ipvs: IPv6 tunnel mode
[deliverable/linux.git] / include / net / ip_vs.h
1 /*
2 * IP Virtual Server
3 * data structure and functionality definitions
4 */
5
6 #ifndef _NET_IP_VS_H
7 #define _NET_IP_VS_H
8
9 #include <linux/ip_vs.h> /* definitions shared with userland */
10
11 /* old ipvsadm versions still include this file directly */
12 #ifdef __KERNEL__
13
14 #include <asm/types.h> /* for __uXX types */
15
16 #include <linux/sysctl.h> /* for ctl_path */
17 #include <linux/list.h> /* for struct list_head */
18 #include <linux/spinlock.h> /* for struct rwlock_t */
19 #include <asm/atomic.h> /* for struct atomic_t */
20 #include <linux/compiler.h>
21 #include <linux/timer.h>
22
23 #include <net/checksum.h>
24 #include <linux/netfilter.h> /* for union nf_inet_addr */
25 #include <linux/ip.h>
26 #include <linux/ipv6.h> /* for struct ipv6hdr */
27 #include <net/ipv6.h> /* for ipv6_addr_copy */
28 #ifdef CONFIG_IP_VS_NFCT
29 #include <net/netfilter/nf_conntrack.h>
30 #endif
31
32 /* Connections' size value needed by ip_vs_ctl.c */
33 extern int ip_vs_conn_tab_size;
34
35
36 struct ip_vs_iphdr {
37 int len;
38 __u8 protocol;
39 union nf_inet_addr saddr;
40 union nf_inet_addr daddr;
41 };
42
43 static inline void
44 ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
45 {
46 #ifdef CONFIG_IP_VS_IPV6
47 if (af == AF_INET6) {
48 const struct ipv6hdr *iph = nh;
49 iphdr->len = sizeof(struct ipv6hdr);
50 iphdr->protocol = iph->nexthdr;
51 ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr);
52 ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr);
53 } else
54 #endif
55 {
56 const struct iphdr *iph = nh;
57 iphdr->len = iph->ihl * 4;
58 iphdr->protocol = iph->protocol;
59 iphdr->saddr.ip = iph->saddr;
60 iphdr->daddr.ip = iph->daddr;
61 }
62 }
63
64 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
65 const union nf_inet_addr *src)
66 {
67 #ifdef CONFIG_IP_VS_IPV6
68 if (af == AF_INET6)
69 ipv6_addr_copy(&dst->in6, &src->in6);
70 else
71 #endif
72 dst->ip = src->ip;
73 }
74
75 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
76 const union nf_inet_addr *b)
77 {
78 #ifdef CONFIG_IP_VS_IPV6
79 if (af == AF_INET6)
80 return ipv6_addr_equal(&a->in6, &b->in6);
81 #endif
82 return a->ip == b->ip;
83 }
84
85 #ifdef CONFIG_IP_VS_DEBUG
86 #include <linux/net.h>
87
88 extern int ip_vs_get_debug_level(void);
89
90 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
91 const union nf_inet_addr *addr,
92 int *idx)
93 {
94 int len;
95 #ifdef CONFIG_IP_VS_IPV6
96 if (af == AF_INET6)
97 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]",
98 &addr->in6) + 1;
99 else
100 #endif
101 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
102 &addr->ip) + 1;
103
104 *idx += len;
105 BUG_ON(*idx > buf_len + 1);
106 return &buf[*idx - len];
107 }
108
109 #define IP_VS_DBG_BUF(level, msg, ...) \
110 do { \
111 char ip_vs_dbg_buf[160]; \
112 int ip_vs_dbg_idx = 0; \
113 if (level <= ip_vs_get_debug_level()) \
114 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
115 } while (0)
116 #define IP_VS_ERR_BUF(msg...) \
117 do { \
118 char ip_vs_dbg_buf[160]; \
119 int ip_vs_dbg_idx = 0; \
120 pr_err(msg); \
121 } while (0)
122
123 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
124 #define IP_VS_DBG_ADDR(af, addr) \
125 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
