a816c37417bbbe3adc70a364cf0e84f99be3a164
[linux-2.6.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
29
30 /* Connections' size value needed by ip_vs_ctl.c */
31 extern int ip_vs_conn_tab_size;
32
33
34 struct ip_vs_iphdr {
35         int len;
36         __u8 protocol;
37         union nf_inet_addr saddr;
38         union nf_inet_addr daddr;
39 };
40
41 static inline void
42 ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
43 {
44 #ifdef CONFIG_IP_VS_IPV6
45         if (af == AF_INET6) {
46                 const struct ipv6hdr *iph = nh;
47                 iphdr->len = sizeof(struct ipv6hdr);
48                 iphdr->protocol = iph->nexthdr;
49                 ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr);
50                 ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr);
51         } else
52 #endif
53         {
54                 const struct iphdr *iph = nh;
55                 iphdr->len = iph->ihl * 4;
56                 iphdr->protocol = iph->protocol;
57                 iphdr->saddr.ip = iph->saddr;
58                 iphdr->daddr.ip = iph->daddr;
59         }
60 }
61
62 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
63                                    const union nf_inet_addr *src)
64 {
65 #ifdef CONFIG_IP_VS_IPV6
66         if (af == AF_INET6)
67                 ipv6_addr_copy(&dst->in6, &src->in6);
68         else
69 #endif
70         dst->ip = src->ip;
71 }
72
73 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
74                                    const union nf_inet_addr *b)
75 {
76 #ifdef CONFIG_IP_VS_IPV6
77         if (af == AF_INET6)
78                 return ipv6_addr_equal(&a->in6, &b->in6);
79 #endif
80         return a->ip == b->ip;
81 }
82
83 #ifdef CONFIG_IP_VS_DEBUG
84 #include <linux/net.h>
85
86 extern int ip_vs_get_debug_level(void);
87
88 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
89                                          const union nf_inet_addr *addr,
90                                          int *idx)
91 {
92         int len;
93 #ifdef CONFIG_IP_VS_IPV6
94         if (af == AF_INET6)
95                 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]",
96                                &addr->in6) + 1;
97         else
98 #endif
99                 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
100                                &addr->ip) + 1;
101
102         *idx += len;
103         BUG_ON(*idx > buf_len + 1);
104         return &buf[*idx - len];
105 }
106
107 #define IP_VS_DBG_BUF(level, msg, ...)                                  \
108         do {                                                            \
109                 char ip_vs_dbg_buf[160];                                \
110                 int ip_vs_dbg_idx = 0;                                  \
111                 if (level <= ip_vs_get_debug_level())                   \
112                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
113         } while (0)
114 #define IP_VS_ERR_BUF(msg...)                                           \
115         do {                                                            \
116                 char ip_vs_dbg_buf[160];                                \
117                 int ip_vs_dbg_idx = 0;                                  \
118                 pr_err(msg);                                            \
119         } while (0)
120
121 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
122 #define IP_VS_DBG_ADDR(af, addr)                                        \
123         ip_vs_dbg_addr(af, ip_vs_dbg_buf,                               \
124                        sizeof(ip_vs_dbg_buf), addr,                     \
125                        &ip_vs_dbg_idx)
126
127 #define IP_VS_DBG(level, msg, ...)                                      \
128         do {                                                            \
129                 if (level <= ip_vs_get_debug_level())                   \
130                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
131         } while (0)
132 #define IP_VS_DBG_RL(msg, ...)                                          \
133         do {                                                            \
134                 if (net_ratelimit())                                    \
135                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
136         } while (0)
137 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)                         \
138         do {                                                            \
139                 if (level <= ip_vs_get_debug_level())                   \
140                         pp->debug_packet(pp, skb, ofs, msg);            \
141         } while (0)
142 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)                      \
143         do {                                                            \
144                 if (level <= ip_vs_get_debug_level() &&                 \
145                     net_ratelimit())                                    \
146                         pp->debug_packet(pp, skb, ofs, msg);            \
