IPVS: Backup, Change sending to Version 1 format
[linux-flexiantxendom0-3.2.10.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 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
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, af, pp, skb, ofs, msg)                     \
140         do {                                                            \
141                 if (level <= ip_vs_get_debug_level())                   \
142                         pp->debug_packet(af, pp, skb, ofs, msg);        \
143         } while (0)
144 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg)                  \
145         do {                                                            \
146                 if (level <= ip_vs_get_debug_level() &&                 \
147                     net_ratelimit())                                    \
148                         pp->debug_packet(af, 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, af, pp, skb, ofs, msg)     do {} while (0)
156 #define IP_VS_DBG_RL_PKT(level, af, 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)(int af, 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         __u32                    fwmark;         /* Fire wall mark from skb */
386         __be16                   cport;
387         __be16                   vport;
388         __be16                   dport;
389         __u16                   protocol;       /* Which protocol (TCP/UDP) */
390
391         /* counter and timer */
392         atomic_t                refcnt;         /* reference count */
393         struct timer_list       timer;          /* Expiration timer */
394         volatile unsigned long  timeout;        /* timeout */
395
396         /* Flags and state transition */
397         spinlock_t              lock;           /* lock for state transition */
398         volatile __u16          state;          /* state info */
399         volatile __u16          old_state;      /* old state, to be used for
400                                                  * state transition triggerd
401                                                  * synchronization
402                                                  */
403
404         /* Control members */
405         struct ip_vs_conn       *control;       /* Master control connection */
406         atomic_t                n_control;      /* Number of controlled ones */
407         struct ip_vs_dest       *dest;          /* real server */
408         atomic_t                in_pkts;        /* incoming packet counter */
409
410         /* packet transmitter for different forwarding methods.  If it
411            mangles the packet, it must return NF_DROP or better NF_STOLEN,
412            otherwise this must be changed to a sk_buff **.
413            NF_ACCEPT can be returned when destination is local.
414          */
415         int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
416                            struct ip_vs_protocol *pp);
417
418         /* Note: we can group the following members into a structure,
419            in order to save more space, and the following members are
420            only used in VS/NAT anyway */
421         struct ip_vs_app        *app;           /* bound ip_vs_app object */
422         void                    *app_data;      /* Application private data */
423         struct ip_vs_seq        in_seq;         /* incoming seq. struct */
424         struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
425
426         const struct ip_vs_pe   *pe;
427         char                    *pe_data;
428         __u8                    pe_data_len;
429 };
430
431
432 /*
433  *      Extended internal versions of struct ip_vs_service_user and
434  *      ip_vs_dest_user for IPv6 support.
435  *
436  *      We need these to conveniently pass around service and destination
437  *      options, but unfortunately, we also need to keep the old definitions to
438  *      maintain userspace backwards compatibility for the setsockopt interface.