126 sizeof(ip_vs_dbg_buf), addr, \
127 &ip_vs_dbg_idx)
128
129 #define IP_VS_DBG(level, msg, ...) \
130 do { \
131 if (level <= ip_vs_get_debug_level()) \
132 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
133 } while (0)
134 #define IP_VS_DBG_RL(msg, ...) \
135 do { \
136 if (net_ratelimit()) \
137 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
138 } while (0)
139 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg) \
140 do { \
141 if (level <= ip_vs_get_debug_level()) \
142 pp->debug_packet(pp, skb, ofs, msg); \
143 } while (0)
144 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg) \
145 do { \
146 if (level <= ip_vs_get_debug_level() && \
147 net_ratelimit()) \
148 pp->debug_packet(pp, skb, ofs, msg); \
149 } while (0)
150 #else /* NO DEBUGGING at ALL */
151 #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
152 #define IP_VS_ERR_BUF(msg...) do {} while (0)
153 #define IP_VS_DBG(level, msg...) do {} while (0)
154 #define IP_VS_DBG_RL(msg...) do {} while (0)
155 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg) do {} while (0)
156 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg) do {} while (0)
157 #endif
158
159 #define IP_VS_BUG() BUG()
160 #define IP_VS_ERR_RL(msg, ...) \
161 do { \
162 if (net_ratelimit()) \
163 pr_err(msg, ##__VA_ARGS__); \
164 } while (0)
165
166 #ifdef CONFIG_IP_VS_DEBUG
167 #define EnterFunction(level) \
168 do { \
169 if (level <= ip_vs_get_debug_level()) \
170 printk(KERN_DEBUG \
171 pr_fmt("Enter: %s, %s line %i\n"), \
172 __func__, __FILE__, __LINE__); \
173 } while (0)
174 #define LeaveFunction(level) \
175 do { \
176 if (level <= ip_vs_get_debug_level()) \
177 printk(KERN_DEBUG \
178 pr_fmt("Leave: %s, %s line %i\n"), \
179 __func__, __FILE__, __LINE__); \
180 } while (0)
181 #else
182 #define EnterFunction(level) do {} while (0)
183 #define LeaveFunction(level) do {} while (0)
184 #endif
185
186 #define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); }
187
188
189 /*
190 * The port number of FTP service (in network order).
191 */
192 #define FTPPORT cpu_to_be16(21)
193 #define FTPDATA cpu_to_be16(20)
194
195 /*
196 * TCP State Values
197 */
198 enum {
199 IP_VS_TCP_S_NONE = 0,
200 IP_VS_TCP_S_ESTABLISHED,
201 IP_VS_TCP_S_SYN_SENT,
202 IP_VS_TCP_S_SYN_RECV,
203 IP_VS_TCP_S_FIN_WAIT,
204 IP_VS_TCP_S_TIME_WAIT,
205 IP_VS_TCP_S_CLOSE,
206 IP_VS_TCP_S_CLOSE_WAIT,
207 IP_VS_TCP_S_LAST_ACK,
208 IP_VS_TCP_S_LISTEN,
209 IP_VS_TCP_S_SYNACK,
210 IP_VS_TCP_S_LAST
211 };
212
213 /*
214 * UDP State Values
215 */
216 enum {
217 IP_VS_UDP_S_NORMAL,
218 IP_VS_UDP_S_LAST,
219 };
220
221 /*
222 * ICMP State Values
223 */
224 enum {
225 IP_VS_ICMP_S_NORMAL,
226 IP_VS_ICMP_S_LAST,
227 };
228
229 /*
230 * SCTP State Values
231 */
232 enum ip_vs_sctp_states {
233 IP_VS_SCTP_S_NONE,
234 IP_VS_SCTP_S_INIT_CLI,
235 IP_VS_SCTP_S_INIT_SER,
236 IP_VS_SCTP_S_INIT_ACK_CLI,
237 IP_VS_SCTP_S_INIT_ACK_SER,
238 IP_VS_SCTP_S_ECHO_CLI,
239 IP_VS_SCTP_S_ECHO_SER,
240 IP_VS_SCTP_S_ESTABLISHED,
241 IP_VS_SCTP_S_SHUT_CLI,
242 IP_VS_SCTP_S_SHUT_SER,
243 IP_VS_SCTP_S_SHUT_ACK_CLI,
244 IP_VS_SCTP_S_SHUT_ACK_SER,
245 IP_VS_SCTP_S_CLOSED,
246 IP_VS_SCTP_S_LAST
247 };
248
249 /*
250 * Delta sequence info structure
251 * Each ip_vs_conn has 2 (output AND input seq. changes).