147         } while (0)
148 #else   /* NO DEBUGGING at ALL */
149 #define IP_VS_DBG_BUF(level, msg...)  do {} while (0)
150 #define IP_VS_ERR_BUF(msg...)  do {} while (0)
151 #define IP_VS_DBG(level, msg...)  do {} while (0)
152 #define IP_VS_DBG_RL(msg...)  do {} while (0)
153 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)         do {} while (0)
154 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)      do {} while (0)
155 #endif
156
157 #define IP_VS_BUG() BUG()
158 #define IP_VS_ERR_RL(msg, ...)                                          \
159         do {                                                            \
160                 if (net_ratelimit())                                    \
161                         pr_err(msg, ##__VA_ARGS__);                     \
162         } while (0)
163
164 #ifdef CONFIG_IP_VS_DEBUG
165 #define EnterFunction(level)                                            \
166         do {                                                            \
167                 if (level <= ip_vs_get_debug_level())                   \
168                         printk(KERN_DEBUG                               \
169                                pr_fmt("Enter: %s, %s line %i\n"),       \
170                                __func__, __FILE__, __LINE__);           \
171         } while (0)
172 #define LeaveFunction(level)                                            \
173         do {                                                            \
174                 if (level <= ip_vs_get_debug_level())                   \
175                         printk(KERN_DEBUG                               \
176                                pr_fmt("Leave: %s, %s line %i\n"),       \
177                                __func__, __FILE__, __LINE__);           \
178         } while (0)
179 #else
180 #define EnterFunction(level)   do {} while (0)
181 #define LeaveFunction(level)   do {} while (0)
182 #endif
183
184 #define IP_VS_WAIT_WHILE(expr)  while (expr) { cpu_relax(); }
185
186
187 /*
188  *      The port number of FTP service (in network order).
189  */
190 #define FTPPORT  cpu_to_be16(21)
191 #define FTPDATA  cpu_to_be16(20)
192
193 /*
194  *      TCP State Values
195  */
196 enum {
197         IP_VS_TCP_S_NONE = 0,
198         IP_VS_TCP_S_ESTABLISHED,
199         IP_VS_TCP_S_SYN_SENT,
200         IP_VS_TCP_S_SYN_RECV,
201         IP_VS_TCP_S_FIN_WAIT,
202         IP_VS_TCP_S_TIME_WAIT,
203         IP_VS_TCP_S_CLOSE,
204         IP_VS_TCP_S_CLOSE_WAIT,
205         IP_VS_TCP_S_LAST_ACK,
206         IP_VS_TCP_S_LISTEN,
207         IP_VS_TCP_S_SYNACK,
208         IP_VS_TCP_S_LAST
209 };
210
211 /*
212  *      UDP State Values
213  */
214 enum {
215         IP_VS_UDP_S_NORMAL,
216         IP_VS_UDP_S_LAST,
217 };
218
219 /*
220  *      ICMP State Values
221  */
222 enum {
223         IP_VS_ICMP_S_NORMAL,
224         IP_VS_ICMP_S_LAST,
225 };
226
227 /*
228  *      Delta sequence info structure
229  *      Each ip_vs_conn has 2 (output AND input seq. changes).
230  *      Only used in the VS/NAT.
231  */
232 struct ip_vs_seq {
233         __u32                   init_seq;       /* Add delta from this seq */
234         __u32                   delta;          /* Delta in sequence numbers */
235         __u32                   previous_delta; /* Delta in sequence numbers
236                                                    before last resized pkt */
237 };
238
239
240 /*
241  *      IPVS statistics objects
242  */
243 struct ip_vs_estimator {
244         struct list_head        list;
245
246         u64                     last_inbytes;
247         u64                     last_outbytes;
248         u32                     last_conns;
249         u32                     last_inpkts;
250         u32                     last_outpkts;
251
252         u32                     cps;
253         u32                     inpps;
254         u32                     outpps;
255         u32                     inbps;
256         u32                     outbps;
257 };
258
259 struct ip_vs_stats {
260         struct ip_vs_stats_user ustats;         /* statistics */
261         struct ip_vs_estimator  est;            /* estimator */
262
263         spinlock_t              lock;           /* spin lock */
264 };
265
266 struct dst_entry;
267 struct iphdr;
268 struct ip_vs_conn;
269 struct ip_vs_app;
270 struct sk_buff;
271
272 struct ip_vs_protocol {
273         struct ip_vs_protocol   *next;
274         char                    *name;
275         u16                     protocol;
276         u16                     num_states;
277         int                     dont_defrag;
278         atomic_t                appcnt;         /* counter of proto app incs */
279         int                     *timeout_table; /* protocol timeout table */
280
281         void (*init)(struct ip_vs_protocol *pp);
282
283         void (*exit)(struct ip_vs_protocol *pp);
284
285         int (*conn_schedule)(int af, struct sk_buff *skb,
286                              struct ip_vs_protocol *pp,
287                              int *verdict, struct ip_vs_conn **cpp);
288