439  */
440 struct ip_vs_service_user_kern {
441         /* virtual service addresses */
442         u16                     af;
443         u16                     protocol;
444         union nf_inet_addr      addr;           /* virtual ip address */
445         u16                     port;
446         u32                     fwmark;         /* firwall mark of service */
447
448         /* virtual service options */
449         char                    *sched_name;
450         char                    *pe_name;
451         unsigned                flags;          /* virtual service flags */
452         unsigned                timeout;        /* persistent timeout in sec */
453         u32                     netmask;        /* persistent netmask */
454 };
455
456
457 struct ip_vs_dest_user_kern {
458         /* destination server address */
459         union nf_inet_addr      addr;
460         u16                     port;
461
462         /* real server options */
463         unsigned                conn_flags;     /* connection flags */
464         int                     weight;         /* destination weight */
465
466         /* thresholds for active connections */
467         u32                     u_threshold;    /* upper threshold */
468         u32                     l_threshold;    /* lower threshold */
469 };
470
471
472 /*
473  *      The information about the virtual service offered to the net
474  *      and the forwarding entries
475  */
476 struct ip_vs_service {
477         struct list_head        s_list;   /* for normal service table */
478         struct list_head        f_list;   /* for fwmark-based service table */
479         atomic_t                refcnt;   /* reference counter */
480         atomic_t                usecnt;   /* use counter */
481
482         u16                     af;       /* address family */
483         __u16                   protocol; /* which protocol (TCP/UDP) */
484         union nf_inet_addr      addr;     /* IP address for virtual service */
485         __be16                  port;     /* port number for the service */
486         __u32                   fwmark;   /* firewall mark of the service */
487         unsigned                flags;    /* service status flags */
488         unsigned                timeout;  /* persistent timeout in ticks */
489         __be32                  netmask;  /* grouping granularity */
490
491         struct list_head        destinations;  /* real server d-linked list */
492         __u32                   num_dests;     /* number of servers */
493         struct ip_vs_stats      stats;         /* statistics for the service */
494         struct ip_vs_app        *inc;     /* bind conns to this app inc */
495
496         /* for scheduling */
497         struct ip_vs_scheduler  *scheduler;    /* bound scheduler object */
498         rwlock_t                sched_lock;    /* lock sched_data */
499         void                    *sched_data;   /* scheduler application data */
500
501         /* alternate persistence engine */
502         struct ip_vs_pe         *pe;
503 };
504
505
506 /*
507  *      The real server destination forwarding entry
508  *      with ip address, port number, and so on.
509  */
510 struct ip_vs_dest {
511         struct list_head        n_list;   /* for the dests in the service */
512         struct list_head        d_list;   /* for table with all the dests */
513
514         u16                     af;             /* address family */
515         union nf_inet_addr      addr;           /* IP address of the server */
516         __be16                  port;           /* port number of the server */
517         volatile unsigned       flags;          /* dest status flags */
518         atomic_t                conn_flags;     /* flags to copy to conn */
519         atomic_t                weight;         /* server weight */
520
521         atomic_t                refcnt;         /* reference counter */
522         struct ip_vs_stats      stats;          /* statistics */
523
524         /* connection counters and thresholds */
525         atomic_t                activeconns;    /* active connections */
526         atomic_t                inactconns;     /* inactive connections */
527         atomic_t                persistconns;   /* persistent connections */
528         __u32                   u_threshold;    /* upper threshold */
529         __u32                   l_threshold;    /* lower threshold */
530
531         /* for destination cache */
532         spinlock_t              dst_lock;       /* lock of dst_cache */
533         struct dst_entry        *dst_cache;     /* destination cache entry */
534         u32                     dst_rtos;       /* RT_TOS(tos) for dst */
535         u32                     dst_cookie;
536 #ifdef CONFIG_IP_VS_IPV6
537         struct in6_addr         dst_saddr;
538 #endif
539
540         /* for virtual service */
541         struct ip_vs_service    *svc;           /* service it belongs to */
542         __u16                   protocol;       /* which protocol (TCP/UDP) */
543         union nf_inet_addr      vaddr;          /* virtual IP address */
544         __be16                  vport;          /* virtual port number */
545         __u32                   vfwmark;        /* firewall mark of service */
546 };
547
548
549 /*
550  *      The scheduler object
551  */
552 struct ip_vs_scheduler {
553         struct list_head        n_list;         /* d-linked list head */
554         char                    *name;          /* scheduler name */
555         atomic_t                refcnt;         /* reference counter */
556         struct module           *module;        /* THIS_MODULE/NULL */
557
558         /* scheduler initializing service */
559         int (*init_service)(struct ip_vs_service *svc);
560         /* scheduling service finish */
561         int (*done_service)(struct ip_vs_service *svc);
562         /* scheduler updating service */
563         int (*update_service)(struct ip_vs_service *svc);
564
565         /* selecting a server from the given service */
566         struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
567                                        const struct sk_buff *skb);
568 };
569
570 /* The persistence engine object */
571 struct ip_vs_pe {
572         struct list_head        n_list;         /* d-linked list head */
573         char                    *name;          /* scheduler name */
574         atomic_t                refcnt;         /* reference counter */
575         struct module           *module;        /* THIS_MODULE/NULL */
576
577         /* get the connection template, if any */
578         int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
579         bool (*ct_match)(const struct ip_vs_conn_param *p,
580                          struct ip_vs_conn *ct);
581         u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
582                            bool inverse);
583         int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
584 };
585
586 /*
587  *      The application module object (a.k.a. app incarnation)
588  */
589 struct ip_vs_app {
590         struct list_head        a_list;         /* member in app list */
591         int                     type;           /* IP_VS_APP_TYPE_xxx */
592         char                    *name;          /* application module name */
593         __u16                   protocol;
594         struct module           *module;        /* THIS_MODULE/NULL */
595         struct list_head        incs_list;      /* list of incarnations */
596
597         /* members for application incarnations */
598         struct list_head        p_list;         /* member in proto app list */
599         struct ip_vs_app        *app;           /* its real application */
600         __be16                  port;           /* port number in net order */
601         atomic_t                usecnt;         /* usage counter */
602
603         /*
604          * output hook: Process packet in inout direction, diff set for TCP.