252 * Only used in the VS/NAT.
253 */
254 struct ip_vs_seq {
255 __u32 init_seq; /* Add delta from this seq */
256 __u32 delta; /* Delta in sequence numbers */
257 __u32 previous_delta; /* Delta in sequence numbers
258 before last resized pkt */
259 };
260
261
262 /*
263 * IPVS statistics objects
264 */
265 struct ip_vs_estimator {
266 struct list_head list;
267
268 u64 last_inbytes;
269 u64 last_outbytes;
270 u32 last_conns;
271 u32 last_inpkts;
272 u32 last_outpkts;
273
274 u32 cps;
275 u32 inpps;
276 u32 outpps;
277 u32 inbps;
278 u32 outbps;
279 };
280
281 struct ip_vs_stats {
282 struct ip_vs_stats_user ustats; /* statistics */
283 struct ip_vs_estimator est; /* estimator */
284
285 spinlock_t lock; /* spin lock */
286 };
287
288 struct dst_entry;
289 struct iphdr;
290 struct ip_vs_conn;
291 struct ip_vs_app;
292 struct sk_buff;
293
294 struct ip_vs_protocol {
295 struct ip_vs_protocol *next;
296 char *name;
297 u16 protocol;
298 u16 num_states;
299 int dont_defrag;
300 atomic_t appcnt; /* counter of proto app incs */
301 int *timeout_table; /* protocol timeout table */
302
303 void (*init)(struct ip_vs_protocol *pp);
304
305 void (*exit)(struct ip_vs_protocol *pp);
306
307 int (*conn_schedule)(int af, struct sk_buff *skb,
308 struct ip_vs_protocol *pp,
309 int *verdict, struct ip_vs_conn **cpp);
310
311 struct ip_vs_conn *
312 (*conn_in_get)(int af,
313 const struct sk_buff *skb,
314 struct ip_vs_protocol *pp,
315 const struct ip_vs_iphdr *iph,
316 unsigned int proto_off,
317 int inverse);
318
319 struct ip_vs_conn *
320 (*conn_out_get)(int af,
321 const struct sk_buff *skb,
322 struct ip_vs_protocol *pp,
323 const struct ip_vs_iphdr *iph,
324 unsigned int proto_off,
325 int inverse);
326
327 int (*snat_handler)(struct sk_buff *skb,
328 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
329
330 int (*dnat_handler)(struct sk_buff *skb,
331 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
332
333 int (*csum_check)(int af, struct sk_buff *skb,
334 struct ip_vs_protocol *pp);
335
336 const char *(*state_name)(int state);
337
338 int (*state_transition)(struct ip_vs_conn *cp, int direction,
339 const struct sk_buff *skb,
340 struct ip_vs_protocol *pp);
341
342 int (*register_app)(struct ip_vs_app *inc);
343
344 void (*unregister_app)(struct ip_vs_app *inc);
345
346 int (*app_conn_bind)(struct ip_vs_conn *cp);
347
348 void (*debug_packet)(struct ip_vs_protocol *pp,
349 const struct sk_buff *skb,
350 int offset,
351 const char *msg);
352
353 void (*timeout_change)(struct ip_vs_protocol *pp, int flags);
354
355 int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to);
356 };
357
358 extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto);
359
360 struct ip_vs_conn_param {
361 const union nf_inet_addr *caddr;
362 const union nf_inet_addr *vaddr;
363 __be16 cport;
364 __be16 vport;
365 __u16 protocol;
366 u16 af;
367
368 const struct ip_vs_pe *pe;
369 char *pe_data;
370 __u8 pe_data_len;
371 };
372
373 /*
374 * IP_VS structure allocated for each dynamically scheduled connection
375 */
376 struct ip_vs_conn {
377 struct list_head c_list; /* hashed list heads */
378
379 /* Protocol, addresses and port numbers */
380 u16 af; /* address family */
381 union nf_inet_addr caddr; /* client address */
382 union nf_inet_addr vaddr; /* virtual address */
383 union nf_inet_addr daddr; /* destination address */
384 volatile __u32 flags; /* status flags */
385 __be16 cport;
386 __be16 vport;
387 __be16 dport;
388 __u16 protocol; /* Which protocol (TCP/UDP) */
389
390 /* counter and timer */
391 atomic_t refcnt; /* reference count */
392 struct timer_list timer; /* Expiration timer */
393 volatile unsigned long timeout; /* timeout */
394
395 /* Flags and state transition */
396 spinlock_t lock; /* lock for state transition */
397 volatile __u16 state; /* state info */
398 volatile __u16 old_state; /* old state, to be used for
399 * state transition triggerd
400 * synchronization
401 */
402
403 /* Control members */
404 struct ip_vs_conn *control; /* Master control connection */
405 atomic_t n_control; /* Number of controlled ones */
406 struct ip_vs_dest *dest; /* real server */
407 atomic_t in_pkts; /* incoming packet counter */
408
409 /* packet transmitter for different forwarding methods. If it
410 mangles the packet, it must return NF_DROP or better NF_STOLEN,
411 otherwise this must be changed to a sk_buff **.
