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
147 const union nf_inet_addr *b)
148{
149#ifdef CONFIG_IP_VS_IPV6
150 if (af == AF_INET6)
151 return ipv6_addr_equal(&a->in6, &b->in6);
152#endif
153 return a->ip == b->ip;
154}
155
Linus Torvalds1da177e2005-04-16 15:20:36 -0700156#ifdef CONFIG_IP_VS_DEBUG
Arnaldo Carvalho de Melo14c85022005-12-27 02:43:12 -0200157#include <linux/net.h>
158
Linus Torvalds1da177e2005-04-16 15:20:36 -0700159extern int ip_vs_get_debug_level(void);
Julius Volzc842a3a2008-09-02 15:55:35 +0200160
161static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
162 const union nf_inet_addr *addr,
163 int *idx)
164{
165 int len;
166#ifdef CONFIG_IP_VS_IPV6
167 if (af == AF_INET6)
Harvey Harrison5b095d9892008-10-29 12:52:50 -0700168 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]",
Harvey Harrison0c6ce782008-10-28 16:09:23 -0700169 &addr->in6) + 1;
Julius Volzc842a3a2008-09-02 15:55:35 +0200170 else
171#endif
Harvey Harrison3685f25d2008-10-31 00:56:49 -0700172 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
173 &addr->ip) + 1;
Julius Volzc842a3a2008-09-02 15:55:35 +0200174
175 *idx += len;
176 BUG_ON(*idx > buf_len + 1);
177 return &buf[*idx - len];
178}
179
Hannes Eder9aada7a2009-07-30 14:29:44 -0700180#define IP_VS_DBG_BUF(level, msg, ...) \
181 do { \
182 char ip_vs_dbg_buf[160]; \
183 int ip_vs_dbg_idx = 0; \
184 if (level <= ip_vs_get_debug_level()) \
185 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
186 } while (0)
187#define IP_VS_ERR_BUF(msg...) \
188 do { \
189 char ip_vs_dbg_buf[160]; \
190 int ip_vs_dbg_idx = 0; \
191 pr_err(msg); \
192 } while (0)
Julius Volzc842a3a2008-09-02 15:55:35 +0200193
194/* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
Hannes Eder9aada7a2009-07-30 14:29:44 -0700195#define IP_VS_DBG_ADDR(af, addr) \
196 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
197 sizeof(ip_vs_dbg_buf), addr, \
198 &ip_vs_dbg_idx)
Julius Volzc842a3a2008-09-02 15:55:35 +0200199
Hannes Eder9aada7a2009-07-30 14:29:44 -0700200#define IP_VS_DBG(level, msg, ...) \
201 do { \
202 if (level <= ip_vs_get_debug_level()) \
203 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
204 } while (0)
205#define IP_VS_DBG_RL(msg, ...) \
206 do { \
207 if (net_ratelimit()) \
208 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
209 } while (0)
Julian Anastasov0d796412010-10-17 16:46:17 +0300210#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
Hannes Eder9aada7a2009-07-30 14:29:44 -0700211 do { \
212 if (level <= ip_vs_get_debug_level()) \
Julian Anastasov0d796412010-10-17 16:46:17 +0300213 pp->debug_packet(af, pp, skb, ofs, msg); \
Hannes Eder9aada7a2009-07-30 14:29:44 -0700214 } while (0)
Julian Anastasov0d796412010-10-17 16:46:17 +0300215#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
Hannes Eder9aada7a2009-07-30 14:29:44 -0700216 do { \
217 if (level <= ip_vs_get_debug_level() && \
218 net_ratelimit()) \
Julian Anastasov0d796412010-10-17 16:46:17 +0300219 pp->debug_packet(af, pp, skb, ofs, msg); \
Hannes Eder9aada7a2009-07-30 14:29:44 -0700220 } while (0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700221#else /* NO DEBUGGING at ALL */
Julius Volzc842a3a2008-09-02 15:55:35 +0200222#define IP_VS_DBG_BUF(level, msg...) do {} while (0)
223#define IP_VS_ERR_BUF(msg...) do {} while (0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700224#define IP_VS_DBG(level, msg...) do {} while (0)
225#define IP_VS_DBG_RL(msg...) do {} while (0)
Julian Anastasov0d796412010-10-17 16:46:17 +0300226#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
227#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700228#endif
229
230#define IP_VS_BUG() BUG()
Hannes Eder1e3e2382009-08-02 11:05:41 +0000231#define IP_VS_ERR_RL(msg, ...) \
Hannes Eder9aada7a2009-07-30 14:29:44 -0700232 do { \
233 if (net_ratelimit()) \
Hannes Eder1e3e2382009-08-02 11:05:41 +0000234 pr_err(msg, ##__VA_ARGS__); \
Hannes Eder9aada7a2009-07-30 14:29:44 -0700235 } while (0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700236
237#ifdef CONFIG_IP_VS_DEBUG
238#define EnterFunction(level) \
Hannes Eder9aada7a2009-07-30 14:29:44 -0700239 do { \
240 if (level <= ip_vs_get_debug_level()) \
241 printk(KERN_DEBUG \
242 pr_fmt("Enter: %s, %s line %i\n"), \
243 __func__, __FILE__, __LINE__); \
244 } while (0)
245#define LeaveFunction(level) \
246 do { \
247 if (level <= ip_vs_get_debug_level()) \
248 printk(KERN_DEBUG \
249 pr_fmt("Leave: %s, %s line %i\n"), \
250 __func__, __FILE__, __LINE__); \
251 } while (0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700252#else
253#define EnterFunction(level) do {} while (0)
254#define LeaveFunction(level) do {} while (0)
255#endif
256
257#define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); }
258
259
260/*
261 * The port number of FTP service (in network order).