605          * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
606          *         2=Mangled but checksum was not updated
607          */
608         int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
609                        struct sk_buff *, int *diff);
610
611         /*
612          * input hook: Process packet in outin direction, diff set for TCP.
613          * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
614          *         2=Mangled but checksum was not updated
615          */
616         int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
617                       struct sk_buff *, int *diff);
618
619         /* ip_vs_app initializer */
620         int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
621
622         /* ip_vs_app finish */
623         int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
624
625
626         /* not used now */
627         int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
628                          struct ip_vs_protocol *);
629
630         void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
631
632         int *                   timeout_table;
633         int *                   timeouts;
634         int                     timeouts_size;
635
636         int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
637                              int *verdict, struct ip_vs_conn **cpp);
638
639         struct ip_vs_conn *
640         (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
641                        const struct iphdr *iph, unsigned int proto_off,
642                        int inverse);
643
644         struct ip_vs_conn *
645         (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
646                         const struct iphdr *iph, unsigned int proto_off,
647                         int inverse);
648
649         int (*state_transition)(struct ip_vs_conn *cp, int direction,
650                                 const struct sk_buff *skb,
651                                 struct ip_vs_app *app);
652
653         void (*timeout_change)(struct ip_vs_app *app, int flags);
654 };
655
656
657 /*
658  *      IPVS core functions
659  *      (from ip_vs_core.c)
660  */
661 extern const char *ip_vs_proto_name(unsigned proto);
662 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
663 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
664
665 #define IP_VS_APP_TYPE_FTP      1
666
667 /*
668  *     ip_vs_conn handling functions
669  *     (from ip_vs_conn.c)
670  */
671
672 enum {
673         IP_VS_DIR_INPUT = 0,
674         IP_VS_DIR_OUTPUT,
675         IP_VS_DIR_INPUT_ONLY,
676         IP_VS_DIR_LAST,
677 };
678
679 static inline void ip_vs_conn_fill_param(int af, int protocol,
680                                          const union nf_inet_addr *caddr,
681                                          __be16 cport,
682                                          const union nf_inet_addr *vaddr,
683                                          __be16 vport,
684                                          struct ip_vs_conn_param *p)
685 {
686         p->af = af;
687         p->protocol = protocol;
688         p->caddr = caddr;
689         p->cport = cport;
690         p->vaddr = vaddr;
691         p->vport = vport;
692         p->pe = NULL;
693         p->pe_data = NULL;
694 }
695
696 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
697 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
698
699 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
700                                             struct ip_vs_protocol *pp,
701                                             const struct ip_vs_iphdr *iph,
702                                             unsigned int proto_off,
703                                             int inverse);
704
705 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
706
707 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
708                                              struct ip_vs_protocol *pp,
709                                              const struct ip_vs_iphdr *iph,
710                                              unsigned int proto_off,
711                                              int inverse);
712
713 /* put back the conn without restarting its timer */
714 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
715 {
716         atomic_dec(&cp->refcnt);
717 }
718 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
719 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
720
721 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
722                                   const union nf_inet_addr *daddr,
723                                   __be16 dport, unsigned flags,