289         struct ip_vs_conn *
290         (*conn_in_get)(int af,
291                        const struct sk_buff *skb,
292                        struct ip_vs_protocol *pp,
293                        const struct ip_vs_iphdr *iph,
294                        unsigned int proto_off,
295                        int inverse);
296
297         struct ip_vs_conn *
298         (*conn_out_get)(int af,
299                         const struct sk_buff *skb,
300                         struct ip_vs_protocol *pp,
301                         const struct ip_vs_iphdr *iph,
302                         unsigned int proto_off,
303                         int inverse);
304
305         int (*snat_handler)(struct sk_buff *skb,
306                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
307
308         int (*dnat_handler)(struct sk_buff *skb,
309                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
310
311         int (*csum_check)(int af, struct sk_buff *skb,
312                           struct ip_vs_protocol *pp);
313
314         const char *(*state_name)(int state);
315
316         int (*state_transition)(struct ip_vs_conn *cp, int direction,
317                                 const struct sk_buff *skb,
318                                 struct ip_vs_protocol *pp);
319
320         int (*register_app)(struct ip_vs_app *inc);
321
322         void (*unregister_app)(struct ip_vs_app *inc);
323
324         int (*app_conn_bind)(struct ip_vs_conn *cp);
325
326         void (*debug_packet)(struct ip_vs_protocol *pp,
327                              const struct sk_buff *skb,
328                              int offset,
329                              const char *msg);
330
331         void (*timeout_change)(struct ip_vs_protocol *pp, int flags);
332
333         int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to);
334 };
335
336 extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto);
337
338 /*
339  *      IP_VS structure allocated for each dynamically scheduled connection
340  */
341 struct ip_vs_conn {
342         struct list_head        c_list;         /* hashed list heads */
343
344         /* Protocol, addresses and port numbers */
345         u16                      af;            /* address family */
346         union nf_inet_addr       caddr;          /* client address */
347         union nf_inet_addr       vaddr;          /* virtual address */
348         union nf_inet_addr       daddr;          /* destination address */
349         __be16                   cport;
350         __be16                   vport;
351         __be16                   dport;
352         __u16                   protocol;       /* Which protocol (TCP/UDP) */
353
354         /* counter and timer */
355         atomic_t                refcnt;         /* reference count */
356         struct timer_list       timer;          /* Expiration timer */
357         volatile unsigned long  timeout;        /* timeout */
358
359         /* Flags and state transition */
360         spinlock_t              lock;           /* lock for state transition */
361         volatile __u16          flags;          /* status flags */
362         volatile __u16          state;          /* state info */
363         volatile __u16          old_state;      /* old state, to be used for
364                                                  * state transition triggerd
365                                                  * synchronization
366                                                  */
367
368         /* Control members */
369         struct ip_vs_conn       *control;       /* Master control connection */
370         atomic_t                n_control;      /* Number of controlled ones */
371         struct ip_vs_dest       *dest;          /* real server */
372         atomic_t                in_pkts;        /* incoming packet counter */
373
374         /* packet transmitter for different forwarding methods.  If it
375            mangles the packet, it must return NF_DROP or better NF_STOLEN,
376            otherwise this must be changed to a sk_buff **.
377          */
378         int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
379                            struct ip_vs_protocol *pp);
380
381         /* Note: we can group the following members into a structure,
382            in order to save more space, and the following members are
383            only used in VS/NAT anyway */
384         struct ip_vs_app        *app;           /* bound ip_vs_app object */
385         void                    *app_data;      /* Application private data */
386         struct ip_vs_seq        in_seq;         /* incoming seq. struct */
387         struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
388 };
389
390
391 /*
392  *      Extended internal versions of struct ip_vs_service_user and
393  *      ip_vs_dest_user for IPv6 support.
394  *
395  *      We need these to conveniently pass around service and destination
396  *      options, but unfortunately, we also need to keep the old definitions to
397  *      maintain userspace backwards compatibility for the setsockopt interface.