262 */
Harvey Harrisonf3a7c662009-02-14 22:58:35 -0800263#define FTPPORT cpu_to_be16(21)
264#define FTPDATA cpu_to_be16(20)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700265
266/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700267 * TCP State Values
268 */
269enum {
270 IP_VS_TCP_S_NONE = 0,
271 IP_VS_TCP_S_ESTABLISHED,
272 IP_VS_TCP_S_SYN_SENT,
273 IP_VS_TCP_S_SYN_RECV,
274 IP_VS_TCP_S_FIN_WAIT,
275 IP_VS_TCP_S_TIME_WAIT,
276 IP_VS_TCP_S_CLOSE,
277 IP_VS_TCP_S_CLOSE_WAIT,
278 IP_VS_TCP_S_LAST_ACK,
279 IP_VS_TCP_S_LISTEN,
280 IP_VS_TCP_S_SYNACK,
281 IP_VS_TCP_S_LAST
282};
283
284/*
285 * UDP State Values
286 */
287enum {
288 IP_VS_UDP_S_NORMAL,
289 IP_VS_UDP_S_LAST,
290};
291
292/*
293 * ICMP State Values
294 */
295enum {
296 IP_VS_ICMP_S_NORMAL,
297 IP_VS_ICMP_S_LAST,
298};
299
300/*
Venkata Mohan Reddy2906f662010-02-18 12:31:05 +0100301 * SCTP State Values
302 */
303enum ip_vs_sctp_states {
304 IP_VS_SCTP_S_NONE,
305 IP_VS_SCTP_S_INIT_CLI,
306 IP_VS_SCTP_S_INIT_SER,
307 IP_VS_SCTP_S_INIT_ACK_CLI,
308 IP_VS_SCTP_S_INIT_ACK_SER,
309 IP_VS_SCTP_S_ECHO_CLI,
310 IP_VS_SCTP_S_ECHO_SER,
311 IP_VS_SCTP_S_ESTABLISHED,
312 IP_VS_SCTP_S_SHUT_CLI,
313 IP_VS_SCTP_S_SHUT_SER,
314 IP_VS_SCTP_S_SHUT_ACK_CLI,
315 IP_VS_SCTP_S_SHUT_ACK_SER,
316 IP_VS_SCTP_S_CLOSED,
317 IP_VS_SCTP_S_LAST
318};
319
320/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700321 * Delta sequence info structure
322 * Each ip_vs_conn has 2 (output AND input seq. changes).
323 * Only used in the VS/NAT.
324 */
325struct ip_vs_seq {
326 __u32 init_seq; /* Add delta from this seq */
327 __u32 delta; /* Delta in sequence numbers */
328 __u32 previous_delta; /* Delta in sequence numbers
329 before last resized pkt */
330};
331
Hans Schillstromb17fc992011-01-03 14:44:56 +0100332/*
333 * counters per cpu
334 */
335struct ip_vs_counters {
336 __u32 conns; /* connections scheduled */
337 __u32 inpkts; /* incoming packets */
338 __u32 outpkts; /* outgoing packets */
339 __u64 inbytes; /* incoming bytes */
340 __u64 outbytes; /* outgoing bytes */
341};
342/*
343 * Stats per cpu
344 */
345struct ip_vs_cpu_stats {
346 struct ip_vs_counters ustats;
347 struct u64_stats_sync syncp;
348};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700349
350/*
Sven Wegener3a14a3132008-08-10 18:24:41 +0000351 * IPVS statistics objects
Linus Torvalds1da177e2005-04-16 15:20:36 -0700352 */
Sven Wegener3a14a3132008-08-10 18:24:41 +0000353struct ip_vs_estimator {
354 struct list_head list;
355
356 u64 last_inbytes;
357 u64 last_outbytes;
358 u32 last_conns;
359 u32 last_inpkts;
360 u32 last_outpkts;
361
362 u32 cps;
363 u32 inpps;
364 u32 outpps;
365 u32 inbps;
366 u32 outbps;
367};
368
Eric Dumazetfd2c3ef2009-11-03 03:26:03 +0000369struct ip_vs_stats {
Hans Schillstromb17fc992011-01-03 14:44:56 +0100370 struct ip_vs_stats_user ustats; /* statistics */
Sven Wegenere9c0ce22008-09-08 13:39:04 +0200371 struct ip_vs_estimator est; /* estimator */
Hans Schillstromb17fc992011-01-03 14:44:56 +0100372 struct ip_vs_cpu_stats *cpustats; /* per cpu counters */
373 spinlock_t lock; /* spin lock */
Julian Anastasov55a3d4e2011-03-14 01:37:49 +0200374 struct ip_vs_stats_user ustats0; /* reset values */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700375};
376
Arnaldo Carvalho de Melo14c85022005-12-27 02:43:12 -0200377struct dst_entry;
378struct iphdr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700379struct ip_vs_conn;
380struct ip_vs_app;
Arnaldo Carvalho de Melo14c85022005-12-27 02:43:12 -0200381struct sk_buff;
Hans Schillstrom252c6412011-01-03 14:44:46 +0100382struct ip_vs_proto_data;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700383
384struct ip_vs_protocol {
385 struct ip_vs_protocol *next;
386 char *name;
Julian Anastasov2ad17de2008-04-29 03:21:23 -0700387 u16 protocol;
388 u16 num_states;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700389 int dont_defrag;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700390
391 void (*init)(struct ip_vs_protocol *pp);
392
393 void (*exit)(struct ip_vs_protocol *pp);
394
Hans Schillstrom252c6412011-01-03 14:44:46 +0100395 void (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
396
397 void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
398
Julius Volz51ef3482008-09-02 15:55:40 +0200399 int (*conn_schedule)(int af, struct sk_buff *skb,
Hans Schillstrom93304192011-01-03 14:44:51 +0100400 struct ip_vs_proto_data *pd,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700401 int *verdict, struct ip_vs_conn **cpp);
402
403 struct ip_vs_conn *
Julius Volz51ef3482008-09-02 15:55:40 +0200404 (*conn_in_get)(int af,
405 const struct sk_buff *skb,
Julius Volz51ef3482008-09-02 15:55:40 +0200406 const struct ip_vs_iphdr *iph,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700407 unsigned int proto_off,
408 int inverse);
409
410 struct ip_vs_conn *
Julius Volz51ef3482008-09-02 15:55:40 +0200411 (*conn_out_get)(int af,
412 const struct sk_buff *skb,
Julius Volz51ef3482008-09-02 15:55:40 +0200413 const struct ip_vs_iphdr *iph,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700414 unsigned int proto_off,
415 int inverse);
416
Herbert Xu3db05fe2007-10-15 00:53:15 -0700417 int (*snat_handler)(struct sk_buff *skb,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700418 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
419
Herbert Xu3db05fe2007-10-15 00:53:15 -0700420 int (*dnat_handler)(struct sk_buff *skb,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700421 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
422
Julius Volz51ef3482008-09-02 15:55:40 +0200423 int (*csum_check)(int af, struct sk_buff *skb,
424 struct ip_vs_protocol *pp);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700425
426 const char *(*state_name)(int state);
427
Simon Horman4a516f12011-09-16 14:11:49 +0900428 void (*state_transition)(struct ip_vs_conn *cp, int direction,
429 const struct sk_buff *skb,
430 struct ip_vs_proto_data *pd);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700431
Hans Schillstromab8a5e82011-01-03 14:44:53 +0100432 int (*register_app)(struct net *net, struct ip_vs_app *inc);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700433
Hans Schillstromab8a5e82011-01-03 14:44:53 +0100434 void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700435
436 int (*app_conn_bind)(struct ip_vs_conn *cp);
437
Julian Anastasov0d796412010-10-17 16:46:17 +0300438 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700439 const struct sk_buff *skb,
440 int offset,
441 const char *msg);
442
Hans Schillstrom93304192011-01-03 14:44:51 +0100443 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700444};
445
Hans Schillstrom252c6412011-01-03 14:44:46 +0100446/*
447 * protocol data per netns
448 */
449struct ip_vs_proto_data {
450 struct ip_vs_proto_data *next;
451 struct ip_vs_protocol *pp;
452 int *timeout_table; /* protocol timeout table */
453 atomic_t appcnt; /* counter of proto app incs. */
454 struct tcp_states_t *tcp_state_table;
455};
456
457extern struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
458extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
459 unsigned short proto);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700460
Simon Hormanf11017e2010-08-22 21:37:52 +0900461struct ip_vs_conn_param {