412 */
413 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
414 struct ip_vs_protocol *pp);
415
416 /* Note: we can group the following members into a structure,
417 in order to save more space, and the following members are
418 only used in VS/NAT anyway */
419 struct ip_vs_app *app; /* bound ip_vs_app object */
420 void *app_data; /* Application private data */
421 struct ip_vs_seq in_seq; /* incoming seq. struct */
422 struct ip_vs_seq out_seq; /* outgoing seq. struct */
423
424 char *pe_data;
425 __u8 pe_data_len;
426 };
427
428
429 /*
430 * Extended internal versions of struct ip_vs_service_user and
431 * ip_vs_dest_user for IPv6 support.
432 *
433 * We need these to conveniently pass around service and destination
434 * options, but unfortunately, we also need to keep the old definitions to
435 * maintain userspace backwards compatibility for the setsockopt interface.
436 */
437 struct ip_vs_service_user_kern {
438 /* virtual service addresses */
439 u16 af;
440 u16 protocol;
441 union nf_inet_addr addr; /* virtual ip address */
442 u16 port;
443 u32 fwmark; /* firwall mark of service */
444
445 /* virtual service options */
446 char *sched_name;
447 char *pe_name;
448 unsigned flags; /* virtual service flags */
449 unsigned timeout; /* persistent timeout in sec */
450 u32 netmask; /* persistent netmask */
451 };
452
453
454 struct ip_vs_dest_user_kern {
455 /* destination server address */
456 union nf_inet_addr addr;
457 u16 port;
458
459 /* real server options */
460 unsigned conn_flags; /* connection flags */
461 int weight; /* destination weight */
462
463 /* thresholds for active connections */
464 u32 u_threshold; /* upper threshold */
465 u32 l_threshold; /* lower threshold */
466 };
467
468
469 /*
470 * The information about the virtual service offered to the net
471 * and the forwarding entries
472 */
473 struct ip_vs_service {
474 struct list_head s_list; /* for normal service table */
475 struct list_head f_list; /* for fwmark-based service table */
476 atomic_t refcnt; /* reference counter */
477 atomic_t usecnt; /* use counter */
478
479 u16 af; /* address family */
480 __u16 protocol; /* which protocol (TCP/UDP) */
481 union nf_inet_addr addr; /* IP address for virtual service */
482 __be16 port; /* port number for the service */
483 __u32 fwmark; /* firewall mark of the service */
484 unsigned flags; /* service status flags */
485 unsigned timeout; /* persistent timeout in ticks */
486 __be32 netmask; /* grouping granularity */
487
488 struct list_head destinations; /* real server d-linked list */
489 __u32 num_dests; /* number of servers */
490 struct ip_vs_stats stats; /* statistics for the service */
491 struct ip_vs_app *inc; /* bind conns to this app inc */
492
493 /* for scheduling */
494 struct ip_vs_scheduler *scheduler; /* bound scheduler object */
495 rwlock_t sched_lock; /* lock sched_data */
496 void *sched_data; /* scheduler application data */
497
498 /* alternate persistence engine */
499 struct ip_vs_pe *pe;
500 };
501
502
503 /*
504 * The real server destination forwarding entry
505 * with ip address, port number, and so on.