398  */
399 struct ip_vs_service_user_kern {
400         /* virtual service addresses */
401         u16                     af;
402         u16                     protocol;
403         union nf_inet_addr      addr;           /* virtual ip address */
404         u16                     port;
405         u32                     fwmark;         /* firwall mark of service */
406
407         /* virtual service options */
408         char                    *sched_name;
409         unsigned                flags;          /* virtual service flags */
410         unsigned                timeout;        /* persistent timeout in sec */
411         u32                     netmask;        /* persistent netmask */
412 };
413
414
415 struct ip_vs_dest_user_kern {
416         /* destination server address */
417         union nf_inet_addr      addr;
418         u16                     port;
419
420         /* real server options */
421         unsigned                conn_flags;     /* connection flags */
422         int                     weight;         /* destination weight */
423
424         /* thresholds for active connections */
425         u32                     u_threshold;    /* upper threshold */
426         u32                     l_threshold;    /* lower threshold */
427 };
428
429
430 /*
431  *      The information about the virtual service offered to the net
432  *      and the forwarding entries
433  */
434 struct ip_vs_service {
435         struct list_head        s_list;   /* for normal service table */
436         struct list_head        f_list;   /* for fwmark-based service table */
437         atomic_t                refcnt;   /* reference counter */
438         atomic_t                usecnt;   /* use counter */
439
440         u16                     af;       /* address family */
441         __u16                   protocol; /* which protocol (TCP/UDP) */
442         union nf_inet_addr      addr;     /* IP address for virtual service */
443         __be16                  port;     /* port number for the service */
444         __u32                   fwmark;   /* firewall mark of the service */
445         unsigned                flags;    /* service status flags */
446         unsigned                timeout;  /* persistent timeout in ticks */
447         __be32                  netmask;  /* grouping granularity */
448
449         struct list_head        destinations;  /* real server d-linked list */
450         __u32                   num_dests;     /* number of servers */
451         struct ip_vs_stats      stats;         /* statistics for the service */
452         struct ip_vs_app        *inc;     /* bind conns to this app inc */
453
454         /* for scheduling */
455         struct ip_vs_scheduler  *scheduler;    /* bound scheduler object */
456         rwlock_t                sched_lock;    /* lock sched_data */
457         void                    *sched_data;   /* scheduler application data */
458 };
459
460
461 /*
462  *      The real server destination forwarding entry
463  *      with ip address, port number, and so on.
464  */
465 struct ip_vs_dest {
466         struct list_head        n_list;   /* for the dests in the service */
467         struct list_head        d_list;   /* for table with all the dests */
468
469         u16                     af;             /* address family */
470         union nf_inet_addr      addr;           /* IP address of the server */
471         __be16                  port;           /* port number of the server */
472         volatile unsigned       flags;          /* dest status flags */
473         atomic_t                conn_flags;     /* flags to copy to conn */
474         atomic_t                weight;         /* server weight */
475
476         atomic_t                refcnt;         /* reference counter */
477         struct ip_vs_stats      stats;          /* statistics */
478
479         /* connection counters and thresholds */
480         atomic_t                activeconns;    /* active connections */
481         atomic_t                inactconns;     /* inactive connections */
482         atomic_t                persistconns;   /* persistent connections */
483         __u32                   u_threshold;    /* upper threshold */
484         __u32                   l_threshold;    /* lower threshold */
485
486         /* for destination cache */
487         spinlock_t              dst_lock;       /* lock of dst_cache */
488         struct dst_entry        *dst_cache;     /* destination cache entry */
489         u32                     dst_rtos;       /* RT_TOS(tos) for dst */
490
491         /* for virtual service */
492         struct ip_vs_service    *svc;           /* service it belongs to */
493         __u16                   protocol;       /* which protocol (TCP/UDP) */
494         union nf_inet_addr      vaddr;          /* virtual IP address */
495         __be16                  vport;          /* virtual port number */
496         __u32                   vfwmark;        /* firewall mark of service */
497 };
498
499
500 /*
501  *      The scheduler object
502  */
503 struct ip_vs_scheduler {
504         struct list_head        n_list;         /* d-linked list head */
505         char                    *name;          /* scheduler name */
506         atomic_t                refcnt;         /* reference counter */