Hans Schillstrom6e67e582011-01-03 14:44:57 +0100462 struct net *net;
Simon Hormanf11017e2010-08-22 21:37:52 +0900463 const union nf_inet_addr *caddr;
464 const union nf_inet_addr *vaddr;
465 __be16 cport;
466 __be16 vport;
467 __u16 protocol;
468 u16 af;
Simon Horman85999282010-08-22 21:37:53 +0900469
470 const struct ip_vs_pe *pe;
471 char *pe_data;
472 __u8 pe_data_len;
Simon Hormanf11017e2010-08-22 21:37:52 +0900473};
474
Linus Torvalds1da177e2005-04-16 15:20:36 -0700475/*
476 * IP_VS structure allocated for each dynamically scheduled connection
477 */
478struct ip_vs_conn {
Changli Gao731109e2011-02-19 18:05:08 +0800479 struct hlist_node c_list; /* hashed list heads */
Hans Schillstrom6e67e582011-01-03 14:44:57 +0100480#ifdef CONFIG_NET_NS
481 struct net *net; /* Name space */
482#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700483 /* Protocol, addresses and port numbers */
Hans Schillstrom6e67e582011-01-03 14:44:57 +0100484 u16 af; /* address family */
485 __be16 cport;
486 __be16 vport;
487 __be16 dport;
488 __u32 fwmark; /* Fire wall mark from skb */
489 union nf_inet_addr caddr; /* client address */
490 union nf_inet_addr vaddr; /* virtual address */
491 union nf_inet_addr daddr; /* destination address */
492 volatile __u32 flags; /* status flags */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700493 __u16 protocol; /* Which protocol (TCP/UDP) */
494
495 /* counter and timer */
496 atomic_t refcnt; /* reference count */
497 struct timer_list timer; /* Expiration timer */
498 volatile unsigned long timeout; /* timeout */
499
500 /* Flags and state transition */
501 spinlock_t lock; /* lock for state transition */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700502 volatile __u16 state; /* state info */
Rumen G. Bogdanovskiefac5272007-11-07 02:36:55 -0800503 volatile __u16 old_state; /* old state, to be used for
504 * state transition triggerd
505 * synchronization
506 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700507
508 /* Control members */
509 struct ip_vs_conn *control; /* Master control connection */
510 atomic_t n_control; /* Number of controlled ones */
511 struct ip_vs_dest *dest; /* real server */
512 atomic_t in_pkts; /* incoming packet counter */
513
514 /* packet transmitter for different forwarding methods. If it
515 mangles the packet, it must return NF_DROP or better NF_STOLEN,
516 otherwise this must be changed to a sk_buff **.
Julian Anastasovfc604762010-10-17 16:38:15 +0300517 NF_ACCEPT can be returned when destination is local.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700518 */
519 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
520 struct ip_vs_protocol *pp);
521
522 /* Note: we can group the following members into a structure,
523 in order to save more space, and the following members are
524 only used in VS/NAT anyway */
525 struct ip_vs_app *app; /* bound ip_vs_app object */
526 void *app_data; /* Application private data */
527 struct ip_vs_seq in_seq; /* incoming seq. struct */
528 struct ip_vs_seq out_seq; /* outgoing seq. struct */
Simon Horman85999282010-08-22 21:37:53 +0900529
Simon Hormane9e5eee2010-11-08 20:05:57 +0900530 const struct ip_vs_pe *pe;
Simon Horman85999282010-08-22 21:37:53 +0900531 char *pe_data;
532 __u8 pe_data_len;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700533};
534
Hans Schillstrom6e67e582011-01-03 14:44:57 +0100535/*
536 * To save some memory in conn table when name space is disabled.
537 */
538static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
539{
540#ifdef CONFIG_NET_NS
541 return cp->net;
542#else
543 return &init_net;
544#endif
545}
546static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
547{
548#ifdef CONFIG_NET_NS
549 cp->net = net;
550#endif
551}
552
553static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
554 struct net *net)
555{
556#ifdef CONFIG_NET_NS
557 return cp->net == net;
558#else
559 return 1;
560#endif
561}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700562
563/*
Julius Volzc860c6b2008-09-02 15:55:36 +0200564 * Extended internal versions of struct ip_vs_service_user and
565 * ip_vs_dest_user for IPv6 support.
566 *
567 * We need these to conveniently pass around service and destination
568 * options, but unfortunately, we also need to keep the old definitions to
569 * maintain userspace backwards compatibility for the setsockopt interface.
570 */
571struct ip_vs_service_user_kern {
572 /* virtual service addresses */
573 u16 af;
574 u16 protocol;
575 union nf_inet_addr addr; /* virtual ip address */
576 u16 port;
577 u32 fwmark; /* firwall mark of service */
578
579 /* virtual service options */
580 char *sched_name;
Simon Horman0d1e71b2010-08-22 21:37:54 +0900581 char *pe_name;
Julius Volzc860c6b2008-09-02 15:55:36 +0200582 unsigned flags; /* virtual service flags */
583 unsigned timeout; /* persistent timeout in sec */
584 u32 netmask; /* persistent netmask */
585};
586
587
588struct ip_vs_dest_user_kern {
589 /* destination server address */
590 union nf_inet_addr addr;
591 u16 port;
592
593 /* real server options */
594 unsigned conn_flags; /* connection flags */
595 int weight; /* destination weight */
596
597 /* thresholds for active connections */
598 u32 u_threshold; /* upper threshold */
599 u32 l_threshold; /* lower threshold */
600};
601
602
603/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700604 * The information about the virtual service offered to the net
605 * and the forwarding entries
606 */
607struct ip_vs_service {
608 struct list_head s_list; /* for normal service table */
609 struct list_head f_list; /* for fwmark-based service table */
610 atomic_t refcnt; /* reference counter */
611 atomic_t usecnt; /* use counter */
612
Julius Volze7ade462008-09-02 15:55:33 +0200613 u16 af; /* address family */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700614 __u16 protocol; /* which protocol (TCP/UDP) */
Julius Volze7ade462008-09-02 15:55:33 +0200615 union nf_inet_addr addr; /* IP address for virtual service */
Al Viro014d7302006-09-28 14:29:52 -0700616 __be16 port; /* port number for the service */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700617 __u32 fwmark; /* firewall mark of the service */
618 unsigned flags; /* service status flags */
619 unsigned timeout; /* persistent timeout in ticks */
Al Viro014d7302006-09-28 14:29:52 -0700620 __be32 netmask; /* grouping granularity */
Hans Schillstromfc723252011-01-03 14:44:43 +0100621 struct net *net;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700622
623 struct list_head destinations; /* real server d-linked list */
624 __u32 num_dests; /* number of servers */
625 struct ip_vs_stats stats; /* statistics for the service */
626 struct ip_vs_app *inc; /* bind conns to this app inc */
627
628 /* for scheduling */
629 struct ip_vs_scheduler *scheduler; /* bound scheduler object */
630 rwlock_t sched_lock; /* lock sched_data */
631 void *sched_data; /* scheduler application data */
Simon Horman85999282010-08-22 21:37:53 +0900632
633 /* alternate persistence engine */
634 struct ip_vs_pe *pe;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700635};
636
637
638/*
639 * The real server destination forwarding entry
640 * with ip address, port number, and so on.