724                                   struct ip_vs_dest *dest, __u32 fwmark);
725 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
726
727 extern const char * ip_vs_state_name(__u16 proto, int state);
728
729 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
730 extern int ip_vs_check_template(struct ip_vs_conn *ct);
731 extern void ip_vs_random_dropentry(void);
732 extern int ip_vs_conn_init(void);
733 extern void ip_vs_conn_cleanup(void);
734
735 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
736 {
737         struct ip_vs_conn *ctl_cp = cp->control;
738         if (!ctl_cp) {
739                 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
740                               "%s:%d to %s:%d\n",
741                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
742                               ntohs(cp->cport),
743                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
744                               ntohs(cp->vport));
745
746                 return;
747         }
748
749         IP_VS_DBG_BUF(7, "DELeting control for: "
750                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
751                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
752                       ntohs(cp->cport),
753                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
754                       ntohs(ctl_cp->cport));
755
756         cp->control = NULL;
757         if (atomic_read(&ctl_cp->n_control) == 0) {
758                 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
759                               "%s:%d to %s:%d\n",
760                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
761                               ntohs(cp->cport),
762                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
763                               ntohs(cp->vport));
764
765                 return;
766         }
767         atomic_dec(&ctl_cp->n_control);
768 }
769
770 static inline void
771 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
772 {
773         if (cp->control) {
774                 IP_VS_ERR_BUF("request control ADD for already controlled: "
775                               "%s:%d to %s:%d\n",
776                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
777                               ntohs(cp->cport),
778                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
779                               ntohs(cp->vport));
780
781                 ip_vs_control_del(cp);
782         }
783
784         IP_VS_DBG_BUF(7, "ADDing control for: "
785                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
786                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
787                       ntohs(cp->cport),
788                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
789                       ntohs(ctl_cp->cport));
790
791         cp->control = ctl_cp;
792         atomic_inc(&ctl_cp->n_control);
793 }
794
795
796 /*
797  *      IPVS application functions
798  *      (from ip_vs_app.c)
799  */
800 #define IP_VS_APP_MAX_PORTS  8
801 extern int register_ip_vs_app(struct ip_vs_app *app);
802 extern void unregister_ip_vs_app(struct ip_vs_app *app);
803 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
804 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
805 extern int
806 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
807 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
808 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
809
810 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
811 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
812 extern int ip_vs_app_init(void);
813 extern void ip_vs_app_cleanup(void);
814
815 void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
816 void ip_vs_unbind_pe(struct ip_vs_service *svc);
817 int register_ip_vs_pe(struct ip_vs_pe *pe);
818 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
819 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
820 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
821
822 static inline void ip_vs_pe_get(const struct ip_vs_pe *pe)
823 {
824         if (pe && pe->module)
825                 __module_get(pe->module);
826 }
827
828 static inline void ip_vs_pe_put(const struct ip_vs_pe *pe)
829 {
830         if (pe && pe->module)
831                 module_put(pe->module);
832 }
833
834 /*
835  *      IPVS protocol functions (from ip_vs_proto.c)
836  */
837 extern int ip_vs_protocol_init(void);
838 extern void ip_vs_protocol_cleanup(void);
839 extern void ip_vs_protocol_timeout_change(int flags);