507         struct module           *module;        /* THIS_MODULE/NULL */
508
509         /* scheduler initializing service */
510         int (*init_service)(struct ip_vs_service *svc);
511         /* scheduling service finish */
512         int (*done_service)(struct ip_vs_service *svc);
513         /* scheduler updating service */
514         int (*update_service)(struct ip_vs_service *svc);
515
516         /* selecting a server from the given service */
517         struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
518                                        const struct sk_buff *skb);
519 };
520
521
522 /*
523  *      The application module object (a.k.a. app incarnation)
524  */
525 struct ip_vs_app {
526         struct list_head        a_list;         /* member in app list */
527         int                     type;           /* IP_VS_APP_TYPE_xxx */
528         char                    *name;          /* application module name */
529         __u16                   protocol;
530         struct module           *module;        /* THIS_MODULE/NULL */
531         struct list_head        incs_list;      /* list of incarnations */
532
533         /* members for application incarnations */
534         struct list_head        p_list;         /* member in proto app list */
535         struct ip_vs_app        *app;           /* its real application */
536         __be16                  port;           /* port number in net order */
537         atomic_t                usecnt;         /* usage counter */
538
539         /* output hook: return false if can't linearize. diff set for TCP.  */
540         int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
541                        struct sk_buff *, int *diff);
542
543         /* input hook: return false if can't linearize. diff set for TCP. */
544         int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
545                       struct sk_buff *, int *diff);
546
547         /* ip_vs_app initializer */
548         int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
549
550         /* ip_vs_app finish */
551         int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
552
553
554         /* not used now */
555         int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
556                          struct ip_vs_protocol *);
557
558         void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
559
560         int *                   timeout_table;
561         int *                   timeouts;
562         int                     timeouts_size;
563
564         int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
565                              int *verdict, struct ip_vs_conn **cpp);
566
567         struct ip_vs_conn *
568         (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
569                        const struct iphdr *iph, unsigned int proto_off,
570                        int inverse);
571
572         struct ip_vs_conn *
573         (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
574                         const struct iphdr *iph, unsigned int proto_off,
575                         int inverse);
576
577         int (*state_transition)(struct ip_vs_conn *cp, int direction,
578                                 const struct sk_buff *skb,
579                                 struct ip_vs_app *app);
580
581         void (*timeout_change)(struct ip_vs_app *app, int flags);
582 };
583
584
585 /*
586  *      IPVS core functions
587  *      (from ip_vs_core.c)
588  */
589 extern const char *ip_vs_proto_name(unsigned proto);
590 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
591 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
592
593 #define IP_VS_APP_TYPE_FTP      1
594
595 /*
596  *     ip_vs_conn handling functions
597  *     (from ip_vs_conn.c)
598  */
599
600 enum {
601         IP_VS_DIR_INPUT = 0,
602         IP_VS_DIR_OUTPUT,
603         IP_VS_DIR_INPUT_ONLY,
604         IP_VS_DIR_LAST,
605 };
606
607 extern struct ip_vs_conn *ip_vs_conn_in_get
608 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
609  const union nf_inet_addr *d_addr, __be16 d_port);
610
611 extern struct ip_vs_conn *ip_vs_ct_in_get
612 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
613  const union nf_inet_addr *d_addr, __be16 d_port);
614
615 extern struct ip_vs_conn *ip_vs_conn_out_get
616 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
617  const union nf_inet_addr *d_addr, __be16 d_port);
618
619 /* put back the conn without restarting its timer */
620 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
621 {
622         atomic_dec(&cp->refcnt);
623 }
624 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
625 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
626
627 extern struct ip_vs_conn *
628 ip_vs_conn_new(int af, int proto, const union nf_inet_addr *caddr, __be16 cport,
629                const union nf_inet_addr *vaddr, __be16 vport,
630                const union nf_inet_addr *daddr, __be16 dport, unsigned flags,
631                struct ip_vs_dest *dest);
632 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
633
634 extern const char * ip_vs_state_name(__u16 proto, int state);
635
636 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
637 extern int ip_vs_check_template(struct ip_vs_conn *ct);
638 extern void ip_vs_random_dropentry(void);
639 extern int ip_vs_conn_init(void);
640 extern void ip_vs_conn_cleanup(void);