641 */
642struct ip_vs_dest {
643 struct list_head n_list; /* for the dests in the service */
644 struct list_head d_list; /* for table with all the dests */
645
Julius Volze7ade462008-09-02 15:55:33 +0200646 u16 af; /* address family */
Al Viro014d7302006-09-28 14:29:52 -0700647 __be16 port; /* port number of the server */
Hans Schillstromf2431e62011-01-03 14:45:00 +0100648 union nf_inet_addr addr; /* IP address of the server */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700649 volatile unsigned flags; /* dest status flags */
650 atomic_t conn_flags; /* flags to copy to conn */
651 atomic_t weight; /* server weight */
652
653 atomic_t refcnt; /* reference counter */
654 struct ip_vs_stats stats; /* statistics */
655
656 /* connection counters and thresholds */
657 atomic_t activeconns; /* active connections */
658 atomic_t inactconns; /* inactive connections */
659 atomic_t persistconns; /* persistent connections */
660 __u32 u_threshold; /* upper threshold */
661 __u32 l_threshold; /* lower threshold */
662
663 /* for destination cache */
664 spinlock_t dst_lock; /* lock of dst_cache */
665 struct dst_entry *dst_cache; /* destination cache entry */
666 u32 dst_rtos; /* RT_TOS(tos) for dst */
Hans Schillstrom714f0952010-10-19 10:38:48 +0200667 u32 dst_cookie;
Julian Anastasovc92f5ca2011-05-10 12:46:05 +0000668 union nf_inet_addr dst_saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700669
670 /* for virtual service */
671 struct ip_vs_service *svc; /* service it belongs to */
672 __u16 protocol; /* which protocol (TCP/UDP) */
Al Viro014d7302006-09-28 14:29:52 -0700673 __be16 vport; /* virtual port number */
Hans Schillstromf2431e62011-01-03 14:45:00 +0100674 union nf_inet_addr vaddr; /* virtual IP address */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700675 __u32 vfwmark; /* firewall mark of service */
676};
677
678
679/*
680 * The scheduler object
681 */
682struct ip_vs_scheduler {
683 struct list_head n_list; /* d-linked list head */
684 char *name; /* scheduler name */
685 atomic_t refcnt; /* reference counter */
686 struct module *module; /* THIS_MODULE/NULL */
687
688 /* scheduler initializing service */
689 int (*init_service)(struct ip_vs_service *svc);
690 /* scheduling service finish */
691 int (*done_service)(struct ip_vs_service *svc);
692 /* scheduler updating service */
693 int (*update_service)(struct ip_vs_service *svc);
694
695 /* selecting a server from the given service */
696 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
697 const struct sk_buff *skb);
698};
699
Simon Horman85999282010-08-22 21:37:53 +0900700/* The persistence engine object */
701struct ip_vs_pe {
702 struct list_head n_list; /* d-linked list head */
703 char *name; /* scheduler name */
704 atomic_t refcnt; /* reference counter */
705 struct module *module; /* THIS_MODULE/NULL */
706
707 /* get the connection template, if any */
708 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
709 bool (*ct_match)(const struct ip_vs_conn_param *p,
710 struct ip_vs_conn *ct);
711 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
712 bool inverse);
Simon Hormana3c918a2010-08-22 21:37:53 +0900713 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
Simon Horman85999282010-08-22 21:37:53 +0900714};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700715
716/*
717 * The application module object (a.k.a. app incarnation)
718 */
Eric Dumazetfd2c3ef2009-11-03 03:26:03 +0000719struct ip_vs_app {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700720 struct list_head a_list; /* member in app list */
721 int type; /* IP_VS_APP_TYPE_xxx */
722 char *name; /* application module name */
723 __u16 protocol;
724 struct module *module; /* THIS_MODULE/NULL */
725 struct list_head incs_list; /* list of incarnations */
726
727 /* members for application incarnations */
728 struct list_head p_list; /* member in proto app list */
729 struct ip_vs_app *app; /* its real application */
Al Viro014d7302006-09-28 14:29:52 -0700730 __be16 port; /* port number in net order */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700731 atomic_t usecnt; /* usage counter */
732
Julian Anastasov8b27b102010-10-17 16:17:20 +0300733 /*
734 * output hook: Process packet in inout direction, diff set for TCP.
735 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
736 * 2=Mangled but checksum was not updated
737 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700738 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
Herbert Xu3db05fe2007-10-15 00:53:15 -0700739 struct sk_buff *, int *diff);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700740
Julian Anastasov8b27b102010-10-17 16:17:20 +0300741 /*
742 * input hook: Process packet in outin direction, diff set for TCP.