840 extern int *ip_vs_create_timeout_table(int *table, int size);
841 extern int
842 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
843                         const char *name, int to);
844 extern void
845 ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
846                           const struct sk_buff *skb,
847                           int offset, const char *msg);
848
849 extern struct ip_vs_protocol ip_vs_protocol_tcp;
850 extern struct ip_vs_protocol ip_vs_protocol_udp;
851 extern struct ip_vs_protocol ip_vs_protocol_icmp;
852 extern struct ip_vs_protocol ip_vs_protocol_esp;
853 extern struct ip_vs_protocol ip_vs_protocol_ah;
854 extern struct ip_vs_protocol ip_vs_protocol_sctp;
855
856 /*
857  *      Registering/unregistering scheduler functions
858  *      (from ip_vs_sched.c)
859  */
860 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
861 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
862 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
863                                 struct ip_vs_scheduler *scheduler);
864 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
865 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
866 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
867 extern struct ip_vs_conn *
868 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
869                struct ip_vs_protocol *pp, int *ignored);
870 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
871                         struct ip_vs_protocol *pp);
872
873
874 /*
875  *      IPVS control data and functions (from ip_vs_ctl.c)
876  */
877 extern int sysctl_ip_vs_cache_bypass;
878 extern int sysctl_ip_vs_expire_nodest_conn;
879 extern int sysctl_ip_vs_expire_quiescent_template;
880 extern int sysctl_ip_vs_sync_threshold[2];
881 extern int sysctl_ip_vs_nat_icmp_send;
882 extern int sysctl_ip_vs_conntrack;
883 extern int sysctl_ip_vs_snat_reroute;
884 extern struct ip_vs_stats ip_vs_stats;
885 extern const struct ctl_path net_vs_ctl_path[];
886
887 extern struct ip_vs_service *
888 ip_vs_service_get(int af, __u32 fwmark, __u16 protocol,
889                   const union nf_inet_addr *vaddr, __be16 vport);
890
891 static inline void ip_vs_service_put(struct ip_vs_service *svc)
892 {
893         atomic_dec(&svc->usecnt);
894 }
895
896 extern struct ip_vs_dest *
897 ip_vs_lookup_real_service(int af, __u16 protocol,
898                           const union nf_inet_addr *daddr, __be16 dport);
899
900 extern int ip_vs_use_count_inc(void);
901 extern void ip_vs_use_count_dec(void);
902 extern int ip_vs_control_init(void);
903 extern void ip_vs_control_cleanup(void);
904 extern struct ip_vs_dest *
905 ip_vs_find_dest(int af, const union nf_inet_addr *daddr, __be16 dport,
906                 const union nf_inet_addr *vaddr, __be16 vport, __u16 protocol,
907                 __u32 fwmark);
908 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
909
910
911 /*
912  *      IPVS sync daemon data and function prototypes
913  *      (from ip_vs_sync.c)
914  */
915 extern volatile int ip_vs_sync_state;
916 extern volatile int ip_vs_master_syncid;
917 extern volatile int ip_vs_backup_syncid;
918 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
919 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
920 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
921 extern int stop_sync_thread(int state);
922 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
923
924
925 /*
926  *      IPVS rate estimator prototypes (from ip_vs_est.c)
927  */
928 extern int ip_vs_estimator_init(void);
929 extern void ip_vs_estimator_cleanup(void);
930 extern void ip_vs_new_estimator(struct ip_vs_stats *stats);
931 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
932 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
933
934 /*
935  *      Various IPVS packet transmitters (from ip_vs_xmit.c)
936  */
937 extern int ip_vs_null_xmit
938 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
939 extern int ip_vs_bypass_xmit
940 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
941 extern int ip_vs_nat_xmit
942 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
943 extern int ip_vs_tunnel_xmit
944 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
945 extern int ip_vs_dr_xmit
946 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
947 extern int ip_vs_icmp_xmit
948 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
949 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
950
951 #ifdef CONFIG_IP_VS_IPV6
952 extern int ip_vs_bypass_xmit_v6
953 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
954 extern int ip_vs_nat_xmit_v6
955 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