641
642 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
643 {
644         struct ip_vs_conn *ctl_cp = cp->control;
645         if (!ctl_cp) {
646                 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
647                               "%s:%d to %s:%d\n",
648                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
649                               ntohs(cp->cport),
650                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
651                               ntohs(cp->vport));
652
653                 return;
654         }
655
656         IP_VS_DBG_BUF(7, "DELeting control for: "
657                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
658                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
659                       ntohs(cp->cport),
660                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
661                       ntohs(ctl_cp->cport));
662
663         cp->control = NULL;
664         if (atomic_read(&ctl_cp->n_control) == 0) {
665                 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
666                               "%s:%d to %s:%d\n",
667                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
668                               ntohs(cp->cport),
669                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
670                               ntohs(cp->vport));
671
672                 return;
673         }
674         atomic_dec(&ctl_cp->n_control);
675 }
676
677 static inline void
678 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
679 {
680         if (cp->control) {
681                 IP_VS_ERR_BUF("request control ADD for already controlled: "
682                               "%s:%d to %s:%d\n",
683                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
684                               ntohs(cp->cport),
685                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
686                               ntohs(cp->vport));
687
688                 ip_vs_control_del(cp);
689         }
690
691         IP_VS_DBG_BUF(7, "ADDing control for: "
692                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
693                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
694                       ntohs(cp->cport),
695                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
696                       ntohs(ctl_cp->cport));
697
698         cp->control = ctl_cp;
699         atomic_inc(&ctl_cp->n_control);
700 }
701
702
703 /*
704  *      IPVS application functions
705  *      (from ip_vs_app.c)
706  */
707 #define IP_VS_APP_MAX_PORTS  8
708 extern int register_ip_vs_app(struct ip_vs_app *app);
709 extern void unregister_ip_vs_app(struct ip_vs_app *app);
710 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
711 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
712 extern int
713 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
714 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
715 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
716
717 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
718 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
719 extern int ip_vs_skb_replace(struct sk_buff *skb, gfp_t pri,
720                              char *o_buf, int o_len, char *n_buf, int n_len);
721 extern int ip_vs_app_init(void);
722 extern void ip_vs_app_cleanup(void);
723
724
725 /*
726  *      IPVS protocol functions (from ip_vs_proto.c)
727  */
728 extern int ip_vs_protocol_init(void);
729 extern void ip_vs_protocol_cleanup(void);
730 extern void ip_vs_protocol_timeout_change(int flags);
731 extern int *ip_vs_create_timeout_table(int *table, int size);
732 extern int
733 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
734                         const char *name, int to);
735 extern void
736 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb,
737                           int offset, const char *msg);
738
739 extern struct ip_vs_protocol ip_vs_protocol_tcp;
740 extern struct ip_vs_protocol ip_vs_protocol_udp;
741 extern struct ip_vs_protocol ip_vs_protocol_icmp;
742 extern struct ip_vs_protocol ip_vs_protocol_esp;
743 extern struct ip_vs_protocol ip_vs_protocol_ah;
744
745
746 /*
747  *      Registering/unregistering scheduler functions
748  *      (from ip_vs_sched.c)
749  */
750 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
751 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
752 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
753                                 struct ip_vs_scheduler *scheduler);
754 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
755 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
756 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
757 extern struct ip_vs_conn *
758 ip_vs_schedule(struct ip_vs_service *svc, const struct sk_buff *skb);
759 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
760                         struct ip_vs_protocol *pp);
761
762
763 /*
764  *      IPVS control data and functions (from ip_vs_ctl.c)
765  */
766 extern int sysctl_ip_vs_cache_bypass;
767 extern int sysctl_ip_vs_expire_nodest_conn;
768 extern int sysctl_ip_vs_expire_quiescent_template;
769 extern int sysctl_ip_vs_sync_threshold[2];
770 extern int sysctl_ip_vs_nat_icmp_send;
771 extern struct ip_vs_stats ip_vs_stats;