743 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
744 * 2=Mangled but checksum was not updated
745 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700746 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
Herbert Xu3db05fe2007-10-15 00:53:15 -0700747 struct sk_buff *, int *diff);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700748
749 /* ip_vs_app initializer */
750 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
751
752 /* ip_vs_app finish */
753 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
754
755
756 /* not used now */
757 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
758 struct ip_vs_protocol *);
759
760 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
761
762 int * timeout_table;
763 int * timeouts;
764 int timeouts_size;
765
766 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
767 int *verdict, struct ip_vs_conn **cpp);
768
769 struct ip_vs_conn *
770 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
771 const struct iphdr *iph, unsigned int proto_off,
772 int inverse);
773
774 struct ip_vs_conn *
775 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
776 const struct iphdr *iph, unsigned int proto_off,
777 int inverse);
778
779 int (*state_transition)(struct ip_vs_conn *cp, int direction,
780 const struct sk_buff *skb,
781 struct ip_vs_app *app);
782
783 void (*timeout_change)(struct ip_vs_app *app, int flags);
784};
785
Julian Anastasov2553d062011-03-04 12:18:07 +0200786/* IPVS in network namespace */
787struct netns_ipvs {
788 int gen; /* Generation */
Hans Schillstrom7a4f0762011-05-03 22:09:31 +0200789 int enable; /* enable like nf_hooks do */
Julian Anastasov2553d062011-03-04 12:18:07 +0200790 /*
791 * Hash table: for real service lookups
792 */
793 #define IP_VS_RTAB_BITS 4
794 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
795 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
796
797 struct list_head rs_table[IP_VS_RTAB_SIZE];
798 /* ip_vs_app */
799 struct list_head app_list;
Hans Schillstromc74c0bf2011-05-24 14:11:05 +0200800 /* ip_vs_ftp */
801 struct ip_vs_app *ftp_app;
Julian Anastasov2553d062011-03-04 12:18:07 +0200802 /* ip_vs_proto */
803 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
804 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
805 /* ip_vs_proto_tcp */
806#ifdef CONFIG_IP_VS_PROTO_TCP
807 #define TCP_APP_TAB_BITS 4
808 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
809 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
810 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
811 spinlock_t tcp_app_lock;
812#endif
813 /* ip_vs_proto_udp */
814#ifdef CONFIG_IP_VS_PROTO_UDP
815 #define UDP_APP_TAB_BITS 4
816 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
817 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
818 struct list_head udp_apps[UDP_APP_TAB_SIZE];
819 spinlock_t udp_app_lock;
820#endif
821 /* ip_vs_proto_sctp */
822#ifdef CONFIG_IP_VS_PROTO_SCTP
823 #define SCTP_APP_TAB_BITS 4
824 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
825 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
826 /* Hash table for SCTP application incarnations */
827 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
828 spinlock_t sctp_app_lock;
829#endif
830 /* ip_vs_conn */
831 atomic_t conn_count; /* connection counter */
832
833 /* ip_vs_ctl */
Julian Anastasov2a0751a2011-03-04 12:20:35 +0200834 struct ip_vs_stats tot_stats; /* Statistics & est. */
Julian Anastasov2553d062011-03-04 12:18:07 +0200835
836 int num_services; /* no of virtual services */
Simon Hormanf2247fb2011-02-04 18:33:02 +0900837
Julian Anastasov2553d062011-03-04 12:18:07 +0200838 rwlock_t rs_lock; /* real services table */
Julian Anastasov2553d062011-03-04 12:18:07 +0200839 /* Trash for destinations */
840 struct list_head dest_trash;
841 /* Service counters */
842 atomic_t ftpsvc_counter;
843 atomic_t nullsvc_counter;
844
Simon Hormanf2247fb2011-02-04 18:33:02 +0900845#ifdef CONFIG_SYSCTL
846 /* 1/rate drop and drop-entry variables */
847 struct delayed_work defense_work; /* Work handler */
848 int drop_rate;
849 int drop_counter;
850 atomic_t dropentry;
851 /* locks in ctl.c */
852 spinlock_t dropentry_lock; /* drop entry handling */
853 spinlock_t droppacket_lock; /* drop packet handling */
854 spinlock_t securetcp_lock; /* state and timeout tables */
855
Julian Anastasov2553d062011-03-04 12:18:07 +0200856 /* sys-ctl struct */
857 struct ctl_table_header *sysctl_hdr;
858 struct ctl_table *sysctl_tbl;
Simon Hormanf2247fb2011-02-04 18:33:02 +0900859#endif
860
Julian Anastasov2553d062011-03-04 12:18:07 +0200861 /* sysctl variables */
862 int sysctl_amemthresh;
863 int sysctl_am_droprate;
864 int sysctl_drop_entry;
865 int sysctl_drop_packet;
866 int sysctl_secure_tcp;
867#ifdef CONFIG_IP_VS_NFCT
868 int sysctl_conntrack;
869#endif
870 int sysctl_snat_reroute;
871 int sysctl_sync_ver;
872 int sysctl_cache_bypass;
873 int sysctl_expire_nodest_conn;
874 int sysctl_expire_quiescent_template;
875 int sysctl_sync_threshold[2];
876 int sysctl_nat_icmp_send;
877
878 /* ip_vs_lblc */
879 int sysctl_lblc_expiration;
880 struct ctl_table_header *lblc_ctl_header;
881 struct ctl_table *lblc_ctl_table;
882 /* ip_vs_lblcr */
883 int sysctl_lblcr_expiration;
884 struct ctl_table_header *lblcr_ctl_header;
885 struct ctl_table *lblcr_ctl_table;
886 /* ip_vs_est */
887 struct list_head est_list; /* estimator list */
888 spinlock_t est_lock;
889 struct timer_list est_timer; /* Estimation timer */
890 /* ip_vs_sync */
891 struct list_head sync_queue;
892 spinlock_t sync_lock;
893 struct ip_vs_sync_buff *sync_buff;
894 spinlock_t sync_buff_lock;
895 struct sockaddr_in sync_mcast_addr;
896 struct task_struct *master_thread;
897 struct task_struct *backup_thread;
898 int send_mesg_maxlen;
899 int recv_mesg_maxlen;
900 volatile int sync_state;
901 volatile int master_syncid;
902 volatile int backup_syncid;
Hans Schillstromae1d48b2011-10-11 10:54:35 +0900903 struct mutex sync_mutex;
Julian Anastasov2553d062011-03-04 12:18:07 +0200904 /* multicast interface name */
905 char master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
906 char backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
907 /* net name space ptr */
908 struct net *net; /* Needed by timer routines */
909};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700910
Simon Horman59e03502011-02-04 18:33:01 +0900911#define DEFAULT_SYNC_THRESHOLD 3
912#define DEFAULT_SYNC_PERIOD 50
Simon Horman7532e8d2011-02-04 18:33:01 +0900913#define DEFAULT_SYNC_VER 1
Simon Horman59e03502011-02-04 18:33:01 +0900914
915#ifdef CONFIG_SYSCTL
916
917static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
918{
919 return ipvs->sysctl_sync_threshold[0];
920}
921
922static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
923{
924 return ipvs->sysctl_sync_threshold[1];
925}
926
Simon Horman7532e8d2011-02-04 18:33:01 +0900927static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
928{
929 return ipvs->sysctl_sync_ver;
930}
931
Simon Horman59e03502011-02-04 18:33:01 +0900932#else
933
934static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
935{
936 return DEFAULT_SYNC_THRESHOLD;
937}
938
939static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
940{
941 return DEFAULT_SYNC_PERIOD;
942}