506 */
507 struct ip_vs_dest {
508 struct list_head n_list; /* for the dests in the service */
509 struct list_head d_list; /* for table with all the dests */
510
511 u16 af; /* address family */
512 union nf_inet_addr addr; /* IP address of the server */
513 __be16 port; /* port number of the server */
514 volatile unsigned flags; /* dest status flags */
515 atomic_t conn_flags; /* flags to copy to conn */
516 atomic_t weight; /* server weight */
517
518 atomic_t refcnt; /* reference counter */
519 struct ip_vs_stats stats; /* statistics */
520
521 /* connection counters and thresholds */
522 atomic_t activeconns; /* active connections */
523 atomic_t inactconns; /* inactive connections */
524 atomic_t persistconns; /* persistent connections */
525 __u32 u_threshold; /* upper threshold */
526 __u32 l_threshold; /* lower threshold */
527
528 /* for destination cache */
529 spinlock_t dst_lock; /* lock of dst_cache */
530 struct dst_entry *dst_cache; /* destination cache entry */
531 u32 dst_rtos; /* RT_TOS(tos) for dst */
532 u32 dst_cookie;
533 #ifdef CONFIG_IP_VS_IPV6
534 struct in6_addr dst_saddr;
535 #endif
536
537 /* for virtual service */
538 struct ip_vs_service *svc; /* service it belongs to */
539 __u16 protocol; /* which protocol (TCP/UDP) */
540 union nf_inet_addr vaddr; /* virtual IP address */
541 __be16 vport; /* virtual port number */
542 __u32 vfwmark; /* firewall mark of service */
543 };
544
545
546 /*
547 * The scheduler object
548 */
549 struct ip_vs_scheduler {
550 struct list_head n_list; /* d-linked list head */
551 char *name; /* scheduler name */
552 atomic_t refcnt; /* reference counter */
553 struct module *module; /* THIS_MODULE/NULL */
554
555 /* scheduler initializing service */
556 int (*init_service)(struct ip_vs_service *svc);
557 /* scheduling service finish */
558 int (*done_service)(struct ip_vs_service *svc);
559 /* scheduler updating service */
560 int (*update_service)(struct ip_vs_service *svc);
561
562 /* selecting a server from the given service */
563 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
564 const struct sk_buff *skb);
565 };
566
567 /* The persistence engine object */
568 struct ip_vs_pe {
569 struct list_head n_list; /* d-linked list head */
570 char *name; /* scheduler name */
571 atomic_t refcnt; /* reference counter */
572 struct module *module; /* THIS_MODULE/NULL */
573
574 /* get the connection template, if any */
575 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
576 bool (*ct_match)(const struct ip_vs_conn_param *p,
577 struct ip_vs_conn *ct);
578 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
579 bool inverse);
580 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
581 };
582
583 /*
584 * The application module object (a.k.a. app incarnation)
585 */
586 struct ip_vs_app {
587 struct list_head a_list; /* member in app list */
588 int type; /* IP_VS_APP_TYPE_xxx */
589 char *name; /* application module name */
590 __u16 protocol;
591 struct module *module; /* THIS_MODULE/NULL */
592 struct list_head incs_list; /* list of incarnations */
593
594 /* members for application incarnations */
595 struct list_head p_list; /* member in proto app list */
596 struct ip_vs_app *app; /* its real application */
597 __be16 port; /* port number in net order */
598 atomic_t usecnt; /* usage counter */
599
600 /* output hook: return false if can't linearize. diff set for TCP. */
601 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
602 struct sk_buff *, int *diff);
603
604 /* input hook: return false if can't linearize. diff set for TCP. */
605 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
606 struct sk_buff *, int *diff);
607
608 /* ip_vs_app initializer */
609 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
610
611 /* ip_vs_app finish */
612 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
613
614
615 /* not used now */
616 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
617 struct ip_vs_protocol *);
618
619 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
620
621 int * timeout_table;
622 int * timeouts;
623 int timeouts_size;
624
625 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
626 int *verdict, struct ip_vs_conn **cpp);
627
628 struct ip_vs_conn *
629 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
630 const struct iphdr *iph, unsigned int proto_off,
631 int inverse);
632
633 struct ip_vs_conn *
634 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