772 extern const struct ctl_path net_vs_ctl_path[];
773
774 extern struct ip_vs_service *
775 ip_vs_service_get(int af, __u32 fwmark, __u16 protocol,
776                   const union nf_inet_addr *vaddr, __be16 vport);
777
778 static inline void ip_vs_service_put(struct ip_vs_service *svc)
779 {
780         atomic_dec(&svc->usecnt);
781 }
782
783 extern struct ip_vs_dest *
784 ip_vs_lookup_real_service(int af, __u16 protocol,
785                           const union nf_inet_addr *daddr, __be16 dport);
786
787 extern int ip_vs_use_count_inc(void);
788 extern void ip_vs_use_count_dec(void);
789 extern int ip_vs_control_init(void);
790 extern void ip_vs_control_cleanup(void);
791 extern struct ip_vs_dest *
792 ip_vs_find_dest(int af, const union nf_inet_addr *daddr, __be16 dport,
793                 const union nf_inet_addr *vaddr, __be16 vport, __u16 protocol);
794 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
795
796
797 /*
798  *      IPVS sync daemon data and function prototypes
799  *      (from ip_vs_sync.c)
800  */
801 extern volatile int ip_vs_sync_state;
802 extern volatile int ip_vs_master_syncid;
803 extern volatile int ip_vs_backup_syncid;
804 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
805 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
806 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
807 extern int stop_sync_thread(int state);
808 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
809
810
811 /*
812  *      IPVS rate estimator prototypes (from ip_vs_est.c)
813  */
814 extern int ip_vs_estimator_init(void);
815 extern void ip_vs_estimator_cleanup(void);
816 extern void ip_vs_new_estimator(struct ip_vs_stats *stats);
817 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
818 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
819
820 /*
821  *      Various IPVS packet transmitters (from ip_vs_xmit.c)
822  */
823 extern int ip_vs_null_xmit
824 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
825 extern int ip_vs_bypass_xmit
826 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
827 extern int ip_vs_nat_xmit
828 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
829 extern int ip_vs_tunnel_xmit
830 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
831 extern int ip_vs_dr_xmit
832 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
833 extern int ip_vs_icmp_xmit
834 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
835 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
836
837 #ifdef CONFIG_IP_VS_IPV6
838 extern int ip_vs_bypass_xmit_v6
839 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
840 extern int ip_vs_nat_xmit_v6
841 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
842 extern int ip_vs_tunnel_xmit_v6
843 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
844 extern int ip_vs_dr_xmit_v6
845 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
846 extern int ip_vs_icmp_xmit_v6
847 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
848  int offset);
849 #endif
850
851 /*
852  *      This is a simple mechanism to ignore packets when
853  *      we are loaded. Just set ip_vs_drop_rate to 'n' and
854  *      we start to drop 1/rate of the packets
855  */
856 extern int ip_vs_drop_rate;
857 extern int ip_vs_drop_counter;
858
859 static __inline__ int ip_vs_todrop(void)
860 {
861         if (!ip_vs_drop_rate) return 0;
862         if (--ip_vs_drop_counter > 0) return 0;
863         ip_vs_drop_counter = ip_vs_drop_rate;
864         return 1;
865 }
866
867 /*
868  *      ip_vs_fwd_tag returns the forwarding tag of the connection
869  */
870 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
871
872 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
873 {
874         char fwd;
875
876         switch (IP_VS_FWD_METHOD(cp)) {
877         case IP_VS_CONN_F_MASQ:
878                 fwd = 'M'; break;
879         case IP_VS_CONN_F_LOCALNODE:
880                 fwd = 'L'; break;
881         case IP_VS_CONN_F_TUNNEL:
882                 fwd = 'T'; break;
883         case IP_VS_CONN_F_DROUTE:
884                 fwd = 'R'; break;
885         case IP_VS_CONN_F_BYPASS:
886                 fwd = 'B'; break;
887         default:
888                 fwd = '?'; break;
889         }
890         return fwd;
891 }
892
893 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
894                            struct ip_vs_conn *cp, int dir);
895
896 #ifdef CONFIG_IP_VS_IPV6
897 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
898                               struct ip_vs_conn *cp, int dir);
899 #endif
900
901 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
902
903 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
904 {
905         __be32 diff[2] = { ~old, new };
906
907         return csum_partial(diff, sizeof(diff), oldsum);
908 }
909
910 #ifdef CONFIG_IP_VS_IPV6
911 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
912                                         __wsum oldsum)
913 {
914         __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
915                             new[3],  new[2],  new[1],  new[0] };
916
917         return csum_partial(diff, sizeof(diff), oldsum);
918 }
919 #endif
920
921 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
922 {
923         __be16 diff[2] = { ~old, new };
924
925         return csum_partial(diff, sizeof(diff), oldsum);
926 }
927
928 #endif /* __KERNEL__ */
929
930 #endif  /* _NET_IP_VS_H */