943
Simon Horman7532e8d2011-02-04 18:33:01 +0900944static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
945{
946 return DEFAULT_SYNC_VER;
947}
948
Simon Horman59e03502011-02-04 18:33:01 +0900949#endif
950
Linus Torvalds1da177e2005-04-16 15:20:36 -0700951/*
952 * IPVS core functions
953 * (from ip_vs_core.c)
954 */
955extern const char *ip_vs_proto_name(unsigned proto);
956extern void ip_vs_init_hash_table(struct list_head *table, int rows);
Sven Wegenerafdd6142008-08-10 09:18:01 +0000957#define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700958
Linus Torvalds1da177e2005-04-16 15:20:36 -0700959#define IP_VS_APP_TYPE_FTP 1
960
961/*
962 * ip_vs_conn handling functions
963 * (from ip_vs_conn.c)
964 */
965
Linus Torvalds1da177e2005-04-16 15:20:36 -0700966enum {
967 IP_VS_DIR_INPUT = 0,
968 IP_VS_DIR_OUTPUT,
969 IP_VS_DIR_INPUT_ONLY,
970 IP_VS_DIR_LAST,
971};
972
Hans Schillstrom6e67e582011-01-03 14:44:57 +0100973static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
Simon Hormanf11017e2010-08-22 21:37:52 +0900974 const union nf_inet_addr *caddr,
975 __be16 cport,
976 const union nf_inet_addr *vaddr,
977 __be16 vport,
978 struct ip_vs_conn_param *p)
979{
Hans Schillstrom6e67e582011-01-03 14:44:57 +0100980 p->net = net;
Simon Hormanf11017e2010-08-22 21:37:52 +0900981 p->af = af;
982 p->protocol = protocol;
983 p->caddr = caddr;
984 p->cport = cport;
985 p->vaddr = vaddr;
986 p->vport = vport;
Simon Horman85999282010-08-22 21:37:53 +0900987 p->pe = NULL;
988 p->pe_data = NULL;
Simon Hormanf11017e2010-08-22 21:37:52 +0900989}
Julius Volz28364a52008-09-02 15:55:43 +0200990
Simon Hormanf11017e2010-08-22 21:37:52 +0900991struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
992struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
Julius Volz28364a52008-09-02 15:55:43 +0200993
Simon Horman5c0d2372010-08-02 17:12:44 +0200994struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
Simon Horman5c0d2372010-08-02 17:12:44 +0200995 const struct ip_vs_iphdr *iph,
996 unsigned int proto_off,
997 int inverse);
998
Simon Hormanf11017e2010-08-22 21:37:52 +0900999struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001000
Simon Horman5c0d2372010-08-02 17:12:44 +02001001struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
Simon Horman5c0d2372010-08-02 17:12:44 +02001002 const struct ip_vs_iphdr *iph,
1003 unsigned int proto_off,
1004 int inverse);
1005
Linus Torvalds1da177e2005-04-16 15:20:36 -07001006/* put back the conn without restarting its timer */
1007static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1008{
1009 atomic_dec(&cp->refcnt);
1010}
1011extern void ip_vs_conn_put(struct ip_vs_conn *cp);
Al Viro014d7302006-09-28 14:29:52 -07001012extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001013
Simon Hormanf11017e2010-08-22 21:37:52 +09001014struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
1015 const union nf_inet_addr *daddr,
1016 __be16 dport, unsigned flags,
Hans Schillstrom0e051e62010-11-19 14:25:07 +01001017 struct ip_vs_dest *dest, __u32 fwmark);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001018extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1019
1020extern const char * ip_vs_state_name(__u16 proto, int state);
1021
Hans Schillstrom4a85b962011-01-03 14:44:47 +01001022extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001023extern int ip_vs_check_template(struct ip_vs_conn *ct);
Hans Schillstromf6340ee2011-01-03 14:44:59 +01001024extern void ip_vs_random_dropentry(struct net *net);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001025extern int ip_vs_conn_init(void);
1026extern void ip_vs_conn_cleanup(void);
1027
1028static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1029{
1030 struct ip_vs_conn *ctl_cp = cp->control;
1031 if (!ctl_cp) {
Julius Volzcfc78c52008-09-02 15:55:53 +02001032 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1033 "%s:%d to %s:%d\n",
1034 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1035 ntohs(cp->cport),
1036 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1037 ntohs(cp->vport));
1038
Linus Torvalds1da177e2005-04-16 15:20:36 -07001039 return;
1040 }
1041
Julius Volzcfc78c52008-09-02 15:55:53 +02001042 IP_VS_DBG_BUF(7, "DELeting control for: "
1043 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1044 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1045 ntohs(cp->cport),
1046 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1047 ntohs(ctl_cp->cport));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001048
1049 cp->control = NULL;
1050 if (atomic_read(&ctl_cp->n_control) == 0) {
Julius Volzcfc78c52008-09-02 15:55:53 +02001051 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1052 "%s:%d to %s:%d\n",
1053 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1054 ntohs(cp->cport),
1055 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1056 ntohs(cp->vport));
1057
Linus Torvalds1da177e2005-04-16 15:20:36 -07001058 return;
1059 }
1060 atomic_dec(&ctl_cp->n_control);
1061}
1062
1063static inline void
1064ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1065{
1066 if (cp->control) {
Julius Volzcfc78c52008-09-02 15:55:53 +02001067 IP_VS_ERR_BUF("request control ADD for already controlled: "
1068 "%s:%d to %s:%d\n",
1069 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1070 ntohs(cp->cport),
1071 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1072 ntohs(cp->vport));
1073
Linus Torvalds1da177e2005-04-16 15:20:36 -07001074 ip_vs_control_del(cp);
1075 }
1076
Julius Volzcfc78c52008-09-02 15:55:53 +02001077 IP_VS_DBG_BUF(7, "ADDing control for: "
1078 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1079 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1080 ntohs(cp->cport),
1081 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1082 ntohs(ctl_cp->cport));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001083
1084 cp->control = ctl_cp;
1085 atomic_inc(&ctl_cp->n_control);
1086}
1087
Hans Schillstrom7a4f0762011-05-03 22:09:31 +02001088/*
1089 * IPVS netns init & cleanup functions
1090 */
Hans Schillstrom503cf152011-05-01 18:50:16 +02001091extern int ip_vs_estimator_net_init(struct net *net);
1092extern int ip_vs_control_net_init(struct net *net);
1093extern int ip_vs_protocol_net_init(struct net *net);
1094extern int ip_vs_app_net_init(struct net *net);
1095extern int ip_vs_conn_net_init(struct net *net);
1096extern int ip_vs_sync_net_init(struct net *net);
1097extern void ip_vs_conn_net_cleanup(struct net *net);
1098extern void ip_vs_app_net_cleanup(struct net *net);
1099extern void ip_vs_protocol_net_cleanup(struct net *net);
1100extern void ip_vs_control_net_cleanup(struct net *net);
1101extern void ip_vs_estimator_net_cleanup(struct net *net);
1102extern void ip_vs_sync_net_cleanup(struct net *net);
1103extern void ip_vs_service_net_cleanup(struct net *net);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001104
1105/*
1106 * IPVS application functions
1107 * (from ip_vs_app.c)
1108 */
1109#define IP_VS_APP_MAX_PORTS 8
Hans Schillstromab8a5e82011-01-03 14:44:53 +01001110extern int register_ip_vs_app(struct net *net, struct ip_vs_app *app);
1111extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001112extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1113extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