635 const struct iphdr *iph, unsigned int proto_off,
636 int inverse);
637
638 int (*state_transition)(struct ip_vs_conn *cp, int direction,
639 const struct sk_buff *skb,
640 struct ip_vs_app *app);
641
642 void (*timeout_change)(struct ip_vs_app *app, int flags);
643 };
644
645
646 /*
647 * IPVS core functions
648 * (from ip_vs_core.c)
649 */
650 extern const char *ip_vs_proto_name(unsigned proto);
651 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
652 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
653
654 #define IP_VS_APP_TYPE_FTP 1
655
656 /*
657 * ip_vs_conn handling functions
658 * (from ip_vs_conn.c)
659 */
660
661 enum {
662 IP_VS_DIR_INPUT = 0,
663 IP_VS_DIR_OUTPUT,
664 IP_VS_DIR_INPUT_ONLY,
665 IP_VS_DIR_LAST,
666 };
667
668 static inline void ip_vs_conn_fill_param(int af, int protocol,
669 const union nf_inet_addr *caddr,
670 __be16 cport,
671 const union nf_inet_addr *vaddr,
672 __be16 vport,
673 struct ip_vs_conn_param *p)
674 {
675 p->af = af;
676 p->protocol = protocol;
677 p->caddr = caddr;
678 p->cport = cport;
679 p->vaddr = vaddr;
680 p->vport = vport;
681 p->pe = NULL;
682 p->pe_data = NULL;
683 }
684
685 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
686 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
687
688 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
689 struct ip_vs_protocol *pp,
690 const struct ip_vs_iphdr *iph,
691 unsigned int proto_off,
692 int inverse);
693
694 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
695
696 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
697 struct ip_vs_protocol *pp,
698 const struct ip_vs_iphdr *iph,
699 unsigned int proto_off,
700 int inverse);
701
702 /* put back the conn without restarting its timer */
703 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
704 {
705 atomic_dec(&cp->refcnt);
706 }
707 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
708 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
709
710 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
711 const union nf_inet_addr *daddr,
712 __be16 dport, unsigned flags,
713 struct ip_vs_dest *dest);
714 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
715
716 extern const char * ip_vs_state_name(__u16 proto, int state);
717
718 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
719 extern int ip_vs_check_template(struct ip_vs_conn *ct);
720 extern void ip_vs_random_dropentry(void);
721 extern int ip_vs_conn_init(void);
722 extern void ip_vs_conn_cleanup(void);
723
724 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
725 {
726 struct ip_vs_conn *ctl_cp = cp->control;
727 if (!ctl_cp) {
728 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
729 "%s:%d to %s:%d\n",
730 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
731 ntohs(cp->cport),
732 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
733 ntohs(cp->vport));
734
735 return;
736 }
737
738 IP_VS_DBG_BUF(7, "DELeting control for: "
739 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
740 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
741 ntohs(cp->cport),
742 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
743 ntohs(ctl_cp->cport));
744
745 cp->control = NULL;
746 if (atomic_read(&ctl_cp->n_control) == 0) {
747 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
748 "%s:%d to %s:%d\n",
749 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
750 ntohs(cp->cport),
751 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
752 ntohs(cp->vport));
753
754 return;
755 }
756 atomic_dec(&ctl_cp->n_control);
757 }
758
759 static inline void
760 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
761 {
762 if (cp->control) {
763 IP_VS_ERR_BUF("request control ADD for already controlled: "
764 "%s:%d to %s:%d\n",
765 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
766 ntohs(cp->cport),
767 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
768 ntohs(cp->vport));
769
770 ip_vs_control_del(cp);
771 }
772
773 IP_VS_DBG_BUF(7, "ADDing control for: "
774 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
775 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
776 ntohs(cp->cport),
777 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
778 ntohs(ctl_cp->cport));
779
780 cp->control = ctl_cp;
781 atomic_inc(&ctl_cp->n_control);
782 }
783
784
785 /*
786 * IPVS application functions
787 * (from ip_vs_app.c)
788 */
789 #define IP_VS_APP_MAX_PORTS 8
790 extern int register_ip_vs_app(struct ip_vs_app *app);