956 extern int ip_vs_tunnel_xmit_v6
957 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
958 extern int ip_vs_dr_xmit_v6
959 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
960 extern int ip_vs_icmp_xmit_v6
961 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
962  int offset);
963 #endif
964
965 /*
966  *      This is a simple mechanism to ignore packets when
967  *      we are loaded. Just set ip_vs_drop_rate to 'n' and
968  *      we start to drop 1/rate of the packets
969  */
970 extern int ip_vs_drop_rate;
971 extern int ip_vs_drop_counter;
972
973 static __inline__ int ip_vs_todrop(void)
974 {
975         if (!ip_vs_drop_rate) return 0;
976         if (--ip_vs_drop_counter > 0) return 0;
977         ip_vs_drop_counter = ip_vs_drop_rate;
978         return 1;
979 }
980
981 /*
982  *      ip_vs_fwd_tag returns the forwarding tag of the connection
983  */
984 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
985
986 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
987 {
988         char fwd;
989
990         switch (IP_VS_FWD_METHOD(cp)) {
991         case IP_VS_CONN_F_MASQ:
992                 fwd = 'M'; break;
993         case IP_VS_CONN_F_LOCALNODE:
994                 fwd = 'L'; break;
995         case IP_VS_CONN_F_TUNNEL:
996                 fwd = 'T'; break;
997         case IP_VS_CONN_F_DROUTE:
998                 fwd = 'R'; break;
999         case IP_VS_CONN_F_BYPASS:
1000                 fwd = 'B'; break;
1001         default:
1002                 fwd = '?'; break;
1003         }
1004         return fwd;
1005 }
1006
1007 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1008                            struct ip_vs_conn *cp, int dir);
1009
1010 #ifdef CONFIG_IP_VS_IPV6
1011 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1012                               struct ip_vs_conn *cp, int dir);
1013 #endif
1014
1015 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1016
1017 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1018 {
1019         __be32 diff[2] = { ~old, new };
1020
1021         return csum_partial(diff, sizeof(diff), oldsum);
1022 }
1023
1024 #ifdef CONFIG_IP_VS_IPV6
1025 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1026                                         __wsum oldsum)
1027 {
1028         __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1029                             new[3],  new[2],  new[1],  new[0] };
1030
1031         return csum_partial(diff, sizeof(diff), oldsum);
1032 }
1033 #endif
1034
1035 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1036 {
1037         __be16 diff[2] = { ~old, new };
1038
1039         return csum_partial(diff, sizeof(diff), oldsum);
1040 }
1041
1042 /*
1043  * Forget current conntrack (unconfirmed) and attach notrack entry
1044  */
1045 static inline void ip_vs_notrack(struct sk_buff *skb)
1046 {
1047 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1048         enum ip_conntrack_info ctinfo;
1049         struct nf_conn *ct = ct = nf_ct_get(skb, &ctinfo);
1050
1051         if (!ct || !nf_ct_is_untracked(ct)) {
1052                 nf_reset(skb);
1053                 skb->nfct = &nf_ct_untracked_get()->ct_general;
1054                 skb->nfctinfo = IP_CT_NEW;
1055                 nf_conntrack_get(skb->nfct);
1056         }
1057 #endif
1058 }
1059
1060 #ifdef CONFIG_IP_VS_NFCT
1061 /*
1062  *      Netfilter connection tracking
1063  *      (from ip_vs_nfct.c)
1064  */
1065 static inline int ip_vs_conntrack_enabled(void)
1066 {
1067         return sysctl_ip_vs_conntrack;
1068 }
1069
1070 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1071                                    int outin);
1072 extern int ip_vs_confirm_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp);
1073 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1074                                       struct ip_vs_conn *cp, u_int8_t proto,
1075                                       const __be16 port, int from_rs);
1076 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1077
1078 #else
1079
1080 static inline int ip_vs_conntrack_enabled(void)
1081 {
1082         return 0;
1083 }
1084
1085 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1086                                           struct ip_vs_conn *cp, int outin)
1087 {
1088 }
1089
1090 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb,
1091                                           struct ip_vs_conn *cp)
1092 {
1093         return NF_ACCEPT;
1094 }
1095
1096 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1097 {
1098 }
1099 /* CONFIG_IP_VS_NFCT */
1100 #endif
1101
1102 #endif /* __KERNEL__ */
1103
1104 #endif  /* _NET_IP_VS_H */