Hans Schillstromab8a5e82011-01-03 14:44:53 +01001114extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1115 __u16 proto, __u16 port);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001116extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1117extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1118
Herbert Xu3db05fe2007-10-15 00:53:15 -07001119extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1120extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001121
Simon Horman8be67a62010-08-22 21:37:54 +09001122void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
1123void ip_vs_unbind_pe(struct ip_vs_service *svc);
1124int register_ip_vs_pe(struct ip_vs_pe *pe);
1125int unregister_ip_vs_pe(struct ip_vs_pe *pe);
Simon Hormane9e5eee2010-11-08 20:05:57 +09001126struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
Hans Schillstromfe5e7a12010-11-19 14:25:12 +01001127struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
Simon Hormane9e5eee2010-11-08 20:05:57 +09001128
Paul Gortmaker69e7dae2011-08-29 12:40:15 -04001129/*
1130 * Use a #define to avoid all of module.h just for these trivial ops
1131 */
1132#define ip_vs_pe_get(pe) \
1133 if (pe && pe->module) \
Simon Hormane9e5eee2010-11-08 20:05:57 +09001134 __module_get(pe->module);
Simon Hormane9e5eee2010-11-08 20:05:57 +09001135
Paul Gortmaker69e7dae2011-08-29 12:40:15 -04001136#define ip_vs_pe_put(pe) \
1137 if (pe && pe->module) \
Simon Hormane9e5eee2010-11-08 20:05:57 +09001138 module_put(pe->module);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001139
1140/*
1141 * IPVS protocol functions (from ip_vs_proto.c)
1142 */
1143extern int ip_vs_protocol_init(void);
1144extern void ip_vs_protocol_cleanup(void);
Hans Schillstrom93304192011-01-03 14:44:51 +01001145extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001146extern int *ip_vs_create_timeout_table(int *table, int size);
1147extern int
Jan Engelhardt36cbd3d2009-08-05 10:42:58 -07001148ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1149 const char *name, int to);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001150extern void
Julian Anastasov0d796412010-10-17 16:46:17 +03001151ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1152 const struct sk_buff *skb,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001153 int offset, const char *msg);
1154
1155extern struct ip_vs_protocol ip_vs_protocol_tcp;
1156extern struct ip_vs_protocol ip_vs_protocol_udp;
1157extern struct ip_vs_protocol ip_vs_protocol_icmp;
1158extern struct ip_vs_protocol ip_vs_protocol_esp;
1159extern struct ip_vs_protocol ip_vs_protocol_ah;
Venkata Mohan Reddy2906f662010-02-18 12:31:05 +01001160extern struct ip_vs_protocol ip_vs_protocol_sctp;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001161
1162/*
1163 * Registering/unregistering scheduler functions
1164 * (from ip_vs_sched.c)
1165 */
1166extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1167extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1168extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1169 struct ip_vs_scheduler *scheduler);
1170extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
1171extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1172extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1173extern struct ip_vs_conn *
Julian Anastasov190ecd22010-10-17 16:24:37 +03001174ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
Hans Schillstrom93304192011-01-03 14:44:51 +01001175 struct ip_vs_proto_data *pd, int *ignored);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001176extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
Hans Schillstrom93304192011-01-03 14:44:51 +01001177 struct ip_vs_proto_data *pd);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001178
Patrick Schaaf41ac51e2011-02-11 14:01:12 +01001179extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1180
Linus Torvalds1da177e2005-04-16 15:20:36 -07001181
1182/*
1183 * IPVS control data and functions (from ip_vs_ctl.c)
1184 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001185extern struct ip_vs_stats ip_vs_stats;
Sven Wegener5587da52008-08-10 18:24:40 +00001186extern const struct ctl_path net_vs_ctl_path[];
Hans Schillstromb880c1f2010-11-19 14:25:14 +01001187extern int sysctl_ip_vs_sync_ver;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001188
Hans Schillstromf1313152011-01-03 14:44:55 +01001189extern void ip_vs_sync_switch_mode(struct net *net, int mode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001190extern struct ip_vs_service *
Hans Schillstromfc723252011-01-03 14:44:43 +01001191ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
Julius Volz3c2e0502008-09-02 15:55:38 +02001192 const union nf_inet_addr *vaddr, __be16 vport);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001193
1194static inline void ip_vs_service_put(struct ip_vs_service *svc)
1195{
1196 atomic_dec(&svc->usecnt);
1197}
1198
1199extern struct ip_vs_dest *
Hans Schillstromfc723252011-01-03 14:44:43 +01001200ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol,
Julius Volz7937df12008-09-02 15:55:48 +02001201 const union nf_inet_addr *daddr, __be16 dport);
1202
Linus Torvalds1da177e2005-04-16 15:20:36 -07001203extern int ip_vs_use_count_inc(void);
1204extern void ip_vs_use_count_dec(void);
1205extern int ip_vs_control_init(void);
1206extern void ip_vs_control_cleanup(void);
Rumen G. Bogdanovski1e356f92007-11-07 02:35:54 -08001207extern struct ip_vs_dest *
Hans Schillstromfc723252011-01-03 14:44:43 +01001208ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1209 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1210 __u16 protocol, __u32 fwmark);
Rumen G. Bogdanovski1e356f92007-11-07 02:35:54 -08001211extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001212
1213
1214/*
1215 * IPVS sync daemon data and function prototypes
1216 * (from ip_vs_sync.c)
1217 */
Hans Schillstromf1313152011-01-03 14:44:55 +01001218extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1219 __u8 syncid);
1220extern int stop_sync_thread(struct net *net, int state);
1221extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001222
1223
1224/*
1225 * IPVS rate estimator prototypes (from ip_vs_est.c)
1226 */
Julian Anastasov6ef757f2011-03-14 01:44:28 +02001227extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1228extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001229extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
Julian Anastasovea9f22c2011-03-14 01:41:54 +02001230extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst,
1231 struct ip_vs_stats *stats);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001232
1233/*
1234 * Various IPVS packet transmitters (from ip_vs_xmit.c)
1235 */
1236extern int ip_vs_null_xmit
1237(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1238extern int ip_vs_bypass_xmit
1239(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1240extern int ip_vs_nat_xmit
1241(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1242extern int ip_vs_tunnel_xmit
1243(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1244extern int ip_vs_dr_xmit
1245(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1246extern int ip_vs_icmp_xmit