791 extern void unregister_ip_vs_app(struct ip_vs_app *app);
792 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
793 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
794 extern int
795 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
796 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
797 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
798
799 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
800 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
801 extern int ip_vs_app_init(void);
802 extern void ip_vs_app_cleanup(void);
803
804 void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
805 void ip_vs_unbind_pe(struct ip_vs_service *svc);
806 int register_ip_vs_pe(struct ip_vs_pe *pe);
807 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
808 extern struct ip_vs_pe *ip_vs_pe_get(const char *name);
809 extern void ip_vs_pe_put(struct ip_vs_pe *pe);
810
811 /*
812 * IPVS protocol functions (from ip_vs_proto.c)
813 */
814 extern int ip_vs_protocol_init(void);
815 extern void ip_vs_protocol_cleanup(void);
816 extern void ip_vs_protocol_timeout_change(int flags);
817 extern int *ip_vs_create_timeout_table(int *table, int size);
818 extern int
819 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
820 const char *name, int to);
821 extern void
822 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb,
823 int offset, const char *msg);
824
825 extern struct ip_vs_protocol ip_vs_protocol_tcp;
826 extern struct ip_vs_protocol ip_vs_protocol_udp;
827 extern struct ip_vs_protocol ip_vs_protocol_icmp;
828 extern struct ip_vs_protocol ip_vs_protocol_esp;
829 extern struct ip_vs_protocol ip_vs_protocol_ah;
830 extern struct ip_vs_protocol ip_vs_protocol_sctp;
831
832 /*
833 * Registering/unregistering scheduler functions
834 * (from ip_vs_sched.c)
835 */
836 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
837 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
838 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
839 struct ip_vs_scheduler *scheduler);
840 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
841 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
842 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
843 extern struct ip_vs_conn *
844 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb);
845 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
846 struct ip_vs_protocol *pp);
847
848
849 /*
850 * IPVS control data and functions (from ip_vs_ctl.c)
851 */
852 extern int sysctl_ip_vs_cache_bypass;
853 extern int sysctl_ip_vs_expire_nodest_conn;
854 extern int sysctl_ip_vs_expire_quiescent_template;
855 extern int sysctl_ip_vs_sync_threshold[2];
856 extern int sysctl_ip_vs_nat_icmp_send;
857 extern int sysctl_ip_vs_conntrack;
858 extern int sysctl_ip_vs_snat_reroute;
859 extern struct ip_vs_stats ip_vs_stats;
860 extern const struct ctl_path net_vs_ctl_path[];
861
862 extern struct ip_vs_service *
863 ip_vs_service_get(int af, __u32 fwmark, __u16 protocol,
864 const union nf_inet_addr *vaddr, __be16 vport);
865
866 static inline void ip_vs_service_put(struct ip_vs_service *svc)
867 {
868 atomic_dec(&svc->usecnt);
869 }
870
871 extern struct ip_vs_dest *
872 ip_vs_lookup_real_service(int af, __u16 protocol,
873 const union nf_inet_addr *daddr, __be16 dport);
874
875 extern int ip_vs_use_count_inc(void);
876 extern void ip_vs_use_count_dec(void);
877 extern int ip_vs_control_init(void);
878 extern void ip_vs_control_cleanup(void);
879 extern struct ip_vs_dest *
880 ip_vs_find_dest(int af, const union nf_inet_addr *daddr, __be16 dport,
881 const union nf_inet_addr *vaddr, __be16 vport, __u16 protocol);
882 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
883
884
885 /*
886 * IPVS sync daemon data and function prototypes
887 * (from ip_vs_sync.c)
888 */
889 extern volatile int ip_vs_sync_state;
890 extern volatile int ip_vs_master_syncid;
891 extern volatile int ip_vs_backup_syncid;
892 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
893 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
894 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
895 extern int stop_sync_thread(int state);
896 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
897
898
899 /*
900 * IPVS rate estimator prototypes (from ip_vs_est.c)
901 */
902 extern int ip_vs_estimator_init(void);
903 extern void ip_vs_estimator_cleanup(void);
904 extern void ip_vs_new_estimator(struct ip_vs_stats *stats);