Julian Anastasovc92f5ca2011-05-10 12:46:05 +00001247(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1248 int offset, unsigned int hooknum);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001249extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
1250
Julius Volzb3cdd2a72008-09-02 15:55:45 +02001251#ifdef CONFIG_IP_VS_IPV6
1252extern int ip_vs_bypass_xmit_v6
1253(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1254extern int ip_vs_nat_xmit_v6
1255(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1256extern int ip_vs_tunnel_xmit_v6
1257(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1258extern int ip_vs_dr_xmit_v6
1259(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1260extern int ip_vs_icmp_xmit_v6
1261(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
Julian Anastasovc92f5ca2011-05-10 12:46:05 +00001262 int offset, unsigned int hooknum);
Julius Volzb3cdd2a72008-09-02 15:55:45 +02001263#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001264
Simon Horman3a1bbf12011-02-04 18:33:02 +09001265#ifdef CONFIG_SYSCTL
Linus Torvalds1da177e2005-04-16 15:20:36 -07001266/*
1267 * This is a simple mechanism to ignore packets when
1268 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1269 * we start to drop 1/rate of the packets
1270 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001271
Hans Schillstroma0840e22011-01-03 14:44:58 +01001272static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001273{
Hans Schillstroma0840e22011-01-03 14:44:58 +01001274 if (!ipvs->drop_rate)
1275 return 0;
1276 if (--ipvs->drop_counter > 0)
1277 return 0;
1278 ipvs->drop_counter = ipvs->drop_rate;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001279 return 1;
1280}
Simon Horman3a1bbf12011-02-04 18:33:02 +09001281#else
1282static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1283#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001284
1285/*
1286 * ip_vs_fwd_tag returns the forwarding tag of the connection
1287 */
1288#define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1289
Adrian Bunk732db652005-09-01 17:40:26 -07001290static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001291{
1292 char fwd;
1293
1294 switch (IP_VS_FWD_METHOD(cp)) {
1295 case IP_VS_CONN_F_MASQ:
1296 fwd = 'M'; break;
1297 case IP_VS_CONN_F_LOCALNODE:
1298 fwd = 'L'; break;
1299 case IP_VS_CONN_F_TUNNEL:
1300 fwd = 'T'; break;
1301 case IP_VS_CONN_F_DROUTE:
1302 fwd = 'R'; break;
1303 case IP_VS_CONN_F_BYPASS:
1304 fwd = 'B'; break;
1305 default:
1306 fwd = '?'; break;
1307 }
1308 return fwd;
1309}
1310
Linus Torvalds1da177e2005-04-16 15:20:36 -07001311extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
Julius Volzb3cdd2a72008-09-02 15:55:45 +02001312 struct ip_vs_conn *cp, int dir);
1313
1314#ifdef CONFIG_IP_VS_IPV6
1315extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1316 struct ip_vs_conn *cp, int dir);
1317#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001318
Al Virob1550f22006-11-14 21:37:50 -08001319extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001320
Al Virof9214b22006-11-16 02:41:18 -08001321static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001322{
Al Virof9214b22006-11-16 02:41:18 -08001323 __be32 diff[2] = { ~old, new };
Linus Torvalds1da177e2005-04-16 15:20:36 -07001324
Joe Perches07f07572008-11-19 15:44:53 -08001325 return csum_partial(diff, sizeof(diff), oldsum);
Al Virof9214b22006-11-16 02:41:18 -08001326}
1327
Julius Volz0bbdd422008-09-02 15:55:42 +02001328#ifdef CONFIG_IP_VS_IPV6
1329static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1330 __wsum oldsum)
1331{
1332 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1333 new[3], new[2], new[1], new[0] };
1334
Joe Perches07f07572008-11-19 15:44:53 -08001335 return csum_partial(diff, sizeof(diff), oldsum);
Julius Volz0bbdd422008-09-02 15:55:42 +02001336}
1337#endif
1338
Al Virof9214b22006-11-16 02:41:18 -08001339static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1340{
1341 __be16 diff[2] = { ~old, new };
1342
Joe Perches07f07572008-11-19 15:44:53 -08001343 return csum_partial(diff, sizeof(diff), oldsum);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001344}
1345
Julian Anastasovcf356d62010-10-17 16:21:07 +03001346/*
1347 * Forget current conntrack (unconfirmed) and attach notrack entry
1348 */
1349static inline void ip_vs_notrack(struct sk_buff *skb)
1350{
1351#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1352 enum ip_conntrack_info ctinfo;
Jesper Juhl06b69392011-03-09 22:55:05 +01001353 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
Julian Anastasovcf356d62010-10-17 16:21:07 +03001354
1355 if (!ct || !nf_ct_is_untracked(ct)) {
1356 nf_reset(skb);
1357 skb->nfct = &nf_ct_untracked_get()->ct_general;
1358 skb->nfctinfo = IP_CT_NEW;
1359 nf_conntrack_get(skb->nfct);
1360 }
1361#endif
1362}
1363
Julian Anastasovf4bc17c2010-09-21 17:35:41 +02001364#ifdef CONFIG_IP_VS_NFCT
1365/*
1366 * Netfilter connection tracking
1367 * (from ip_vs_nfct.c)
1368 */
Hans Schillstroma0840e22011-01-03 14:44:58 +01001369static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
Julian Anastasovf4bc17c2010-09-21 17:35:41 +02001370{
Simon Hormana4e2f5a2011-02-04 18:33:02 +09001371#ifdef CONFIG_SYSCTL
Hans Schillstroma0840e22011-01-03 14:44:58 +01001372 return ipvs->sysctl_conntrack;
Simon Hormana4e2f5a2011-02-04 18:33:02 +09001373#else
1374 return 0;
1375#endif
Julian Anastasovf4bc17c2010-09-21 17:35:41 +02001376}
1377
Julian Anastasov6523ce12010-09-05 18:02:29 +00001378extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1379 int outin);
Simon Horman3c2de2a2011-09-16 14:02:19 +09001380extern int ip_vs_confirm_conntrack(struct sk_buff *skb);
Julian Anastasovf4bc17c2010-09-21 17:35:41 +02001381extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1382 struct ip_vs_conn *cp, u_int8_t proto,
1383 const __be16 port, int from_rs);
1384extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1385
1386#else
1387
Hans Schillstroma0840e22011-01-03 14:44:58 +01001388static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
Julian Anastasovf4bc17c2010-09-21 17:35:41 +02001389{
1390 return 0;
1391}
1392
1393static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1394 struct ip_vs_conn *cp, int outin)
1395{
1396}
1397
Krzysztof Wilczynskie23ebf02011-10-20 13:18:04 +01001398static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
Julian Anastasovf4bc17c2010-09-21 17:35:41 +02001399{
1400 return NF_ACCEPT;
1401}
1402
1403static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1404{
1405}
1406/* CONFIG_IP_VS_NFCT */
1407#endif
Julian Anastasov6523ce12010-09-05 18:02:29 +00001408
Changli Gaob552f7e2011-02-19 17:32:28 +08001409static inline unsigned int
1410ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1411{
1412 /*
1413 * We think the overhead of processing active connections is 256
1414 * times higher than that of inactive connections in average. (This
1415 * 256 times might not be accurate, we will change it later) We
1416 * use the following formula to estimate the overhead now:
1417 * dest->activeconns*256 + dest->inactconns
1418 */
1419 return (atomic_read(&dest->activeconns) << 8) +
1420 atomic_read(&dest->inactconns);
1421}
1422
Julius Volzbc4768e2008-07-31 20:45:24 -07001423#endif /* _NET_IP_VS_H */