905 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
906 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
907
908 /*
909 * Various IPVS packet transmitters (from ip_vs_xmit.c)
910 */
911 extern int ip_vs_null_xmit
912 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
913 extern int ip_vs_bypass_xmit
914 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
915 extern int ip_vs_nat_xmit
916 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
917 extern int ip_vs_tunnel_xmit
918 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
919 extern int ip_vs_dr_xmit
920 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
921 extern int ip_vs_icmp_xmit
922 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
923 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
924
925 #ifdef CONFIG_IP_VS_IPV6
926 extern int ip_vs_bypass_xmit_v6
927 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
928 extern int ip_vs_nat_xmit_v6
929 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
930 extern int ip_vs_tunnel_xmit_v6
931 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
932 extern int ip_vs_dr_xmit_v6
933 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
934 extern int ip_vs_icmp_xmit_v6
935 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
936 int offset);
937 #endif
938
939 /*
940 * This is a simple mechanism to ignore packets when
941 * we are loaded. Just set ip_vs_drop_rate to 'n' and
942 * we start to drop 1/rate of the packets
943 */
944 extern int ip_vs_drop_rate;
945 extern int ip_vs_drop_counter;
946
947 static __inline__ int ip_vs_todrop(void)
948 {
949 if (!ip_vs_drop_rate) return 0;
950 if (--ip_vs_drop_counter > 0) return 0;
951 ip_vs_drop_counter = ip_vs_drop_rate;
952 return 1;
953 }
954
955 /*
956 * ip_vs_fwd_tag returns the forwarding tag of the connection
957 */
958 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
959
960 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
961 {
962 char fwd;
963
964 switch (IP_VS_FWD_METHOD(cp)) {
965 case IP_VS_CONN_F_MASQ:
966 fwd = 'M'; break;
967 case IP_VS_CONN_F_LOCALNODE:
968 fwd = 'L'; break;
969 case IP_VS_CONN_F_TUNNEL:
970 fwd = 'T'; break;
971 case IP_VS_CONN_F_DROUTE:
972 fwd = 'R'; break;
973 case IP_VS_CONN_F_BYPASS:
974 fwd = 'B'; break;
975 default:
976 fwd = '?'; break;
977 }
978 return fwd;
979 }
980
981 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
982 struct ip_vs_conn *cp, int dir);
983
984 #ifdef CONFIG_IP_VS_IPV6
985 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
986 struct ip_vs_conn *cp, int dir);
987 #endif
988
989 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
990
991 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
992 {
993 __be32 diff[2] = { ~old, new };
994
995 return csum_partial(diff, sizeof(diff), oldsum);
996 }
997
998 #ifdef CONFIG_IP_VS_IPV6
999 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1000 __wsum oldsum)
1001 {
1002 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1003 new[3], new[2], new[1], new[0] };
1004
1005 return csum_partial(diff, sizeof(diff), oldsum);
1006 }
1007 #endif
1008
1009 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1010 {
1011 __be16 diff[2] = { ~old, new };
1012
1013 return csum_partial(diff, sizeof(diff), oldsum);
1014 }
1015
1016 #ifdef CONFIG_IP_VS_NFCT
1017 /*
1018 * Netfilter connection tracking
1019 * (from ip_vs_nfct.c)
1020 */
1021 static inline int ip_vs_conntrack_enabled(void)
1022 {
1023 return sysctl_ip_vs_conntrack;
1024 }
1025
1026 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1027 int outin);
1028 extern int ip_vs_confirm_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp);
1029 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1030 struct ip_vs_conn *cp, u_int8_t proto,
1031 const __be16 port, int from_rs);
1032 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1033
1034 #else
1035
1036 static inline int ip_vs_conntrack_enabled(void)
1037 {
1038 return 0;
1039 }
1040
1041 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1042 struct ip_vs_conn *cp, int outin)
1043 {
1044 }
1045
1046 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb,
1047 struct ip_vs_conn *cp)
1048 {
1049 return NF_ACCEPT;
1050 }
1051
1052 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1053 {
1054 }
1055 /* CONFIG_IP_VS_NFCT */
1056 #endif
1057
1058 #endif /* __KERNEL__ */
1059
1060 #endif /* _NET_IP_VS_H */
This page took 0.051277 seconds and 5 git commands to generate.